Zero Alcohol and Other Options: Limits for Truck and Bus Drivers -- Special Report 216 (1987)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

ALCOHOL Limits for Truck and Bus Drivers TranspottitiResearch1BöäidI [NATIIONALRESERCHtOUNCI L

1987 TRANSPORTATION RESEARCH BOARD EXECUTIVE COMMITTEE OFFICERS Chairman: Lowell B. Jackson, Executive Director, Colorado Department of Highways, Denver Vice Chairman: Herbert H. Richardson, Deputy Chancellor and Dean of Engineering, Texas A&M University System, College Station Executive Director: Thomas B. Deco, Transportation Research Board MEMBERS Ray A. Barnhart, Administrator, Federal Highway Administration, U.S. Department of Transporta- tion (cx officio) John A. Clements, Program Director/Highways, Parsons, Brinckerhoff, Quade, and Douglas, Boston, Massachusetts Alfred A. Delliflovi, Deputy Administrator, Urban Mass Transportation Adminitration, U.S. De- partment of Transportation (Cx officio) Francis B. Francois, Executive Director, American Association of State Highway and Transporta- tion Officials, Washington, D.C. (cx officio) E. R (Vald) Hciberg ill, Chief of Engineers and Commander, U.S. Army Corps of Engineers, Washington, D.C. (cx officio) Lester A. Hod, Hamilton Professor and Chairman, Department of Civil Engineering, University of Virginia, Charlottesville (cx officio, Past Chairman, 1986) T. Allan McArtor, Administrator, Federal Aviation Administration, U.S. Department of Transporta- tion (cx officio) Diane Steed, Administrator, National Highway Traffic Safety Administration, U.S. Department of Transportation (cx officio) George H. Way, Jr., Vice President, Research and Test Department, Association of American Rail- roads, Washington, D.C. (cx officio) Alan A. Alishuler, Dean, Graduate School of Public Administration, New York University, New York John R. Borchert, Regents Professor, Department of Geography, University of Minnesota, Minneapolis Robert D. llugher, Executive Director, American Public Works Association, Chicago, Illinois Dana F. Connors, Commissioner, Maine Department of Transportation, Augusta C. Leslie Dawson, Secretary, Kentucky Transportation Cabinet, Frankfort Paul B. Gaines, Director of Aviation, City of Houston Aviation Department, Texas Louis J. Gambaccini, Assistant Executive Director/Trans-Hudson Transportation of The Port Au- thority of New York and New Jersey, New York Jack R. Gilstrap, Executive Vice President, American Public Transit Association, Washington, D.C. William J. Hrris Snc'ad Distinguished Professor of Transportation Engineering. Department of Civil Engineering, Texas A&M University, College Station Raymond H. 1-logrefe, Director-State Engineer, Nebraska Department of Roads, Lincoln Thomas L. Mainwaring, Consultant, Trucking Industry Affairs, Ryder System, Inc., Miami, Florida James E. Martin, President and Chief Operating Officer, Illinois Central Gulf Raiiroad, Chicago Denman K. McNear, Chairman, President and Chief Executive Officer, Southern Pacific Transpor- tation Company, San Francisco, California Leno Menghini, Superintendent and Chief Engineer, Wyoming Highway Department, Cheyenne William W. Miilar, Executive Director, Port Authority of Allegheny County, Pittsburgh, Pennsylvania Milton Pikarsky, Distinguished Professor of Civil Engineering, The City College of New York, New York James P. Pitz, Director, Michigan Department of Transportation, Lansing Joe G. Rideoutte, Chief Commissioner, South Carolina Department of Highways and Public Trans- portation, Columbia Ted Tedesco, Vice President, Resource Planning, American Airlines, Inc., Dallas/Fort Worth Air- port, Texas Carl S. Young, County Executive, Broome County, Binghamton, New York

SPECIAL REPORT 216 ERO ALCOHOL AND OTHER OPTIONS Limits for Truck and Bus Drivers COMMITTEE ON BENEFITS AND COSTS OF ALTERNATIVE FEDERAL BLOOD ALCOHOL CONCENTRATION STANDARDS FOR COMMERCIAL VEHICLE OPERATORS Transportation Research Board National Research Council Washington, D.C. 1987

Transportation Research Board Special Report 216 modes 1 highway transportation 2 public transit subject areas 51 transportation safety 52 human factors 70 transportation law Transportation Research Board publications are available by ordering directly from TRB. They may also be obtained on a regular basis througi organizational or individ- ual affiliation with TRB; affiliates or library subscribers are eligible for substantial discounts. For further information, write to the Transportation Research Board, Na- tional Research Council, 2101 Constitution Avenue, N.W., Washington, D.C. 20418. Printed in the United States of America NOTICE: The project that is the subject of this report was approved by the Governing Board of the National Research Council, whose members are drawn from the councils of the National Academy of Sciences, the National Academy of Engineering, and the Institute of Medicine. The members of the committee responsible for the report were chosen for their special competencies and with regard for appropriate balance. This report has been reviewed by a group other than the authors according to procedures approved by a Report Review Committee consisting of members of the National Academy of Sciences, the National Academy of Engineering, and the In- stitute of Medicine. This study was sponsored by the Federal Highway Administration of the U.S. Department of Transportation. Library of Congress Cataloging-in-Publication Data National Research Council (U.S.). Transportation Research Board. Zero alcohol and other options : limits for truck and bus drivers. p. cm. - (Special report I Transportation Research Board, National Research Council ; 216) "August 1987." Includes bibliographies. ISBN 0-309-04468-5 1. Drinking and traffic accidents—United States. 2. Truck drivers—United States. 3. Bus drivers—United States. 4. Breath tests—United States. 5. Drunk driving—United States. I. Title. II. Series: Special report (National Research Council (U.S.). Transportation Research Board) ; 216. HE5620.D7N3 1987 363.1'251—cIcl9 ISSN 0360-859x 87-30296 CIP

Committee on the Benefits and Costs of Alternative Federal Blood Alcohol Concentration Standards for Commercial Vehicle Operators M. W. PERRINE, Chairman, Boston University, Burlington, Vermont JAMES D. BEARD, University of Tennessee, Memphis ROBERT J. FORMAN, Greyhound Lines, Dallas, Texas ROBERT B. FORNEY, Indiana University School of Medicine, Indianapolis GERALD J. FRIEDMAN, United Panel Service, New York City LARRY G. MAJERUS, Montana Division of Motor Vehicles, Helena KIMBALL I. MAuLL, University of Tennessee, Knoxville LAIMU-rIs NARGELANAS, Illinois State Police, Springfield OLGA J. PENDLETON, Texas Transportation Institute, College Station ROBERT H. REEDER, Northwestern University, Evanston, Illinois THOMAS H. ROCKWELL, Ohio State University, Columbus ALIs0N SMILEY, Human Factors North, Toronto, Ontario, Canada SCOrr D. SOLDON, Previant, Goldberg, Ucirnan, Gratz, Miller and Breuggeman, Milwaukee, Wisconsin ROBERT B. VOAS, ERIM, Inc., Arlington, Virginia Liaison Representatives JACK BURKERT, American Bus Association NEIL DARMSTADTER, American Trucking Associations RALPH V. DURhAM, International Brotherhood of Teamsters DONALD J. DZ!NSKI, American Public Transit Association MARK GREEN, National Institute on Alcohol Abuse and Alcoholism ROBERT HOLLEYMAN, Senate Commerce Committee

JAMES JOHNSTON, Owner-Operators Independent Drivers Association ALLEN D. MANESS, Senate Commerce Committee CARYLL F. RINEhART, Committee on Public Works and Transportation, U.S. House of Representatives R. H. SOSTKOWSKI, International Association of Chiefs of Police JOHN G. VINER, Federal Highway Administration Study Staff ROBERT E. SKINNER, JR., Director of Special Projects, Transportation Re- search Board (TRB) STEPHEN R. GODWIN, Senior Program Officer, TRB DEAN R. GERSTEIN, Study Director, Institute of Medicine and Commission on Behavioral and Social Sciences and Education NANCY HUMPhREY, Senior Program Officer, TRB JOHN A. DEACON, Consultant RICHARD MARGIOTTA, Research Associate, TRB MALCOLM M. QuINT, Research Associate, TRB THOMAS MENZIES, Research Associate, TRB NAOMI C. KASSABIAN, Associate Editor, TRB

Preface In enacting the Commercial Motor Vehicle Safety Act of 1986, the Congress established a framework for a national driver's license for operators of medium and heavy trucks and buses. The framework establishes a national standard for such licenses that would define, in terms of blood alcohol concentration (BAC), the offense of driving while deemed to be under the influence of alcohol. The act requires the states to pass legislation to imple- ment these provisions or risk loss of a fixed percentage of their federal highway aid. Among the provisions that the states must adopt is the mandated penalty for violating the new BAC standard for commercial vehicle drivers. On the first offense the state will revoke the driver's license for one year. For a second offense, the state will revoke the license permanently. The new BAC standard will be set by the Secretary of Transportation by October 1988. The act allows the Secretary to use discretion in setting the standard between zero and 0.10 percent BAC (the latter being the standard that prevails in most states). In the absence of a completed rule making within two years, the standard will automatically become 0.04 percent BAC. Con- gress also instructed the Secretary of Transportation to ask the National Academy of Sciences to study the benefits and costs that might attach to alternative BAC standards. This report is the outcome of that study. The study committee that addressed this issue had less than complete information for analyzing many of the issues of concern. Although consider- able research has been conducted on the problem of drinking and driving, most of this research has focused on automobile driving. For example, virtually all studies of the incidence of drinking and driving are based on

vi PREFACE drivers of passenger vehicles. Driving-related skills have often been examined under controlled dosage conditions in instrumented cars or in closed-course experiments where subjects operated automobiles at selected BACs. Comparable research is not available for drivers of large, heavy commercial trucks and buses. Large-scale comparisons of the BACs of drivers involved in crashes compared with the BACs of drivers in the general population have always excluded commercial vehicles. Despite the many gaps in our knowl- edge about alcohol-impaired commercial vehicle driving, the committee be- lieves it possible to arrive at judgments regarding the relative costs and benefits of a lower BAC standard for such drivers. These judgments, however, involve extrapolating beyond current knowledge. Recognizing the limited research and data, the committee has been prudent in its estimates, which, though approximate, are of sufficient accuracy to guide public policy making. ACKNOWLEDGMENTS This report is the result of the contributions and efforts of many individuals. A dedicated study committee devoted a considerable amount of time in a concentrated period to meet the mandated deadline. The chairman, M. W. Perrine, brought to his task a broad background in research on alcohol and behavior and a thorough understanding of the role of alcohol in motor vehicle crashes. The liaison representatives to the study provided additional back- ground information and helped the committee appreciate the context in which its recommendations would apply. The study was conducted under the supervision of Robert E. Skinner, Jr., Director of Special Projects for the Transportation Research Board. Stephen R. Godwin, Senior Program Officer, managed the study and drafted the Executive Summary, Chapters 1 through 3 and Chapter 7 under the guidance of the committee. Dean R. Gerstein, Study Director, Institute of Medicine and Commission on Behavioral and Social Sciences and Education of the National Research Council, collaborated on the Executive Summary, drafted Chapter 5, and-was instrumental in the preparation of the entire report. Nan Humphrey, Senior Program Officer, drafted Chapters 4 and 6, and John A. Deacon, Consultant, assisted in the preparation of Chapter 3. Richard Margiotta, Malcolm M. Quint, and Thomas Menzies, Research Associates for the study, assisted in the preparation of technical appendices. Special appreciation is expressed to Nancy A. Ackerman, TRB Publications Manager, and Naomi Kassabian, Associate Editor, for publication of the final report, and to Mar- guerite E. Schneider, Frances B. Holland, and Connie Woldu for word pro- cessing support. The committee was ably assisted by John Viner of the Federal Highway Administration, who made every effort to ensure that the committee had all

PREFACE vii the information necessary to carry out its charge. John Eicher of the Federal Highway Administration facilitated the early discussions about the study and helped initiate it at an early date. National Highway Traffic Safety Administration (NHTSA) personnel also made contributions. James C. Fell greatly assisted the staff by providing insight into interpreting data from the Fatal Accident Reporting System (FARS). Terry M. Kline, a consultant to NHTSA, also assisted the staff in interpreting FARS data. Along with his own insights, Alfred J. Farina provided the committee with information from ongoing research. Grace B. Hazzard greatly helped the staff by providing special analyses of National Accident Sampling System and FARS data. Ronald Engle and Henry Rockel also provided the staff with guidance. Researchers and administrators outside the federal government were also quite helpful. Richard R. Mudge, President of Apogee Research, critiqued the methodology developed for estimating economic costs associated with enfor- cing a more stringent BAC standard. Adrian Lund, with the Insurance Institute for Highway Safety, briefed the committee on his research on alcohol- and drug-impaired driving by commercial vehicle operators. Herbert Moskowitz provided the staff with detailed comments on background materials prepared for the committee. Ted Miller, of the Urban Institute, shared papers and data on the costs associated with motor vehicle crashes. Kurt M. Dubowski, with the University of Oklahoma Health Sciences Center, provided constructive comments on BAC measurement and testing. John Goslin, with the Tennessee Public Service Commission, provided considerable information regarding Tennessee's new commercial safety enforcement program. Mark Schock, California Highway Patrol, helped in developing public sector enforcement costs. Donald Young of the Texas Transportation Institute provided extensive programming assistance.

Contents ExEcUTIVE SUMMARY . 1 1 INTRODUCTION . 4 Request for Report, 5 Scope of Report, 6 2 ALCOHOL-INVOLVED TRUCK AND Bus CIsHEs ................ 10 Commercial Trucks and Buses, 11 Intercity Motor Carriers, 16 Commercial Vehicle Drivers, 19 Truck and Bus Crashes, 27 Summary, 35 3 ALCOHOL, PERFORMANCE, AND CRASH RISK.................. 39 Effects of Alcohol on Behavior, 40 Alcohol and the Driving Task, 44 Alcohol and Heavy-Truck Driving, 56 Summary, 62

4 APPREHENDING THE IMPAIRED DRIVER: LEGAL IssuEs AND TESTING TECHNOLOGIES....................................67 Overview of the Legal System and Enforcement Process, 67 Legal Issues, 75 Alcohol Concentrations in the Blood and Breath, 80 Testing Techniques and Their Effectiveness, 81 Summary, 86 5 PREVENTING ALCOHOL-IMPAIRED DRIVING THROUGH DETERRENCE..............................................90 The Deterrence Model, 91 Evaluations of Deterrence Policies, 92 Lessons of Deterrence Research, 106 Scenarios of Reform, 108 Benefits of Deterrence Scenarios, 111 6 COSTS AND BENEFITS OF PUBLIC AND PRIVATE DUI ENFORCEMENT PRACTICES.................................118 Public Enforcement, 118 Private Enforcement, 131 Benefits and Costs of Alternative BAC Standards, 136 Summary, 141 7 SUMMARY ASSESSMENT.................................... 145 Findings, 145 Recommendations, 154 APPENDIX A Alcohol Involvement in Fatal Truck Crashes.......... 157 APPENDIX B FARS and Texas Data on Fatalities for Drivers of Medium and Heavy Trucks........................164 APPENDIX C Description and Costs of Testing Techniques..........168 APPENDIX D Estimating the Costs of a Comprehensive Commercial Safety Enforcement Option........................173 APPENDIX E Estimating the Economic Costs of Shipment Delay.....178

APPENDIX F Estimating the Costs of Post-Crash Testing...........182 APPENDIX G Cost of Traffic Delays Caused by Accidents .......... 188 STUDY COMMITTEE BIOGRAPHICAL INFORMATION................ 192

Executive Summary Driving a heavy truck is hazardous work. The occupational death rate of heavy-truck drivers is five times greater than the average for all workers in the United States. Because the commercial vehicle driver's work takes place on public highways, other motorists share in the hazards and their share is substantial. Each year, the drivers of medium and heavy trucks and buses are involved in about 5,000 fatal crashes, the majority of which are collisions between heavy trucks (gross weight in excess of 13 tons) and other vehicles and pedestrians. About 80 percent of the 5,750 persons killed in those fatal crashes are passenger vehicle occupants, cyclists, or pedestrians rather than the commercial vehicle drivers themselves. The extent to which alcohol- impaired driving contributes to the frequency of commercial vehicle crashes is a matter of serious concern. At the request of Congress and the Secretary of Transportation, the Trans- portation Research Board appointed a study committee to examine the avail- able evidence on the alcohol problem in commercial vehicle driving and to weigh the benefits and costs of alternative blood alcohol concentration (B AC) standards. SUMMARY OF FINDINGS The study committee's findings are summarized as follows: About 750 fatal crashes occur annually in which a commercial vehicle driver had been drinking. Although this is a considerable number of fatal

2 ZERO ALCOHOL AND OTHER OPTIONS crashes, 15 percent of commercial vehicle drivers in fatal crashes had been drinking compared with 45 percent of all drivers in fatal crashes. Performance on driving-related tasks decreases at any BAC above zero and crash risk increases sharply as BAC rises. The technical ability to detect and measure low BACs (less than 0.05 percent) is available with current screening and testing devices. The legal ability of public authorities to enforce a low BAC standard with breath-screening devices has not been adjudicated at a definitive level, and opinions are divided on whether and how routine breath screening can occur consistent with constitutional rights. Alternative BAC standards for commercial vehicle drivers could be efficiently enforced by extensive screening of drivers at truck weigh stations and as part of vehicle safety inspections, along with mandatory blood tests following crashes that result in injuries. Estimating the potential benefits of alternative BAC limits for commer- cial vehicle drivers requires extrapolating from a small and imperfect data base. Recognizing the degree of uncertainty that attaches to any estimate based on limited data, roughly 80-140 lives could be saved annually by vigorously enforcing the 0.10 percent BAC that prevails in most states. Roughly 110-190 lives could be saved by vigorously enforcing 0.04 percent BAC, and 130-250 lives might be saved by vigorously enforcing a zero BAC limit. The total public and private costs of vigorous enforcement would be about $30 million to enforce a 0.10 percent BAC standard, about $40 million to enforce 0.04 percent BAC, and about $50 million to enforce zero BAC. SUMMARY OF RECOMMENDATIONS The Commercial Motor Vehicle Safety Act of 1986 requires the Secretary of Transportation to determine the BAC at which drivers would be deemed to be under the influence of alcohol and mandates a penalty of one year's loss of license on the first offense and lifetime revocation on the second offense. The phrase "deemed to be under the influence of alcohol" indicates that Congress did not intend that the standard be set at the level at which all drivers are impaired. Rather, Congress expressed its desire to reduce the risks to the public even if not all drivers would be impaired at the standard established. The study committee believes that any consumption of alcohol on the job by commercial vehicle drivers is inappropriate for the workplace and incompat- ible with traffic safety. The majority (three-fourths) of the committee recom- mends that the penalties required by the Commercial Motor Vehicle Safety Act be applied to violations of 0.04 percent BAC. In addition, the majority

EXECUTIVE SUMMARY 3 recommends that drivers detected with BACs greater than zero but less than 0.04 percent be penalized with a license revocation of up to 30 days on the first offense and from 30 days to one year on second and subsequent offenses. These drivers should also be referred for diagnosis and treatment as appropri- ate. In short, the majority of the committee believes that the BAC standard for commercial vehicle drivers should be zero and that the penalties should be graduated to match the offense. A minority of the committee believes that current state BAC limits are appropriate for the severe penalties required by the Commercial Motor Vehi- cle Safety Act of 1986. An additional argument against lowering the BAC limit below 0.10 percent is that the steps needed to enforce lower BACs (notably use of breath-screening devices) may be ruled unconstitutional in terms of the Fourth Amendment. The entire committee recommends that the BAC standard adopted by the Secretary be applied initially to drivers of buses, trucks that have a gross vehicle weight rating of 26,001 lb or more, and vehicles transporting haz- ardous cargoes. Drivers of trucks weighing between 10,001 and 26,000 lb should be covered as soon as is practical. Adoption of lower BAC limits and experimentation with different enforce- ment strategies and sanctions should be carefully evaluated. This evaluation should be carried out by independent researchers. To improve the quality of data for such an evaluation, the Department of Transportation should continue to emphasize the importance of reporting BACs of commercial vehicle drivers involved in crashes and should support research to determine the incidence of drinking and driving among commercial vehicle drivers. Experience with the costs of alternative BAC limits and estimates of their deterrent effects will provide the best basis for deciding whether the BAC limits and sanctioning patterns recommended in this report are appropriate or should be adjusted.

1 Introduction A driver, his judgment and responses dulled by alcohol, turns and brakes too late to avoid a crash. He kills himself or some innocent person. This story is repeated with relentless regularity thousands of times each year, year after year. As early as 1904 a report on "automobile wagon" crashes pointed out that alcohol had been consumed by a driver within 1 hr in 19 Out of 25 fatal crashes: Today, 83 years later, alcohol is involved in about half of 38,000 fatal crashes and contributes to the deaths of 25,000 individuals in the United States each year. Surprisingly, however, this seemingly intractable problem, which has failed to yield to dozens of different safety measures over recent decades, has begun to shift in a positive direction. The incidence of alcohol involve- ment in traffic fatalities declined 11 percent between 1982 and 1985 (NH'TSA 1987, v). The reasons for this safety gain are impossible to pinpoint; nonethe- less a heightened social awareness about drugs, the crusades of organizations like Mothers Against Drunk Driving, various state and local enforcement crackdowns, and an increase in the minimum drinking age are probably among the most influential measures (Fell and Klein 1986). Perhaps heartened by these successes, public officials and concerned citizens have sought other access points on the alcohol problem from which to leverage additional safety benefits. Alcohol impairment in commercial transportation has received particular attention. The Federal Aviation Administration recently required that com- mercial aviation crews—including flight attendants and maintenance engi- neers—must adhere to the same blood alcohol concentration (BAC) standard as that set for pilots (0.04 percent). (In the United States, the grams of alcohol 4

INTRODUCTION 5 per milliliter of blood, which can also be expressed as "percent BAC," is the most widely used legal measure of blood alcohol concentration.) Similarly, in 1986, the Federal Railroad Administration issued a regulation requiring that railroad crews adhere to the 0.04 percent standard as well. The BAC estab- lished in these regulations is well below the 0.10 percent BAC now used by most states as the standard for legal intoxication while driving on the highways. Even as stricter alcohol standards were applied in aviation and rail Iranspor- tation, public officials and trucking industry leaders expressed concern about the apparent rise in heavy-truck crashes. Public and industry efforts to crack down on problem drivers has long been stymied by the ease with which drivers obtained multiple licenses. Because few state licensing agencies in the past shared driver records, a driver could lose his license in one state but readily obtain another in a different state. Alter years of consensus building, Congress acted to end this practice by requiring a uniform national license in the Commeitial Motor Vehicle Safety Act of 1986. REQUEST FOR REPORT The Commercial Motor Vehicle Safety Act [Title XII of Public Law 99-570 (Oct. 27, 1986)] calls for a National Academy of Sciences study as follows (Section 12009): (t) BLOOD ALCOHOL CONCENTRATION LEVEL— (1) STUDY— NATIONAL ACADEMY OF SCIENCES—Not later than 30 days after the date of the enactment of this title, the Secretary shall undertake to enter into appropriate arrangements with the National Academy of Sciences to conduct a study of the appropriateness of reducing the blood alcohol con- centration level at or above which a person when operating a commercial motor vehicle is deemed to be driving while under the influence of alcohol from 0.10 to 0.04 percent. REPORT—In entering into any arrangements with the National Acad- emy of Sciences for conducting the study under this subsection, the Secretary shall request the National Academy of Sciences to submit, not later than 1 year after the date of the enactment of this title, to the Secretary a report on the results of such study. The same section provides that the Secretary of Transportation must com- plete rule making within 2 years to establish a BAC limit for commercial vehicle drivers. If the rule has not been issued within 2 years, the limit

6 ZERO ALCOHOL AND OTHER OPTIONS automatically becomes 0.04 percent. Under the provisions of the act, a commercial vehicle driver convicted of driving under the influence of alcohol would lose his commercial license for one year on the first offense. Two offenses would result in lifetime revocation [Sec. 12008(a)]. In order for the regulated BAC to be implemented, the act requires each state to pass legisla- tion that defines the BAC for commercial vehicle drivers as established in the federal regulation. States failing to comply will have 5 percent of federal-aid highway apportionments withheld in 1994 and 10 percent withheld thereafter (Sec. 12011). The legislation also opens the door for regulating drivers of commercial vehicles weighing less than 26,000 lb (the traditional definition of a heavy truck in federal regulations). The act leaves to the Secretary of Transportation the determination of whether vehicles weighing more than 10,000 lb should be subject to regulation. The legislation deliberately uses the phrase "deemed to be driving under the influence of alcohol." Senator Danforth, when introducing the bill, elabo- rated: "It is not necessary," he said, "to establish that everyone is impaired at the 0.04 BAC level. Rather, if some drivers' driving abilities are reduced, that is enough. The federal regulations are clear: any alcohol [for commercial vehicle drivers] is too much alcohol" (Congressional Record 1986). This phrase implies that the BAC standard should be set at a level where the effects of alcohol on driving can be demonstrated for some, but not necessarily all, drivers. The meaning of "deemed to be driving under the influence" will be discussed in more detail in Chapter 3 after the introduction of statistics on the trucking industry and the prevalence of alcohol involvement in highway crashes in Chapter 2. SCOPE OF REPORT Alcohol Involvement in Commercial Vehicle Crashes The trucks and buses defined as commercial vehicles in the Commercial Motor Vehicle Safety Act are used by public and private organizations engaged in all forms of commerce. The drivers of more than 3.6 million vehicles, who operate nearly 100 billion vehicle-mi annually, would be sub- ject to the more stringent blood alcohol regulation. These drivers are involved in about 5,000 fatal crashes annually and hundreds of. thousands of nonfatal crashes. Because of imperfect and incomplete testing for alcohol involvement in crashes, the percentage of commercial vehicle crashes in which the driver had alcohol in his bloodstream can only be approximated. A conservative estimate of alcohol involvement indicates that about 15 percent of commercial

ThTTRODUCTJON 7 truck drivers involved in fatal crashes have a measurable amount of alcohol in their blood. More detail about the truck and bus industries and the number of crashes in which commercial drivers had been drinking is presented in Chapter 2. Effects of Alcohol on Performance Alter alcohol has been ingested, human performance of a demanding task degrades. Dozens of different studies have demonstrated this effect. These same studies, however, have also indicated that the effects of alcohol vary across individuals, and even for a given individual the effect will vary depending on such factors as experience with alcohol and fatigue. The key issue, of course, is how to define the amount of alcohol in the blood that is sufficient for regulating driver behavior. During the last few years, almost all states adopted a BAC of 0.10 percent as the standard for laws involving driving under the influence (DUI). Most persons are visibly impaired—they slur their words, stumble when walking, and drive erratically—at 0.10 percent BAC. Studies conducted in laboratories with driving simulators and on closed driving courses, however, indicate that the average person's ability to perform mental and physical tasks degrades at BACs well below 0.10 percent. Given the variability in response to alcohol, use of the average response fails to capture the idiosyncratic response of some individuals. In this context, the phrase "deemed to be driving under the influence of alcohol" becomes critical because it implies that for administra- tive ease the average response to alcohol is adequate for setting the BAC standard. The studies estimating the effects of low and moderate amounts of alcohol on human performance and how they are interpreted are discussed in Chapter 3. Testing for Alcohol-Impaired Driving Testing drivers for alcohol impairment at BACs as low as 0.04 percent depends partly on the availability of accurate technologies for doing so, but in terms of deterrence, the legal standards under which testing can occur are more important. Portable and reliable breath-testing devices are widely used in many nations. In Australia, for example, a police officer can stop any driver and request a breath test. The Netherlands uses portable breath-testing devices to enforce its 0.05 percent BAC standard, and a variety of devices are being experimented with in the United States. In the United States, however, drivers cannot be tested simply because of the intuition of a police officer. Instead, the officer must have a reasonable

8 ZERO ALCOHOL AND OThER OPTIONS suspicion. that a driver is intoxicated before the driver can be stopped, and the officer must establish probable cause for taking a test before he can request a breath or blood test. Erratic driving, excessive speed, slurred speech, and the odor of alcohol are among the cues that an officer must have to establish reasonable suspicion. Once assured that reasonable suspicion exists, officers frequently require that a suspected driver demonstrate impairment by walking a straight line or performing some other physical task to establish probable cause. A driver who fails to perform adequately can then be arrested, and, in most states, it is at this point that a breath test is taken to measure the driver's BAC. This test becomes the evidence for violation of the state BAC limit. Obviously, at veiy low BACs, officers will have a more difficult task estab- lishing reasonable suspicion and probable cause. The legal framework within which testing occurs and the devices used for testing BACs are described in Chapter 4. Effectiveness of Enforcement Perhaps the most important issue in setting a more stringent BAC limit for drivers is whether enforcement of the rather severe penalty imposed will deter drinking by commercial vehicle drivers. As is described in Chapter 5, the deterrence effect of many different alcohol measures has been evaluated and the findings from these studies provide some insight into the potential effec- tiveness of a more stringent BAC for commercial vehicle drivers. Enforcing DU! Violations by Commercial Vehicle Drivers The successful model for deterring drinking and driving described in Chapter 5 is applied to the context of commercial driving in Chapter 6, in which a public-sector strategy for deterring DUT by commercial vehicle drivers is outlined. Private industry also has a role to play in enforcing laws against drinking and driving. Some large truck and bus companies prohibit any alcohol con- sumption by drivers and others are contemplating various forms of testing. The public and private approaches to regulating drinking and driving and the costs of these approaches are presented in Chapter 6. Chapter 6 also sum ma- rizes the probable range of costs and benefits at alternative BAC limits. The findings from these analyses indicate that vigorous enforcement can have positive benefits at BAC standards lower than those that prevail in state law. As described in Chapter 6, achieving these benefits will require a coordinated and concerted national effort.

ThTRODUCT!ON 9 The findings and recommendations of the study committee are summarized in Chapter 7. REFERENCES Congressional Record. 1986. Vol. 132, No. 144 (daily ed., Oct. 17, 1986), S16919. Fell, J., and T. Klein. 1986. The Nature of the Reduction in Akohol in U. Fatal Crashes. SAE Technical Paper 860038. Society of Automotive Engineers, Warren- dale, Pa. NHTSA. 1987. Fatal Accident Reporting System 1985. U.S. Department of Transpor- tation.

2 Alcohol-Involved Truck and Bus Crashes The Commercial Motor Vehicle Safety Act of 1986 covers drivers of tractor- semitrailers, intercity buses, transit buses, vehicles carrying hazardous mate- rials, and, potentially, drivers of all commercial trucks weighing more than 10,000 lb. With the exception of most pickup trucks and some lightweight delivery vans, this includes virtually all drivers of trucks and buses engaged in commerce. Although the alcohol-related provisions of the act are meant to influence driver behavior, the act itself rests on classifications of vehicles (hence the title). The act defines commercial vehicles as those with a gross vehicle weight rating of more than 26,000 lb used in commerce to move passengers or goods. The legislation also directs the Secretary of Transportation to deter- mine whether vehicles weighing between 10,001 and 26,000 lb should also be covered. To assist the Secretary in this determination, this report provides separate estimates of vehicles, crashes, and alcohol involvement by vehicle weight class where available data permit such estimates. Throughout this report, heavy trucks are defined as those weighing more than 26,000 lb and medium trucks as those weighing between 10,000 and 26,000 lb. The vehicle classifications notwithstanding, the request for determining an appropriate BAC for commercial vehicle drivers aims at the behavior of truck and bus drivers. The incidence of drinking and driving by commercial vehicle drivers will be explored in this chapter and the number of crashes that result from alcohol-impaired driving quantified. The available statistics indicate that in roughly 15 percent of medium-and heavy-truck crashes resulting in a fatality and in about 7 percent of crashes resulting in an injury, the truck drivers involved had been drinking. In 1.7 percent of medium- and heavy- truck crashes resulting in property damage only (PDO), the truck driver had 10

ALCOHOL-INVOLVED TRUCK AND BUS CRASHES 11 been drinking. Although these are relatively low percentages compared with the involvement of alcohol in passenger-car, light-truck, and motorcycle crashes, the resulting number of crashes is roughly 750 causing a fatality, 7,700 causing injury, and 4,750 causing property damage only. Before the discussion of statistical techniques that provide estimates of alcohol-involved truck and bus crashes, a brief review is given of the vehicles covered by the Commercial Motor Vehicle Safety Act of 1986. The industry segments that generate the bulk of intercity truck and bus mileage are re- viewed next, along with recent changes in federal regulation of these indus- tries that have greatly increased the competitive environment of commercial driving. Next the drivers themselves are described and the results of surveys of drinking and driving in the intercity trucking industry and available infor- mation about drinking practices in the transit industry are summarized. In the final section an estimate isgiven of the number of crashes in which commer- cial vehicle drivers under the influence of alcohol are involved. COMMERCIAL TRUCKS AND BUSES Trucks Some 30 million trucks operate on the nation's roads, but most of these vehicles weigh less than 10,000 lb and are therefore not covered by the Commercial Motor Vehicle Safety Act (Table 2-1). About 3.6 million trucks TABLE 2-I TRUCKS AND TRUCK MILES BY WEIGHT, 1982 (Census Bureau 1984,7) Trucks Annual Mileage No. Weight (lb) (thousands) Percent Miles (millions) Percent Light (less than 10,000) 30,228.8 89.4 299,540.5 79.5 Medium (10,000-24,000) 1,990.9 5.8 18,838.5 4.9 He:re than 26,000) 1,620.7 4.8 58,896.8 15.6 Total 33,840.4 377,275.8 are covered, or potentially covered, by the act. Trucks classified as medium trucks (10,000 to 26,000 lb) are usually straight trucks (power unit and cargo unit on a single chassis), as illustrated by the examples for Classes 3 to 6 shown in Figure 2-1. Medium trucks come in a wide variety of body types: wreckers, garbage haulers, some heavy-duty pickups, tank trucks, public

F"l Class 1 utilIty van compact van 6,000 lb & Less pick-up mum-purpose utIlity van compact van Class 2 06 6,001-10,000 lb pick-up walk-in Class 3 milk bread walk-in 10,001-14,000 lb oft compact van conveniionai van Class 4 14,001-16,000 lb large walk-in Class 5 rack tree specialist 16,001-1 9,500 lb large walk-tn ___________________________ I I I - - j Class6 furniture school - wp 19,501-26,000 lb coe van single axle van TIID home fuel transit Class 7 4% #* 26,001-33,000 lb medium conventional trash Class 8 extra heavy dump tandem conventional 33,001 lb & Over ** 4A - cement coe sleeper FIGURE 2-1 Vehicles by weight class. (Reprinted from Commercial Carrier Journal, June 1985, copyright 1985 Chilton Co.)

ALCOHOL-INVOLVED TRUCK AND BUS CRASHES 13 utility trucks, and cranes. Some tractor-semitrailer combination trucks (a truck with a power unit hauling a separate cargo unit) weigh less than 26,000 lb, particularly those with fewer than five axles. Most five-axle combinations, however, have an empty weight of 26,000 lb or more. In addition to tractor- semitrailer combinations, concrete mixers, dump trucks, and some cranes are classified as heavy trucks. TABLE 2-2 TRUCKS AND TRUCK MILES BY MAJOR USE, 1982 (Census Bureau 1984, 12, 74) Major Use Medium Trucks No. (thousands) Miles (millions) Heavy Trucks No. (thousands) Miles (millions) Agriculture 675.5 3,548.71 270.5 4,136.1 Forestry and lumbering 39.5 312.4 49.1 1,546.6 Mining 28.3 317.4 33.3 1,287.2 Construction 304.8 2,833.9 261.4 4,802.3 Manufacturing 80.9 1,165.9 121.7 5,886.9 Wholesale trade 195.8 3,408.0 160.5 6,117.8 Retail trade 174.9 2,253.5 92.0 2,887.0 For hire 115.4 1,867.5 493.1 2,248.4 Utilities 90.6 854.1 29.8 344.6 Service 117.5 1,171.0 44.3 943.5 Rental 62.1 696.8 31.6 1,573.0 Other 35.4 61.9 10.8 33.1 Total 1,920.7 18,490.4 1,620.7 58,878.4 NOTE: Trucks used primarily for personal transportation have been excluded. Columns may not add to totals because of rounding errors. Businesses of all types use trucks (Table 2-2). The largest users of medium trucks are in agriculture, construction, and the wholesale and retail trade. About one-third of heavy trucks are classified as being in the for-hire segment of the intercity trucking industry. These trucks-typically tractor-semitrailer combinations-are the type that motorists are most likely to encounter on major highways. Though they account for 30 percent of heavy trucks, these vehicles are responsible for more than half of the heavy-truck vehicle miles annually. Because for-hire carriers account for a large share of the total number of heavy trucks and truck miles, they are discussed in more detail in the next section. Although the intercity trucking industry accounts for a large share of heavy-truck miles, as shown in Table 2-2, heavy trucks are also extensively operated by private companies engaging in most forms of commerce (but not directly engaged in transportation).

14 ZERO ALCOHOL AND OThER OPTIONS In the review of statistics about trucks, it is useful to anticipate the discus- sion in Chapter 6 regarding private-sector enforcement of laws regulating alcohol-impaired driving and to note that companies of sufficient size to mount their own alcohol and drug abuse enforcement efforts are responsible for less than one-third of the heavy trucks using the highways. Just over one- half of trucks are operated by firms or individuals with five or fewer vehicles. Firms owning 20 trucks or more own and operate about 28 percent of the trucks on the road (Census Bureau 1984a, 24). More than 180 million shipments of hazardous materials cross the United States each year, and about one-half move by truck (OTA 1984, 3-4). Accurate statistics on these movements are difficult to come by because so many commodities are classified as hazardous. More than 2,400 substances have been so classified in the Code of Federal Regulations and nearly 70,000 chemicals have not even been reviewed (OTA 1984, 3-4). The commodities classified as hazardous range from petroleum products (fuel oil and gasoline) that are routinely transported by tank trucks to low-level radioactive waste that is moved in specially designed containers. Intercity and Transit Buses Intercity Buses In 1984, roughly 25,000 buses operated by 3,000 companies were engaged in intercity bus operations. These vehicles accounted for more than 3.3 billion vehicle-mi and 25.5 billion passenger-mi (TSC 1986, 25-26). A relatively small number of firms accounts for most scheduled bus operations. Just over 40 firms with more than $3 million in annual revenues (classified by the Interstate Commerce Commission as Class I carriers) account for about 60 percent of passenger miles. About 2,950 smaller firms offer charter service and tours and account for about 40 percent of passenger miles (TSC 1986; ABA 1982). Although intercity buses serve a vital passenger transportation function, the scale of the industry is much smaller than that of intercity trucking. Total intercity vehicle miles by buses equal only 4.5 percent of intercity vehicle miles by heavy trucks. (Vehicle mileage comparisons under- state the exposure to risk of highway users because buses carry many pas- sengers per vehicle mile. On a person-mile basis, intercity buses account for almost half as many miles of travel as do heavy trucks.) Transit Buses More than 430 agencies provide transit service in cities and towns across the nation (TSC 1986, 31-33). Most of these agencies (379) provide bus service as well as other transit services (e.g., rail, streetcar, and trolley) and employ

ALCOHOL-INVOLVED TRUCK AND BUS CRASHES 15 about 73,000 drivers to operate 55,000 motor buses in active service (UMTA 1986, Tables 2.13, 2.14). These buses accounted for about 1.6 billion vehicle mi in 1984, or roughly 19 billion passenger-mi (UMTA 1986). Nonrevenue buses and school buses are also covered by the Commercial Motor Vehicle Safety Act. Roughly 480,000 school buses are in operation and account for about 3.4 billion vehicle-mi each year (TSC 1986, 25). Summary Among the various classes of medium and heavy trucks and buses using the nation's roads, heavy trucks dominate the statistics; they account for 70 percent of annual vehicle miles traveled (Figure 2-2). Medium trucks are a distant second at 22 percent of annual vehicle miles. Intercity buses account for about 3 percent of total vehicle miles traveled by trucks and buses, as do school buses. Transit buses account for less than 2 percent. School Bus 3.4 Transit bus 1.6 rtercity bus 3.3 FIGURE 2-2 Travel by vehicle type (billions of annual vehicle miles). Drivers of heavy trucks operating on intercity highways account for the majority of annual truck and bus vehicle miles, and are likely to be the most affected by the licensing initiatives of the Commercial Motor Vehicle Safety Act. Intercity trucking and bus companies have shared a common regulatory framework during the last several decades and have been similarly affected by the policies of deregulation that began in the late 1970s.

16 ZERO ALCOHOL AND OTHER OPTIONS INTERCITY MOTOR CARRIERS Intercity Trucking Intercity trucking is classified into the following types: firms that transport goods for the public on the basis of published rates (common carriers), firms that transport goods for specific companies under contract (contract carriers), drivers who own their own equipment but who often lease their services to other carriers (owner-operators), firms such as agricultural cooperatives that haul cargoes not regulated by the Interstate Commerce Commission (exempt carriers), and companies whose primary business is not trucking but who have their own fleets to move the goods they produce or sell (private carriers) (Figure 2-3). The Interstate Commerce Commission (ICC) began regulating the industry in 1935 and has had a powerful influence over the types of carriers that have evolved. With substantial deregulation of the industry, however, the distinc- tions among carrier types have blurred and a continued evolution of the industry structure can be expected. For example, common carriers now fre- quently move goods under contract and private carriers have begun competing in markets from which they used to be excluded. Although the deregulation of the motor carrier industry has eroded the distinctions among segments, the available information about the industry is still classified in these traditional definitions and the available survey information about truck drivers (discussed in the next section) uses ICC definitions to classify drivers. Private carriers account for about 44 percent of the total combination-truck miles traveled annually (Table 2-3). Many national wholesale and retail stores, large grocery chains, utilities, manufacturers, governments, and oil companies own and operate private fleets. These fleets vary widely in size, from one tractor-semitrailer to hundreds (and in some cases, thousands) of vehicles. Independent owner-operators make up a relatively small but important share of the industry. Various estimates suggest that there may be as many as 100,000 to 300,000 independent operators. Recent statistics indicate that owner-operators account for 11 percent of combination vehicle miles (Table 2-3). According to estimates provided by the Owner-Operator Independent Drivers Association, about three-fourths of independent drivers regularly lease their services to regulated for-hire carriers. About 10 percent of indepen- dent drivers lease their services to exempt carriers, for example, agricultural cooperatives. Contract carriers transport goods classified by the ICC as special com- modities, but contract carriage is also a means to assure a given level of service (under contract) to a particular shipper. Contract carriers account for

ICC REGULA INTERSTATE MOTOR CARRIER INDUSTRY FQR HIRE REGULAR ROUTE GeneralFreighT - COMMON ATE Special Commodities Machinery - Petroleum IRREGULAR - Refrigerated Products ROUTE - Motor Vehicles Building Materials - Household Goods - Other - CONTRACT PRIVATE EXEMPT Exempt Agricultural Products FIGURE 2-3 The structure of the motor carrier industry, 1984 (TRB 1986).

18 ZERO ALCOHOL AND OTHER OPTIONS TABLE 2-3 COMPOSITION OF MOTOR CARRIER INDUSTRY, 1983 (TRB 1986) Carrier Type Share of Total for Combination Trucks Miles (millions) Percent For hire Common 18,290 27.7 Contract 4,094 6.2 Exempt 2,245 3.4 Owner-operator 7,461 11.3 Private 29,252 44.3 Other (local, rental not reported) 4,688 7.1 Total 66,030 about 11 percent of the combination vehicle miles of regulated carriers and about 6 percent of all combination vehicle miles (Table 2-3). Common carriers also transport special commodities but are principally in the business of hauling general freight. General freight is an ICC classification that includes most goods moved by truck except truckload-sized shipments of special commodities. General freight common carriers include most of the largest for-hire trucking firms. Common carriers account for about one-third of intercity truck miles of travel (Table 2-3). The Motor Carrier Act of 1980 lifted many of the regulatory constraints on the trucking industry. Among the more important changes, the ICC simplified entry into the industry, allowed private fleets to operate more like for-hire carriers, expanded the classification of exempt commodities, and eased the requirement that common carriers operate on regular routes. Even several years after the advent of deregulation the full effects on the industry remain unclear. The industry sank into a severe recession in 1980 (following the trend in the general economy) from which it had only partially emerged by 1985. During this period hundreds of trucking firms failed, among them some major, well-known companies (TRB 1986). These failures probably resulted from both the competitive pressures en- couraged by deregulation and the decline in demand during the recession, but the changes ushered in with deregulation should have other, lasting effects. Transportation companies have been given more flexibility to provide ship- pers with more efficient service. One apparent consequence of deregulation to date has been a large number of new firms seeking authorization to enter the business. In 1982, a year with the lowest level of shipper demand in recent history, 6,000 firms applied to the ICC for permanent operating rights. This compares with only 700 new applicants in 1979 (TRB 1986). Many of these firms may have operated as exempt carriers or owner-operators before deregulation.

ALCOHOL-INVOLVED TRUCK AND BUS CRASHES 19 The ICC no longer places heavy burdens on applicants to prove the need for new service. Instead, it routinely approves almost all applications. Many new applicants have entered the industry under contract carrier operating rights, but the majority have applied to enter the common carrier market (TRB 1986). Intercity Bus Industry Along with intercity trucking, intercity busing has been deregulated. For the decade preceding deregulation, the industry was fairly stable. Between 1,000 and 1,200 firms provided service with about 20,000 buses, and revenue passenger miles hovered around 25 billion annually (ABA 1982). During this period Class I carriers accounted for 60 percent or more of total passenger miles, and a large share of this mileage was generated by two firms— Trailways and Greyhound. Deregulation has upset this stability. Many small firms have entered the charter and tour business, the only growing segment of the industry. Mean- while, after taldng severe losses Greyhound Lines was sold to investors and restructured. The Class I carriers face stiff competition for charter business from smaller, lower-cost firms and are also having difficulty competing with discount air fares in some markets. For example, Greyhound ridership in 1986 was less than half the levels of the mid-1960s (Metro 1987) and the company lost money between 1982 and 1985 (Washington Post 1987a). Trailways has also suffered losses in ridership in recent years and Trailways Lines was sold to the newly restructured Greyhound Lines to avoid bankruptcy (Washington Post 1987b). Many Class I carriers have reported a loss in passenger miles and revenues. In 1981 (the last year for which data are available) about 14,000 drivers served the Class I carriers. Although these regulatory changes are still working their way through the intercity transportation industry, one implication for intercity drivers is clear: intense competition is forcing firms and individual owner-operators to provide improved service, lower cost, or both. In this context some drivers may well be pressed to drive up to—and in some cases beyond—the daily limit (10 hr) imposed by federal regulation. As discussed in the next section, fatigue— occasionally mixed with alcohol use but more often with use of stimulant drugs—is a problem some intercity drivers face regularly. COMMERCIAL VEHICLE DRIVERS The distinctions between industry segments described in the previous section become important when driver behavior is considered. For example, truck

TABLE 2-4 AGE OF DRIVER BY TYPE OF OPERATION AND REGULATORY STATUS (Wyckoff 1979, 10) Company Drivers (%) Owner-Operators (%) Exempt Private Contract Common Exempt Contract Common Driver's Age (n=89) (n=432) (n=482) (n=5,860) (n= 100) Private (n=599) (n=710) Less than 25 16.85 5.80 4.80 0.78 11.0 NA 9.58 7.63 25-34 37.08 36.20 29.02 18.56 31.00 NA 35.46 28.25 35-44 33.71 35.50 30.48 31.90 33.00 NA 29.92 32.48 45-54 8.99 17.40 26.72 35.70 20.00 NA 18.82 22.46 Age55orolder 3.37 5.10 8.98 13.05 5.0 NA 6.22 9.18 NOTE: NA = data not available

ALCOHOL-INVOLVED TRUCK AND BUS CRASHES 21 drivers working for private, contract, and common carriers tend to be older and more experienced than exempt drivers and owner-operators (Table 2-4). These age distributions, along with most of the statistics about drivers that follow, come from a nationwide stratified survey taken in late 1977 and early 1978 (9,630 usable responses out of 65,000 questionnaires distributed) (Wy- ckoff 1979, 10). Caution must be exercised in interpreting the Wyckoff survey. The usable response rate is fairly low (15 percent), the data are 10 years old, and the survey is not based on a random sample. Although it is not clear how representative the survey is of all drivers, it provides a rare glimpse of nearly 10,000 drivers' perceptions of their working environment and provides the only available estimates of self-reported drinking and driving habits by commercial truck drivers. Intercity Truck Drivers General The differences among drivers employed in the various industry segments appear from the Wyckoff survey to be determined as much by driver age as by the type of firm worked for. For example, according to their own reports, drivers under the age of 25 in this survey drive faster, are much more likely than their older counterparts to violate the 10-hr driving limit, and are much more likely to misrepresent their log books (Table 2-5). Younger drivers also report having more moving violations and more reportable accidents per 100,000 mi driven. TABLE 2-5 DRIVER SAFETY PRACTICES AND PERFORMANCE BY AGE (Wyckoff 1979, 15) Item Driver's Age (n=9,360) Less than 25 25-50 Age 50 + Cruising speed (mph) 62.0 59.8 58.1 Regularly misrepresent logs (%) 39.0 16.0 4.1 Regularly drive beyond the 10-hr limit (%) 36.1 12.0 2.7 Moving violations per 100,000 mi per year (no.) 1.3 0.7 0.3 Reportable accidents per 100,000 mi per year (no.) 0.4 0.2 0.2 Unionized drivers, who tend to work for the larger private, contract, or common carriers, are older than exempt drivers and owner-operators. Only 20 percent of union drivers are age 34 or younger, whereas slightly more than 46

22 ZERO ALCOHOL AND OTHER OPTIONS percent of nonunion drivers are age 34 or younger (Wyckoff 1979, 28). This difference is partially explained by the unwillingness of large, unionized firms to hire drivers without experience. Union seniority also plays a part. In unionized firms senior drivers are able to obtain over-the-road long hauls, whereas younger drivers have less regular hours. In common carriage less experienced drivers may operate pickup and delivery trucks to gain experi- ence. Wyckoff speculates that some of the younger drivers operating as independents or as exempt often choose to do so in order to obtain the experience needed to be hired by an established firm. The older, more experienced union drivers generally operate their vehicles at lower average speeds and rarely misrepresent their log books or violate the 10-hr driving limit (Table 2-6). More than 20 percent of owner-operators and more than 30 percent of nonunion drivers report that they regularly drive beyond the 10-hr limit (Table 2-6). In conjunction with their lower likelihood of violating the regulation about hours of service, union drivers report that they work fewer hours per 7-day week than their nonunion counterparts. Nonetheless, many intercity drivers work long hours. The average union driver works 9 hr a day. The average nonunion driver works more than 10 hr a day (Wyckoff 1979, 36). TABLE 2-6 DRIVING RECORDS AND SAFETY PRACTICES OF UNION DRIVERS VERSUS NONUNION DRIVERS (Wyckoff 1979, 30) Item Union Company Drivers (n = 6,532) Owner- Operators (n = 1,017) Nonunion Company Drivers (n = 745) Owner- Operators (n = 888) Cruising speed (mph) 58.8 60.1 61.8 60.8 Use multiple log books (%) 1.6 9.1 17.3 15.4 Regularly misrepresent log books (%) 4.8 27.9 35.7 37.8 Regularly drive beyond the 10-hr limit (%) 2.6 20.7 31.9 31.5 Moving violations per 100,000 mi per year (no.) 0.4 0.7 1.2 1.0 Reportable accidents per 100,000 mi per year (no.) 0.2 0.3 0.3 0.3 Drug and Alcohol Use on the Job Wyckoff Survey The long hours worked by intercity drivers contribute to fatigue and drug use. For example, among the 10 percent of drivers who report that they regularly exceed the 10-hr limit, 35 percent admit to occasionally falling asleep or dozing at the wheel (Wyckoff 1979, 58). Truck and bus

ALCOHOL-INVOLVED TRUCK AND BUS CRASHES 23 drivers do not have to violate the 10-hr limit to suffer from fatigue. In two major evaluations of driver fatigue it was found that in drivers' own reports of feeling tired, fatigue becomes more pronounced after 6 or 7 hr on the mad and that the actual frequency of crashes exceeds the expected frequency of crashes after about 7 hr of driving time (Harris and Mackie 1972, viii; Mackie and Miller 1978, xii). Not surprisingly, a high percentage of drivers in the Wyckoff survey who occasionally or regularly violate the 10-hr rule report using amphetamines or other, nonprescription stimulants to stay awake (Table 2-7). Nearly half of the drivers under the age of 25 report taking amphetamines or nonprescription stimulants when they feel drowsy (Wyckoff 1979, 61). TABLE 2-7 EFFECTIVE METHODS OF STAYING ALERT BY SEVERITY OF DOZING PROBLEM (Wyckoff 1979, 61) Method Percent of Drivers by Severity of Dozing Problem (n=9,360) Once or Never Twice Occasionally Regularly Stopping, walking, etc. 90.3 89.6 84.7 72.3 Washing face 75.7 77.5 72.7 60.0 Opening window 86.4 83.5 80.8 64.5 Listening to AM and FM radio 53.1 52.2 48.5 39.4 Using CB radio 80.8 85.7 84.9 74.8 Refreshment 91.2 87.9 85.5 69.7 Chewing gum or tobacco 41.7 40.0 38.6 30.3 Smoking 39.5 37.0 35.2 20.7 Singing or talking 35.9 65.9 66.9 51.6 Changing seat position 59.4 51.9 46.0 31.0 Nonprescription stimulant 6.4 10.5 13.1 18.7 Amphetamines 14.0 22.9 31.0 44.5 In Wyckoff's survey, which was filled out and mailed in anonymously, drivers were asked, "How many hours do you wait to drive after having any alcoholic beverages such as beer, wine, or whiskey?" There were several possible responses: (1) I do not drink; (2) I can drive satisfactorily without waiting; (3)1 wait about 1 hr; (4)1 wait about 2 hr; (5) I wait about 3 hr; or (6) I wait about 4 hr or more (Wyckoff 1979,28). Almost half (47.5 percent) of all intercity truck drivers in the Wyckoff survey reported that they did not drink. This percentage is considerably different from the 72 percent of adult males who reported alcohol consumption to the Census Bureau in 1979 (Census Bureau 1983, 123). In many respects this result is incongruous. Several studies of alcohol involvement in crashes have suggested that male drivers from lower occupational levels (typically blue collar) are overrepresented

24 ZERO ALCOHOL AND OTHER OFIIONS (Perrine 1971). In addition, frequent beer drinkers are overrepresented in fatal crashes (Perrine 1975), and beer is the alcoholic beverage of choice among blue-collar workers. Thus, on the basis of previous reseaith, truck drivers, who are mostly men, who work in a blue-collar job, and who prefer beer as a beverage, could be expected to be overrepresented among drivers with alcohol violations on their records and be overrepresented in crashes. Despite their lower reported alcohol use, 5 to 6 percent of the drivers in the Wyckoff survey indicated a willingness to violate the prohibition against driving within 4 hr of consuming alcohol (Table 2-8). Nearly 16 percent of drivers younger than 25 admitted a willingness to drive after drinking before 4 hr had expired and almost the same percentage of drivers in the exempt category gave the same response (these two groups overlap). The reported willingness to violate the 4-hr rule decreases with age and for drivers of regulated firms (Table 2-8). Drivers that regularly or sometimes haul haz- ardous materials are less likely than other drivers to report driving within 4 hr of consuming alcohol (Wyckoff 1979, 71). In some respects, a lower reported use of alcohol by intercity truck drivers is not entirely surprising. Some of the drivers may have interpreted the question in the Wyckoff survey about their drinking to apply only to drinking while on the job. In addition, younger drivers might have answered more truthfully. Probably more important, however, is that alcohol—a depressant— exacerbates drowsiness. Under the long and frequently monotonous hours of commercial driving it is logical to assume that drivers are more likely to use stimulants than depressants. Lund Survey The results from urine and blood tests of male commercial vehicle drivers randomly selected at a truck weigh station in Tennessee tend to corroborate the self-reported alcohol use in the Wyckoff survey (Lund et al. 1987). About one-third of the drivers in this study tested positive for the presence of drugs (legal and illegal), but less than 1 percent tested positive for alcohol. In contrast, 12 percent of the drivers were using nonprescription stimulants and 5 percent were using prescription stimulants such as amphetamines. In this study, drivers of combination trucks arriving at a weigh station on Interstate 40 in December 1986 were approached at random and invited to participate in what they were told was an anonymous and voluntary health survey. The survey was staged over several days to cover a 24-hr sampling period. Researchers approached 359 drivers; 317 drivers (88 percent) agreed to participate and provided blood and urine samples. Four drivers agreed to participate but then declined or were unable to provide both blood and urine. Thirty-eight drivers declined to participate, usually stating that they were in a

TABLE 2-8 ATHTUDE TOWARD ALCOHOLIC BEVERAGES BY REGULATORY STATUS AND AGE OF DRIVER (Wyckoff 1979, 65) Percent by Regulatory Status Percent by Driver's Age Attitude Toward All Exempt Private Contract Common Alcoholic Beverages (n=8,319) (n= 189) (n=478) (n= 1,082) (n=6,570) <25 25-49 >50 Do not drink 47.5 50.74 49.18 47.49 47.22 47.28 45.07 56.75 Can drive satisfactorily without waiting 2.1 5.47 4.50 2.99 1.67 7.48 2.48 1.60 Wait about 1 hr to drive 0.99 4.98 2.25 0.88 0.80 2.38 1.13 0.68 Wait about 2 hr to drive 1.6 2.99 2.25 1.85 1.45 4.08 1.86 0.72 Wait about 3 hr to drive 1.0 1.99 0.82 1.76 0.90 2.38 1.15 0.60 Wait 4 hr or more to drive 46.8 33.83 40.98 45.03 47.96 36.40 48.31 39.65

26 ZERO ALCOHOL AND OTHER OPTIONS hurry. Among the 12 percent of drivers who did not participate or did not provide samples, it is possible that a high percentage was under the influence of drugs or alcohol. Of the 317 drivers tested, 15 percent tested positive for marijuana (3 percent for THC), 2 percent for cocaine, 1 percent for amphetamines, 5 percent for other prescription stimulants, 12 percent for nonprescription stimulants, and less than 1 percent for alcohol. The two drivers testing positive for alcohol had BACs below 0.04 percent. An additional driver tested positive for alcohol in his urine but the presence of alcohol was not confirmed in his blood. The relatively high proportion of drivers testing positive for marijuana should be interpreted with caution. THC, the primary psychoactive ingredient in cannabis, can be detected in the bloodstream for a few hours (Hawks and Chiang 1986, 85). The metabolites of THC, however, are often stored in fatty tissue for days or even weeks after marijuana has been used (Hawks and Chiang 1986, 87). As acknowledged by the authors, the principal weakness of the study by Lund et al. is the refusal of 12 percent of the drivers to participate and the lack of comparative results from other truck stops. The distribution of drug use among the 42 drivers who declined (or were unable) to participate is un- known. The highest possible amount of drug use in this group would result from the assumption that all the drivers had consumed an illegal drug and for that reason were unwilling to participate. It can be assumed that the distribu- tion of drug use among the known participants is similar to that of the other drivers. Among the 317 drivers who participated, about 48 had used marijuana or cocaine or both, 16 were using amphetamines (or a closely related stim- ulant), and 2 tested positive for alcohol. Thus among those positive for drugs that either were or might be illegal in this context (the stimulants might have been prescribed and the alcohol might have been consumed 4 hr before driving), 73 percent tested positive for marijuana or cocaine or both, 24 percent tested positive for amphetamines, and 3 percent tested positive for alcohol. If these same percentages are applied to the 42 who refused to participate and it is assumed that all would test positive for these drugs, of the 42 drivers, 31 would be positive for marijuana or cocaine, 10 positive for amphetamines, and 1 positive for alcohol. Even when the projected drug use for the unknown drivers is added to the results for the known drivers, the percentage of drivers positive for alcohol is not increased; only 1 percent would test positive for alcohol. The other weakness of this study is that it is simply a case study and not necessarily a random sample of drivers. These results may overstate or understate the incidence of alcohol consumption by intercity drivers.

ALCOHOL-INVOLVED TRUCK AND BUS CRASHES 27 Swnmary The studies reviewed in this section are both imperfect estimates of drinking and driving by commercial vehicle drivers. The Wyckoff survey was designed to get a large enough response rate to compare driver responses but was not designed to be representative of all drivers. The recent survey taken at a inick weigh station had a relatively high nonresponse rate, and there is no assurance that the results are representative of all commercial vehicle drivers. Nonethe- less, both studies point toward a lower incidence of drinking and driving among commercial vehicle drivers than would be expected on the basis of studies of drinking by passenger car drivers and alcohol-related automobile crashes. Bus Drivers A recent study that examined the degree of stress among transit bus drivers also reported on their drinking practices. The study surveyed a stratified sample of drivers meant to be representative of large, medium, and small transit bus operations (TRB 1985). Just over half of the drivers (57 percent) in this survey drink alcoholic beverages at least weekly. This is considerably higher than the 36 percent of all adults who drink at least weekly (Census Bureau 1984b, 106). Among this 57 percent, the average consumption is fairly high—nine drinks per week (the median is six drinks per week). By way of contrast, only about one-third of the total population consumes more than three alcoholic beverages per week (Moore and Gerstein 1982). Although this survey indicates that transit bus drivers are heavier drinkers than the general population, no information is available regarding the frequency of drinking by transit bus operators while on the job. In addition, no direct information is available regarding drinking practices of intercity bus drivers nor on their frequency of drinking while on the job. TRUCK AND BUS CRASHES Heavy-truck driving is one of the most hazardous occupations in the United States. Only the mining and quarrying group has a higher death rate per 100,000 workers (Table 2-9). Commercial vehicle drivers are involved in about 440,000 crashes each year. In part because of the sheer size and mass of heavy trucks, crashes involving these vehicles are more than twice aslikely to result in fatalities than passenger-car crashes (Table 2-10). Of central interest

28 ZERO ALCOHOL AND OTHER OPTIONS to this report is the proportion of the commercial vehicle drivers involved in these crashes who were under the influence of alcohol. After a brief review of the size of the truck safety problem, the discussion turns to the different data sources and statistical techniques that estimate the role of alcohol in commer- cial vehicle crashes. TABLE 2-9 OCCUPATIONAL FATALITIES, 1984 (NHTSA 1987a, 16) Industry Workers Deaths per Group (thousands) Deaths 100,000 Workers All industries - 104,300 11,500 11 Trade 24,000 1,200 5 Manufacturing 19,000 1,100 6 Service 28,900 1,200 7 Government 15,900 1,400 9 Transportation and public utilities 5,500 1,500 27 Construction 5,700 2,200 39 Agriculture 3,400 1,600 46 Truck drivers 1,876 1,087 58 Mining and quarrying 1,000 600 60 TABLE 2-10 ESTIMATED CRASHES AND CRASH SEVERITY BY VEHICLE TYPE (NHTSA 1987b, 6-9, 6-25, 6-27, 6-36) Vehicle No. of Fatal Crashes Type Crashes No. Percent Passenger car 5,157,000 34,000 0.5 Heavy truck 300,000 4,198 1.4 Medium truck 78,000 642 0.8 Bus' 63,500 337 0.5 Nom: Crash estimates in first column from NHTSA based on 1983-1985 National Accident Sampling System data. a Buses include school, transit, charter, and intercity buses. The crash and travel data pieced together for this section come from a variety of sometimes conflicting sources and should be considered as approx- imations at best. Estimates of involvement of medium and heavy trucks and buses in police-reported crashes and in crashes resulting in injury were provided by the National Highway Traffic Safety Administration (NHTSA) from the National Accident Sampling Survey (NASS) reports of 1983 through 1985. Though the only national sample of crashes, NASS has never reached its intended sampling frame. To improve the reliability of the estimates,

ALCOHOL-INVOLVED TRUCK AND BUS CRASHES 29 NHTSA provided information from the most recent 3 years of NASS data. Because of the lack of specificity in NASS data and assumed underreporting, crash data from individual states are used to estimate alcohol involvement in nonfatal crashes. Fatal accident data from the Fatal Accident Reporting System (FARS) are more reliable than other crash data because FARS data represent an almost complete census of traffic fatalities (NHTSA 1987b). Drivers of medium and heavy trucks are involved in about 378,000 crashes annually (Table 2-11). Crashes in which a truck was hauling hazardous materials are difficult to estimate because of the large number of materials classified as hazardous that are not recognized as such by investigating officers. Reports to the Bureau of Motor Carrier Safety (BMCS) truck acci- dent data base indicate that about 5 percent of total truck crashes reported to BMCS in 1984 and 1985 involve hazardous materials (Table 2-12). TABLE 2-11 COMMERCIAL VEHICLES INVOLVED IN CRASHES (ANNUAL AVERAGE, 1983-1985) No. of Vehicles Involved Vehicle Injury-Producing Total No. Type All Crashes Crashes of Crashes Medium truck 80,000 25,000 78,000 Heavy truck 314,000 90,000 300,000 Busa 63,500 20,000 63,500 NOTE: Estimates provided by NHTSA from 1983-1985 NASS tapes. Some 30,000 crashes involved body types that could not be classified as either medium or heavy. These crashes were apportioned to medium and heavy trucks on the basis of the proportion of known crashes. 0 Buses include school, transit, charter, and intercity buses. TABLE 2-12 TRUCK CRASHES INVOLVING HAZARDOUS MATERIALS, 1984-1985 Hazardous Materials Crashes Material Release Item Total No. Percent Yes No Truck crashes 74,867 3,703 4.95 3,183 520 Fatalities 5,320 326 6.13 273 53 Injuries 57,477 2,955 5.21 2,514 441 SOURCE: Estimates provided by FHWA.

30 ZERO ALCOHOL AND OTHER OPTIONS Alcohol in Motor Vehicle Crashes The role of alcohol in motor vehicle crashes has always been difficult to estimate. Many studies have estimated alcohol involvement from police re- ports, but the police do not detect and record alcohol in many crashes (Voas 1983). Blood alcohol measures are increasingly available from fatal crashes, but usually the driver is tested only if he is the one killed in the crash. BACs are reported to FARS for only 22 percent of surviving drivers involved in fatal crashes (N}ITSA 1987b). In many jurisdictions BAC measures are still not regularly made even for the driver killed in a fatal crash (BAC measures for 83 percent or more of fatally injured drivers were reported in 32 states in 1985). Blood alcohol measures are not regularly made in crashes resulting in injury or in crashes involving only property damage. Many fender benders escape the notice of the police altogether. In the past, the percentage of fatal crashes in which alcohol was involved was estimated by examining the BACs of fatally injured drivers. Fell used these data from 15 states with relatively complete reporting to estimate that alcohol was involved in 59 percent of driver fatalities (Fell 1982). In 49 percent of fatal crashes, the measured BAC was 0.10 or greater. In 10 percent of fatal crashes the BAC ranged between 0.01 and 0.099. A new approach to estimating alcohol-involved fatal crashes that relies on discriminant analysis has been developed (Klein 1986) and described in a recent paper (Fell and Klein 1986). The new approach was used by NHTSA in the 1985 FARS report, but is currently under review. The report of that review will not be available during the time period of this study. As a practical matter, however, the new methodology results in an estimate of alcohol involvement in all fatal crashes that is quite close to that arrived at with the 15-state sample. Using the discriminant analysis approach, Fell has reestimated the role of alcohol in all crashes (Table 2-13). For all crashes of all types (cars, vans, TABLE 2-13 ESTIMATE OF ALCOHOL-INVOLVED CRASHES (Fell 1982) Item Fatal Crashes Injury Crashes PDO Crashes Proportion of alcohol-involved crashes (%) ~0.I0%BAC 41 - ~ 0.01% BAC 51 16 8 Total no. 39,168 2,248,000 17,100,000 No. of fatalities 43,795 - - No. of alcohol-involved fatalities 22,360 - - No. of injured persons - 3,363,000 - No. of alcohol-involved injured persons - 541,000 - No. of alcohol-involved PDO crashes - - 1,368,000 NOTE: Updated by J.C. Fell, unpublished data.

ALCOHOL-INVOLVED TRUCK AND BUS CRASHES 31 trucks, buses, motorcycles), the percentages are similar to those from the estimates based on the 15-state sample to estimate alcohol involvement in fatal crashes. About one-half of fatal crashes involve alcohol, with three- fourths involving BACs greater than or equal to 0.10 percent. Relying on estimated alcohol involvement in the NASS data, Fell estimates that roughly 16 percent of injury crashes and 8 percent of PDO crashes involve alcohol (Table 2-13). Alcohol in Truck and Bus Crashes The available information about commercial vehicle driver impairment by. alcohol and the extent of alcohol involvement in truck crashes indicates that alcohol plays a much smaller role in truck crashes than would be expected on the basis of the estimates of alcohol involvement for all crashes presented in the preceding section. This lower percentage of alcohol-impaired driving for commercial vehicle drivers compared with passenger vehicle drivers is to be expected. Commercial vehicle drivers are at work and alcohol consumption in the work place is not generally accepted. The majority of passenger vehicle drivers are not at work; indeed, most are driving for recreational purposes. Fatal Crashes As is the case for all crashes, the BACs of commercial vehicle drivers involved in fatal crashes are not well reported. BACs are reported for roughly half of the fatally injured medium- and heavy-truck drivers (Table 2-14). BACs are reported for only 11.3 percent of surviving truck drivers (Table 2-15), about 14 percent of whom had some alcohol in their bloodstream. This TABLE 2-14 KNOWN AND UNKNOWN BACs FOR FATALLY INJURED COMMERCIAL VEHICLE DRIVERS, 1982-1985 BAC Truck Type Medium Heavy Unknown All Percent of Total Drivers Percent of Known BACs Known No alcohol 161 1,260 45 1,466 43.0 77.9 0.01-0.03 4 50 4 58 1.7 3.1 0.04-0.09 13 37 6 56 1.6 3.0 0.10 + 67 207 28 302 8.9 16.0 Unknown 204 1,263 63 1,530 44.8 n/a Total 449 2,817 146 3,412 NOTE: The total number of drivers includes 146 drivers fatally injured in commercial trucks of unknown weight. SOURCE: FARS data tapes, 1982-1985.

32 ZERO ALCOHOL AND OTHER OPTIONS TABLE 2-15 KNOWN AND UNKNOWN BACs FOR SURVIVING COMMERCIAL VEHICLE DRIVERS, 1982-1985 BAC Truck Type Medium Heavy Unknown All Percent of Total Drivers Percent of Known BACs Known No alcohol 172 1,378 107 1,657 9.7 85.6 0.01-0.03 5 60 5 70 0.4 3.6 0.04-0.09 11 41 7 59 0.3 3.0 0.10 + 38 96 18 152 2.1 7.8 Unknown 1,959 12,421 721 15,101 88.6 n/a Total 2,158 13,996 858 17,039 SOURCE: FARS data tapes, 1982-1985. low reporting for surviving drivers creates a pmblem in estimating the role of alcohol in truck crashes because in more than 75 percent of fatal crashes involving trucks, the occupant of a passenger car or a pedestrian is usually killed; the truck driver is rarely the victim (NHTSA 1987b). In addition, the distribution of known BACs for surviving drivers differs from that for fatally injured drivers. Only 14 percent of surviving drivers are reported as positive for alcohol and only 8 percent have BACs equal to or more than 0.10 percent. In contrast, 22 percent of fatally injured drivers are reported positive for alcohol, and in 16 percent the BAC is equal to or above 0.10 percent. Involvement of a bus driver in a fatal crash with a reported positive BAC is rare. In 1982 through 1985 FARS data, only 5 bus drivers had a positive BAC out of 1,210 bus drivers involved in fatal crashes. BACs are known for only 8 percent of fatal crashes involving a bus driver (in 95 percent of cases where the BAC is known, it equals zero), but this is much too small a percentage from which to extrapolate to the unknowns. Fortunately in this case the police report—a separate indication on the FARS record of whether the driver had been drinking—is available for 83 percent of the fatal crashes involving bus drivers. In 99 percent of these cases the police officer has indicated that the driver had not been drinking. Application of NHTSA's recently developed discriminant model to predic- tion of alcohol involvement in fatal truck crashes results in a low estimate of alcohol-involved crashes (NHTSA 1987b). For 1985, NHTSA estimated that about 5 percent of heavy-truck drivers and about 10 percent of medium-truck drivers had been drinking. Too few bus drivers have positive BACs to apply this methodology to bus drivers as a separate group. Because of the large proportion of unknown BACs for heavy- and medium-truck drivers, statistical modeling may misrepresent the involvement of alcohol-impaired commercial vehicle drivers in fatal crashes.

ALCOHOL-INVOLVED TRUCK AND BUS CRASHES 33 A more cautious approach to estimating alcohol involvement in crashes is to pool all known BACs for killed and surviving drivers and, having adjusted for the characteristics associated with these crashes, such as vehicles involved, time of day, age of driver, and land use, to predict alcohol involvement among the unknown cases. In effect, this method adjusts the frequency of alcohol involvement among the known BACs to account for potential overrepresenta- tion of known BACs in certain kinds of crashes and then assumes that the unknown BACs are distributed similarly to the known BACs (as adjusted). The method is described in more detail in Appendix A. TABLE 2-16 ESTIMATED BACs BASED ON DISTRIBUTION OF KNOWN BACs FOR FATALLY INJURED AND SURVIVING DRIVERS, 1982-1985 Truck Type Medium Heavy Total BACs No. Percent No. Percent No. Percent No alcohol 2,007 76.2 14,256 86.4 16,263 85.0 0.01-0.03 46 1.7 573 3.4 619. 3.2 0.04-0.09 127 4.8. 406 2.4 533 2.7 0.10 + 454 17.3 1,308 7.8 1,762 9.1 Total 2,634 16,813 19,447 NOTE: Drivers in commercial trucks of unknown weight excluded from calculations. Methodology described in Appendix A. SOURCE: FARS data tapes, 1982-1985. One potential drawback of this approach is the small number of observa- tions for some of the predictor cells. To correct for this, data from 1982 through 1985 were pooled. As shown in Table 2-16, the estimates from the pooled data indicate that medium-truck drivers are positive for alcohol in 24 percent of crashes and heavy-truck drivers are positive for alcohol in 13.6 percent of crashes. According to this estimating technique, 60 percent of the drivers of medium and heavy trucks involved in fatal crashes who had been drinking had BACs higher than the prevailing standard of 0.10 percent. About 40 percent had BACs less than 0.10 percent. Nonfatal Crashes Few studies have investigated the role of alcohol in medium- and heavy-truck crashes, but the reported estimates fall in a fairly narrow range. An early study of 9,000 police-reported crashes involving all kinds of trucks revealed alcohol involvement in 5.5 percent (Ernst and Ernst 1968). Drinking was reported in crashes involving straight trucks (roughly equivalent to the 10,000-26,000-lb

34 ZERO ALCOHOL AND OTHER OPTIONS weight class) and tractor-trailers (more than 26,000 lb) in 3.7 and 2.0 percent of cases, respectively. Using 1973 police reports of accidents involving tractor-trailers and two- and three-axle trucks, Lohman and Wailer (1975) reported that alcohol impairment of the truck driver occurred in 1.2 percent of cases. Similar results were reported by Scott and O'Day (1971) and Li et al. (1979). Reported alcohol impairment in accidents reported to the BMCS was even lower-0.6 percent (BMCS 1986). The NASS data for 1983 through 1985 in which alcohol involvement was based on police reports indicate that about 1.3 percent of medium-truck crashes and about 0.7 percent of heavy- truck crashes involve alcohol. The incidence of one bus driver who tested positive for alcohol in the NASS sample (one driver with alcohol of 280 bus drivers in the NASS sample) is too small to estimate the frequency with which bus drivers with alcohol in their bloodstream are involved in crashes. As noted in the NASS report, the police miss the presence of alcohol in some cases and as a result NASS underestimates the presence of alcohol in crashes (NIITSA 1987c, 8). The highest estimates in these studies of alcohol involvement in medium- and heavy-truck crashes (3.7 and 2 percent, respectively) come from the study by Ernst and Ernst in 1968. Because recent estimates of alcohol involvement suggest that the role of alcohol in crashes has declined (see Table A-5, Appendix A), even if the estimate by Ernst and Ernst was accurate at the time of the study, it is likely to be too high today. The estimate of less than 1 percent in the BMCS reports is likely to be too low because it is based solely on voluntary reports. Reports from individual states, also largely based on police reports, cor- roborate the estimates from NASS. For example, in Washington State alcohol is cited as a contributing factor in 1 percent of crashes in which the truck driver was at fault (NHTSA 1987a, 161). Because other research has supported the reliability of the reports by the Texas state police [Department of Public Safety (DPS)] regarding alcohol involvement (Pendleton et al. 1986), the DPS reports for 7 years of Texas data were examined as part of this study (Appendix B). The Texas police report asks the officer to determine the contributing role of alcohol in crashes, which officers apparently interpret to mean determination of whether the driver is legally intoxicated (BAC of 0.10 percent). For the period 1980 through 1986, DPS reports indicate that the truck driver was intoxicated in 1 percent of the property-damage-only (PDO) crashes involving medium and heavy trucks. In 4.2 percent of the injury-producing crashes, the driver was intoxicated. These percentages, though based on reliable police reports, still underestimate the total amount of commercial vehicle crashes involving alcohol because they are unlikely to include BACs less than 0.10 percent. This underestimate can be approximated by assuming that the distribution of alcohol involvement in

ALCOHOL-INVOLVED TRUCK AND BUS CRASHES 35 fatal commercial vehicle crashes (Table 2-16) is similar to the distribution for injury and PDO crashes. In Table 2-16, 60 percent of the alcohol-involved fatal crashes occurred at BACs of 0.10 percent or more. Assuming that PDO and injury crashes in the Texas data have a similar distribution, it follows that 40 percent of injury and PDO crashes with alcohol involvement are not included in the Texas estimate. Adjusting for this suggests that truck drivers involved in PDO crashes would test positive for alcohol in 1.7 percent of cases, and commercial vehicle drivers involved in injury crashes would test positive in 7 percent of cases. Texas DPS data on bus drivers involved in crashes between 1980 and 1986 provide a report in only 4.4 percent of cases (434 reports or 9,860 involved drivers). In 9 of these 434 cases, the DPS indicated that the driver had been drinking. In other words, information about alcohol is available in less than 5 percent of crashes involving bus drivers, and for only 9 out of 9,860 involved drivers did officers report the presence of alcohol. SUMMARY About 3.6 million trucks and their drivers are affected, or potentially affected, by the Commercial Motor Vehicle Safety Act of 1986. Heavy trucks, those weighing more than 26,000 lb, account for more than half the total trucks affected by the law and for 73 percent of the vehicle miles traveled each year by commercial vehicles. Although businesses of all types use heavy trucks, the intercity trucking industry accounts for the single biggest share of truck travel. Because of the sheer size of the intercity trucking industry and its history as a regulated industry, more information about these vehicles and their drivers is available than for other types of commercial vehicle use. Since deregulation, the intercity trucking industry, which accounts for more than half of the truck miles driven annually, has entered a very competitive environment. Many drivers in the industry work long hours and many admit to driving while fatigued, as evidenced by the 20 to 30 percent of nonunion drivers who admit to violating the 10-hr driving rule and the more than 20 percent of drivers who admit to occasionally falling asleep at the wheel. A significant percentage admit to the use of drugs and alcohol while on the job. Use of stimulants appears more prevalent than alcohol, primarily as a means of combating fatigue. Although it appears that a smaller proportion of intercity truck drivers use alcohol than does the adult population generally, about 6 percent of a national survey of drivers admit to a willingness to violate the federal regulation prohibiting driving within 4 hr of consuming alcohol. A smaller percentage of drivers for large, unionized firms admit to violating the 4-hr rule. The only available roadside survey measuring the presence of alcohol in drivers of

36 ZERO ALCOHOL AND OTHER OPTIONS commercial vehicles estimated that about 1 percent of drivers had been drinking alcohol (Lund et al. 1987). Because 12 percent of drivers randomly approached to participate in the study refused to give a blood or urine sample, this 1 percent estimate is likely to be too low. The study results, though imperfect, do corroborate the Wyckoff survey results (1979), which suggest a much lower rate of alcohol-impaired driving by commercial vehicle drivers than that for the general population of highway users. The available evidence that intercity truck drivers drink and drive less than the general population is supported by the lower incidence of alcohol involve- ment by commercial vehicle drivers in fatal crashes. Whereas about half of the fatally injured noncommercial vehicle drivers had a measurable amount of alcohol in their blood, about 15 percent of the drivers of medium and heavy trucks involved in fatal crashes appeared to have been drinking. Drivers of intercity trucks, which are more likelyto be heavy than medium trucks, have a lower incidence of alcohol-involved fatal crashes than do drivers of medium trucks. About 14 percent of the fatally injured heavy-truck drivers had been drinking as compared with 24 percent of the drivers of medium trucks. Drivers of heavy and medium trucks with positive BACs are involved in about 750 fatal crashes, approximately 7,700 injury-producing crashes, and about 4,750 PDO crashes annually (Table 2-17). The reported incidence of alcohol-im- paired bus drivers involved in crashes is rare, probably considerably less than 1 percent of crashes. TABLE 2-17 ALCOHOL-INVOLVED CRASHES OF MEDIUM AND HEAVY TRUCKS Crash Type No. of Drivers Drivers Under the Influence No. Percent Property damage only 279,000 4,750 1.7 Nonfatal injury 110,000 7,700 7.0 Fatal 5,100 750 15.0 Of the 378,000 medium- and heavy-truck crashes that occur each year, roughly 5 percent involve hazardous materials. No direct information about the role of alcohol in these crashes is available. Although survey results indicate that drivers of vehicles hauling hazardous materials are less likely, to drive within 4 hr of consuming alcohol, about 3 to 4 percent of them report that they occasionally drive within 4 hr of consuming alcohol (Wyckoff 1979, 71). Crashes involving hazardous materials in which the driver had been drinking appear to be low-probability events; however, the consequences could be very severe.

ALCOHOL-INVOLVED TRUCK AND BUS CRASHES 37 REFERENCES ABBREVIATIONS ABA American Bus Association BMCS Bureau of Motor Carrier Safety NHThA National Highway Traffic Safety Administration OTA Office of Technology Assessment TRB Transportation Research Board TSC Transportation Systems Center UMTA Urban Mass Transportation Administration ABA. 1982. Bus Facts: 1982 Edition. BMCS. 1986. Accidents of Motor Carriers of Property 1984. U.S. Department of Transportation. Census Bureau. 1983. Statistical Abstract of the United States. U.S. Department of Commerce. Census Bureau. 1984a. Truck Inventory and Use Survey 1982. U.S. Department of Commerce. Census Bureau. 1984b. Statistical Abstract of the United States. U.S. Department of Commerce. Ernst and Ernst. 1968. Truck Accident Study (prepared for Automobile Manufacturers Association, Cleveland, Ohio). Cited in T. A. Ranney, K. Perchonok, and L Pollack, Identification and Testing of Countermeasures for Specific Akohol Acci- dent Types and Problems, yol. 3: The Heavy Truck Problem, Report DOT- HS-806-651. U.S. Department of Transportation, 1984. Fell, I. C. 1982. Alcohol Involvement in Traffic Accidents. Cited in Akohol and Highway Safety: A Review of the State of the Knowledge. Report DOT-HS-806-269. U.S. Department of Transportation, Feb. 1985. Fell, J. C., and T. Klein. 1986. The Nature of the Reduction in Akohol in U.S. Fatal Crashes. SAE Technical Paper 860038. Society of Automotive Engineers, Warren- dale, Pa. Harris, W., and R. Mackie. 1972. A Study of the Relationships Among Fatigue, Hours of Service, and Safety of Operations of Truck and Bus Drivers. Report BMCS- RD-i 1-2. FHWA, U.S. Department of Transportation. Hawks, R. L, and C. N. Chiang. 1986. Examples of Specific Drug Assays. In Urine Testing for Drugs of Abuse (R.L. Hawks and C.N. Chiang, eds.), DHHS Publication (ADM) 87-1481, Research Monograph Series 73, National Institute on Drug Abuse, U.S. Department of Health and Human Services, pp. 84-114. Klein, T. 1986. A Method for Estimating Posterior BAC Distributions for Persons Involved in Fatal Traffic Accidents. Sigmastat, Washington, D.C. Li, L. K., et al. 1979. Requirements for a Licensing Program for Heavy Duty Vehicle Drivers. Vol. 2: Summary of Literature Review. Highway Safety Research Center, University of North Carolina, Chapel Hill. Lohman, L. S., and P. F. Wailer. 1975. Trucks: An Analysis of Accident Characteristics by Vehicle Weight. Highway Safety Research Center, University of North Carolina, Chapel Hill. Lund, A., D. Preusser, R. Blomberg, and A. Williams. 1987. Drug Use by Tractor- Trailer Drivers. Insurance Institute for Highway Safety, Washington, D.C. Mache, R., and J. Miller. 1978. Effects of Hours of Service, Regularity of Service, and Cargo Loading on Truck and Bus Driver Fatigue. Report DOT HS-5-01 142. NHTSA, U.S. Department of Transportation.

38 ZERO ALCOHOL AND OTHER OPTIONS Metro Magazine, May/June 1987, pp. 38-44. New Owners Take Over Greyhound Lines. Moore, M., and D. Gerstein (eds.). 1982. Alcohol and Public Policy: Beyond the Shadow of Prohibition. National Research Council, Washington, D.C. NHTSA. 1987a. Heavy Truck Safety. Report DOT HS-807-109. U.S. Department of Transportation. NHTSA. 1987b. Fatal Accident Reporting System: 1985. U.S. Department of Transportation. NHTSA. 1987c. National Accident Sampling System 1985. U.S. Department of Transportation. OTA. 1984. Transportation of Hazardous Materials: State and Local Initiatives. Report OTA-SET-301. Pendleton, 0. J., N. Hatfield, and R. Bremer. 1986. Alcohol Involvement in Texas Driver Fatalities: Accident Reports versus Blood Alcohol Concentration. In Alco- hol, Accidents and Injuries, SAE Technical Paper 860037, Society of Automotive Engineers, Warrendale, Pa. Perrine, M. W. 1975. Alcohol Involvement in Highway Crashes: A Review of the Epidemiologic Evidence. Clinics in Plastic Surgery, Vol. 2, No. 1, pp. 11-34. Perrine, M. W., et al. 1971. Cited in M.W. Perrine, Clinics in Plastic Surgery, Vol.2, No. 1, 1975, pp. 11-34. Scott, R. E., and J. O'Day. 1971. Statistical Analysis of Truck Accident Involvements. Report HSRI 001580. Highway Safety Research Institute, University of Michigan, AnnArbor. TRB. 1985. NCTRP Research Results Digest 3: Predicting and Dealing with Transit Bus Operator Stress. National Research Council, Washington, D.C. TRB. 1986. Special Report 211: Twin Trailer Trucks. National Research Council, Washington, D.C. TSC. 1986. National Transportation Statistics. U.S. Department of Transportation. UMTA. 1986. National Urban Mass Transportation Statistics: 1984 Section 15 Annual Report. Report UMTA-IT-06-0310-86-1. U.S. Department of Transportation. Washington Post, Feb. 15, 1987a, p. H3. New Owner Seeks Concessions to Keep Greyhound on Road. Washington Post, June 20, 1987b, p. Al. Greyhound to Purchase Rival Trailways. Wyckoff, D. 1979. Truck Drivers in America. D.C. Heath and Co., Lexington, Mass. Was, R. B. 1983. Police Reporting of Alcohol Involvement on the FARS File. Report DTNH22-83-R-07160. NHTSA, U.S. Department of Transportation.

3 Alcohol, Performance, and Crash Risk Over the last decade, most states have established 0.10 percent BAC as the legal standard for intoxication in highway driving. During the same period, the federal government regulated 0.04 percent BAC as the standard for determining alcohol impairment for commercial aviation and railroad crews. The difference between these two standards stems partly from the different risks to others posed by automobile drivers compared with airline pilots and railroad engineers. Virtually all drivers' skills are significantly impaired at 0.10 percent BAC. The 0.04 percent BAC standard for airline crews was set where "detrimental effects on performance" occurred that were "incompat- ible with flight safety" (Federal Register 1985, 15376). In requesting this study, Congress asked for a determination of the "appro- priateness of reducing the blood alcohol concentration level at or above which a person when operating a commercial motor vehicle is deemed to be driving while under the influence of alcohol from 0.10 to 0.04 percent." The legisla- tion deliberately uses the phrase, "deemed to be driving under the influence of alcohol," as elaborated by Senator Danforth when introducing the bill. "It is not necessary," he said, "to establish that everyone is impaired at the 0.04 percent level. Rather, if some drivers' driving abilities are reduced, that is enough" (Congressional Record 1986). If the Secretary of Transportation does not establish a regulation by October 1988, the BAC for commercial vehicle drivers will automatically become 0.04 percent, the same level estab- lished for aviation crews and railroad engineers. The presumption is that 0.04 represents a threshold after which impairment begins. Determining the appropriateness of a reduced BAC depends partly on the findings• of scientific research examining the relationship between alcohol impairment and crash risk, partly on the ability to legally test and accurately

40 ZERO ALCOHOL AND OThER OPTIONS measure low BACs, and partly on the potential success of such a policy in deterring drinking and driving by commercial vehicle drivers. This chapter provides an analysis of the first of these three areas by reviewing the scientific literature examining the effect of low BACs (at or below 0.04 percent) on behavior and skills resembling driving. In subsequent chapters testing issues and deterrence will be examined. The effect of alcohol on automobile crashes has been the subject of scientific inquiry in this country since the 1930s (Heise 1934; Miles 1934; Holcomb 1938). Despite the large number of studies that have been made over the last half century, a gap must be bridged in applying existing scientific research to determination of the risk of driving a commercial vehicle while under the influence of alcohol. Most studies of driving-related skills and skill decrements due to alcohol relate to automobile drivers and automobile crashes. This is as it should be. The majority of alcohol-related crashes occur in passenger cars. Nonetheless, application of this research to commercial vehicle driving requires extrapolat- ing from existing studies and making judgments about probable, though not certain, consequences. The discussion begins by introducing one of the central problems facing all alcohol and drug research. The human response to alcohol and other drugs varies, depending on many individual and environmental influences (Drew et al. 1959). Not all studies adequately account for this variability. Some well- conducted studies, however, have demonstrated that, for the majority of subjects, human performance decreases at low BACs (at or below 0.05 percent). Epidemiological studies (also called case-control studies) also mdi- cate that the risk of being involved and, more importantly, the responsibility for causing a crash begin increasing at low BACs. As noted, much of this research, particularly the epidemiological studies, is based on automobile driving or the skills closely related to automobile driving. In the final section of this chapter the demanding conditions of commercial vehicle driving are examined to illustrate the probable effect of low BACs on the risk of commer- cial vehicle crashes. EFFECTS OF ALCOHOL ON BEHAVIOR Ethanol, the psychoactive drug in alcoholic beverages, is a central nervous system depressant. Small doses induce a sense of euphoria in many individ- uals, larger doses cause slurred speech and lack of coordination, and very large doses cause coma and ultimately death. The main pharmacological effect of ethanol is probably closely related to the concentration of alcohol in the brain, which can be approximated by measuring the concentration of

ALCOHOL, PERFORMANCE, AND CRASH RISK 41 alcohol in the blood (preferably capillary rather than venous blood). The actual biological and physiological processes by which ethanol affects the brain are not well understood. The effects of ethanol, however, can be observed by examining its effects on behavior at various BACs. In the United States, the most widely used legal measure of BAC is the percentage of grams of alcohol per milliliter of blood. The legal terminology, "percent BAC," however, is at variance with the terminology adopted by most scientific journals. Strictly speaking, calculating a percentage by divid- ing grams of alcohol (weight) by milliliters of blood (volume) is inappropri- ate, because the numerator and denominator are in different units. Most journals use a formulation such as milligrams of alcohol per 100 ml of blood (mg/mI) or grams of alcohol per deciliter of blood (g/dl). Fortunately, a milliliter of blood weighs approximately 1 g; hence the error introduced by "percent BAC" is quite small. Throughout this report, the phrase "percent BAC" is retained because of its widespread use as legal terminology. Rather than being digested, alcohol is absorbed into the blood from the stomach and small intestine. It is then metabolized by the body, mostly by the liver. Absorption occurs rather rapidly, but the process of elimination takes much longer. For example, the BAC for a 150-lb man who consumes two beers on an empty stomach over a half-hour period will peak at about 0.05 percent within half an hour. His BAC will drop to zero roughly 3.5 hr after he began to consume alcohol. This process of absorption and metabolism is shown for four separate doses in Figure 3-1. In this example, the peak BAC for the lowest dose is reached about 30 min after the completion of drinking and for the higher doses after 50 miii. It then falls at a slow, almost linear pace. The shape of the curve showing the speed of absorption and metabolism varies across individuals (Wilson et al. 1984). Variation in Individual Responses to Alcohol The individual response to alcohol varies on two dimensions that are easily confused, but which are completely distinct. The first dimension of variability results from differences in the BAC reached between individuals given a dose of alcohol, even one controlled for body weight. For example, O'Neill et al. (1983) reported that the 95 percent confidence interval for individuals given an ethanol dose of 0.5 g/kg of body weight ranged from 0.036 to 0.095 BAC. The range for individuals given a dose of 0.10 g/kg of body weight was similarly wide. The time required to reach the peak BAC across individuals given a controlled dose and with equal amounts of food in their systems can vary from 15 to 90 miii (Wilson et al. 1984). The sources of these BAC variations include the individual rate of absorption (influenced by food in the digestive tract) and metabolism.

42 ZERO ALCOHOL AND OThER OPTIONS 0.10 0.08 C 0 0.06 0 C 0.04 0.02 0.0 30 60 90 120 Time After Drinking (mm) Note: 10 min allowed for drinking. Dose 1 = 0.20 g/kg body weight Dose 2 = 0.35 g/kg body weight Dose 3 = 0.50 g/kg body weight Dose 4=0.65 g/kg body weight FIGURE 3-1 Mean blood alcohol at four different times after drinking, for four doses. (Drew et al. 1959. Reprinted with pennission.) The second dimension of variability is the different behavioral responses of individuals at the same BAC. These different behaviors result from dif- ferences in individual motivation and experience with alcohol. In most of the experiments discussed in the following sections, subjects have been dosed on an empty stomach according to their body weight to obtain BACs as close as possible to the target level. These procedures control for differences in the physical response to a given dose. Although the be- havioral response to alcohol at a target BAC will still show variability, statistical tests can demonstrate whether the difference in the average group response with and without alcohol is greater than the difference between individuals.

ALCOHOL, PERFORMANCE, AND CRASH RISK 43 Performance Decrements at Low BACs Hundreds of studies have examined at least one facet of behavior or skill at various BACs (Moskowitz and Robinson 1987). In their review of the litera- ture, Moskowitz and Robinson identified nearly 400 studies relating alcohol to performance, 177 of which were accessible, measured a skill related to driving, reported the dose administered or the mean BAC, and tested for statistical significance. In order to make the findings in the literature compara- ble, Moskowitz and Robinson approximated the BACs for all of the studies, including those that only reported the dose administered, by using average body weight, water content ratios, and the dose administered. Three-fourths of the studies examined BACs above the recomputed value of 0.05 percent. Despite the attention of most researchers to high BACs, 45 studies examined BACs at or below the recomputed value of 0.04 percent, and 37 of these studies reported impairment Not all the studies included in this review controlled adequately for individ- ual variability in the rate of metabolism or the peak BAC reached. The rate at which an individual metabolizes alcohol is critical when low BACs are examined. The time period allowed in an experiment for consumption, the time before testing, and the BAC at the time of the test must be closely tracked. As shown in Figure 3-1, the peak BAC is reached rather quickly, but it then falls over a longer period of time. If an experiment allows 20 min for absorption and the experimental trial requires an hour, and if the subject's BAC is measured after absorption and at the end of the trial, his or her BAC would have probably peaked in the interim. In this case, the BAC at the time that performance is measured may have been higher than that at the time the BAC was measured. Researchers over the years have frequently demonstrated that performance is more sensitive to alcohol during the rising phase of the BAC curve than in the declining phase; this phenomenon is known as the Mellanby effect. Few of the studies in the literature adequately report the specific time periods allowed for ingestion and absorption, and the time between trial and BAC measurement. Without this (and additional) informa- tion, whether a subject's BAC at the time of the test is above or below a critical value such as 0.04 percent may be indeterminate. In addition to the weakness in the research just cited, studies of human performance at low BACs are sometimes compromised by other methodologi- cal flaws. Sample sizes tend to be small and nonrandom, thus running the risk of being unrepresentative. The small sample size can cut both ways. A significant effect is difficult to achieve with a small sample; thus when significance can be demonstrated, the reliability of the results is enhanced (Blalock 1972, 163). In contrast, the liability of a small sample occurs when the results from a minority of subjects greatly affect the mean response. The

44 ZERO ALCOHOL AND OTHER OPTIONS consistency in significant results across several studies, however, compensates for the potential of a biased sample in a single study. Some studies fail to systematically relate the effects of alcohol to the time duration of the task; thus fatigue effects may potentially be interpreted as alcohol effects. In addition, the ability of subjects to learn and master a task during the experiment is sometimes not controlled for. The committee reviewed the 37 articles reporting significant impairment in performance at or below 0.04 percent identified by Moskowitz and Robinson and from these selected illustrative research on performance effects at low BACs. In general, these studies met the following conditions: the BAC was accurately measured and reported; the time periods between consumption, experimentation, and BAC testing were controlled and reported; and the study tested for statistical significance. These studies form the basis of the following analysis of the effects of low BACs on driving and driving-related skills. ALCOHOL AND THE DRIVING TASK In daylight on a well-built and well-maintained highway with moderate traffic, a driver can scan the lane ahead and behind, occasionally glance at the instrument panel, steer, and maintain speed with relatively little effort. To respond to a potential accident situation, perhaps caused by an abrupt lane change by another driver, a driver must perform another set of tasks almost simultaneously. These tasks can be categorized by using Smith's four-phase information-processing model. As this model is described by Huntley (1974), a driver must (1) see a situation developing (stimulus registered and sampled at the perceptual level) and (2) recognize it (stimulus recognition at the cognitive level). He must then (3) decide how to respond (cognitive level), and (4) perform the required physical tasks, for example, braking or steering (motor response). An alert driver can execute these four phases in less than 2 sec when confronted with an unexpected hazard in the roadway (Olson et al. 1984). But how well will he perform at 0.04 percent BAC? The nonalert, inattentive driver is at a higher risk of crash involvement (Shinar et al. 1978). How do the soporific effects of low BACs contribute to this risk? These questions, unfortunately, cannot be answered directly. Experiments cannot be conducted that place subjects at risk, especially the risk of driving an auto- mobile under the influence of alcohol at speeds typical of those on high-speed highways. To provide answers to this question, or at least part of it, re- searchers have examined individual elements of psychomotor performance in laboratories, have simulated driving conditions with visual display panels and driving simulators, and have conducted research at low speeds on closed driving courses or on isolated stretches of highway.

ALCOHOL, PERFORMANCE, AND CRASH RISK 45 Laboratory and Simulator Studies of Performance Visual Peiformance Most of the information necessary for driving must be processed by the human visual system. Wilson and Mitchell (1983) examined the visual perfor- mance of 10 healthy subjects aged 19 to 28 at BACs that peaked at 0.06 percent. Testing continued for an hour, and BACs were measured 40 and 60 min following ingestion, by which time the mean BAC had fallen to within 0.04 to 0.05 percent. They found a statistically significant decrease in the visual focus necessary for accurate depth perception (near and distant visual deviations) and in static visual acuity at a distance of 6 in (when the mean BACs had fallen to between 0.04 and 0.05 percent). This implies that at a low BAC a driver would not discriminate fine visual detail as clearly as he would without alcohol, nor would he judge the distance between himself and another vehicle as accurately. Wilson and Mitchell, however, report a significant decline in static acuity at a much lower level than do other researchers (Perrine 1973). In a 1963 study, Mortimer speculated that low BACs might not affect visual acuity. In his experiment, acuity was significantly impaired at an estimated BAC of 0.068 percent (j)eak reached during the test), but not impaired at 0.01 percent BAC. Using an early version of a driving simulator, Mortimer tested the tracking skills of 16 men aged 22 to 39 with 20-20 vision. Tracking ability was tested under normal illumination, low illumination to simulate nighttime driving, and low illumination with glare caused by simulated oncoming headlights. At the very low BAC of 0.01 percent, tracking was not affected by alcohol under normal and low illumination, but when low illumination was combined with glare, the number of tracking errors increased significantly (by 14.4 percent). Tracking was significantly impaired at the higher BAC under all illumination conditions. More recently, Adams and Brown (1975) examined the time period needed to recover from glare for nine subjects at doses of 0.5 and 1.0 mI/kg of body weight. Recovery time was examined during the rising and falling limbs of the BAC curve at five different time periods. The time period needed for recovery from glare was significantly increased at both doses, and the effect continued as long as 3 hr after ingestion, by which time the BACs had fallen to 0.01 percent. The glare effects studied were similar to those that might be experi- enced by a driver because of the reflection of sunlight from a rear window or from the windshield of an oncoming vehicle. These studies, with the limitations noted, provide some evidence that the visual system is affected at low BACs in that static acuity, focus needed for depth perception, and the ability to locate visual targets following the effects of glare are decreased.

46 ZERO ALCOHOL AND OThER OPTIONS Cognitive Peiformance The effect of alcohol on the recognition and decision-making steps of Smith's information-processing model has been studied by various tests that divide the subjects' attention. Moskowitz et al. (1985) tested the ability of 10 moderate drinkers to combine a tracking task with a visual search task at very low BACs. These skills resemble those required for driving, that is, the ability to steer and maintain lane position while simultaneously scanning the roadway, intermittently scanning the instrument panel, and being able to respond to peripheral visual stimuli. In this particular study, the subject had to track a fluctuating bar on a visual display in the center of his visual field and attend to a random pattern of numbers appearing on four video screens placed around the central display. The subject used a displacement stick to track the bar displayed. When the number 2 appeared on one of the four surrounding video screens, the subject responded by pushing a four-way button to indicate which of the four displays showed that number. Error scores in tracking were noted as were errors in responding to the visual displays and the time required to respond. The subjects were tested at controlled doses designed to reach the target BACs of zero, 0.015, 0.030, 0.045, and 0.060 percent. The subjects received only one dose per session. The mean BACs obtained for the tests were zero, 0.015, 0.029, 0.044, and 0.059 percent. To control for the variability in individual metabolism rates, the subject's BAC was monitored with a breath tester (Mark IV Intoximeter) until his BAC had fallen to just 0.005 percent above the target BAC. The number of errors made by subjects increased in a linear fashion with increasing BACs. At even the lowest BAC, 0.015 percent, subjects made significantly more errors than they did without alcohol. This study combined the results from the divided-attention task with a measure of information processing referred to as visual backward masking. The authors point out that the divided-attention task was more sensitive to BAC than the rate of informa- tion processing. Moskowitz had theorized earlier that mental processes are less affected by alcohol when an individual has to attend to only a single task (Moskowitz 1973). When he is required to process two kinds of information arriving from different sources, the effect of alcohol becomes apparent at low BACs. In a 1974 study, Moskowitz and Sharma demonstrated this effect at BACs of 0.06 and 0.09 percent. Under the influence of alcohol, the 12 subjects in the experiment did not show a performance decrement when they were able to focus their attention on a single task. Response to peripheral signals was significantly reduced when subjects tried to divide their attention between a central visual task and a peripheral visual task. These findings suggest that a

ALCOHOL, PERFORMANCE, AND CRASH RISK 47 driver at a low BAC may have little difficulty when able to devote his attention to a single task such as steering, but his response to peripheral stimuli (such as those needed to detect and avoid crossing traffic) might be slowed. Other researchers have shown performance decrements at low BACs when tracking is combined with other demands to create a divided-attention task. Connors and Maisto (1980) used a large sample (64 men) of experienced drinkers. According to the index of alcohol use developed by Callahan et al. (1969), 53 of the men were classified as heavy drinkers and 11 as moderate drinkers. The subjects were given controlled doses and required to perform a pursuit rotor tracking task while responding to a choice reaction task. BAC was approximated with a breath test (Alco-Analyzer) immediately before and immediately after the task. Before the 15-min trial, the subjects had.a mean BAC of 0.041 percent and afterwards a BAC of 0.04 percent. During the test the subjects' tracking skill, reaction to a visual display, and time required to respond (reaction time) were all monitored. Tracking errors alone and combined with choice reaction errors increased significantly with alcohol use compared with the control condition (placebo). Simple reaction time, however, was not affected. These findings suggest that of the skills needed for driving in the four phases described earlier, the last skill, simple motor response, may not be affected at low BACs, whereas the earlier steps may well be affected. In this particular study, the initial breath test was administered only 20 min after alcohol consumption; thus the BACs peaked at a higher point than 0.04 percent during the experiment. Because the trial lasted only 15 min (trials began 20 min after the breath test), the average mean BAC would probably have been in the 0.04 to 0.05 percent range. Evans et al. (1974) also demonstrated significant impairment on a tracking task combined with a visual stimulus (similar to a test of divided attention) with 14 young men at BACs of 0.024, 0.047, 0.061, and 0.089 percent. Performance decreased linearly with increasing BAC. Although the research just summarized demonstrates reduced performance on perceptual, cognitive, and psychomotor tasks at low BACs, not all studies have consistently shown performance decrements at low BACs. For example, Palva et al. (1982) studied the effects of diazepam and alcohol separately and together in 31 subjects of various ages in a three-phase experiment. At a dose of 0.5 g/kg of body weight, which would result in a mean BAC between 0.04 and 0.06 percent, no significant effects were observed in a choice reaction test. The lack of significant effect may be explained by the extensive amount of training the subjects received before the trials. Huntley (1974) has shown that at BACs of 0.08 percent, choice reaction time increases when subjects are presented with a novel stimulus but not with a familiar one. Cherry et al. (1983), in a study using eight subjects, also reported no significant impairment

48 ZERO ALCOHOL AND OTHER OPTIONS in choice reaction time or simple reaction time at a mean BAC of 0.05 percent. Nevertheless, this study showed a significant increase in errors made in a visual search task at 0.05 percent BAC. Other researchers have demonstrated a decline in tracking skills at low BACs when the tracking task is moderately demanding. Drew et al. (1959) tested 40 subjects at a series of doses resulting in mean BACs of 0.023, 0.036, 0.058, and 0.074 percent. In this tracking task (using an early driving simula- tor), the number of errors increased in a linear fashion as BACs increased. Landauer and Howat (1983) tested tracking skills of 24 subjects with a visual display panel. Four BACs were examined: zero, 0.021, 0.050, and 0.073 percent. Although simple reaction time actually improved, the total errors made in tracking increased significantly at all BACs, and the number of errors increased with increasing BAC. Summary The studies reviewed in this section suggest that three of the four phases in the information-processing model described earlier appear to be affected by low BACs. In the first phase, simply seeing a potential hazard, low BACs appear to decrease visual performance (Wilson and Mitchell 1983; MacArthur and Sekuler 1982; Adams and Brown 1975). In the second and third phases, tests of divided attention have shown that errors in tracking and choice reaction increase at BACs as low as 0.015 percent (Moskowitz et al. 1985). Perfor- mance in tracking alone diminishes at BACs between 0.02 and 0.05 percent if the task is demanding enough (Drew et al. 1959; Landauer and Howat 1983). The final phase, simple reaction time, is either unaffected or becomes faster; that is, the subjects apparently trade off speed for accuracy (Connors and Maisto 1980; Landauer and Howat 1983). Studies that examine several BACs report that performance decrements increase linearly as BAC increases, with significant effects occurring at the lowest BAC (Moskowitz et al. 1985; Adams and Brown 1975; Evans et al. 1974; Landauer and Howat 1983; Drew et al. 1959). These findings imply that some psychomotor performance de- creases after very small amounts of alcohol have been ingested and contra- dicts the presumption that a threshold BAC exists after which performance decreases. Closed-Course Studies of Low BACs The advantage of laboratory tests of motor skills and tests of performance in driving simulators is that they allow researchers to control for many potential

ALCOHOL, PERFORMANCE, AND CRASH RISK 49 influences and to therefore focus on the individual variables of interest. The disadvantage with such studies is their inability to replicate actual driving conditions. This leaves the possibility open that such studies do not directly test the combined Set of skills needed for driving (Huntley 1973). In response to this criticism, many researchers have tried to approximate actual driving conditions by having subjects tested in cars on closed driving courses or on isolated stretches of highway. Closed-course experiments usu- ally take place in a test vehicle, which is sometimes instrumented to measure a variety of responses. Impairment in some driving abilities in these experi- ments shows up at fairly low BACs. One of the first of these studies, conducted in 1950 by Bjerver and Gold- berg, compared the driving skills of a control group with those of an experi- mental group with BACs ranging from 0.016 to 0.074 percent (mean of 0.048 percent BAC). All 37 subjects were experienced drivers; most were driving school instructors. The subjects made a number of precision manuevers— parking, backing, and starting—under a time constraint. Improperly executed maneuvers were repeated. Each subject practiced all maneuvers before ingest- ing alcohol or the placebo. The subjects were then randomly allocated to the control or experimental groups and the maneuvers were repeated. After 50 to 80 mm (which included a driving trial lasting 7 to 8 mm), three samples of capillary blood were taken immediately after completion of the trial. When the time taken to complete the practice maneuvers and that for the test maneuvers were compared, the control group performed about 25 to 30 percent better than the group that had ingested alcohol (difference statistically significant at 0.05). In a critical review of the work by Bjerver and Goldberg, Huntley (1973) disagrees with their conclusion that the threshold of impairment is between 0.035 and 0.04 percent. Instead, he notes that the findings only allow one to conclude that in comparison with the control group, the test group's "driving is impaired at BACs approximating 0.048 percent." One shortcoming of the study by Bjerver and Goldberg is the combination of results from subjects who had consumed spirits and subjects who had consumed beer. These beverages have different alcohol contents and the subjects were allowed different time periods for ingestion. When the BACs were measured, however, the mean BACs of both groups were roughly equivalent (spirits, 0.049 percent; beer, 0.046 percent). The principal short- coming of this study, however, is that it only measures the time required to perform a task, not the number of errors committed. Although Bjerver and Goldberg demonstrate increased performance time in their experiment, the actual tasks required of the drivers more closely resemble low-speed precision maneuvers than the skills needed for driving at high speeds in traffic. Laurell (1977) more closely replicated a high-speed driving task to deter- mine the effect of a low dose of alcohol on the ability to brake and steer in a

50 ZERO ALCOHOL AND OThER OPTIONS potential accident situation. In this study, 26 subjects drove a lane of roadway while maintaining a speed of 30 mph. (The subjects were involved in three different phases of the experiment: 6 in the pilot, 10 in Experiment A, and 10 in Experiment B.) When given a signal, the subject attempted to brake his speed and steer into the adjoining lane (as if steering onto a shoulder) and then come to a full stop. The number of pylons moved or hit in the control versus alcohol-influenced conditions and the overall stopping distance were com- pared. The subjects' BACs were closely monitored immediately before and after the trial. Before the trial the BACs ranged between 0.026 and 0.065 percent with a mean of 0.045 to 0.046 percent (the results for Experiments A and B were reported separately but are combined here for convenience). Immediately after the trial, their BACs ranged between 0.018 and 0.057 percent, with a mean of 0.036 to 0.037 percent. Two subjects were found to have rising BACs during the trials, whereas for the others the BACs were declining. In the alcohol-influenced condition, drivers knocked over signifi- cantly more pylons and took significantly longer to stop than in the control condition. Laurell's study also included a surprise that subjects were not expecting. While expecting to perform one evasive maneuver, subjects were surprised by having a cardboard figure of a man suddenly appear in their path. Weather conditions and an improperly functioning apparatus precluded successful surprises for all subjects, but results were obtained for 10 drivers in the control group and 10 drivers in the alcohol-influenced group. Five of the 10 drivers with low BACs collided with the figure, whereas only 1 of the 10 control drivers was unable to avoid hitting it. Laurell later repeated a similar experiment to test the performance of subjects with hangovers (Laurell and TOmros 1983). In this study, the subjects were allowed to consume as much alcohol as they wished over a 6-hr period in the evening. The mean peak BAC obtained was 0.147 percent. The subjects were then put to bed and awakened 8 hr later, when their BACs were approximated at regular intervals with a breath tester (Alcolmeter) until they reached zero. The test of emergency braking and steering was then performed. No driver was able to match his performance when the same trial was conducted without alcohol. The combined performance score decreased sig- nificantly-19 percent—from the alcohol to the no-alcohol trials. In contrast to these findings regarding hangover effects, Collins's research (1980) on pilots in control versus hangover conditions did not show signifi- cant effects. However, in his study the subjects reached a lower peak BAC (0.09 percent compared with 0.148 percent) and the greater skill and perfor- mance of the pilots may have been sufficient to override any hangover effect. In addition, the pilots were not tested in an accident situation. A subsequent study of pilots with hangovers (Yesavage and Leirer 1986) did show significant decreases in performance between the no-alcohol and the

ALCOHOL, PERFORMANCE, AND CRASH RISK 51 hangover conditions. This study used a more demanding task and 14 hr before the trial the subjects had peak BACs between 0.101 and 0.121 percent. The results from these driving-course studies can be interpreted by referring to the example of a driver on the highway and the information-processing model described earlier. The laboratory and simulator studies suggest that three of the four phases in this model—the ability to see a situation, to recognize it, and then to decide how to respond—are affected at low BACs. Laurell's research on a closed driving course confirms these results. At low BACs and at zero BAC with a hangover, a driver's performance degrades in a simulated accident situation. The consistency of the results from these studies helps compensate for the flaws inherent in many of them. Indeed, the results from this research prob- ably yield conservative estimates of the effects of alcohol at low BACs. The subjects in these experiments know they are being observed and typically try very hard to perform well. In doing so they may compensate for the effects of low doses of alcohol, which they may be less able to override under normal conditions (Perrine 1976). Laurell actually encouraged the subjects in his closed-course experiments to try hard by providing larger monetary rewards for good performance. Even so, his work demonstrates poorer performance under the influence of alcohol and with a hangover. Case-Control Studies Though very useful for demonstrating the effects of alcohol on performance, Iaboztory, simulator, and closed-course experiments nonetheless imperfectly measure driving performance. Although the studies reviewed so far indicate that some physical and mental processes degrade at low BACs, they do not indicate how much the risk of accident occurrence increases because of this performance deficit (Moskowitz et al. 1985). Case-control studies are used to estimate the role of alcohol in crashes by comparing the BACs of a comparison group of drivers with the BACs of accident-involved drivers. Assuming that comparative information can be collected from drivers with a similar exposure to risk as those involved in crashes, the contributing role of alcohol can be inferred. Various kinds of roadside surveys are used to collect comparative exposure data by requesting breath tests from passing drivers. The resulting BACs, when collected at the time of day and locations where crashes have occurred, are compared with the BACs of crash-involved drivers. The first such study was conducted in the United States in 1938 by Holcomb, and several more studies have been conducted in the United States and abroad since that time. These studies, as summarized by Perrine (1975) and updated by Hurst (1985, 1-2) to include subsequent studies, demonstrate a consistent pattern.

52 ZERO ALCOHOL AND OThER OPTIONS The relative probability of being involved in a crash is defined as the ratio of the BACs of comparison drivers to those of drivers involved in crashes. This probability remains roughly equal for drinking drivers compared with drivers who have not been drinking up to about 0.08 percent BAC (Figure 3-2). (However, as described in the next paragraph, these relative risk curves understate the risk of involvement at low BACs.) Although the rate of increased risk varies across studies—in part because some studies examine all crashes and some only fatal crashes—in all cases the risk increases after about 0.08 percent BAC and in most studies increases dramatically after 0.10 percent BAC. The curves depicted in Figure 3-2 are based on groups of drivers of different ages with varying experience with alcohol and with driving. Because of the heterogeneity of control groups and the lack of perfect comparability, the effect of alcohol at low BACs is masked by other variables. For example, the major shortcoming of the Grand Rapids study (among the most cited case- control study) is the lack of comparability between the drivers involved in crashes and the control drivers regarding the frequency of consuming alcohol. This lack of comparability is the source of the apparent improvement in crash risk at low BACs in the Grand Rapids data (the much debated "Grand Rapids dip") and the understatement of the risk of crash involvement at low BACs. Hurst (1973) noted that the control group had a higher proportion of drivers who were regular consumers of alcohol. Their apparently greater tolerance for alcohol had made them safer drivers at low BACs than the drivers involved in crashes, presumably because the latter had less experience as drinkers. Hurst recalculated the relative risk of crash involvement in the Grand Rapids data based on the drivers' self-reported frequency of alcohol consumption (Figure 3-3). He drew three conclusions from the results. First, drivers with experi- ence as drinkers are less likely to be involved in crashes than light and moderate drinkers at comparable BACs. Second, regardless of the tolerance for alcohol, the risk of crash involvement increases with BAC. Third, the curves greatly underestimate the risk for the average driver at any BAC; they only demonstrate the relative hazard to drivers who regularly drink and drive. The curvilinear relationship between relative risk of crash involvement and BAC is therefore partially caused by the comparison of drivers with varying degrees of experience as drinkers and experience driving under the influence of alcohol'. When experience with alcohol is controlled for, the risk of crash involvement increases with BAC without evidence of a threshold effect. As noted by Perrine in his review of the literature, the relative risk of involvement is not the same as evidence of causality. Given the many interact- ing factors that may contribute to a crash (and the lack of data on many of them), the role of any single factor is difficult to isolate. Three of the case- control studies deserve special attention because they also estimate the effect of alcohol on the probability of being responsible for a crash.

U Grand Rapids data (5,985 total crashes) U— - - Grand Rapids data (300 total or serious crashes) D—O Evanston data (270 InJury crashes) D— - —O Toronto data (423 total crashes) • Manhattan data (34 fatal crashes) 30 Estimated approximate extension of Manhattan trace O—O Vermont data (106 fatal crashes) a Huntsville data (615 InJury crashes) 26 a-- - - Adelaide data (299 inJury crashes) / / 22 / / 18 / 14 : / / I 10 4' 6 2 0 .02 .04 .06 .08 .10 .12 .14 .16 .18 .20 .22 .24 BAC %WN FIGURE 3-2 Relative probability of crash involvement as a function of BAC. (Hurst 1985. Reprinted with permission.)

54 ZERO ALCOHOL AND OTHER OPTIONS 10.0 - SELF-REPORTED DRINKING FREQUENCY i-. 8:0 : ....... YEARLY OR LESS - • MONTHLY 6.0 - - - WEEKLY ••. / - 3X/WEEK DAILY 4.0 - LL .•/ • .. 2.0- ••, •••..•,• 0.4 - cc I I I 0 .02 .04 .06 .08 BAC, %WN FIGURE 3-3 Relative probability of crash involvement by self-reported drinking frequency(Hurst 1973. Reprinted with permission from the Journal of Safely Research, a joint publication with the National Safety Council and Pergamon Press, Ltd.) The methodology for estimating crash responsibility was first developed by McCarrol and Haddon (1962) in their case-control study of fatal crashes in Manhattan. They categorized the crashes into five classes, the first three of which were assigned responsibility: (1) only one vehicle involved, (2) two vehicles involved but only one moving, and (3) more than one vehicle involved and in motion with responsibility assigned based on circumstances of the crash (cases in which there was any doubt were excluded from this category). The Manhattan study is based on a sample of 43 drivers fatally injured in crashes that occurred between June 1950 and June 1960. For the 26 drivers in the assigned responsibility classes, 19 (65 percent) had positive BACs, of which 14 (46 percent) had BACs greater than 0.10 percent. Of these 14 drivers, 12 had BACs of 0.25 or greater. Of 156 drivers randomly selected as controls at or near the sites of the crashes, 39 (25 percent) had positive BACs, of which only 8 (5 percent) were at or above 0.10 percent. In a study in Grand Rapids, Michigan, by far the largest sample of all the case-control studies was used (5,985 crashes of all types) (Borkenstein et al. 1964). By comparison, the next largest sample, in Toronto, has 423 cases

ALCOHOL, PERFORMANCE, AND CRASH RISK 55 (Hurst 1985). Using McCarrol and Haddon's method for assigning respon- sibility, Borkenstein et al. then estimated that 3,305 of the involved drivers were responsible for the crash that occurred. They used the innocent drivers as contmls. In the third study to estimate crash responsibility, made with data from Vermont (Penine et al. 1971), there were 106 cases (all fatal crashes). These crashes resulted in 113 fatalities, and 97 of the drivers were assigned respon- sibility, again relying on the method developed by McCarrol and Haddon. Of the drivers judged responsible, 60 percent had positive BACs and 46 percent had BACs at or above 0.10 percent. Perrine et al. (1971) also calculated a crash responsibility curve, but in contrast to the Grand Rapids study, the drivers stopped at roadblocks were used as controls. Hurst (1973) replotted the curves from these three studies on a logarithmic scale to facilitate comparison (Figure 3-4). Although in all three studies risk of crash responsibility increases as BAC increases, several disadvantages with the underlying data should be noted. The trend in the Manhattan data is based on a very small number of crashes (25 responsible drivers with positive BACs). In addition, the trend at the higher BACs is greatly understated. For the fatal crashes in which the driver had a BAC of 0.25 percent or more (about half of the drivers in the driver- responsible category) no driver in the control group had an equivalent BAC. "Hence, the relative hazard calculated from the case/control ratio would be infinite within the range, were it possible to graph it" (Hurst 1973). One of the shortcomings of the relative risk curve estimated in the Grand Rapids study is the inclusion of drivers involved in single-vehicle crashes in the group of responsible drivers. Although the responsibility of the driver is not in question, because of the nature of the crash, a control driver is not available. The published data do not provide sufficient detail to allow the curve to be completely recalculated without the single-vehicle crashes to determine the effect of including these crashes, but the available data suggest that the curve would shift to the right. It would still accelerate after 0.04 percent BAC and at an exponential rate, but the curve would not rise as quickly as shown in Figure 3-4. One problem with the Vermont data is the small number of crashes in the sample. In the comparison of crash risk as BAC rises, one or two drivers are responsible in some of the BAC ranges. Chance occurrence could distort the results when so few drivers are the basis of the calculations. Despite the weaknesses in the case-control studies, some important conclu- sions can be drawn. In several of the case-control studies done in the United States and abroad, a consistent increase in risk in crash involvement has been shown. When experience with drinking is controlled for, this risk increases

40 30 I .02 .04 .06 .08 .10 .12 .14 .16 .18 .20 56 ZERO ALCOHOL AND OTHER OPTIONS U GRAND RAPIDS DATA, 3305 CRASHES, DRIVER ASSUMED RESPONSIBLE • MANHAUAN DATA, 24 FATAL CRASHES, DRIVER ASSUMED RESPONSIBLE O-O VERMONT DATA, 75 FATAL CRASHES, DRIVER ASSUMED RESPONSIBLE BAC %(WN)OF DRIVER FIGURE 3-4 Relative crash responsibility for drivers assumed responsible and those not assumed responsible as a function of BAC where 1.0 = relative probability at zero alcohol. (Hurst 1973. Reprinted with permission from the Journal of Safety Research, a joint publication with the National Safety Council and Pergamon Press, Ltd.) with BAC without any evidence of a threshold effect. The three studies that attempt to estimate crash responsibility show that the risk of causing a crash increases even more rapidly than the risk of crash involvement as BAC increases. ALCOHOL AND HEAVY-TRUCK DRIVING Laboratory and simulator studies, the closed-course driving studies, and all of the case-contml studies reviewed document the effects of alcohol on drivers

ALCOHOL, PERFORMANCE, AND CRASH RISK 57 of automobiles. Is this research applicable to the effects of alcohol on truck and bus drivers as well? Evidence for and against application of such research to commercial vehicle driving is reviewed in the following sections. Highly Skilled Performance Many intercity truck and bus drivers are far more skilled at driving than the subjects used in studies of the effects of alcohol (who are typically students or young adults). Some research has suggested, though not proved, that persons highly skilled at certain tasks are less impaired by alcohol when performing these tasks than less skilled persons. For example, Lovibond and Bird (1970) compared the abilities of 16 racing drivers with the abilities of 26 ordinary drivers on a closed-course track at BACs of 0.05, 0.08 and 0.10 percent. For the tracking, steering, and backing maneuvers required in the tests, the overall driving skill of ordinary drivers was impaired at 0.05 percent BAC, but the skilled drivers did not become impaired until 0.08 percent BAC. This study did not estimate statistical significance or provide statistics to allow others to do so. In a similar study, Fomey et al. (1961) were also unable to show significant impairment for racing drivers compared with a control group at BACs of 0.05 percent (except for the task of precision backing). A comparison of nonprofessional and professional pilots also suggests that persons with greater skill and experience are better able to compensate for the effect of alcohol (Billings et al. 1972). In actual flights (with a safety copilot) at mean BACs of 0.04, 0.08, and 0.12 percent, eight nonprofessional pilots accumulated many errors in tracking and in takeoff and landing procedures. In contrast, eight professional pilots demonstrated their ability to compensate for the effects of alcohol. Even at the highest dosage their tracking abilities did not decrease. At the lowest BACs, however, the professional pilots committed hazardous procedural errors not associated with tracking skills. As has been shown in many studies, simple tasks that receive the subject's undivided attention, such as tracking, tend to require high BACs (0.10 or more) before degrading. The studies mentioned earlier, though not conclusive, suggest that more skilled persons are better able to compensate for the effects of alcohol than less skilled persons. The increase in procedural errors by the skilled pilots at low BACs, however, implies that when their attention is divided between two or more tasks, attention is focused on one task at the expense of the others (Moskowitz 1973). Skilled drivers may be less affected at low BACs than unskilled drivers during relatively routine driving. Nevertheless, even skilled drivers show a decreased ability to handle divided-attention tasks. These skills are precisely those required in a potential accident situation. As demonstrated by Laurell's

58 ZERO ALCOHOL AND OTHER OPTIONS closed-course experiment, performance decreases at low BACs in potential accident situations. Although skilled drivers might be expected to perform better than unskilled drivers in such situations, as shown by the effect of low BAC on divided-attention tasks, their abilities would probably be impaired even at low BACs. It is also important to remember that whereas many commercial vehicle drivers employed by large truck and bus firms have considerable experience and safe driving records, some drivers of medium and heavy trucks have no more training and experience than the average passenger car driver. Though the provisions of the Commercial Motor Vehicle Safety Act of 1986 will require testing for licensure, in the past, few states required drivers to demonstrate ability or complete a driving course before obtaining a commercial license. Fatigue Fatigue may also offset the ability of'skilled drivers to perform better than average drivers at low BACs. As noted in Chapter 2, commercial vehicle drivers work long hours and many admit to occasionally dozing at the wheel. Actual crash frequency for commercial vehicle drivers compared with the expected rate tends to increase after about 7 hr of driving (Harris and Mackie 1972, viii; Mackie and Miller 1978, xii) and most drivers work 9 to 10 hr a day. The few studies done on alcohol and fatigue interaction have had equivocal results because of small samples, wide variability in individual responses, and inadequate controls (Ryder et al. 1981). Nevertheless, one result that consistently emerges is that subjects report feeling more fatigued under the influence of even low doses of alcohol than in the control tests (Landauer and Howat 1983; Lister and File 1983). Any increase in fatigue that leads to dozing at the wheel obviously increases the risk of crash involvement. Demands of Commercial Vehicle Driving The task of truck driving is simply much more demanding than driving a passenger car. Commercial vehicle driving requires more attention and de- mands greater skills than automobile driving, both in response to potential hazards as well as under normal driving situations (Eicher et al. 1982; Ranney et al. 1984). Commercial vehicles—particularly large trucks—are much less maneuverable than the automobiles with which they share the road. They require greater distances for passing, stopping, turning, and accelerating. As a result, the commercial vehicle driver must anticipate approaching danger much sooner than the car driver and, when confronted with it, is less able to

ALCOHOL, PERFORMANCE, AND CRASH PJSK 59 take effective evasive action. These demands on driver alertness will continue to increase with the projected increase in traffic on major rural and urban highways. Furthermore, basic highway design standards, often predicated on automobile characteristics, can result in significantly less margin for error in the routine driving of commercial vehicles than in the driving of automobiles (Ervin et al. 1986, 77-8). Some of the specific characteristics of commercial vehicles that intensify the driving task are summarized in the following paragraphs. Offiracking Seldom do the rear wheels of any turning vehicle follow in the tracks of the forward wheels. This phenomenon is greatly exaggerated for vehicles with a long wheelbase, making it difficult for commercial vehicle drivers to execute tight right turns at intersections (Jackson 1986; Billing and Mercer 1986) and mandating extra outside clearance when small-radius curves must be negoti- ated at relatively high speeds, such as when an off ramp is used (Ervin et al. 1986). Acceleration In part because of their large weight-to-power ratios, commercial vehicles are unable to accelerate as rapidly as passenger cars. Safety-related decisions, such as to enter an intersection (Jackson 1986), to cross at-grade rail tracks (Jackson 1986), to begin a passing maneuver on a two-lane road (Gordon 1979; Khasnabis 1986), or to perform a freeway merge (Ervin et al. 1986), must be made much sooner by the commercial vehicle driver. Center of Gravity The center of gravity of a large truck, measured from the roadway, is much higher than that of a passenger car. This higher center of gravity introduces overturning as a major mode of instability and a major cause of truck accidents (Ervin et al. 1983). Trucks with high center of gravity will overturn on some curves considered by current standards to have been properly de- signed (Ervin et al. 1986). On horizontal curves, the truck driver must adjust travel speed to reflect not only outward sliding—as must the car driver—but also the potential for overturn.

60 ZERO ALCOHOL AND OThER OPTIONS Deceleration Commercial vehicles cannot stop as quickly as cars can, their brakes are often out of adjustment or need maintenance, and loss of brakes on long down- grades is a particularly truck-related problem (Radlinski 1987). To compen- sate for deficient braking, the truck driver must scan farther down the road for potential danger, critically assess the severity of downhill grades, maintain a larger-than-normal gap to leading vehicles, and check brake adjustment more carefully and more often. Sensory Feedback The driver of articulated vehicles, such as tractor-semitrailers and full-trailer trucks (those with axles at both ends of the trailer), is somewhat isolated from the trailing units and less well able than drivers of nonarticulated vehicles such as cars and straight trucks to sense impending sliding, rollover, or even bouncing of these units (TRB 1986). As partial compensation, the driver must use rear-view mirrors to sense impending instabilities. As a result, attention is partly diverted from the view ahead, a potentially critical problem in emer- gency situations (Radlinski 1987; Billing 1982). Length Commercial vehicles are longer than passenger cars. This exacerbates the difficulty in lane changing and merging because a larger gap is required in which to move and because of impairment in the driver's ability to judge the clearance for trailing vehicles (California Business and Transportation Agency 1972; Michigan Department of State Highways and Transportation 1976). Overtaking on two-lane roads is also a more critical task in longer vehicles. Rearward Ampl/ication Abrupt steering in multiple-trailer trucks—as might occur in evasive maneu- vering—induces a transient motion in trailing units with exaggerated lateral acceleration. One unwelcome consequence is sometimes rollover of the rear- most trailer (Ervin et al. 1983). Drivers of multiple-trailer trucks must be aware of this type of instability and 'be especially vigilant to avoid situations demanding immediate and sharp turns.

ALCOHOL, PERFORMANCE, AND CRASH RISK 61 Controls and Displays Safe driving requires that drivers be able to quickly and accurately locate displays and reach controls of the vehicle. This task is much more difficult for truck drivers than car drivers. The truck driver is much more likely to switch from vehicle to vehicle than passenger vehicle drivers and hence be unac- customed to the layout and operational features of importance. In addition, the truck driver is presented with more displays (about 34 in a truck versus 17 in a car) and more controls (about 52 versus 33) than the typical light truck or car driver (Eicher et al. 1982). The foregoing summaries, although excluding for the sake of brevity other potentially significant features such as reduced rearward and sideward vis- ibility (Eicher et al. 1982) and increased vehicle width, demonstrate the added complexity of driving commercial vehicles in comparison with the smaller and more maneuverable automobile. This added complexity occurs in all of the three dimensions typically used in highway safety studies: driver, vehicle, and roadway and environment. Commercial drivers often drive for long hours and complain of fatigue and drowsiness, which alcohol, even at low BACs, exacerbates. The monotony and fatigue of driving increase the risk of crash occurrence after about 7 hr and low BACs would increase that risk even more. As just shown, large vehicles are much more difficult to drive and have longer stopping distances, slower acceleration capability, wider turning radii, more controls and displays, and other features that place more demands on the driver. Finally, large trucks operate close to the design limits of major highways, and in some cases (e.g., ramps) exceed them. Ranney et al. (1984) summarized the relative effects of alcohol impairment on heavy-truck and automobile driving as follows: It is apparent that the effects of alcohol are more likely to influence the driving of a heavy truck than of a passenger car, given an identical situation. In general, heavy trucks must operate closer to the design limits of both the vehicle and the roadway, the result of which is that smaller margins of error exist, particularly for the recovezy of an errant vehicle. Because of the increased attentional demands and the increased precision required of heavy truck drivers, it can be concluded that in almost any driving situation, the effects of alcohol on a heavy truck driver would prove more debilitating than for drivers of other vehicles. Although skilled drivers are able to partly compensate for low BACs when performing routine driving tasks, it appears that, given the complexity of the commercial driving task, commercial vehicle drivers cannot count on skill to significantly mitigate the effects of alcohol.

62 ZERO ALCOHOL AND OTHER OPTIONS SUMMARY Human performance in driving-related tasks decreases at low BACs. Al- though some research suggests that performance decrements occur within the visual system (Wilson and Mitchell 1983; Adams and Brown 1975), more conclusive effects for driving-related skills and performance have been dem- onstrated at the cognitive level. BACs below 0.05 percent cause significant increases in errors in tasks of divided attention (Connors and Maisto 1980; Evans et al. 1974; Moskowitz et al. 1985). Studies that have examined performance decrements over a range of BACs suggest that performance decreases in a linear fashion starting with the lowest BAC examined (Moskowitz et al. 1985; Evans et al. 1974; Drew et al. 1959; Landauer and Howat 1983; Adams and Brown 1975). As BACs increase, the ability of most persons to recognize a stimulus and decide on an appropriate response de- creases. In potential accident situations, low BACs increase stopping distance and increase errors in steering (Laurell 1977). These effects also occur in hangover conditions (Laurell and TOmros 1983). Given the variability in individual responses to alcohol, not all persons are impaired at these low BACs, but it appears that cognitive performance is decreased for most individuals at BACs of 0.04 percent or less. Case-control studies show that across large groups of drivers the responsibility for crashes increases when the driver judged culpable for the crash has a BAC greater than 0.04 percent (Hurst 1973, 1986; Perrine et al. 1971; Perrine 1975). When the case-control data are controlled for experience with alcohol, the risk of crash involvement begins increasing at the lowest BACs measured (Hurst 1973). Without specific empirical research on the effect of low BACs on commer- cial vehicle drivers, the effect of low BACs on accident risk cannot be estimated with complete confidence; nonetheless, a number of findings sug- gest that at BACs well below 0.10 percent, the abilities of most commercial vehicle drivers would be reduced and the risk of crash occurrence would be increased. Although routine driving by an alert, skilled driver may not be affected at low BACs, even professional pilots commit more procedural errors when presented with divided-attention tasks. Furthermore, drivers are not always alert. Federal regulations permit drivers to work up to 10 hr per day and crash risk begins increasing after about 7 hr of driving (Harris and Mackie 1972; Mackie and Miller 1978). Commercial vehicle drivers report suffering from fatigue (Wyckoff 1979), and the frequency of falling asleep at the wheel may be increased by even low BACs (Landauer and Howat 1983; Lister and File 1983). The demands on drivers of heavy commercial vehicles are much greater than the demands on drivers of passenger cars. Heavy trucks must be operated close to the margin of the capability of the vehicles and the design of

ALCOHOL, PERFORMANCE, AND CRASH RISK 63 the highway. These findings indicate that at any BAC above zero most commercial vehicle drivers would experience a degradation in skill that would increase the risk of crash involvement. REFERENCES Adams, A. J., and B. Brown. 1975. Alcohol Prolongs Time Course of Glare Recovery. Nature, Vol. 257, pp. 481-483. Billing, A. M. 1982. Rollover Tests of Double Trailer Combinations. Report TVS- CV-82-1 14. Transport Technology and Energy Division, Ministry of Transporta- tion and Communications, Downsview, Ontario, Canada. Billing, J. R., and W. R. J. Mercer. 1986. Swept Paths of Large Trucks in Right Turns of Small Radius. In Transportation Research Record 1052, TRB, National Re- search Council, Washington, D.C., pp. 116-119. Billings, C. E., R. L. Wick, R. J. Gerke, and R. C. Chase. 1972. The Effects ofAlcohol on Pilot Performance During Instrument Flight. Report FAA-AM-72-4. Federal Aviation Administration, U.S. Department of Transportation. Cited by H. Moskowitz (1973). Bjerver, K., and L. Goldberg. 1950. Effect of Alcohol Ingestion on Driving Ability: Results of Practical Road Tests and Laboratory Experiments. Quarterly Journal on Studies on Alcohol, Vol. 11, pp. 1-30. Blalock, H. M. 1972. Social Statistics, 2nd ed. McGraw Hill, New York. Borkenstein, R. F., R. F. Crowther, R. P. Shumate, W. W. Zeil, and R. Zylinan. 1964. The Role of the Drinking Driver in Traffic Accidents. Department of Police Admin- istration, Indiana University, Bloomington. California Business and Transportation Agency. 1972. Triple Trailer Study in California. Callahan, D., I. H. Cisin, and H. M. Crossley. 1969. American Drinking Practices: A National Study of Drinking Behavior and Attitudes. Monograph 6. Rutgers Center for Alcohol Studies, New Brunswick, N.J. Cherry, N., et al. 1983. The Effects of Toluene and Alcohol on Psychomotor Perfor- mance. Ergonomics, Vol. 26, pp. 108 1-1087. Collins, W.E. 1980. Performance Effects of Alcohol Intoxication and Hangover at Ground Level and at Simulated Altitude. Aviation, Space, and Environmental Medicine, Vol. 51, pp. 327-335. Congressional Record. 1986. Vol. 132, No. 144 (daily ed., Oct. 17, 1986), S 16919. Connors, G. J., and S. A. Maisto. 1980. Effects of Alcohol, Instructions and Consump- tion Rate on Motor Performance. Journal of Studies on Alcohol, Vol. 41, pp. 509-517. Drew, G. C., W. P. Colquhoun, and H. A. Long. 1959. Effect of Small Doses ofAlcohol on a Skill Resembling Driving. Her Majesty's Stationery Office, London. Eicher, J. P., H. D. Robertson, and C. R. Toth. 1982. Large Truck Accident Causation. Report DOT HS 806 300. National Highway Traffic Safety Administration, U.S. Department of Transportation. Ervin, R. D., et al. 1983. Influence of Size and Weight Variables on the Stability and Control Properties of Heavy Truckc. Report FHWA-RD-83-029. University of Michigan Transportation Research Institute, Ann Arbor.

64 ZERO ALCOHOL AND OTHER OPTIONS Ervin, R. D., C. C. MacAdam, and M. Barnes. 1986. Influence of the Geometric Design of Highway Ramps on the Stability and Control of Heavy-Duty Trucks. In Transportation Research Record 1052, TRB, National Research Council, Wash- ington, D.C., pp. 77-89. Evans, M. A., et al. 1974. Quantitative Relationship Between Blood Alcohol Con- centration and Psychomotor Performance. Clinical Pharmacology and Therapeu- tics, Vol. 15, No. 3, pp. 253-260. Federal Register. 1985. Use of Alcohol or Drugs. Vol. 50, No. 74, April 17, pp. 15376-15383. Fomey, R. B., F. W. Hughes, H. R. Hulpieu, and C. A. Davis. 1961. Performance in a Gymkhana Sports Car Event with Low Levels of Blood Alcohol. Traffic Safety Research Review, Vol. 5, No. 3, pp. 5-12. Gordon, D. A. 1979. Highway Sight-Distance Requirements: Truck Applications. Re- port FHWA-RD-79-26. Office of Research, Federal Highway Administration, U.S. Department of Transportation. Harris, W., and R. Mackie. 1972. A Study of the Relationships Among Fatigue, Hours of Service, and Safety of Operations of Truck and Bus Drivers. Report BMCS- RD-71-2. Federal Highway Administration, U.S. Department of Transportation. Heise, H. A. 1934. Alcohol and Automobile Accidents. Journal of the American Medical Association, Vol. 103, pp. 739-741. Holcomb, R. L. 1938. Alcohol in Relation to Traffic Accidents. Journal of the Ameri- can Medical Association, Vol. 107, pp. 1067-1038. Cited by Perrine (1975, 16). Huntley, M. S. 1973. Alcohol Influences Upon Closed Course Driving Performance. Journal of Safety Research, Vol. 5, No. 3, pp. 149-164. Huntley, M. S. 1974. Effects of Alcohol, Uncertainty and Novelty Upon Response Selection. Psychopharmacologia, Vol. 39, pp. 259-266. Hurst, P. 1973. Epidemiological Apects of Alcohol in Driver Crashes and Citations. Journal of Safety Research, Vol. 5, No. 3, pp. 130-148. Hurst, P. 1985. Blood Alcohol Limits and Deterrence: Is There a Rational Basis for Choice? Alcohol, Drugs, and Driving: Abstracts and Reviews, Vol. 1, University of California, Los Angeles. Jackson, L. E. 1986. Truck Accident Studies. In Transportation Research Record 1052, TRB, National Research Council, Washington, D.C., pp. 137-145. Khasnabis, S. 1986. Operational and Safety Problems of Trucks in No-Passing Zones on Two-Lane Rural Highways. In Transportation Research Record 1052, TRB, National Research Council, Washington, D.C., pp. 36-44. Landauer, A. A., and P. Howat. 1983. Low and Moderate Alcohol Doses, Psychomotor Performance and Perceived Drowsiness. Ergonomics, Vol. 26, No. 7, pp. 647-657. Lauren, H. 1977. Effects of Small Doses of Alcohol on Driver Performance in Emergency Traffic Situations. Accident Analysis and Prevention, Vol. 9, pp. 191-201. Laurell, H., and J. Törnros. 1983. Investigation of Alcoholic Hang-over Effects on Driving Performance. Bluralkohol, Vol. 20, pp. 489-499. Lister, R. G., and S. E. File. 1983. Performance Impairment and Increased Anxiety Resulting from the Combination of Alcohol and Lorazepam. Journal of Clinical Psychopharmacology, Vol. 3, pp. 66-71. Lovibond, S. H., and K. Bird. 1970. Effects of Blood Alcohol Level on the Driving Behavior of Competition and Non-Competition Drivers. Presented at 29th Interna- tional Conference on Alcoholism and Drug Dependence, Sydney, Australia.

ALCOHOL, PERFORMANCE, AND CRASH RISK 65 MacArthur, R. D., and R. Sekuler. 1982. Alcohol and Motion Perception. Perception and Psychophysics, Vol. 31, pp. 587-596. McCarrol, J. R., and W. M. Haddon. 1962. A Controlled Study of Fatal Automobile Accidents in New York City. Journal of Chronic Diseases, Vol. 15, pp. 8 11-826. Mackie, R., and J. Miller. 1978. Effects of Hours of Service, Regularity of Service, and Cargo Loading on Truck and Bus Driver Fatigue. Report DOT HS-5-01142. National Highway Traffic Safety Administration, U.S. Department of Transportation. Michigan Department of State Highways and Transportation. 1976. Operational Characteristics of 100-Foot Double Trailer/Tractor Operations in Michigan. Lansing. Miles, W. R. 1934. Alcohol and Motor Vehicle Drivers. HRB Proc., Vol. 13, pp. 362-379. Mortimer, R. G. 1963. Effect of Low Blood-Alcohol Concentration on Simulated Day and Night Driving. Perceptual and Motor Skills, Vol. 17, pp. 399-408. Moskowitz, H. 1973. Laboratory Studies of the Effects of Alcohol on Some Variables Related to Driving. Journal of Safety Research, Vol. 5, No. 3, pp. 185-197. Moskowitz, H., M. Burns, and A. Williams. 1985. Skills Performance at Low Blood Alcohol Concentrations. Journal of Studies on Alcohol, Vol. 46. Moskowitz, H., and C. Robinson. 1987. Effects of Low Doses of Alcohol on Driving Related Skills: A Review of the Evidence (draft). SRA Technologies, Alexandria, Va. Moskowitz, H., and S. Sharma. 1974. Effects of Alcohol on Peripheral Vision as a Function of Attention. Human Factors, Vol. 16, pp. 174-180. Olson, P., D. Cleveland, P. Fancher, L. Kostyniuk, and L. Schneider. 1984. NCHRP Report 270: Parameters Affecting Stopping Sight Distance. TRB, National Re- search Council, Washington, D.C. O'Neill, B., A. Williams, and K. Dubowski. 1983. Variability in Blood Alcohol Concentrations. Journal of Studies on Alcohol, Vol. 44, No. 2, pp. 222-230. Palva, E. S., et al. 1982. Actions and Interaction of Diazepam and Alcohol on Psychomotor Skills in Young and Middle-Aged Subects. Acta Pharmacologia et Toxicologica, Vol. 50, pp. 363-369. Perrine, M. W. 1973. Alcohol Influences Upon Driving Related Behavior: A Critical Review of Laboratory Studies of Neurophysiological, Neuromuscular, and Sensory Activity. Journal of Safety Research, Vol. 5, No. 3, pp. 165-184. Perrine, M. W. 1975. Alcohol Involvement in Highway Crashes: A Review of the Epidemiologic Evidence. Clinics in Plastic Surgery, Vol. 2, No. 1, pp. 11-34. Perrine, M. W. 1976. Alcohol and Highway Crashes: Closing the Gap Between Epidemiology and Experimentation. Modern Problems of Pharmacopsychiatry, Vol. 11, pp. 22-41. Perrine, M. W., J. A. Waller, and L. S. Harris. 1971. Alcohol and Highway Safety: Behavioral and Medical Aspects. Report DOT-HS-900-599. National Highway Traffic Safety Administration, U.S. Department of Transportation. Radlinski, R. W. 1987. Braking Performance of Heavy U.S. Vehicles. SAE Technical Paper 870492. Society of Automotive Engineers, Inc., Warrendale, Pa. Ranney, T. A., K. Perchonok, and L. E. Pollack. 1984. Identification and Testing of Countermeasures for Specific Alcohol Accident Types and Problems, Vol. 3: The Heavy Truck Alcohol Problem. Report DOT-HS-806-651. National Highway Traf- fic Safety Administration, U.S. Department of Transportation.

66 ZERO ALCOHOL AND OTHER OPTIONS Ryder, J. M., S. A. Maim, and C. H. Kinsley. 1981. The Effects of Fatigue and Alcohol on Highway Safety. Report DOT HS-805 854. National Highway Traffic Safety Administration, U.S. Department of Transportation. Shinar, D., D. Zaidel, and W. Paarlberg. 1978. Driver Performance and Individual Differences in Attention and I,formation Processing. Report DOT-HS-8-01819. Institute for Research in Public Safety, Indiana University. TRB. 1986. Special Report 211: Twin Trailer Trucks. National Research Council, Washington, D.C. Wilson, G., and R. Mitchell. 1983. The Effect of Alcohol on the Visual and Ocular Motor Systems. Australian Journal of Opthalmology, Vol. 11, pp. 315-319. Wilson, J. R., et al. 1984. Effects of Ethanol: II. Behavioral Sensitivity and Acute Behavioral Tolerance. Alcoholism: Clinical and Experimentai Research, Vol.8, No. 4, pp. 366-374. Wyckoff, D. 1979. Truck Drivers in America. D.C. Heath and Co., Lexington, Mass. Yesavage, I. A., and V. 0. Leirer. 1986. Hangover Effects on Aircraft Pilots 14 Hours After Alcohol Ingestion: A Preliminary Report. American Journal of Psychiatry, pp. 1546-1550.

4 Apprehending the Impaired Driver: Legal Issues and Testing Technologies In the preceding chapter evidence was examined showing driver impairment at BACs lower than the current general enforcement standard of 0.10 percent. The ability to determine impairment and apprehend drivers of commercial vehicles at a more stringent standard will depend on the feasibility of detect- ing and measuring impairment at lower BACs and on the legal framework within which enforcement efforts must be carried Out. In this chapter an overview of the legislative context that guides enforcement practices is provided, the key legal issues that constrain enforcement of a more stringent BAC standard are identified, and the effectiveness of current methods for measuring alcohol impairment, particularly at lower BACs, is reviewed. OVERVIEW OF THE LEGAL SYSTEM AND ENFORCEMENT PROCESS Drinking-Driving Legislation in the United States In the United States control of thinking and driving is enforced through the legal system. The control process is embodied in a system of traffic laws that provide for enforcement, adjudication, and sanctioning (NHTSA 1985, 52). In recent years this system has been in a state of change, with many revisions in the laws guiding enforcement of drinking-driving restrictions. Although en- forcement systems in the United States can still be characterized as predomi- nantly behavioral, that is, based on observable driver behavior, recent legisla- tion is moving in the direction of a more objective system based on verifiable test result.s. 67

68 ZERO ALCOHOL AND OTHER OPTIONS Early state drinking-driving statutes focused entirely on the behavior of the driver. They defined driving while legally drunk as a criminal act and then subsequently broadened the definition to include individuals driving under the influence of alcohol or driving while impaired (Voas 1982, 5). Proving that an individual was guilty of a drinking-driving offense required behavioral evi- dence that the driver was sufficiently lacking in control of his vehicle to be a safety risk on the road, proof that was subjective in nature and frequently open to challenge. The introduction of chemical testing for alcohol in the blood had a major impact on alcohol safety legislation. The test provided an objective means of determining a drinking-driving violation by measuring actual BACs in drivers.1 The BAC limits were based on studies that defined a BAC at which the ability of all drivers to safely operate a vehicle is impaired (Reeder 1974, 179). In 1936 Norway passed the first legislation under which driving at or above a specified BAC was deemed in itself (i.e., per se) to be illegal irrespective of driver behavior. The state of Indiana adopted chemical testing in 1939—the first state to do so—but the tests were used only after the driver had already been arrested and charged with driving under the influence (DUI). Indiana's law provided only a presumption of intoxication and still required behavioral evidence for a court conviction. In 1962 Nebraska was the first state to adopt a per se law. By 1987, 45 of the 50 states plus the District of Columbia and Puerto Rico had adopted such laws, which have removed the requirement of providing behavioral evidence of impairment to prosecute a drinking-driving offense. The majority of these laws make it illegal to drive at a BAC of 0.10 percent or more; in 41 states this is the standard. Two states have a more stringent 0.08 percent standard, and in two other states it is 0.12 and 0.15 percent (NHTSA 1986, 2-1, 2-2). This represents a significant increase from 1983 when only 26 states had enacted "illegal per se" laws. As the use of chemical testing to establish a DUI offense became more prevalent, all states (plus the District of Columbia and Puerto Rico) followed the example of New York State, which in 1953 adopted an implied consent law to facilitate use of the testing procedure. Because cooperation is required to obtain good test results that can be used as evidence in court, implied consent laws encourage voluntary driver compliance with requests for testing, which eliminates the need for any coercion on the part of enforcement officers. These laws state that persons driving on the highways have given their consent to testing for blood alcohol levels if the officer has established probable cause for a DUI offense, arrested the driver-offender, and warned him of the consequences of refusal. In the event of refusal, licenses are suspended or revoked for a specified period (see Table 41).2 Other innovations in testing technology have had a significant impact on drinking-driving legislation. The concurrent development of portable elec- tronic breath-testing devices and the extensive use of prearrest roadside

TABLE 4-1 SUMMARY OF MAJOR STATE DUI LEGISLATION AS OF JANUARY 1, 1987 State Prelim- nary Breath Test implied Consent Law (All States), Mandatory Minimum Licensing Action First Second Refusal Refusal Illegal Per Sc (BAC Limit) Pre- sumptive (BAC Limit) Administrative Per Se Mandatory Minunum Licensing Action BAC First Second Limit Offense Offense Third Offense Alabama S-90 dys 5-1 yr 0.10 0.10 N - - - Alaska X R-90 dys R-1 yr 0.10 - Y-0.10 R-30 dys R-1 yr R-10 yrs Arizona S-12 mos S-12 mos 0.10 0.10 N - - - Arkansas S-6 mos 5-1 yr 0.10 - N - - - California S-6 mos S-2 yrs 0.10 0.10 N - - - Colorado X R-1 yr R-I yr 0.15 .05, .10" Y-0.15 R-I yr R-1 yr - Connecticut X S-6 mos 5-1 yr 0.10 - N - - - Delaware X R-6 mos R-18 mos 0.10 0.I0C Y R-3 mos R-1 yr R-18 mos District of Columbia S-12 mos S-12 mos 0.10 0•05C y - - - Florida X - S-18 mos 0.10 0.10 N - - - Georgia S-6 mos S-6 mos 0.12 0.10 N - - - Hawaii R-12 mos R-2 yrs 0.10 - N - - - Idaho S-120 dys S-120 dys 0.10 >0.08 N - - - Illinois S-30 dysd S-90 dys 0.10 0.10 Y-0.10 S-30 dysd S-90 dys S-90 dys Indiana X S-I yr' S-I yr' 0.10 0.I0C Y-0.10 S-180 dysf S-180 dys-1 - Iowa X R-240 dys R-360 dysg 0.10 - Y-0.10 - - - Kansas X - - 0.10 0.10c N - - - Kentucky X - - - 0.10 Nh - - - Louisiana S-90 dys S-545 dys 0.10 0.10 Y-0.10 5-30 dys S-365 dys S-365 dys Maine . 5-90 dys 5-1 yr 0.10 - Y-0.10 - S-8 mosi 5-16 mos' Maryland X - - - .08, •13CJ N - - - Massachusetts 5-120 dys 5-120 dys - 0.10 Nh - - - Michigan X - 5-1 yr 0.10 .07, .I0' N - - - Minnesota X - - 0.10 - Y-0.10 - - -

TABLE 4-1 continued State Prelim- mary Breath Test Implied Consent Law (All States), Mandatory Minimum Licensing Action First Second Refusal Refusal Illegal Per Se (BAC Limit) Pie- sumptive (BAC Limit) Administrative Per Se Mandatory Minimum Licensing Action BAC First Second Limit Offense Offense Third Offense Mississippi X S-90 dysk S-90 dys" 0.10 - Y-o.ioI - - - Missouri - R-1 yr 0.10 0.10c Y-0.13 R-I yr' R-1 yr' R-I yr' Montana S-90 dys R-I yr 0.10 0.10 N - - - Nebraska X R-60 dys R-6 mos 0.10 - N - - - Nevada X R-1 yr R-3 yrs 0.10 0.10 Y-0.10 - - - New Hampshire X R-90 dys R-1 yr 0.10 0.IOC N - - - New Jersey R-6 mos R-2 yrs 0.10 - N - - - New Mexico R-I yr R-1 yr 0.10 0.10 Y-0.10 R-90 dys R-1 yr R-1 yr New York X R-6 mos R-1 yr 0.10 .08, .l0c Nh - - - North Carolina X R-6 mos R-12 mos 0.10 - Y-0.I0 R-I0 dys R-10 dys R-10 dys North Dakota X R-1 yr R-2 yr 0.10 - Y-0.10 S-30 dys S-364 dys S-2 yrs Ohio - - 0.10 - Nh - - - Oklahoma - - 0.10 0.IOC Y-0.10 - - - Oregon S-90 dys S-I yr 0.08 0.0811 Y-0.08 - S-90 dys S-90 dys Pennsylvania X S-12 mos S-12 mos 0.10 0.10 N - - - Puerto Rico X - - - 0.10 N - - Rhode Island X S-3 mos S-I yr 0.10 - N - - - South Carolina S-90 dys S-90 dys - 0.10 N - - - South Dakota X - - 0.10 0.10 N - - - Tennessee - - - 0.10 N - - - Texas - - 0.10 - N - - - Utah R-I yr R-1 yr 0.08 - Y-0.08 S-90 dys S-120 dys S-120 dys Vermont X S-6 mos S-I8 mos 0.10 0.10 N - - - Virginia X S-6 mos S-I yr 0.10 0.10 N - - - Washington R-I yr R-2 yrs 0.10 - N - - - West Virginia X R-1 yr R-5 yrs 0.10 0.10 Y-0.10 R-90 dys R-5 yrs R-10 yrs

TABLE 4-1 concluded Implied Consent Law (All States), Administrative Per Se Mandatory Minimum Licensing Action Prelim- Illegal Pre- mary Per Se sumptive Mandatory Minimum Licensing Action Breath First Second (BAC (BAC BAC First Second Third State Test Refusal Refnsal Limit) Limit) Limit Offense Offense Offense Wisconsin X R-15 dys R-60 dys 0.10 - Y-0.20 R-15 dys R-60 dys R-90 dys Wyoming S-6 mos S-6 mos — 0.10 Y-0.10 — S-90 dys S-90 dys Total 26 S: 22 S: 22 .08: 2 >08: 1 22 S: 5 S: 8 S: 7 R: 17 R: 20 .10:41 .10:21 R: 8 R: 8 R: 7 .12: 1 .10 prima .15: 1 facie: 7 Other: 5 NoTE: S = suspension; R = revocation; DWI = driving while intoxicated; Y = yes; N = no. Special license suspension/revocation periods for persons who have been involved in a DWI-related accident and who have had a previous DWI-related vehicle homicide conviction. b The lower of the two limits is for driving while impaired; the higher is for driving under the influence. C BAC limit or limits that indicate prima facie evidence. d Based on personal communication with the Illinois Department of Motor Vehicles. A judicial driving permit may be issued after the first 30 days of the suspension period. e If the court finds that it is in the best interests of society, it may terminate all or any part of this suspension. Suspensions up to 180 days or until the DWI charges have been disposed of, whichever occurs first. A restricted license may be issued for an implied consent law violation provided the defendant pleads guilty to a subsequent DWI charge. ' Courts may take licensing action before DWI criminal adjudication. After two-thirds of the suspension period, 1 year for second offense and 2 years subsequently, the DWI offender may be issued a restricted license after completing an alcohol treatment program. -' The lower of the two limits is for driving under the influence; the higher is for driving while intoxicated. k License suspension for 1 year if the driver has had a previous DWI offense conviction. Special provisions and procedures. Applies only to DWI offenses; for illegal per se and administrative per se actions, a restricted hardship license may be granted provided the defendant has not received such a privilege within the past 5 years. ' Not less than 0.08 percent BAC constitutes being under the influence of intoxicating liquor.

72 ZERO ALCOHOL AND OTHER OPTIONS testing in Great Britain in the late 1960s to determine alcohol use before administration of an evidential chemical test spurred interest in preliminary testing in the United States. Over the past 15 years, 26 states have adopted preliminary breath test laws, which define those circumstances under which preliminary prearrest testing can be conducted (Table 4-1). Typically these tests are used as a roadside screening device to assist the police in determining the extent of driver impairment; preliminary test results are rarely admissible in court. States laws vary widely, however, concerning the conditions under which preliminary breath tests can be administered and the consequences of refusal. Another significant milestone in the development of U.S. drinking-driving legislation was the introduction of a separate civil adjudication process that runs parallel with, but independent of, any criminal DUT proceedings. Many states found that court backlogs and plea bargaining had reduced both the swiftness and the severity of DUI enforcement efforts. In response 22 states, starting with Minnesota in 1976, adopted administrative per se laws that provide for swift license suspension or revocation either on receipt of 0.10 percent BAC test results or on test refusal, irrespective of the outcome of any criminal proceedings (see Table 4-1). To prevent delaying tactics through requests for administrative hearings from again curbing enforcement efforts, in 1982 Minnesota, and several states thereafter, changed its legislation to provide for automatic imposition of licensing sanctions after a set period regardless of when the hearing occurs. The courts have upheld such legislation as long as the statutes follow the procedural due process requirements of the Fourth Amendment. Typically, this means that the state must provide the opportunity for a postsuspension hearing in a timely manner (Reese 1986,54). Several state studies have documented the effectiveness of license revocations as a specific deterrent to illegal drinking and driving behavior (Latchaw 1986; Ross 1987). This approach has provided the model for the sanctioning process to accompany the more stringent BAC standards proposed for commercial vehicle operators. This brief review of drinking-driving legislation in the United States shows a trend toward a more objective testing-based system of traffic laws. The majority of states have adopted per se laws that specify a BAC above which it is illegal to drive irrespective of driving behavior. In addition, nearly half of the states have adopted parallel administrative per se laws, accelerating the civil sanctioning process through swift license suspension or revocation for those drivers testing above an objective legal limit or refusing to be tested. Finally, half of the states have adopted preliminary breath test laws to assist the police in prearrest screening of drivers suspected to be alcohol impaired. Despite these apparent changes toward more objective standards, many aspects of the legal system, particularly those stages of the enforcement

APPREHENDING THE IMPAIRED DRIVER 73 process leading up to an arrest, continue to be driven by behavioral standards and evidence. Moreover, differences among state drinking-driving statutes have resulted in considerable variations in actual enforcement practices. Stages in the Enforcement Process Figure 4-1 presents the various stages in the enforcement process. Enforce- ment is generally triggered by suspicion of alcohol impairment through contact with the driver in one of several situations, including acci- dents,3 traffic violations, or generally erratic driving behavior. Visual cues, such as weaving in and out of a lane or jackrabbit starts, account for approx- imately 90 percent of all stops made by officers patrolling for drunk drivers (Harris et al. 1979, 132). More recently, law enforcement officials have introduced the use of sobriety checkpoints at which all drivers are sys- tematically stopped and checked for alcohol abuse as an additional way of observing possible DUT infractions. Once a vehicle has been stopped, an officer must have some reason to believe that the driver has been drinking before he can justify further inves- tigation to establish the extent of impairment Here again the police rely heavily on evidence such as the odor of alcohol on the breath, bloodshot eyes, or slurred speech to make this determination. Although testing devices (e.g., the passive sensor) have been developed to screen for the presence of alcohol, legal issues (discussed in the next section) in part have slowed their wide- spread adoption. If an officer suspects alcohol use, he can begin the process of further testing to establish the extent of impairment. Most law enforcement agencies con- tinue to favor field or behavioral sobriety tests. These are methods of collect- ing evidence of impairment by observing the psychomotor behavior of the driver thmugh such tests as walking a straight line or standing on one leg. Only after initial observation and preliminary measures of impairment have established probable cause of legal intoxication can the driver be placed under arrest and an evidential test for BAC be required under the implied consent statutes. The process of detecting and measuring alcohol impairment is thus still heavily based on behavior. Implications of Lowering Legal BAC Limits The proposal to lower the legal BAC limits for commercial vehicle drivers raises several issues with respect to current enforcement procedures. First, lowering the BAC limit is likely to increase the difficulty of detecting

Criminal Proceedings —jail, fines (all states) Vehicle Selection Accidents Moving violations Erratic driving Checkpoints Determination of Alcohol Use Behavioral cues (odor of alcohol, slurred speech, etc.) Passive sensor Test for Impairment Field or behav- ioral sobriety tests Preliminary breath tests Arrest and Evidential Test Breath test Blood test Urine test Civil Proceedings - license revo- cation, fines (22 states only) HGURE 4-1 Stages in the DUE enforcement process.

APPREHENDING THE IMPAIRED DRIVER 75 impairment at the point both of selecting vehicles to stop and of determining alcohol use. Conventional methods of identifying drunk drivers through erra- tic driving behavior and other behavioral cues and tests are not likely to be effective at lower BACs. Second, successful detection of alcohol impairment at low BACs will require greater use of enforcement methods that are independent of driver behavior. Greater use of sobriety checkpoints provides opportunities for driver contact without depending on behavioral cues of driver impairment. Objective testing devices can determine the presence of alcohol in the absence of behavioral cues. However, legal issues as well as the accuracy of testing technologies will determine the extent to which these measures are likely to be used. LEGAL ISSUES Judicial interpretation of the Fourth Amendment to the U.S. Constitution, which protects citizens against unreasonable search and seizure, provides the basis for enforcing drinking-driving laws. By definition, whenever a police officer stops a motorist, he has in effect "seized" him and thus must have a reasonable basis for the stop (Compton and Engle 1983, 3). Enforcement procedures to control drinking and driving in the United States thus retain a behavioral component, reflecting a careful balancing of individual rights against legitimate government interest in highway safety. In contrast, other countries, such as Australia, which are not bound by such constitutional constraints, use a system that is independent of the driver's behavior. Drivers can be stopped at sobriety checkpoints without visible evidence of alcohol impairment, automatically administered roadside tests to determine alcohol levels, and then given evidential chemical tests if found over the legal 0.05 BAC limit (NIHTSA 1985, 54). Establishing Cause Generally speaking, law enforcement officials in the United States must have cause for stopping and testing a driver for intoxication. (Checkpoints, which are an exception to this rule, will be discussed subsequently.) Once the driver is requested to submit to increasingly lengthy and intrusive procedures to measure alcohol impairment, the reasons for detaining the driver must become increasingly strong. This process of establishing cause can best be understood by referring back to the diagram of the stages in the enforcement process (Figure 4-1).

76 ZERO ALCOHOL AND OTHER ONiONS Reasonable Suspicion Stopping a vehicle, determining alcohol use, and preliminary testing of a driver for alcohol impairment—the first three stages in the enforcement process—require that a law enforcement officer have only a reasonable suspicion of alcohol use. According to the courts, which have defined reason- able suspicion with respect to stopping and frisking a suspect, nonevidential seizures are permitted when "specific and articulate facts which, taken to- gether with rational inferences from these facts, reasonably warrant the intru- sion" [Terry v. Ohio, 392 U.S. 1 (1968)]. The courts, however, have given little direction on what constitutes an adequate basis for determining whether a person has been drinking. Many law enforcement agencies have adopted the 11 symptoms or signs of alcohol intoxication recognized by the State Supreme Court of Oregon: the odor of alcohol on the breath, flushed appearance, lack of muscular coordination, speech difficulties, disorderly or unusual conduct, mental disturbance, visual disorders, sleepiness, muscular tremors, dizziness, and nausea [State v. Clark, P.2d 123 (1979)]. Considerable ambiguity exists concerning the use of devices for prelimi- nary breath testing at this stage in the enforcement process, because these may constitute a seaith under the Fourth Amendment and thus require more than a reasonable suspicion to administer (Manak and Engle 1985,4). Legal restric- tions on the use of preliminary breath tests will be discussed subsequently. Probable Cause Only after initial observation and preliminary investigation of the level of intoxication have given an officer sufficient evidence to establish probable cause of a legally actionable offense can he proceed to the fourth and final stage of the enforcement process—arresting the driver and requiring an evidential test that meets the definition of a search under the Fourth Amend- ment. In 1949 the U.S. Supreme Court (Brinegar v. United States, 338 U.S. 160) defined probable cause as "a reasonable ground for belief of guilt" that is stronger than reasonable suspicion but less strong than "evidence which would justify condemnation" (Ruschmann et al. 1980, 15). In a landmark U.S. Supreme Court case in 1966 (Schmerber v. California, 348 U.S. 757) that applied the Fourth Amendment to alcohol testing, the probable cause standard was upheld for intrusive sobriety tests such as chemical analysis of BAC (lift 1983).

APPREHENDING THE IMPAIRED DRIVER 77 Effect of Lower BAC Limits Actual or perceived legal restrictions on the use of various enforcement practices will have important implications for the feasibility of enforcing more stringent BAC standards for commercial vehicle drivers. In particular, greater use of objective, nonbehavioral enforcement techniques such as sobriety checkpoints and prearrest testing will be required to detect lower levels of alcohol impairment. However, case law is not as well developed regarding the use of many of these techniques as it is with the more traditional behavior- based approaches (Ruschmann et al. 1980). Thus, some states are acting conservatively and opting for the more conventional, less controversial be- havior-based enforcement procedures. Sobriety Checkpoints Sobriety checkpoints may be an exception to the rule that a law enforcement officer must have reasonable suspicion of alcohol use to stop a vehicle on the highway. In a sobriety checkpoint, all drivers, or a systematic sample of drivers, who pass a particular location are stopped and briefly observed for signs of intoxication. Determination of whether a checkpoint is deemed an unreasonable seizure under the Fourth Amendment rests on a balancing of the state interests served by the seizure with the extent of the intrusion on individual rights. The constitutionality of sobriety checkpoints has not been adjudicated by the U.S. Supreme Court. However, in a related case in 1969, Delaware v. Prouse (440 U.S. 648), the court ruled against discretionary selection of drivers to check licenses and registration when there was no evidence of wrongdoing, but indicated that other methods for spot checks, such as road- blocks, might be acceptable if they were conducted in a less intrusive and arbitrary fashion (lift 1983). In a more recent case, the court has upheld use of roadblocks to check licenses and take enforcement action against other crimes that also come to the attention of the police [Texas v. Brown, 460 U.S. 730 (1983)]. In addition, many appellate court decisions on related issues have held that checkpoint operations are permissible. Case law thus indicates that checkpoints can be used in most states as a reasonable law enforcement practice to promote a legitimate government interest, highway safety, if conducted with a minimum of intrusion and in a nonarbitrary manner (LaFave 1987, Vol.4, 70). In some states, however, state constitutional restrictions may prohibit the use of checkpoints as an enforce- ment procedure. The National Highway Traffic Safety Administration (NHTSA) reports that as of 1985, law enforcement agencies were using

78 ZERO ALCOHOL AND OTHER OPTIONS checkpoints in 39 states. Checkpoints should be used as one component of a systematic plan of drinking-driving deterrence and should employ specific field procedures to limit any appearance of discriminatory or overly intrusive police action (Compton and Engle 1983, 8). Prearrest Testing At lower BACs, use of objective testing devices will be critical to screen for alcohol use and ascertain likely levels of impairment that are unlikely to be revealed through more conventional behavioral measures. Field testing pro- cedures would include a preliminary breath test (PBT), administered on the basis of reasonable suspicion of alcohol abuse; test results would provide more substantive evidence to establish probable cause of a DUI offense, although they would not be used as evidence of impairment in an administra- tive hearing or court trial. The primary legal issue is whether the test itself can be considered a search within the meaning of the Fourth Amendment (Manak and Engle 1985,4). If the answer is yes, then a police officer can administer a PBT only after he has established a strong case for probable cause of a DUI infraction, the very task that the PBT was meant to facilitate. If the answer is no, then a lesser standard such as the "stop and frisk" standard of reasonable suspicion would be sufficient grounds for administering the test. Use of PBTs at a checkpoint may present additional problems. 1'pically, the officer has little indication of alcohol impairment at the point of initial contact with the driver and must justify increasing intrusiveness, such as requesting a PBT, or the checkpoint may be deemed an unreasonable search (Ifft 1983). The constitutionality of using PBTs on less than a probable cause standard has not been adjudicated by the U.S. Supreme Court. Many would argue, however, that several recent state high court decisions as well as the balancing argument in the related Terry v. Ohio bode well for a positive decision (Macdonald and Wagner 1981, 2-2 to 2-7). They believe that the court will accept the use of PBTs as a relatively minor intrusion when balanced against an important governmental interest in detecting alcohol-impaired driving behavior (Manak and Engle 1985, 5). Under these circumstances, no more than a reasonable suspicion standard should be required for their use. Others argue that because PBTs require driver cooperation as well as a specimen of deep lung air for accurate readings, their use is considered a search within the meaning of the Fourth Amendment and thus requires probable cause (Ruschmann et al. 1980, 20). In light of this problem, the passive alcohol sensor has been developed to test the ambient air in front of

APPREHENDING THE IMPAIRED DRIVER 79 the driver's mouth for the presence of alcohol. Because this method of testing is considered by many to be less intrusive than a PBT, it may be permissible to use the passive sensor without probable cause and still not violate the sus- pect's Fourth Amendment rights (Fields and Hricko 1986, 50). The courts have not ruled on this issue and passive screening technology has not yet been widely adopted by law enforcement agencies. State laws authorizing the circumstances under which PBTs can be admin- istered reflect these differences in interpretation. Some states require a proba- ble cause standard for use of PBTs; others indicate that accidents, traffic violations, or any other incidents providing reasonable suspicion of alcohol use provide sufficient grounds to administer a test (Macdonald and Wagner 1981, 2-9 to 2-22). The real difficulty arises if the individual refuses the test. Only eight states invoke any penalty for refusal to submit to the test (NHTSA 1986). Minnesota, for example, can require an evidential test if a person has refused a PBT or if the test shows a BAC of 0.10 percent or more (NHTSA 1986, 3-183). Similarly, in Nebraska refusal to take the test is grounds for arrest and a fine (Macdonald and Wagner 1981,2-15). These, however, are the exceptions rather than the rule. Applying the principle of implied consent to commercial vehicle drivers, as described in the following section, could reduce the problem of driver refusal. Adopting a Regulatory Approach One option for relaxing the probable cause standard for testing is to adopt the regulatory approach established in the airline and rail industries. Federal regulations covering these industries have recently been amended to require, as a condition of employment, employee consent to testing and release of test results when there is a reasonable suspicion of alcohol or drug abuse (Federal Register 1985a, 1985b).4 The regulations further mandate a per se standard of alcohol intoxication of 0.04 percent and establish a civil sanction, disqualifica- tion for service for a specified period, for those employees who test positive or refuse the test.5 In this context, in which the test or "search" is clearly administrative, a lesser standard of reasonable suspicion (often referred to as administrative probable cause) rather than probable cause should be sufficient grounds for conducting tests (Federal Register 1985b, 31565). Proposed alcohol regulations for commercial vehicle drivers—a per se BAC limit and a civil sanction—follow the model of the airline and rail industries. To facilitate enforcement efforts, the regulations could further stipulate that commercial vehicle drivers explicitly consent to testing and admission of test results on a minimum standard of reasonable suspicion as a condition of receiving a commercial license. The authority to make these

80 ZERO ALCOHOL AND OTHER OPTIONS regulations derives from a long history of federal regulation of commercial vehicle driver safety under the commerce clause, which now extends to intrastate as well as to interstate drivers. State legislation embodying the federal regulations will be required, however, if state and local law enforce- ment officials, who are likely to administer the tests, are to adopt this lower standard. Summary Many of the enforcement practices that will be required to detect impairment at lower BACs, such as checkpoints and prearrest testing, have been so relatively recently introduced that the legal issues concerning their use have not always been adjudicated. It appears that sobriety checkpoints are likely to be ruled constitutional in most states as a reasonable law enforcement practice as long as they are conducted properly. The circumstances under which PBTs can be administered, however, are likely to remain clouded until the issue of whether a PBT is considered a search under the Fourth Amendment definition is clarified. Regulations stipulating submission to testing on a lesser standard of reasonable suspicion as a condition of receiving a commercial license could be adopted concurrently by federal and state governments as an alternative measure to encourage greater use of testing in cases of suspected DUI offense. One important consideration affecting the acceptability of various testing procedures to law enforcement officials is the perceived accuracy of the testing devices themselves, particularly at low BACs. ALCOHOL CONCENTRATIONS IN THE BLOOD AND BREATH Most enforcement agencies measure violations of DUI laws with breath- testing equipment that approximates the subject's blood alcohol concentration. As pointed out in Dubowski's 1986 review of developments in alcohol analysis, the developers of breath measurement technologies chose to express the measure of alcohol in the breath in terms equivalent to the concentration of alcohol in the blood (Dubowski 1986, 13). This decision introduced an undesirable source of variability in the obtained measure because it requires a factor to adjust the breath alcohol concentration (BrAC) to the blood alcohol concentration (BAC). Most breath-testing equipment is calibrated with a BAC-to-BrAC ratio of 2,100 to 1, but Dubowski's review suggests that the population mean is closer to 2,300 to 1 (Dubowski 1986, 29). As with any other biological measure of alcohol in the body, the ratio of blood alcohol concentration to breath alcohol concentration (BAC/BrAC) is

APPREHENDING THE IMPAIRED DRiVER 81 also subject to interindividual and intraindividual variability. As a result, "population mean BAC/BrAC ratios, no matter how carefully and correctly determined, do not necessarily apply to a given individual at a given time and under given conditions, to the degree of certainty required for some applica- tions, such as criminal law prosecutions" (Dubowski 1986, 29). The vari- ability of BAC/BrAC has proven a convenient route for seeking dismissals of DUI cases in which the defendant's measured BAC is close to the state standard. The simplest way to avoid the variability introduced by BAC/BrAC ratios is to write DUI laws in terms of breath alcohol concentrations. The Uniform Vehicle Code [Section 11-902.1(a)(5), 1968 rev. ed., Supplement IV, 198416 has adopted the following language: "Alcohol concentration shall mean either grams of alcohol per 100 milliliters of blood or grams of alcohol per 210 liters of breath" and several states have adopted this convention (Dubowski 1986, 24). TESTING TECHNIQUES AND THEIR EFFECTIVENESS The appropriateness of reducing BACs for commercial vehicle drivers de- pends in part on the ability to efficiently detect low BACs. The police use a number of techniques to determine whether the driver has been drinking and is under the influence of alcohol. In all cases the officer begins by interviewing the driver. The passive sensor can aid the officer at this stage in the process to detect the presence of alcohol. If the officer decides to proceed with testing, he will use the field sobriety tests, possibly followed by a PBT. The accuracjof these field techniques and of evidential tests is reviewed here. Descriptions of each technique and associated costs are discussed in Appendix C. The Roadside Interview Once a driver has been stopped, the officer uses a roadside interview to establish a basis for testing for intoxication. Although behavioral cues, such as slurred speech, help skilled áfficers assess intoxication, even trained officers correctly identify only about half of the drivers whose BAC is higher than 0.10 percent (Zusman and Huber 1979). Although not yet in widespread use, the passive alcohol sensor can aid the officer in detecting the presence of alcohol.

82 ZERO ALCOhOL AND OTHER OPTIONS Passive Alcohol Sensors The passive alcohol sensor (PAS) has been developed to test the ambient air in front of the driver's mouth for the presence of alcohol (Figures 4-2 and 4-3). In sobriety checkpoints police must assess the impairment level of many drivers in a short period of time, typically less than I min per driver. The PAS is viewed by many as a nonintrusive method of enhancing the senses of the police officer and not as a test. To obtain a BAC reading, the officer must place the device 6 in. from the driver's mouth. The effectiveness of the sensor is significantly reduced when the suspect is outside the vehicle in windy conditions, when the subject has recently used a breath freshener containing alcohol, or when the driver turns his head away from the instrument. FIGURE 4-2 Passive alcohol sensor built into standard police flash- light. (Photograph courtesy Robert B. Voas) In extensive tests the sensor reduced false positives from 20 percent to 10 percent. Use of the device also improved detection of the true positives, especially in testing just below the presumptive level of impairment—from 0.05 percent to 0.09 percent (Jones and Lund 1986). Although the passive sensor is less accurate than portable breath testers in detecting alcohol (and the sensitivity of the devices tends to diminish at the lowest BACs), it is more

APPREhENDING TIlE IMPAIRED DRIVER 83 FIGURE 4-3 Passive alcohol sensor used at Charlottesville, Virginia, checkpoint. (Photograph courtesy Insurance Institute for Highway Safety) accurate than the senses of individual officers. Thus the use of passive sensors holds promise as a technique for establishing reasonable suspicion. Behavioral Sobriety Tests When an officer thinks a driver may be intoxicated, he must establish probable cause before making an arrest. In most cases, he will use a field sobriety test to judge the BAC. Among many techniques used by officers in the field, a test of the involuntary jerking motion of the eyes, referred to as horizontal gaze nystagmus, is one of the best indicators of intoxication (Tharp et al. 1981). In this test an officer asks a suspect to focus his vision on a pencil held directly in front of him. When the officer moves the pencil from side to side, he observes the eyes of the suspect. The eyes of an individual under the influence of some levels of alcohol exhibit an involuntary jerking or fluttering motion as they move from side to side. However, the difference in nystagmus for a person with zero BAC and for one with 0.04 percent BAC is very small, so this test

) 84 ZERO ALCOHOL AND OTHER OPTIONS may have limited value. Used in combination with gaze nystagmus, the traditional tests of walk and turn with divided attention and the one-leg stand with divided attention (see Appendix C) are almost as effective as portable breath testers in reducing false arrests (Anderson et al. 1983). Nevertheless, even the best of the behavioral indicators are not very dependable below 0.10 percent BAC (Compton 1984, 3). Preliminary Breath Testing Portable breath-testing devices are most commonly used to screen out false positive suspects (Figure 4-4). Fuel-cell instruments are the most popular devices because of their modest cost, small size, and high level of accuracy. In laboratory tests performed by NHTSA, fuel-cell portable breath testers accu- rately measure known concentrations of alcohol in vapor at BACs equivalent to 0.10 percent and 0.05 percent to within plus or minus 0.005 percent. Use of the portable breath tester in the field, however, will lower its accuracy. Environmental factors affect test results. When the units are cold, the time required to measure the sample increases substantially, compromising accuracy. (Officers are instructed to keep them in an inside pocket to keep FIGURE 44 Portable breath tester (PBT) that provides a prelim- mary reading of BAC. (Photograph courtesy Robert B. Voas)

APPREHENDING THE IMPAIRED DRIVER 85 them warm.) Lack of cooperation from the suspect reduces the accuracy of the reading by 30 percent or more (Gibb et al. 1984; Jacobs and Goodson 1973). The equipment must also be calibrated and used properly by trained police officers. Field tests of portable breath testers indicate that readings obtained by the portable devices in the field were about 15 percent lower than data obtained from laboratory investigations (Yohman and Ebare, undated). Evidential Testing NHTSA tests commercial evidential breath alcohol devices to ensure that they meet guideline specifications. The devices measure a known concentration of alcohol vapor. The guideline is to accurately measure the alcohol content in vapor (equivalent to 0.05 and 0.10 percent BAC) to within plus or minus 0.005 percent. Measurements of breath alcohol are closely correlated with those of blood alcohol. In a Swedish study, correlations ranged from 0.96 to 0.99 between BrAC and BAC depending partly on the time difference between obtaining the two samples (Bonnichsen and Goldberg 1980, Vol. 2, 796-810). Different methods of testing blood samples showed similar correlations. The gas chromatography method was compared with a technique measuring chemical reactions (alcohol dehydrogenase-based oxidation); correlations ranged from 0.99 to 0.93 depending on whether the test was performed once or in triplicate (Dubowski 1986). These high correlations refer to group and not individual results. A BrAC for any given individual that is approximated to a BAC using a group mean for the BAC-to-BrAC ratio of 2,100 to 1 will typically underestimate the BAC by 10 percent or more. Evidential breath-testing technologies have been in use for many years and scientific protocols have been developed to ensure accurate readings. The scientific principles employed in forensic alcohol analysis have withstood numerous court cases (Dubowski, 1986, 24). Strict adherence to these princi- ples is required, however, to obtain reliable results, particularly at low BACs. Summary A number of devices, some quite new on the American scene, exist to improve an officer's abilities to screen and test drivers: passive sensors, portable breath testers, and evidential breath-testing equipment. The accuracy and use of these devices become more important as the legal BAC is lowered from current standards. To enforce a 0.10 percent BAC, which is the standard that prevails

86 ZERO ALCOHOL AND OTHER OV1'IONS in most states, most screening for driving while intoxicated relies on be- havioral cues that will not be manifested by drivers at low BACs. Passive sensors, which would only be used for screening drivers, are more sensitive to the presence of alcohol than the senses of an individual officer. Portable breath testers can provide a reasonable approximation of a suspect's BAC that is sufficiently accurate to assist in establishing probable cause. Proper use of breath-screening devices by officers is required to obtain reliable readings. The passive sensor needs to be held within 6 in. of a driver's face, and an uncooperative driver can reduce the value of the sensor by merely turning his head. The portable breath tester requires a sample of deep lung air that requires the suspected driver to exhale for about 5 sec and requires proper handling. Uncooperative drivers could be expected to reduce the accuracy of this device by as much as 30 percent or more. Evidential breath-testing equipment and portable breath-testing equipment are as accurate at 0.05 percent BAC as they are at 0.10 percent when tests are properly conducted. Thus the ability to measure BACs accurately for eviden- tial purposes does not appear to limit the selection of a lower BAC standard. Adoption of a lower BAC standard, however, may require officers to be more rigorous in following scientific standards for forensic alcohol analysis to ensure that the evidence collected holds up in court. SUMMARY The ability to enforee a more stringent BAC standard for commercial vehicle drivers will depend on the feasibility of detecting and measuring impairment at lower BACs and on the legal framework within which enforcement efforts are carried out. The legal system that provides the context for enforcement efforts is moving toward a more objective testing-based system of traffic laws. The majority of states have adopted per se laws that specify an objective BAC level above which it is illegal to drive irrespective of driving behavior. Twenty-two states have also adopted administrative per se laws that provide for swift license revocation or suspension for those drivers testing above the legal limit or refusing an evidential test subsequent to arrest. Despite adoption of these laws, however, many stages of the enforcement process, particularly those leading up to arrest, continue to be driven by behavioral standards and evidence. Lowering the BAC limit for commercial vehicle drivers will increase the difficulty of detecting impairment, particularly in these prearrest stages of the enforcement process, if conventional behavior-based enforcement methods continue to be relied upon. Greater use of enforcement strategies that are less behaviorally dependent, such as sobriety checkpoints and prearrest testing,

APPREHENDING THE IMPAIRED DRIVER 87 will be required to identify low BACs. Many states, however, have been reluctant to adopt these methods because of legal considerations, the newness of the prearrest tests, and the perceived unreliability of the testing tech- nologies themselves. Judicial interpretation of the Fourth Amendment protection of citizens against unreasonable search and seizure by law enforcement officials provides the basis for enforcing drinking-driving laws. The legality of specific enforce- ment procedures generally hinges on a delicate balancing of individual rights against legitimate government interests such as highway safety. The legality of checkpoints and the use of PBTs without first establishing probable cause have both been challenged as unwarranted searches under the Fourth Amend- ment. Only adjudication will determine whether both practices will be ruled constitutional under the balancing principle that has pruvailed in similar cases. Because some states have been reluctant to adopt these techniques until these issues are resolved, clarification by the courts of the circumstances under which these enforcement measures can be legally implemented is important to successful enforcement of a more stringent BAC standard. Testing equipment is available that, when used properly, can accurately measure low BACs. Prearrest testing devices, such as the passive sensor and the handheld breath tester, are less accurate in the field than in laboratory trials. However, in contrast to behavioral sobriety tests, which cannot accu- rately measure low BACs, these testing devices should improve the efficiency of detecting low BACs in the field. Successful adjudication of per se cases is assisted when states adopt the language recommended by the Uniform Vehicle Code that defines per se limits in terms of both blood and breath alcohol concentrations. Adoption of federal regulations requiring commercial vehicle drivers to submit to testing on a lesser standard of reasonable suspicion as a condition of receiving a commercial license may provide another way to encourage testing when a DUI offense is suspected. Because the test results will lead to a civil sanction, the lesser standard should provide sufficient grounds for conducting the testing. Passage of state legislation embodying the federal regulations, however, will be required if state and local law enforcement officials, who are to conduct the tests, are to adopt these more relaxed standards. NOTES The test, however, does not account for variability in the degree to which a driver's ability is compromised at a given BAC or in the amount of alcohol required to reach a specified BAC. This summary table developed by N1-ITSA is current as of Januazy 1, 1987. In analyzing the table it should be kept in mind that some of the entries may already be outdated because state drinking-driving legislation is constantly being amended. The nuances of each state's laws also make it difficult to provide brief but accurate summary comparisons, which is attempted in this table.

88 ZERO ALCOHOL AND OTHER OPTIONS Not all states, however, test in the event of an accident. See the Federal Register of April 17 and August 2, 1985, for final rules pertaining to alcohol and drug use for the Federal Aviation Administration and the Federal Railroad Administration, respectively. The final rule for the former became effective on June 17,1985, and for the latter, in Februaiy—March 1986. Railroad employees are disqualified from service for 9 months. The term of suspension or revocation of certificates or ratings for airline crews is left up to the enforcement authorities. In the 1987 edition, which is in press, Section 11-902.1(a)(5) will be renumbered 11-903(a)(5). REFERENCES Anderson, T. E., R. M. Schweitz, and M. B. Synder. 1983. Field Evaluation of a Behavioral Test Battery for DWI. Technical Note. NHTSA, U.S. Department of Transportation. Bonnichsen, R., and L. Goldberg. 1980. Large-Scale Breath-Blood Comparisons Under Field Conditions: Methods, Evaluation Techniques and Results. In Alcohol, Drugs and Traffic Safety (L. Goldberg, ed.), Almqvist och Wiskell International, Stock- holm, Sweden. Compton, R. P. 1984. Use of the Gaze Nystagmus Test to Screen Drivers at DWI Sobriety Checkpoints. Research Notes. NHTSA, U.S. Department of Transportation. Compton, R. P., and R. E. Engle. 1983. The Use of Safety Checkpoints for DWI Enforcement. Technical Note. NHTSA, U.S. Department of Transportation. Dubowski, K. M. 1986. Recent Developments in Alcohol Analysis. Alcohol, Drugs, and Driving: Abstracts and Reviews, Vol. 2, No. 2 March/April, pp. 13-46. Federal Register. 1985a. Vol. 50, No. 74, April 17, pp. 15376-15383. Federal Register. 1985b. Vol. 50, No. 149, Aug. 2, pp. 31508-31579. Fields, M., and A. R. Hricko. 1986. Passive Alcohol Sensors—Constitutional Jmnplica- tions. The Prosecutor, Vol. 20, No. 1, pp. 45-52. Gibb, K. A., A. S. Yee, C. C. Johnston, S. D. Martin, and R. M. Nowak. 1984. Accuracy and Usefulness of a Breath Alcohol Analyzer. Annals of Emergency Medicine, Vol. 13, pp. 516-520. Harris, D. H., J. B. Howlett, and R. G. Ridgeway. 1979. In Project Interim Report: Identfi cation of Visual Cues and Development of Detection Methods. Report HS 805-051. NHTSA, U.S. Department of Transportation. lift, R. A. 1983. Curbing the Drunk Driver Under the Fourth Amendment: The Constitutionality of Roadblock Seizures. The Georgetown Law Journal, Vol. 47, pp. 1485-1486. Jacobs, W. B., and L. H. Goodson. 1973. Evaluation of the Intoximeter Fuel Cells for Breath Alcohol Testing, Part 2: The Ako Sensor. Midwest Research Institute, Kansas City, Mo. Jones, I. S., and A. K. Lund. 1986. Detection of Alcohol-Impaired Drivers Using a Passive Alcohol Sensor. Journal of Police Science and Administration, Vol. 14, No. 2. LaFave, W. R. 1987. Search and Seizure. West Publishing Co., St. Paul, Minn., 4 vols. Latchaw, J. 1986. Do's and Don'ts on Administrative License Suspensions, and Other Issues. In Reducing Highway Crashes Through Administrative License Revocation, Report 806-921, NHTSA, U.S. Department of Transportation.

APPREHENDING THE IMPAIRED DRIVEl? 89 Macdonald, D., and M. Wagner. 1981. Report on a National Study of Prelininary Breath Test (PBT) and Illegal Per Se Laws. Report DOT-HS-806-048. NHTSA, U.S. Department of Transportation. Manak, J. P., and R. E. Engle. 1985. The Legal Aspects of the Use of Passive Alcohol Screening Devices as Law Enforcement Tools for DWI Enforcenent (unpublished manuscript). NHTSA, U.S. Department of Transportation. NHTSA. 1985. Alcohol and Highway Safety 1984: A Review of the State of the Knowledge. Report 806-569. U.S. Department of Transportation. NHTSA. 1986. Digest of State Alcohol-Highway Safety Related Legislation. 5th ed. Report 807-062. U.S. Department of Transportation. Reeder, R. H. 1974. Vehicle Traffic Law, rev. ed. The Traffic Institute, Northwestern University, Evanston, Ill. Reese, J. H. 1986. Summary Suspension of Driver Licenses of Drunken Drivers: Constitutional Dimensions. In Reducing Highway Crashes Through Administrative License Revocation, Report 806-921, NHTSA, U.S. Department of Transportation. Ross, H. L. 1987. Administrative License Revocation in New Mexico: An Evaluation. Law and Policy, Vol. 9, No. 1, January 1987, pp. 5-16. Ruschmann, P. A., K. B. Joscelyn, M. Greyson, and H. 0. Carroll. 1980. An Analysis of the Potential Legal Constraints on the Use ofAdvancedAkohol-Testing Technol- ogy. Report 805-453. Highway Safety Research Institute, University of Michigan. Ann Arbor. Tharp, V. K., M. Bums, and H. Moskowitz. 1981. Limited Field Testing of a Standard- ized Sobriety Test Battery. Proc., 25th Annual Conference, American Association for Automotive Medicine, San Francisco, California. Voas, R. B. 1982. Drinking and Driving: Scandanavian Laws, Tough Penalties and United States Alternatives. Report 806-240. NHTSA, U.S. Department of Trans- portation. Yohman, D. T., and W. Ebare. Final Report: Preliminary Breath Testing. Project- 81-108. Maryland Department of Transportation, Baltimore-Washington Interna- tional Airport (undated). Zusman, M. E., and J. D. Huber. 1979. Multiple Measures and the Validity of Re- sponse in Research on the Drinldng Driver. Journal of Safety Research, Vol. 2, No.3.

5 Preventing Alcohol-Impaired Driving Through Deterrence In this chapter the possible benefits are estimated of reforming statutes and law enforcement practices aimed at alcohol-impaired driving by commercial vehicle drivers. First the deterrence model is delineated, a conceptual frame- work that has increasingly formed criminal justice approaches to highway safety over the past two decades. Next a series of empirical findings is reviewed that deals with the capacity of law enforcement reforms to change the patterns of alcohol-impaired driv- ing. Most of these results derive from the study of changes in crush statistics during "natural experiments," that is, fairly abrupt, well-defined changes in the handling of the problem of alcohol-impaired driving by one or more agencies of government in a particular jurisdiction. The changes may involve statutory proscriptions and penalties, police behavior, or judicial, pros- ecutorial, or administrative licensing practices. The bulk of the well-founded empirical knowledge about the effects of law enforcement on alcohol-impaired driving is from studies in which commercial vehicles and their drivers have either been excluded from coverage or com- pose only a small and undifferentiated fraction of the vehicles and drivers studied. To extrapolate such results, dominated by drivers of personal vehi- cles, onto the case of drivers of buses and trucks requires careful analysis of the differences between the two, namely, differences in patterns of alcohol- crash involvement and in how safety laws concerning personal versus com- mercial vehicles are written and enforced. In light of these considerations, scenarios are developed to estimate the likely benefits, in terms of averted deaths, injuries, and property damage, of uniform nationwide enforcement of alternative BAC standards for commer- cial vehicle drivers. These scenarios include extrapolations about the possible 90

PREVENTING ALCOHOL-IMPAIRED DRIVING 91 impact of policies that differ from those for which the evidence of effective- ness is clearest—such as moving to more severe punishment of much lower BACs or using less intensive enforcement procedures. THE DETERRENCE MODEL The idea of using police power to prevent criminal acts appears fairly simple and even obvious. As a matter of theory, anyone can draw from common sense the prediction that a threat of punishment will reduce the likelihood of untoward behavior (e.g., "spare the rod and spoil the child"). As a matter of practice, it is also common knowledge that every jurisdiction in the country uses police power, courts, and jails to enforce the law. However, some highly plausible complications can be readily introduced into this picture. Common sense dictates that a child cannot be raised on a diet of punishment alone; the desire to please others, gain respect, and behave morally are highly important in sustaining law-abiding attitudes and actions among the majority of adults. Practically speaking, the police cannot (and indeed should not) be everywhere at once. Moreover, before U.S. courts will order punishment, they are bound to have either an admission of guilt or admissible evidence of guilt beyond a reasonable doubt. Finally, the jails are now already full to overflowing. In short, there is good reason to think that deterring crime is not so simple a matter after all. The criminological principles of deterrence are parallel to those of common sense knowledge, but they add a critical degree of conceptual specificity and empirical depth (Zimring and Hawkins 1973; Gibbs 1975; Blumstein et at. 1978; Tittle 1980). The fundamental proposition of the classical theory is that "proscribed behavior is deterred by perceptions that legal punishments are swift, sure, and severe" (Ross and LaFree 1986). Upon dissection, there are several parts to this proposition1: Proscribed behavior: The behavior needs to be definable, understand- able, and detectable, not only by the individuals to be deterred but also by those who are required or expected to respond with sanctions (punishments). Perceptions: The behavior of an individual at liberty is not controlled directly or immediately by sanctions and their objective pattern of delivery but by the individual's subjective impression and evaluation of the sanctions and pattern of delivery. Swift, sure, and severe punishment: Sanctions as delivered may vary along the independent dimensions of whether they are prompt or delayed, reliable or erratic, heavy or light the magnitude of the deterrent effect is a joint function of these variations.

92 ZERO ALCOHOL AND OTHER OPTIONS Researchers distinguish specific deterrence from general deterrence. The first of these refers only to individuals who have been apprehended while commit- ting violations and have received punishment. The specific deterrent effect is the resulting attributable change (if any) in the rate at which these individuals repeat the proscribed behavior (which is sometimes called the "recidivism rate"). The general deterrent effect is the extent to which the overall violation rate is affected by the existence of legal proscriptions and violations. In practice, the components of the deterrence proposition become elabo- rated through the specific operations of legislation, enforcement, and ad- judication in the context of existing perceptions, norms, and practices. An announcement by appropriate authorities that they (and their agents) are henceforth prepared to dispense penalties more promptly, reliably, and heavily for a certain behavior may, if widely broadcast, impress a substantial propor- tion of occasional and frequent violators into promptly cutting down on the amount of this behavior. This reduction is due partly to individual fear of punishment, which is the deterrent element proper; partly to the increased public attention and informal discussion focused on the behavior, which convinces some potential violators that other people really do notice it and consider it wrong; and partly to the expansion of informal practices that can make the behavior easier to avoid (such as having a "designated driver"). In the period of time following such an announcement, evidence begins to accumulate about how fully and efficiently the authorities are delivering on their promises. This evidence, through personal observation, word of mouth, and media reports, may serve either to attenuate or to reinfoite the initial impression that things are changing. At the same time, informal talk and practices may move favorably toward or away from the behavior. These complications are sufficiently numerous and variable that theory alone cannot predict the net impact and time course of a specific reform. But enough evidence has accumulated on reforms in legislation, enforcement, and ad- judication concerning alcohol-impaired driving to provide some valuable guidance for responding to the questions at hand. EVALUATIONS OF DETERRENCE POLICIES A substantial proportion of the research literature on deterrence of crime has focused on alcohol-impaired driving, both in the United States and abroad. The basic propositions of the deterrence model have been refined considerably over the past 20 years as increasingly sophisticated data collection and analysis have been completed. The most revealing results derive from assess- ing the impact of incremental changes in legal statutes or enforcement be- havior or both, changes explicitly designed to increase the severity, certainty, or swiftness of response to violations.

PREVENTING ALCOHOL-IMPAIRED DRIVING 93 The most common and powerful method of analysis is to closely follow what happens when a particular shift in policies or procedures occurs, compar- ing it with baseline levels, control sites, or internal statistical controls, or using all three. Researchers typically employ a form of interrupted time-series analysis (e.g., Box and Tiao 1975), which supplements before-and-after mea- sures of the incidence of alcohol-impaired driving or its consequences with statistical techniques designed to assimilate innovations in policies or pro- cedures to the model of treatments or conditions introduced during a con- trolled experiment. Deterrent effects are usually measured through a time series of crash data, and special attention is typically paid to fatal crashes and severe injury crashes, especially those on weekend nights, because alcohol impairment is known to play its largest role in such cases. Crash data are usually statistically adjusted or proportioned to take into account. parallel daytime crash rates, regional or national trends, aggregate miles driven, or seasonal variations. There are three principal types or aspects of such innovations: "Scandinavian-type" laws, under which driving with a higher BAC than that specified is deemed sufficient in itself (per se) to warrant sanctions, usually at the level of arrest, misdemeanor fines, temporary suspension of driver's license, and brief incarceration. The first such law was passed in 1936 in Norway; the BAC limit there was set at 0.05 percent. Judicial "crackdowns" in municipal courtrooms, intended to tighten the slack between the full severity of sanctions permitted by law and the typically much lighter sentences given to violators. Police "blitz" tactics, undertaken by a local jurisdiction to increase the probability that an episode of alcohol-impaired driving will be detected and cited. These shifts in statutory provisions, judicial sentencing, and police behavior have in some cases been undertaken simultaneously and in other cases separately. Although the possible variations are great, the actual episodes observed have been closely related, and a general line of results has emerged with substantial consistency in the United Kingdom, France, Australia, and the United States. United Kingdom, October 1967 The pioneer example of a well-studied Scandinavian-type reform is the British Road Safety Act of October 1967, analyzed by Ross (1973, 1984). This act defined driving with a BAC of 0.08 percent or higher to be an offense, with

20 16 12 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 Year FIGURE 5-1 U.K. fatality rate corrected for month and with seasonal variations removed. (Ross 1973, 33. Copyright 1973 by The University of Chicago. Reprinted with permission.)

PREVFRIJNG ALCOHOL-IMPAIRED DRIVING 95 the BAC to be determined by an Alcotest breath-testing device of German manufacture, 1 million of which were purchased (with considerable fanfare) by the British government. Police were authorized to have drivers submit to the test in the event of any objective reason to believe that the driver was intoxicated, such as a crash, a moving violation, or erratic driving. If the driver refused the test, the officer could require a formal test. Refusal of the driver to cooperate could be punished as if the driver had failed the test. Judges theoretically had no discretion in sentencing; the first offense was to result in a mandatory 1-year suspension of the driver's license. The Road Safety Act had a dramatic impact on drivers in the United Kingdom (Figure 5-1). In the 3 months after it took effect, overall highway fatalities dropped a remarkable 23 percent and injuries 11 percent. By way of an American benchmark, in 1974, the year following the Arab oil embargo and institution of the national 55-mph speed limit, overall highway fatalities in the United States declined by 16 percent. In the first year of the Road Safety Act, the proportion of fatally injured drivers who were intoxicated dropped nearly 40 percent (see Table 5-1). In the United States, the comparable national statistic (fatally injured drivers with BACs > 0.10 percent) has dropped by about 17 percent since 1980, roughly the period since major public concern and related initiatives have been steadily focusing on drinking and driving, including increases in the minimum drinking age, further spread of per se provisions, mandatory minimum sentencing of drunk driving, and substantially greater publicity and preventive education. (The overall number of U.S. highway fatalities was approximately 13 percent lower in 1985 than in 1980; this reduction was mostly accounted for by 3,500 fewer fatally injured drivers annually with BACs higher than 0.10 percent.) TABLE 5-1 FATALLY INJURED DRIVERS IN THE UNITED KINGDOM WHO WERE LEGALLY INTOXICATED (Sabey 1978, 192) Fatally Injured I Fatally Injured Year Drivers (%) Year Drivers (%) 1967 (to Sept.) 32 1972 30 1968 20 1973 33 1969 25 1974 36 1970 23 1975 38 1971 27 1976 38 NOTE: BAC higher than 0.08 percent constitutes legal intoxication. The general trends mask several specific changes in British drinking prac- tices. Research indicates that the act did not significantly change the amount that people in the United Kingdom drank. Rather, it appears to have affected a rather narrow slice of behavior—the custom of driving to and from pubs,

1,800 1,600 1,400 1,200 >. 1,000 800 U. 600 400 200 .JFMAMJJASONDJFMAMJJASONOJFMAMJJASONDJFMAMJJASONDJFMAM,J.JASONO 1966 1967 1968 1969 1970 Year FIGURE 5-2 Fatalities and serious injuries in the United Kingdom combined for Friday nights (10:00 p.m. to midnight), Saturday mornings (midnight to 4:00 a.m.), Saturday nights (10:00 p.m. to midnight), and Sunday mornings (midnight to 4:00 a.m.), corrected for weekend days per month and with seasonal variations removed. (Ross 1973, 33. Copyright 1973 by The University of Chicago. Reprinted with permission.)

PREVEWFING ALCOHOL-IMPAIRED DRIVING 97 especially on weekend nights (Figure 5-2). After the act took effect, many regular customers began walking to pubs. A number of less conveniently located pubs closed, and pub owners raised a considerable outcry (Ross 1973). Unfortunately, the successes of the act were relatively short-lived. 'Within a few years, traffic fatalities again began to climb. By 1973 the percentage of drivers killed who were intoxicated was back to its pre-Road Safety Act level. By 1975, for reasons still unknown, this percentage had risen considerably higher than it had been before the act (see Table 5-1). The explanation for this reversion originally developed by Ross was that drivers eventually realized that their chances of arrest and punishment for drinking and driving were still not very high, though they had perhaps increased. Ross estimated that even after the act, the chance of being breath- tested was once in 2 million mi driven (about 10 driver lifetimes). Much of the initial effectiveness appears to have come from the breath-testing device itself, which had never been used in the United Kingdom before. The Alcotest was expected to revolutionize the workings of the court on drinking-driving cases. A nearly automatic scientific mechanism would replace the old system of patrols and trials. In fact, the Alcotest had nosuch effect. Well-publicized court cases soon established much narrower limits to its evidentiary authority than the original publicity had suggested. Standards for its use took several years to develop, and British police used it much less frequently than did police in other countries, administering in 1970 about one test annually for every 500 licensed drivers. Moreover, of all the tests administered, about half yielded BACs of 0.08 percent or above, indicating great selectivity on the part of most police officers in testing only when there were quite overt signs of intoxication. Ross concluded that the respect for and fear of the Alcotest among the driving population at large declined, and therefore so did the effectiveness of the act. France, July 1978 A second example of a Scandinavian-type reform was a French law of July 1978. France had previously instituted legislation stipulating a 0.08 percent BAC limit with license suspension as the penalty. In addition, the offense structure had a second tier, providing higher penalties for BACs of 0.15 percent or above. The principal innovation of the new law was to provide for breath testing of all drivers passing through roadside checkpoints, which were deployed solely for this purpose. The 1978 law had a significant prompt effect, reducing overall traffic fatalities by about 14 percent in the first few months (Figure 5-3). However, as in the United Kingdom, the deterrent effect receded, in this case even more rapidly, with crash fatalities returning to the earlier level within a year. Ross (1984) attributes this instance as well to

1,800 1,600 1,400 0 1,200 E z 1,000 800 600 1974 1975 1976 1977 1978 1979 1980 Year FIGURE 5-3 Crash-related deaths in France with seasonal variations removed. (Ross et al. 1982. Reprinted with permission,)

PREVEITTING ALCOHOL-IMPAIRED DRIVING 99 "official diffidence in enforcing the legislation." About 335,000 tests were given nationwide in 6 months, which would be roughly one for every 100 licensed drivers on an annual basis. But only about four-tenths of 1 percent were recorded by the police as positive, which is less than one-fourth the proportion of tests recorded as positive in independent roadside research surveys of driver BAC levels conducted in France at the time. Australia, December 1982 A third and particularly telling case is that of random breath testing (RBT) in the Australian state of New South Wales. The new law, introduced in Decem- ber 1982, was a radical departure from previous law enforcement practices in New South Wales, because under RBT police were permitted (indeed, re- quired) to carry Out preliminary breath tests on all motorists stopped at random at DUI chekpoints regardless of whether those motorists exhibited any sign of alcohol impairment or had committed any observed offense (Homel 1986). Under RBT the number of breath tests administered increased virtually overnight by about a factor of 8. Penalties in the form of fines and license disqualification, as well as a per se limit of 0.05 percent BAC, were already in place in New South Wales. The penalty structure introduced in December 1982 had three tiers, depending on the BAC recorded and on previous DUT convictions. Jail sentences occurred rarely (in only 2 percent of arrests). The results of RBT included a substantial decline in overall highway crash fatalities (Figures 5-4 and 5-5). In the 32 months subsequent to passage of RBT, monthly fatalities were lower by an average of 23 percent compared with the 6 previous years, during which the overall fatality trend had been quite stationary. Although there was also a general shift to a lower number of fatalities in the other large Australian states—in one of which, Queensland, a coincident lowering of the BAC limit took place in December 1982—the shift in New South Wales is substantially greater than elsewhere. The fact that the deterrent effect of RBT did not recede in the near term in New South Wales as it had in France or as the Road Safety Act effect had in the United Kingdom is ascribed by Homel partly to the substantial expendi- ture of money to publicize RBT (although this had occurred to a lesser extent initially in the United Kingdom and France) but more substantially to the sheer scope of law enforcement efforts. In the first 12 months of RBT, more than 923,000 tests were conducted, roughly one for every three licensed drivers in New South Wales.

New South Wales 121 RBTINNSW 111 101 61 51 41 t I I I I I I I I I I 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 Year FIGURE 5-4 Fatal crashes for New South Wales for each month from January 1971 to July 1985. (Homel 1986. Reprinted with pemiission.)

PREVENTING ALCOHOL-IMPAIRED DRIVING 101 United States Since 1970 Several reforms in dealing with alcohol-impaired driving in the United States have produced results that may be compared with those of the international cases. In the early 1970s (roughly 1970 to 1976) the National Highway Traffic Safety Administration (NHTSA) funded 35 locally organized and managed Alcohol Safety Action Projects in various parts of the country. Each project sought in its own way to combine an increased risk of DUL arrest, more effective trial and rehabilitation procedures, and public education to reduce the number of accidents caused by alcohol-impaired driving. By increasing surveillance, targeting DUT patrols for specific times and places, and motivat- ing police to bring in more DUI violators, many of the jurisdictions involved were able to double and triple the previous number of DUT arrests. The individual studies that attempted to evaluate these local projects suffered from serious methodological flaws, including noncomparable sites, inadequate con- trols, premature expansion of the program, and the sudden appearance of the oil shortage and national 55-mph speed limit in the picture. But in their final reports (NHTSA 1979a, 1979b, 1979c) the projects' national evaluators calcu- lated that 12 of the 35 programs had produced discernible favorable effects on nighttime automobile fatalities, a typical indicator of alcohol-impaired driv- ing. Statistical analyses suggested that these 12 projects had reduced overall fatalities in their localities by an average of about 12 percent across project durations of 2 to 5 years. Roadside breath test surveys completed by re- searchers in 9 of the 12 projects indicated an average reduction of about 16 percent in the proportion of drivers at 0.10 percent BAC or above. Indepen- dent researchers, however, have argued that the attributable positive effects need to be treated cautiously because of the inefficiency of the project design (Cameron 1979; Reed 1981, 336-387). In addition to these large-scale national studies, a substantial number of smaller-scale shifts in enforcement have been evaluated. Rather than changes in the statutory basis, these programs have generally involved either judicial initiatives to deliver more severe punishments in the case of convictions than had been the practice earlier or aggressively publicized campaigns by local authorities (sometimes called police blitzes) to increase the chance of arrest. Judicial initiatives by themselves appear to have had no measurable effects on indexes of alcohol-impaired driving (Ross 1984). More generally, studies of increases strictly in the severity of formal punishment have led the majority of researchers to view these as ineffective and in some instances—when the police or courts view them as overly Draconian—counterproductive. As Wilson (1975, 170) has observed: The more severe the penalty, the more unlikely that it will be imposed . Prosecutors and judges will tzy to get a guilty plea, and all they can offer in

90.00 82.00 , 74.00 (5 66.00 C) ' 58.00 (5 LI. ' 50.00 0 z 42.00 34.00 26.00 71 73 75 77 79 81 83 85 87 Year VIrTflIA 66.00 60.00 54.00 48.00 42.00 (5 LI. ' 36.00 0 z 30.00 24.00 18.00 71 73 75 77 79 81 83 Year FIGURE 5-5 Fatal crashes for Victoria, Queensland, South Australia, and the whole of Australia excluding New South 85 87

SOUTH AUSTRALIA 43.00 39.00 35.00 31.00 C.) i 27.00 LL 23.00 z 19.00 15.00 11.00 7.00 71 73 75 77 79 81 83 85 87 Year Al lTPAI IA twI pcw 214.0 200.0 186.0 U, 5 172.0 158.0 0 c 144.0 z 130.0 116.0 102.0 71 73 75 77 79 81 83 85 87 Year Wales for each month from January 1971 to 1985. (Homel 1986. Reprinted with permission.)

60 55 0 50 0 U 0 0 45 z 40 35 104 ZERO ALCOHOL AND OTHER OPTIONS return is a lesser sentence. The more severe the sentence, the greater the bargaining power of the accused, and the greater the likelihood that he will be charged with a lesser offense. Extremely long mandatory minimum sentences do not always strengthen the hands of society; they often strengthen the hand of the criminal instead. Early research on police blitzes typically yielded a familiar pattern of transient, short-term results—prompt reductions in measures of alcohol-im- paired driving that quickly returned to baseline rates (a famous instance is shown in Figure 5-6). However, these results were seen in cases where the blitz itself was of short duration, leaving open the question of what a sustained pattern of increased enforcement would yield. The most clearcut studies of extended shifts in police behavior were in two American cities—Stockton, California (Voas and Hause 1987), and Charlottesville, Virginia (Voas et al. 1985). In Stockton, a central California city with a population of about 120,000, a special federally funded enforcement program was implemented, with brief AMJJASONDJFMAMJJASONDJFM 1974 1975 1976 Year FIGURE 5-6 Total crashes during drinking hours (10:00 p.m. to 4:00 a.m.) in Cheshire, England. (Ross and LaFree 1986. Reprinted with permission.)

PREVEJ'T1'JNG ALCOHOL-IMPAIRED DRIVING 105 planned intermissions, from January 1976 to June 1979. California laws did not provide for per se BAC violations or mandatory prearrest breath tests, and roadside checkpoints were not used. Instead, a substantial number of DUI patrols were added to existing police shifts from 8:00 p.m. to 4:00 a.m. on Friday and Saturday nights. These additional patrols had the effect of increasing DUI arrests during these periods by factors of 7 to 10, which tripled the total rate of DUI arrests in Stockton relative to the baseline periods. By the final period of the intensive patrol project, separate roadside surveys showed a 43 percent reduction in the incidence of drivers exceeding 0.10 percent BAC during the patrol hours (see Table 5-2). There was a reduction of 15 percent in all traffic accidents recorded during weekend nights. There was also a reduction of 10 percent in week-night accidents. In the same year, daytime accidents (which seldom involved DUI) increased 12 percent. Com- parisons with statewide trends and trends in several similar California cities showed that the reductions in nighttime accidents were not matched outside Stockton. The numbers of fatal and serious-injury accidents in Stockton were too few to generate statistically reliable conclusions, but the general shift in nighttime accidents attributable to the Friday and Saturday patrols—equiv- alent to a reduction of about 4 percent in the overall crash rate—was clear and significant. TABLE 5-2 BREATH TEST SURVEYS CONDUCTED IN STOCKTON, CALIFORNIA Percentage of Drivers at or Above Mean BAC 0.10 percent BAC Baseline (Oct—Dec. 1975) 0.032 8.8 Experiment 1 (Oct—Dec. 1976) 0.027a 6.3a Experiment 2 (Oct—Dec. 1977) 0.029 59a Experiment 3 (Oct.—Dec. 1978) 0.026a 5a a Change from baseline level statistically significant at the P = .01 level (chi-square test). The Charlottesville project is the most intensive of the checkpoint opera- tions that have been used in the 21 states that have initiated such operations. In 1984, the police in Charlottesville, a regional-crossroads city of 40,000 with a major public university (16,000 full-time students), set up a checkpoint from about 11:30 p.m to 4:00 a.m. nearly every Friday and Saturday night, at which

106. ZERO ALCOhOL AND OTHER OPTIONS officers inspected drivers' licenses and interviewed each driver for 20 to 60 sec, looking for signs of intoxication. About 24,000 vehicles were stopped in one year. About 300 DUI arrests were made (for a BAC of 0.10 percent or above, which is presumptive of DUI in Virginia), and 400 warnings were issued (for a positive BAC below 0.10 percent), a BAC detection rate of about 3 percent. One in six drivers in Charlottesville surveyed at the end of the year reported having been interviewed at a checkpoinL (About one-half of the arrestees lived in Charlottesville; the others were from the surrounding county and more distant areas.) There was a 12 percent reduction in nighttime accidents and a 15 percent reduction in all police-reported alcohol-related accidents in Charlottesville in 1984. But statistical analysis did not permit a conclusion that these differences were significant once concurrent statewide trends and wide seasonal varia- tions were considered. Late in the experimental period, passive sensors were introduced and used to screen all drivers passing through the checkpoints. This resulted in a doubling of the previous arrest rate per driver interviewed at the checkpoints and a trebling of the warning rate. LESSONS OF DETERRENCE RESEARCH Research on the effectiveness of laws against alcohol-impaired driving has occurred in the context of presumptive and per se BAC standards set mostly at 0.05, 0.08, and 0.10 percent in various jurisdictions. There is a general confirmation that the combination of statutes recognizing different BACs and vigorous, well-publicized, and politically popular enforcement efforts can have prompt effects in reducing fatalities and other crash statistics, but unless the enforcement effort is sustained at a high level of intensity, there is often a rebound after a few months or years. The most vigorous enforcement efforts have included routine prearrest breath testing, graduated tiers of punishment severity based on the BAC detected, and checkpoints on a fairly massive scale—for example, the equivalent of the New South Wales program in the United States would be 50 million total annual roadside stops strictly for the purpose of BAC testing.2 Although there are some initial differences across cases in terms of how large a role alcohol impairment plays in vehicle fatalities and the larger crash picture, the best results achieved to date in deterrence-based reforms to prevent alcohol-impaired driving in the general vehicle population have been fairly consistent. It has proven possible to reduce over-limit BACs by about 40 percent and fatalities by as much as 23 percent (the initial result in the United Kingdom and the sustained result in New South Wales) in the period follow- ing a well-publicized reform. In fewer reports and with less exactness, re- searchers have reported reductions of about 11 percent in total highway crash

PREVENTING ALCOHOL-IMPAIRED DRIVING 107 TABLE 5-3 MAJOR CASE EVIDENCE ON SUCCESSFUL DETERRENCE EFFECTS Location BAC Level Enforced (%) Annual Surveillance Intensity Effects United Kingdom 0.08 per se 1/500 Prompt only: fatalities, 23 percent; injuries, II percent; illegal BAC among fatally injured drivers, 40 percent France 0.08 and 1/100 Prompt only: 0.15 per se fatalities, 14 percent New South Wales 0.05, 0.08, and 1/3 Sustained: 0.15 per se fatalities, 23 percent United States Twelve of 35 Varied, mostly Unknown, arrests Fatalities, 12 percent; ASAP 0.10 doubled to positive BAC in programs presumptive tripled roadside surveys, 16 percent Stockton, Calif. 0.10 presumptive Tripling of arrests Nighttime crashes, 15 percent; all crashes (est.), 4 percent; roadside BAC survey above 0.10, 35 percent Charlottesville, 0.0, 0.05, and 1/6 Nighttime crashes, 15 Va. 0.10 percent presumptive NOTE: ASAP = Alcohol Safety Action Project (NHTSA) injuries and 15 percent in nighttime crashes, equivalent to about 4 percent in overall police-reported highway crashes (see Table 5-3). It should be noted that the deterrent effect of intensive enforcement is not confined to reducing the incidence of BACs (or accidents involving BACs) that are above the statutory limit. The deterrent effect diffuses to drinkers who drive at all BACs, leading to an overall reduction in both the average BAC detected and the percentage of positive BACs recorded at checkpoints or patrol stops. The effect almost certainly extends somewhat beyond DUI deterrence as such, because any increased level of enforcement of traffic safety, especially if achieved through increased police time on the highways, tends to make drivers somewhat more careful of their behavior in other respects. Specific reforms have varied, but the common element appears to be creating a public impression that the authorities are newly committed to reducing the problem and mean to bring a much greater weight of law enforcement to bear in a circumstance where there has earlier been an

108 ZERO ALCOHOL AND OTHER OPTIONS assumption of relative tolerance or inefficiency in detecting and sanctioning violations. So long as the reform has the appearance of a change in vigor that has less (if at all) to do with changing the scale of permitted or required penalties than with increasing efforts to put available sanctions into more certain effect—and so long as this appearance is widely communicated— prompt results may be expected. To sustain these results, however, requires a substantial, continuing commitment of enforcement resources over time. The most widely used of the deterrence-producing police procedures on an inter- national basis is the DUI (or sobriety) checkpoint However, intensive patrol strategies with officers trained to detect characteristic signs of impairment among drivers in traffic have also been used with some indications of success. In the only direct comparison of efficiency, checkpoints in Charlottesville (without passive sensors) were about 20 percent more productive of arrests per hour of police time than patrols. However, the efficiency of these kinds of enforcement measures in generat- ing arrests for DUI and in producing deterrence may be lessened for medium and heavy trucks and buses. The advantage of sobriety checkpoints and specialized DUI patrols for policing automobiles, light trucks, and motorey- des is that alcohol-impaired driving of these vehicles is relatively concen- irated in the weekend and nighttime hours and in particular corridors associ- ated with access to public drinking places. In the case of commercial vehicles, the DUI problem is not only much less in extent but is spread out more evenly across the week and the miles of highways. Moreover, and perhaps more seriously for checkpoints, the nearly universal use of CB radios by truck drivers makes the stationary checkpoint a more easily circumvented measure and may force checkpoints to be moved more often or employed in multiples. This would reduce their efficiency on a per-police-hour basis. SCENARIOS OF REFORM The theory and findings reviewed provide a good indication of the specific conditions under which a reform in DUI deterrence has been achieved and sustained in a general driver population, including the magnitude of shift that can be expected under these conditions. Such a reform may be characterized as follows: It should be sufficiently dramatic to generate extensive publicity focusing on DUI enforcement. It should substantially increase the incidence of face-to-face contact between police and drivers, during which an inspection for signs of DUI may take place; if checkpoints are used, this incidence should be at the level of at least one such contact per year for every three to six licensed drivers.

PREVENTING ALCOHOL-IMPAIRED DRIVING 109 It should increase the rate of DUI detection and sanctioning—ordinarily arrest or some other form of detention—by a factor of 3 or more. The routine use of prearrest BAC sensing technology substantially improves the efficiency of testing. The Commercial Motor Vehicle Safety Act provides some mixed pos- sibilities for meeting these conditions. In the first instance, the regulatory process set in motion by the act ensures that a new standard will be promul- gated in connection with the national commercial vehicle operator's license no later than October 1988. Even if the BAC standard were not dramatically revised—for example, if it were set at the level of 0.10 percent BAC, which is typical of most state laws at present—the relatively severe administrative penalties that are mandated (a one-year license revocation for a first offense and lifetime revocation for a second offense) would attract attention, par- ticularly in communications addressed to, and discussions among, commercial vehicle drivers. However, the implementation of the BAC standard will be on a state-by-state basis, and delays can be expected because of the need for legislation to be passed, for compliance with federal rules to be assessed, and for state enforcement and license authorities to develop and acquire the needed capabilities. Moreover, federal sanctions for state noncompliance— withholding of federal highway trust funds—will not become applicable until 1993. In short, the reform will be effected haltingly rather than in one dramatic nationwide change. This condition may to some extent mitigate the potential effectiveness of the policies; it will also complicate the research problem of evaluating them. The second condition, increasing the incidence of contact between enforce- ment agents and commercial vehicle drivers to the requisite level of sur- veillance, is a more difficult matter. The present incidence of such contacts cannot be estimated, but mechanisms can be identified that could be used to ensure that a sufficient level of contact would occur. The most important possibility is the addition of Dill screening to current and prospective pro- grams of truck weight enforcement and motor carrier safety inspection, as detailed in Chapter 6. If DUI observation were routinely instituted as a component of citation writing for truck weight violations as well as becoming a standard part of the safety inspection process, this would likely yield close to 2 million contacts with commercial vehicle drivers annually. Because the number of commercial vehicle operators expected to hold the new national license is variously estimated between 3 and 6 million, this would reach or exceed the highest intensity of surveillance recorded in successful deterrence experience. A second form of opportunity for contact is provided by police investiga- tions at the scene of crashes. Dill screening could be made a standard part of

110 ZERO ALCOHOL AND OTHER OPTIONS such investigations when any driver operating with a license covered in the Commercial Motor Vehicle Safety Act was involved in a crash resulting in a fatality or injury (injury is defined here as one in which the victim was transported from the scene for medical attention). This requirement would generate about 70,000 additional DUI screenings annually, far fewer than would occur with weight and safety inspection. It could be expected to generate some prompt deterrence effects, but the level of surveillance (1 contact annually per 40 to 80 drivers) may not be high enough to sustain those effects over time. The final point is to increase the rate of actual detection and sanctioning for DUI. The number of commercial vehicle operators now arrested for DUI is not known, so it is difficult to develop separate estimates or strategies perti- nent to this point. It is possible that increases in the rate of contact will be sufficient by themselves to generate the requisite rates of detection and sanctioning of impaired BACs. However, evidence from the Charlottesville operation and from experiments reviewed in Chapter 4 suggests that observa- tion alone may not be adequate and that wider use of passive sensing and portable breath-testing technology may be necessary to meet the requisite of more frequent detection and sanctioning. There is a certain unique difficulty in implementation of the crash-inves- tigation screening mechanism. Most bus-involved and truck-involved crashes lead to property damage only and involve two or more vehicles; in a represen- tative compilation (Clarke et al. 1987, 157), fault has been found to be assessed about half the time to the driver of the other (noncommercial) vehicle or vehicles. There might well be some reluctance on the part of investigating officers to rigorously screen the commercial vehicle operators in such in- stances (or to act on very low BAC results) if there were the prospect of sanctions that threatened the commercial vehicle driver's livelihood, because the at-fault drivers could not be held to comparable requirements. Two aspects of enforcement that are of particular interest to this study but are not viewed as salient points in the literature on deterrence are the specific BAC to which sanctions are attached and the specific nature of the sanctions delivered. This is not to suggest that such things do not matter at all. There are, rather, too small a number of cases available to permit positive differentiation between the effects of relatively small variations in BACs and penalty struc- ture observed relative to the large variations recorded in the intensity of surveillance. In other words, there is no firm empirical basis on which to quantify incremental differences in deterrence effects within the BAC ranges and penalties studied: 0.05 to 0.10 percent BACs lead to arrest, misdemeanor fines, and temporary administrative license suspensions in most cases. It is a matter of speculative judgment to say how the observations based on these BAC limits and penalties might be extrapolated to more stringent possibilities such as zero to 0.04 percent BAC and lifetime suspension of license.

PREVEWTING ALCOHOL-IMPAIRED DRIVING 111 One pertinent finding is that actual police behavior and sanctioning tend to vary with the observed BAC, not only within the provisions of statutory codes but also in the discretionary attitude of enforcement and justice agencies. For example, in New South Wales, a first violation at 0.05 percent BAC was penalized less strongly by statute—which permitted fines but not revocation— than a violation at 0.08 percent BAC. Such explicitly graded tiers of sanctions rising with the BAC are in fact rather common statutory provisions outside the United States. Even though such graduated penalties are not often mandated by statute, an informal system of sanctioning by BAC is also prevalent in the United States. In the Charlottesville checkpoint operation, for example, al- though arrests were made only at 0.10 percent BAC (which was presumptive of DUT but not in violation per se), drivers with positive BAC results greater than 0.05 percent but less than 0.10 percent were required to find other means of transportation from the checkpoint site or to obtain another driver for their vehicle; in other words, these drivers were put Out of service. Moreover, the checkpoint officers issued written warnings to drivers for any positive alcohol level up to 0.10 percent. Prosecuting attorneys in many jurisdictions routinely handle cases involv- ing BAC results that are well in excess of the statutory per se or presumptive limit in a different way than cases involving BAC results at or just above the limit, which are often not brought to prosecution, dismissed in trial, or plea- bargained to substantially lesser charges or penalties than prosecutors are willing to strike at higher BACs. Even mandatory minimum sentences may not be handed down or served, because of various kinds of slippage, though they will be served in a majority of convictions (Ross and Foley 1987). These kinds of basically discretionary aspects of the law enforcement and adjudica- tion processes blur the distinctions between the effects of different per se BAC statutes and penalty structures. BENEFITS OF DETERRENCE SCENARIOS Truck and bus crashes are different in a number of important dimensions from other crashes, and these differences necessarily lead to substantial revisions in the estimated expectation of benefit from DUI reforms in affecting truck and bus crashes. For example, fatal accidents that involve medium and heavy trucks involve collisions with other moving vehicles more often than fatal crashes not involving trucks (70 percent versus 50 percent), less often occur on weekends (16 percent versus 37 percent) and during night hours (39 percent versus 55 percent), and, most significantly, far less often yield positive BACs among the drivers involved (14 percent of heavy-truck drivers and 24 percent of medium-truck drivers versus 45 percent of other drivers). There are very few cases of positive BAC among bus drivers involved in fatal crashes.

112 ZERO ALCOHOL AND OTHER OPTIONS From 1983 through 1985, there were estimated to be about 441,500 report- able crashes annually involving medium and heavy trucks and buses (see Table 2-11). These included about 135,000 commercial vehicles involved in injury-producing crashes, of which 5,434 were recorded in crashes leading to one or more fatalities. (Because of multiple-fatality crashes, the total number of deaths is 10 percent greater than the number of crash involvements.) In the case of fatal crashes, where the data are most complete, the incidence of any BAC among the drivers of commercial vehicles in these crashes is substan- tially below that recorded in the overall population of drivers involved in fatal crashes. It is estimated to be about 30 percent as often in the case of heavy trucks and 52 percent as often in the case of medium trucks; the incidence of any BAC in bus drivers is so small as to be equated to zero for the purpose of this discussion. Compilations of driver responsibility assessment indicate that in roughly half the fatal crashes in which trucks are involved, the commercial vehicle driver is deemed at fault by the investigation. These findings can be used to adjust the loss reductions that have been observed previously with state-of-the-art reforms to increase the deterrence of alcohol-impaired driving in general vehicle populations. The rough estimates of preventable accidents that might result from such reforms applied to commercial vehicle operators are shown in Table 5-4. (The apparent precision of the estimates in Table 5-4 should not be misconstrued. Unrounded figures are presented merely to avoid rounding errors in this table and subsequent tables based on it. Each amount in the final column of the table should be interpreted as the midpoint of a range that is narrowest in the estimates of fatalities and widest in the estimates of property damage involvements.) The deterrence effect on drivers of medium trucks is unlikely to be as large as it would be for drivers of heavy trucks. The enforcement strategy developed in Chapter 6 is based on checking drivers at weigh stations and as part of vehicle inspections. This strategy would thus take place on major intercity highways where most heavy-truck travel occurs. Medium trucks, however, are also extensively used within cities, where they are less likely to be caught in the enforcement net. Lacking an empirical basis for estimating the extent to which the deterrence effect would be less for medium-truck drivers, the effect is assumed to be half as large. With these adjustments it can be calculated that if the DUI reforms in- stituted by the Commercial Motor Vehicle Safety Act of 1986 are enacted and implemented with due regard to the points of deterrence just emphasized, this is likely to prevent about 170 truck involvements in fatal crashes, about 1,760 involvements in non-fatal-injury crashes, and about 1,600 involvements in property-damage-only crashes. The estimates developed here on the basis of previous empirical results indicate the size of deterrence effects that might be achieved with a national

PREVENTING ALCOHOL-IMPAIRED DRIVING 113 TABLE 5-4 PRELIMINARY ESTIMATES OF CRASHES PREVENTABLE BY APPLICATION OF SUCCESSFUL DUI REFORMS TO COMMERCIAL VEHICLE OPERATORS Fraction of Crashes Deemed Preventable No. of All-Vehicle Positive BAC Proportion No. of Vehicle and Vehicles Deterrence Relative to Responsible Preventable Crash Type Involveda Results" All Driversc for Crash'' Involvementse Heavy trucks PDO 224,000 (0.04) (0.30) (0.5)=0.0030 1,340 NI 85,500 (0.11) (0.30) (0.5)=0.0165 1,410 F 4,448 (0.23) (0.30) (0.5)=0.0345 153 Medium trucks PDO 55,000 (0.020) (0.52) (0.5)=0.0026 290 NI 24,400 (0.055) (0.52) (0.5)=0.0143 350 F 649 (0.115) (0.52) (0.5)=0.0299 19 Buses PDO 43,500 - - NI 19,700 - - F 337 - - NOTE: PDO= property damage only; NI= nonfatal injury; F=fatal. a Data from NHTSA (1987) and Table 2-1I. b Based on best results achieved in major cases of successful deterrence (Table 5-3). As noted in the text, deterrence benefits for medium trucks assumed to be half as large as for heavy trucks. C Based on Table 2-16 and FARS data tapes, 1982-1985. d Based on data by Clarke et al. (1987) for all crashes; see also Boyer et al. (1985). e The estimates in this column are approximations. The precision in the calculations is retained to avoid introducing rounding errors in this table and subsequent tables based on it. implementation of an intensive checkpoint surveillance strategy (or a DUI patrol strategy capable of yielding equivalent numbers of detections and sanctions) coupled with the threat of administrative license revocation at a BAC of 0.05 to 0.10 percent. It is probably the case that the lower end of the BAC range is more difficult to enforce with the patrol strategy because of the reduced availability of driver-behavioral cues at these levels. The checkpoint strategy, however, could be made just as effective at detecting lower BACs as higher ones if passive sensor technology were routinely used. It may be speculated that, with comparably intensive levels of enforcement, invocation of sanctions against drivers at 0.04 percent B AC—which is similar to the 0.05 percent BAC provided by statute in New South Wales and informally sanctioned in the Charlottesville demonstration—would be most likely to yield the deterrence levels indicated in Table 54, whereas a 0.10 percent limit, with no sanctions at all applied below this level, might fall short of these levels by one-fourth to one-third, as summarized in Table 5-5.

114 ZERO ALCOHOL AND OThER OPTIONS TABLE 5-5 CRASHES, INJURIES, AND FATALITIES PREVENTABLE BY BEST DETERRENCE SCENARIO AT ALTERNATIVE BAC STANDARDS Type of No. Preventable by BAC Standard Reduction 0.00 004 0.10 PDO crashes 2,040-2,170 1,630 1,070-1,220 Injuries in NI crashes 3,070-3,270 2,460 1,620-1,840 Fatalities 230-250 190 125-140 NoTe: Data are for medium and heavy trucks; use of passive sensors is assumed. PDO = property damage only; NI = nonfatal injury. a These data were drawn from Table 5-4 with the additional coef- ficients of 1.396 persons injured per NI crash and 1.10 persons killed per fatal crash. The data for 0.00 and 0.10 percent BAC are based on multiplication of the 0.04 percent data by 1.25 to 1.33 and 0.66 to 0.75, respectively. By a similar line of reasoning, these results might be extrapolated to a zero BAC proscription. This exercise is even more speculative, however. A zero BAC limit would probably place about twice as many current commercial vehicle drivers at risk of sanction as would a 0.04 percent BAC limit, though it is doubtful that the sanction for such low levels would realistically be deliv- ered in as extreme a form as that now being contemplated for higher limits. A liberal estimate, it seems, is that a zero BAC standard would yield one-third to one-half again as much deterrence effect as might be seen with a 0.04 BAC limit if sanctions were delivered with the same intensity (Table 5-5). The rough estimates of preventable losses in Table 5-5 depend on the assumption of intense enforcement with efficient detection of violations and consistent application of sanctions. These assumptions almost certainly re- quire that new checkpoint strategies be employed and—if BACs below 0.10 percent are to be effectively enforced—the routine use of passive sensor technology. Without passive sensors the police have greater difficulty detecting drinldng drivers with low BACs during checkpoint screening. In the Charlottesville experiment, the police detected 45 percent of drivers with BACs between 0.05 and 0.099 when passive sensors were used but only 24 percent without use of passive sensors (Jones and Lund 1986, 156). (The actual BACs were deter- mined by researchers requesting breath samples from these same drivers after they had passed through the checkpoint.) Passive sensors also assist in the detection of BACs greater than or equal to 0.10 percent. Police detected 68 percent of drivers with high BACs when sensors were used and 45 percent when they were not used (Jones and Lund 1986, 157). Without the passive sensor, detection efficiency was thus reduced 34 percent at high BACs and 50

PREVENTING ALCOHOL-IMPAIRED DRIVING 115 TABLE 5-6 PREVENTABLE CRASHES, INJURIES, AND FATALITIES WITHOUT USE OF PASSIVE SENSORS Type of No. Preventable by BAC Standard Reduction 0.00 0.04 0.10 PDO crashes 1,190-1,280 1,000 710-805 Injuries in NI crashes 1,700-1,800 1,525 1,070-1,220 Fatalities 130-140 115 80-90 NOTE: PDO = property damage only; NI = nonfatal injury. TABLE 5-7 PREVENTABLE CRASHES, INJURIES, AND FATALITIES BASED ON POST-CRASH TESTING ENFORCEMENT OPTION Type of No. Preventable by BAC Standard Reduction 0.00 0.04" 0.10 PDO Crashes 600-640 480 320-360 Injuries in NI crashes 910-970 730 480-540 Fatalities 55-60 45 30-35 NOTE: PDO = property damage only; NI = nonfatal injury a See footnote, Table 5-5. percent for BACs between 0.05 and 0.10. Using these efficiency decrements as benchmarks, the benefits of DUT enforcement without passive sensors can be approximated in rough form. These estimates are shown in Table 5-6 in which the benefits from Table 5-5 have been reduced 34 percent at a BAC of 0.10, the incremental benefit at 0.04 has been reduced 50 percent, and the incremental benefit at zero BAC has been reduced by an assumed 66 percent. If crash-based testing is the only type of enforcement added to current patrol-based strategies, the incremental deterrence effects of the statutory reforms will likely be smaller. In this event these effects might be roughly one-fourth as large at the outset as the benefits derived from the estimates in Table 5-4. The estimate for medium trucks in Table 5-4 has to be doubled in this scenario because the post-crash testing enforcement effort will affect drivers of medium and heavy trucks more equally than enforcement at weigh stations and as part of vehicle safety inspections. Because of the much lower level of driver surveillance, post-crash testing alone is not as likely to yield sustained benefits. The estimates derived from Table 5-4 are extended down- ward to reflect the prompt effect of post-crash testing as shown in Table 5-7.

116 ZERO ALCOHOL AND OTHER OPTIONS NOTES Deterrence falls within a class of contingent propositions that involve the exchange of behavioral performances and sanctions. The central idea of such theories is "if I do X, that is, choose to perform behavior X. the other party (or parties) will probably do Y, that is, respond with sanction Y." Under this contingency, the probability that one will do X depends on how one evaluates the actual probability that Y will follow, along with the desirability of Y. This kind of theoretical proposition is not only contingent, that is, intrinsically probabilistic, but also necessarily voluntaristic. That is, if one cannot do X or cannot avoid doing X, the contingent response, sanction Y, is irrelevant. Less drastically, if there are other forces impinging on the choice of X, such as qualms about its moral goodness or the high prices of resources needed to get X done, the power of the contingent sanction Y to determine that choice is weakened. No estimates are available for roadside stops as such in the United States. But as a related benchmark, in 1985 the FBI Uniform Crime Reports recorded approximately 1.8 million arrests for DUI annually (out of 11.6 million police arrests). If 3 percent of random roadside tests led to additional arrests--a low ballpark figure—that would nearly double the overall Dlii arrest rate. Of course, if the deterrent effect led to a reduction in the numbers of Dlii episodes overall, the total DUI arrests might not change as much despite the greater surveillance. REFERENCES Blumstein, A., J. Cohen, and D. Nagin, eds. 1978. Deterrence and Incapacftation: Estimating the Effects of Criminal Sanctions on Crime Rates. National Academy Press, Washington, D.C. Box, G. E. P., and G. C. Tiao. 1975. Intervention Analysis with Applications to Economic and Environmental Problems. Journal of the American Statistical Asso- ciation, Vol. 70, pp. 70-79. Boyer, V. W., D. A. Couts, A. J. Joshi, and T. M. Klein. 1985. Jdentfication of Preventable Commercial Vehicle Accidents and Their Causes. Mandex, Inc., Vienna, Va. Cameron, T. 1979. A Review of Drinking-Driving Countermeasures: A Review and Education. Contemporary Drug Problems, Vol. 8, No. 4, pp. 495-566. Clarke, R. M., W. A. Leasure, Jr., R. W. Radlinski, and M. Smith. 1987. Heavy Truck Safety Study; Prepared in Response to Section 216: PL. 98-544, October 30, 1984, Motor Ca,rier Safety Act of 1984. Report DOT HS 807-109. U.S. Department of Transportation. Available from NTIS. Gibbs, J. 1975. Crime, Punish,nent, and Deterrence. Elsevier Scientific Publishing Company, New York. Homel, R. 1986. Policing the Drinking Driver: Random Breath Testing and the Process of Deterrence. Federal Office of Road Safety, Sydney, Australia. Jones, I. S., and A. K. Lund. 1986. Detection of Alcohol-Impaired Drivers Using Passive Alcohol Sensor. Journal of Police Science and Administration, Vol. 14, No. 2, pp. 153-160. NHTSA. 1979a. Alcohol Safety Action Projects Evaluation Methodology and Overall Program Impact. Report DOT HS 803896. U.S. Department of Transportation. NHTSA. 1979b. Results of National Alcohol Safety Action Projects. Report DOT- HS-804-033. U.S. Department of Transportation.

PREVEWfING ALCOHOL-IMPAIRED DRIVING 117 NHTSA. 1979c. Alcohol Safety Action Project Evaluation of Operations: Data, Tables of Results, and Formulation. Report DOT 85 804 085. U.S. Department of Transportation. Reed, D. S. 1981. Reducing of the Costs of Drinking and Driving. In Alcohol and Public Policy: Beyond the Shadow of Prohibition (M. H. Moore and D. R. Gerstein, eds.), National Academy Press. Ross, H. L. 1973. Law, Science, and Accidents: The British Road Safety Act of 1967. Journal of Legal Studies, Vol. 2, pp. 1-78. Ross, H. L. 1984. Deterring the Drinking Driver: Legal Policy and Social Control. D.C. Heath and Company, Lexington, Mass. Ross, H. L., and J. P. Foley. 1987. Judicial Disobedience of the Mandate to Imprison Drunk Drivers. Law and Society Review, Vol. 21, No. 2, pp. 315-323. Ross, H. L., and G. D. LaFree. 1986. Deterrence in Criminology and Social Policy. In Behavioral and Social Sciences: Fifty Years of Discovery (M. S. Smelser and D. R. Gerstein, eds.), National Academy Press, pp. 129-152. Ross, H. L., R. McCleary, and T. Epperlein. 1982. Deterrence of Drinking and Driving in France. Law and Society Review, Vol. 16, pp. 345-374. Sabey, B. A. 1978. A Review of Drinking and Drug-Taking in Road Accidents in Great Britain. Presented to the American Association of Automotive Medicine and International Association for Accident and Traffic Medicine, Ann Arbor, Michigan. Tittle, C. 1980. Sanctions and Social Deviance: The Question of Deterrence. Praeger Publishers, New York. Voas, R. B., and J. M. Hause. 1987. Deterring the Drinking Driver. The Stockton Experience. Accident Analysis and Prevention, Vol. 19, No. 2, pp. 81-90. Was, R. B., A. E. Rhodenizer, Jr., and C. Lynn. 1985. Evaluation of Charlottesville Checkpoint Operation: Final Report. NHTSA, U.S. Department of Transportation. Wilson, J. Q. 1975. Thinking AbOUt Crime. Basic Books, New York. Zimring, F., and G. Hawkins. 1973. Deterrence: The Legal Threat in Crime Control. University of Chicago Press, Chicago, ill.

6 Costs and Benefits of Public and Private DUI Enforcement Practices Deterring drinking and driving by commercial vehicle operators, as indicated in the previous chapter, will require vigorous and sustained enforcement. Successful enforcement of a more stringent BAC standard, in turn, will depend on developing cost-effective enforcement strategies directed toward the commercial vehicle driver. In this chapter current drinking-driving en- forcement practices in both public and private settings are reviewed, the problems of adapting these practices to accommodate a new standard for commercial vehicle drivers are examined, workable enforcement strategies for implementing the new standard are identified, and the costs and benefits at alternative BAC standards are assessed. PUBLIC ENFORCEMENT All levels of government are involved in enforcement of drinking-driving laws, reflecting the federalist nature of the U.S. governmental system. State and local law enforcement agencies carry out state alcohol safety laws under the police authority conferred by the state. In addition, federal inspectors and state agencies responsible for enforcing federal motor carrier safety regula- lions have the authority to administer tests for cause to commercial vehicle drivers suspected of DUI infractions. Actual arrests, however, are typically handled by state police. Most law enforcement officials have the authority to carry out DUI legisla- lion, but not all have the training to correctly identify alcohol impairment in the drinking driver. Recognizing this, many police departments have begun to provide special training in DIlL enforcement and in sothe states have created 118

COSTS AND BFJ'IEF!TS 119 special teams, often referred to as "Tiger Teams," whose primary role is DUI enforcement. These special efforts are based on evidence that lack of know!- edge among police officers concerning the relationship between alcohol use and impaired driving behavior has a negative effect on alcohol-related en- forcement and that specialized training and specialized assignments can have a strong positive influence on alcohol-related arrest rates (Oates 1974, vii). Current Enforcement Practices Drinking-driving enforcement efforts are generally keyed to the passenger vehicle population. Most police target their DUI enforcement activities on the drinking-driving patterns of the general population. Enforcement efforts are generally concentrated during times of known high DUI incidence, such as weekend nights and late night and early morning hours. In recent years many police departments have also begun to target the location of their enforcement efforts, using sobriety checkpoints as a highly visible means of screening the general driving population for alcohol use at strategic locations. As noted in the previous chapter, however, these kinds of enforcement measures may be less effective in reaching the commercial vehicle driver. First, the incidence of drinking and driving by such drivers is likely to be spread out over the week rather than concentrated in the weekend late night hours. Moreover, drinking is unlikely to be concentrated at particular loca- tions, with the possible exception of truck stops, because the commercial vehicle driver frequently covers long distances and often has his own drink cooler on board. Second, the space needed to pull over large commercial vehicles without causing traffic backups or unsafe through-driving conditions now often results in their exclusion from DUI enforcement efforts, par- ticularly in heavily congested urban areas. Commercial vehicles, for example, are frequently waved through roadside sobriety checkpoints. Finally, the prevalence of CB radios and the availability of alternative routes for commer- cial vehicle drivers may thwart efforts to concentrate DUI enforcement ac- tivities at fixed times or locations. Opportunities exist for targeting DUI enforcement more directly on the commercial vehicle driver. Currently, commercial safety enforcement efforts are focused on vehicle safety inspections and truck weight standards. These could readily be broadened to include more aggressive DUI enforcement. Many states with heavy-truck traffic have trained and dedicated special state highway patrols to enforce all aspects of the motor carrier safety regulations. For example, in Illinois 45 of approximately 1,600 state troopers work full time on commercial vehicle duiver safety enforcement. Because these teams are knowledgeable about the specific safety problems of motor vehicle car- riers and also have considerable opportunities for interaction with the drivers, they are well positioned to conduct alcohol enforcement activities as well.

120 ZERO ALCOHOL AND OTHER OPTIONS In the following section strategies for targeting DUI enforcement more directly on commercial vehicle drivers are discussed and their costs are analyzed. Developing Workable Commercial DUI Enforcement Strategies From a deterrence perspective, the most effective DUI enforcement strategies are those that sustain a high level of intensity. Successful enforcement efforts increase the probability that drivers will be stopped for observation and testing, include a large publicity element to heighten driver awareness of risk, and assure swift penalties for those identified over the legal BAC limit. A workable enforcement strategy must also be cost-effective; that is, it must result in efficient detection of positive BACs in the target population. Comprehensive Commercial Safety Enforcement One way to meet the criteria just outlined would be to combine DUI enforce- ment activities with existing safety programs that provide opportunities for contact with commercial vehicle drivers. A deterrence strategy modeled on this approach would have four critical elements: Alcohol enforcement as part of commercial vehicle safety inspections, Use of weigh stations as an opportunity to check for driver sobriety, A public information campaign, and License revocation through an administrative process handled by the state Department of Motor Vehicles. This approach provides the opportunity for contact with two out of every five commercial vehicle drivers on the road each year, a contact ratio well within the range of the successful deterrence strategies discussed in the previous chapter. Moreover, by combining DUI enforcement with existing commercial safety programs, enforcement costs can be reduced. How this enforcement strategy would be implemented as well as its cost are outlined in more detail in the following discussion. Safety Inspections Considerable attention has been given recently to the problem of the increasing number of mechanically defective commercial

COSTS AND BF]s'EFITS 121 vehicles on the nation's highways. Thirty-seven states, with the assistance of the federal Motor Carrier Safety Assistance Program (MCSAP), are now in the process of developing safety inspection programs for commercial vehicles (personal communication, FHWA Office of Motor Carrier Safety). Because the program was established by the Surface Transportation Assistance Act of 1982, federal funding has increased from $8 million in 1984 to $44 million in 1987 and is expected to grow to $53 million by 1991.1 Thus, considerable funding should be available to expand the coverage and improve the effective- ness of state safety inspection efforts. To examine the feasibility of combining DUI enforcement with safety inspections, an understanding of how these inspections are currently per- formed is needed. Safety inspections are conducted at fixed locations, such as ports of entry at state borders and scale houses for truck weighing, and on a random basis on the highways. In the latter case, trucks are selected for inspection if the safety inspector suspects a problem. When inspections are conducted at fixed locations, states attempt to select sites where bypass opportunities are limited. To avoid delaying passengers, bus inspections are generally not conducted on the highway, but rather at the vehicle's origin or destination. The inspections themselves typically take between 20 min and 1 hr, de- pending on the extent of the problem. Because of the length of time involved, the primary difficulty with combining safety inspections with DUI enforce- ment is coverage. In 1986, for example, state and federal inspectors performed safety inspections on 470,000 commercial vehicles, representing only about 13 percent of the estimated vehicle population (FHWA Office of Motor Carrier Safety).2 However, the number of inspections is expected to rise to 1.2 million in 1987 as states expand their safety inspection programs. If DUI enforcement is to be combined with state inspection programs, probably the best approaches are to train existing safety inspectors in alcohol enforcement or to add a limited number of specialists in DUI enforcement to participate as members of a dedicated commercial vehicle safety enforcement team. In order to maximize personnel use, those trained to identify DUI infractions should also be able to conduct other safety-related activities. Federal funds from the MCSAP program may be available to defray a portion of the cost if states include DUI enforcement as part of their overall commer- cial vehicle safety programs. The state of Tennessee provides an excellent example of a recent program to combine commercial motor vehicle safety inspections with alcohol enforce- ment in which the lead agency is the Public Service Commission. The state has installed evidential breath-testing units in vans at four fixed weigh stations where safety inspections as well as truck weighing are conducted. All safety enforcement efforts are focused heavily on driver observation. For example,

122 ZERO ALCOHOL AND OTHER OPTIONS safety inspectors actually get into the cab to check license, registration, log books, and other records associated with the safety inspection. If alcohol use is suspected, either from behavioral cues or from the presence of an alcoholic beverage in the cab (which is forbidden by federal regulation), the safety inspector can administer a breath test to establish the level of alcohol impair- ment. If test results are positive, the state highway patrol is contacted and the driver is taken to the police station for further processing.3 The program director believes that the combined approach has increased DUI enforcement, although he stressed that this more targeted effort should not replace regular DUI enforcement activities (personal communication, Tennessee Public Ser- vice Commission). Truck Weigh: Enforcement Alcohol enforcement efforts directed toward the commercial vehicle driver could be expanded considerably if drivers were also observed for alcohol impairment as part of state vehicle weight enforce- ment programs. In 1985, for example, 106 million trucks were weighed—an average rate of 29 times per vehicle per year (FHWA Office of Motor Carrier Safety). Truck weight inspections are conducted by states to ensure that commercial vehicles are in compliance with federal regulations regarding weight limits on Interstate highways. The majority (more than 90 percent) of commercial vehicle weighing activity takes place at fixed scales located in permanent installations along major highways (FHWA 1985, 17). Some 500 scale houses are generally manned by one person each, working an 8-hr shift. Weighing activity is typically rotated among the different scale house locations, in part to minimize avoidance by the drivers. Portable scales that can be set up at random locations along the roadside are also being used by many states to limit drivers' ability to predict where and when inspections will occur. Practical difficulties may somewhat limit the option of combining truck weighing with DUI enforcement. First, despite the large number-of vehicles weighed, actual opportunities for contact with the driver are fewer than the numbers would suggest. The trend is to automate the truck weighing process, so that trucks can be inspected while in motion.4 Only those drivers of trucks that register over the limit must actually interact with the inspector to receive a citation. In 1984, the number of citations was less than 1 percent of the commercial vehicles weighed, but the absolute number-674,000---was still larger than the total number of safety inspections conducted (FHWA 1985, 20). Second, because of communication between truckers on CB radios and bypass opportunities, trucks can avoid scale houses and may have even greater incentive to do so should DUI enforcement activities be combined with truck

COSTS AND BFJVEF!TS 123 weighing. States may be able to circumvent at least a portion of this problem by greater use of portable scales on bypass routes and by locating fixed scales where alternative routing options are limited. Third, combining DUI enforcement with truck weighing may require addi- tional personnel. It is unlikely that the scale operator would have sufficient time to handle both types of activities. Finally, the majority of trucks that weigh less than 26,000 lb will not exceed the weight limits.5 Thus, this enforcement approach will be directed primarily toward large trucks operating on Interstate roads at locations away from congested urban areas. It appears that the best approach from the perspective of coverage of the target population and opportunities for driver contact may be a strategy that piggybacks DUI enforcement efforts onto both vehicle safety inspections and truck weight enforcement activities. Checking for valid commercial licenses is an additional low-cost way to target the commercial vehicle driver for DUI enforcement. Illinois is training its regular state highway patrol to conduct such commercial vehiôle driver evaluations. These involve stopping a driver briefly to check his license, medical records, and log book.6 At the same time this provides an opportunity to observe for any signs of alcohol impairment. Public Information Campaign A successful enforcement strategy must also be accompanied by a vigorous public information effort to alert commer- cial vehicle drivers to the greater likelihood of their being stopped and tested for alcohol use. Ideally, a publicity campaign would involve a national effort to develop a campaign theme, advertising materials, and a promotional strat- egy, as well as accompanying state programs to develop promotional materials tailored to local audiences. Creating driver expectation of a significant crackdown on drinking and driving would require at least two major cam- paigns each year to sustain driver awareness of the likelihood of being apprehended and tested for driving under the influence of alcohol. License Revocation The fourth and final element of an effective deter- rence strategy is a swift penalty through revocation of the licenses of those commercial vehicle drivers with alcohol concentration levels at or above the legal limit. Following the mOdel of many existing state administrative per se laws, those drivers in violation of the legal BAC limit would automatically lose their commercial licenses. Due process requirements of the Fourth Amendment would be met by the right to an administrative review, admin- istrative hearing, and, if requested, a judicial review. Revocation of the license, however, would not be stayed pending the outcome of any appeals.

124 ZERO ALCOHOL AND OTHER OPTIONS Many studies have documented the strong deterrent effect of license revoca- lion relative to other sanctioning methods (Latchaw 1986, 8). Costs To estimate the costs of this intensive enforcement strategy, several simplifying assumptions were made. First, the costs at the current BAC legal limit of 0.10 percent were assumed as the baseline. However, the level of enforcement activity was assumed to represent an increment above current enforcement efforts. Certain activities, such as driver evaluations, could be undertaken with existing personnel. Combining DUT enforcement with safety inspections and truck weight enforcement, however, would represent major new activities requiring additional manpower. Second, enforcement efforts were assumed to apply to drivers of both heavy and medium trucks. Finally, similar enforcement procedures were assumed for all commercial vehicle drivers stopped for suspected DUT infractions: a brief roadside interview to determine the presence of alcohol, field testing for alcohol impairment, and, where cause can be established, an evidential test at a police station. (Figure 6-1 shows the various steps, time allocated for each step, and expected number of incidents involved in each step annually.) A summary of the major costs of this option is provided in the text box. The costs are categorized into four types: those associated with testing for alcohol involvement, those associated with the sanctioning process for drivers who test at 0.10 percent BAC or greater, those associated with a national publicity campaign, and those associated with economic losses from shipment delays. When all of these elements are added together, the comprehensive safety enforcement option would add $22 million to current DUT enforcement costs. A more detailed explanation of the calculations underlying this cost estimate is given in Appendixes D and E. Post-Crash Testing Another way to target Dill enforcement on commercial vehicle drivers is to require testing for alcohol use in serious crashes involving a medium or heavy truck or bus. This is the approach adopted by the airline and rail industries, which have recently implemented new regulations regarding control of alco- hol and drug use (see Chapter 4). Enforcement efforts in these industries are based on the special safety risks that alcohol-impaired airline and rail crews pose to the public. As a condition of employment, airline and rail crews must agree to testing in those circumstances, such as crashes and rule violations, in which the safety risk and likelihood of human error are high. This approach could be extended to commercial vehicle drivers. To sim- plify enforcement, testing could be limited to those crashes involving a fatality

Safety Truck Weight Inspections Citations (1,158,000) 0.05 hr (675,000) Roadside 1 interviews (1,833,000) 0.95 hr [ieid Tests for impairment (32,990) .2hr Evldent!ai Tests Positive; Administrative Sanctions Required (11,000) Administrative Reviews (11,000) Administrative Hearings (5,500) Judicial Reviews (2,750) FIGURE 6-1 Enforcement scenario for comprehensive commercial safety enforcement option.

126 ZERO ALCOHOL AND OThER OPTIONS Comprehensive Commercial Safety Enforcement Cost Estimate Based on 0.10 Percent BAC Alcohol testing Personnel $5,192,000 Equipment 304,000 Training 4,000 Testing 330,000 Towing 2,750,000 Subtotal $8,580,000 Sanctioning Administrative review $275,000 Administrative hearing 1,782,000 Judicial hearing 3,114,000 Subtotal $5,171,000 Publicity campaign $5,000,000 Economic losses from shipment delay Driver $2,136,000 Truck 882,000 Inventory 186,000 Subtotal $3,204,000 Total $21,955,000 NoTE: See Appendixes D and E for a more detailed ex- planation of the calculations underlying the cost estimate. or injury requiring transport to a hospital. Fifteen percent, or 67,000, of the truck- and bus-involved crashes each year fall into this category. Post-crash testing would represent an increase in current DU! enforcement efforts and may result in increased arrests. As pointed out in Chapter 2, in

COSTS AND BENEFITS 127 many jurisdictions BACs are not regularly measured in the case of a fatal accident or in accidents resulting in injury. Although more states have begun to test for blood alcohol in the event of fatal accidents, they frequently exclude the surviving driver, who, in the case of truck-involved fatal accidents, is the truck driver in 80 percent of such crashes. There are some limitations, however, with the post-crash testing option. First, the deterrence benefits of post-crash testing have not been studied. Beyond the specific deterrent on alcohol-involved drivers of the 67,000 trucks and buses involved in serious crashes each year,7 it is difficult to gauge the general deterrent effect on the drinking and driving behavior of other commer- cial vehicle operators. The likelihood of being in a crash and thus being required to take a blood alcohol test may not be strong enough to deter the average commercial vehicle driver from drinking and driving. In addition to the unknown benefits, the costs per driver tested are more than twice as high as those under the comprehensive commercial safety enforcement option at a 0.10 percent BAC. Appendix F provides more detail on the costs of the post- crash testing option, which have been estimated at approximately $10.9 million at 0.10 percent BAC. Finally, for this option to be practical, the crash itself must provide sufficient legal grounds for alcohol testing. This may be accomplished, as suggested in Chapter 4, by requiring explicit consent to testing and cooperation with law enforcement officials as a condition of a commercial license. Otherwise, if it is necessary to establish reasonable suspicion of alcohol impairment, this may limit the use of testing in practice because so many other factors than establishing a case for alcohol testing may appear more important at the accident scene. Total Enforcement Costs If a combined enforcement approach is adopted in which commercial vehicle drivers are tested for alcohol use after serious crashes as well as on the road during vehicle safety inspections and truck weight enforcement activities, then the total costs of enforcement can be calculated. As shown in the following tables, total enforcement costs are somewhat less than the sum of the individ- ual enforcement approaches because certain costs, such as publicity, alcohol testing equipment, and training, are common to both approaches and thus could be scaled back in a combined enforcement program. These costs assume widespread use of passive sensors and portable breath- testing devices as part of an intensive enforcement strategy. Their use will be particularly critical at lower BACs where detection of alcohol use is likely to prove difficult without objective screening and testing devices. If use of passive sensors and PBTs is ruled unconstitutional, then enforcement costs

128 ZERO ALCOHOL AND OTHER OPTIONS would decline by $2 million to $5 million, which reflects savings on equip- ment purchases. However, the benefits would also be reduced accordingly, particularly at low BACs, as shown later in this chapter. The cost estimates are most sensitive to assumptions about the alcohol levels of the commercial vehicle drivers tested. If the assumptions concerning driver alcohol involvement are underestimated by 50 percent, then enforce- ment costs would increase by nearly 40 percent to $40 million. If the under- estimate is as much as 100 percent, then enforcement costs would increase by 60 percent to $48 million. The estimates used here reflect current best judgment concerning the size of the alcohol problem among commercial vehicle drivers. Enforcement costs were also analyzed to determine the incremental costs at alternative BACs, to apportion the costs by truck size, and to allocate the costs by likely payment source. Cost Allocation by Enforcement Standard Table 6-1 summarizes the costs of enforcement at the current legal limit and at more stringent alcohol standards-0.04 percent BAC and 0.00 percent BAC. Costs at the 0.10 BAC standard assume little change in the current enforcement environment. Police officers are likely to use behavioral sobriety tests to establish alcohol impair- ment, and the testing process itself, from stopping the driver initially to arresting the driver and conducting an evidential test, is likely to take up to 3 TABLE 6-I TOTAL ENFORCEMENT COSTS ALLOCATED BY ENFORCEMENT STANDARD Current Proposed "No Alcohol" Standard Standard Standard Item 0.10% BAC) 0.04% BAC) (>0.00% BAC) Comprehensive commercial safety enforcement ($ millions) 21.9 29.2 41.1 Post-crash testing ($ millions) 10.9 15.3 16.0 Combined approach ($ millions) Testing 15.2 21.6 26.4 Sanctioning 6.4 11.1 17.6 Publicity 5.0 5.0 5.0 Delays 3.2 17 5] Total 29.8 41.4 54.1 Increase over current standard(%) - 38.9 81.5 NOTE: The combined approach does not equal the sum of its components because certain shared costs, such as publicity, could be reduced in a combined program.

COSTS AND niiin's 129 hr. At lower BACs greater use of objective testing devices, such as passive sensors and PBTs, was assumed necessary to detect alcohol impairment. This resulted in higher equipment costs, but was offset in part by a reduction in the time required to conduct the testing process from 3 to 21/4 hr. As Table 6-1 shows, nearly half of total enforcement costs can be attributed to the testing process regardless of the legal BAC limiL Lowering the legal limit from the current standard to 0.04 percent increases total enforcement costs by nearly 40 percent. At zero BAC, the cost of enforcement increases by more than three-fourths, reflecting the fact that there are an estimated three times the number of commercial vehicle drivers on the road with some alcohol in their bloodstreams (i.e., > 0.00 percent BAC) than there are with 0.10 percent BAC. Cost Allocation by Vehicle Size The proposed alcohol standards will only cover heavy trucks weighing more than 26,000 lb and buses. Nevertheless, the committee was asked to examine the costs of extending the regulations to medium trucks (those weighing between 10,000 and 26,000 ib). Table 6-2 provides a breakdown of the costs attributable to each truck size category at 0.10 BAC, showing that 70 percent of the costs are attributable to enforcement actions directed toward heavy trucks and buses.8 For example, the entire cost of DUI enforcement activities associated with truck weighing was allocated to heavy trucks because the majority of trucks that are likely to register over- weight and receive a citation will fall into the heavy-truck category. The remaining costs were allocated between heavy and medium trucks and buses on the basis of their representation in the total bus and truck fleet (Census Bureau 1985, 121-123) and on the basis of their representation in crushes.9 TABLE 6-2 TOTAL ENFORCEMENT COSTS ALLOCATED BY TRUCK SIZE AT 0.10 PERCENT BAC Total Heavy Trucks Medium Trucks All Trucks (:~26,000 Ib) (10,000— (2:10,000 Ib) Item and Buses 26,000 Ib) and Buses Comprehensive commercial safety enforcement ($ millions) 14.6 7.4 22.0 Post-crash testing ($ millions) 8.9 2.0 10.9 Combined approach ($ millions) 21.0 8.8 29.8 Share of total cost (%) 70 30 NOTE: The combined approach does not equal the sum of its components because certain shared costs, such as publicity, could be reduced in a combined program.

130 ZERO ALCOHOL AND OTHER OPTIONS Cost Allocation by Payment Source Costs were also analyzed from the perspective of who would pay. As Table 6-3 shows, the public sector would bear nearly four-fifths of the cost (78 percent) at 0.10 percent BAC.'° The estimates assume that the costs associated with the testing process, with the possible exception of the towing costs, would be paid by state and local law enforcement agencies. Similarly, the publicity costs plus a majority of the costs of the sanctioning process would be borne by the public sector, although a small portion of the latter might be defrayed by the commercial vehicle drivers at fault, a common policy now in those states with administrative per se laws.1' The economic costs would fall on the private sector. The majority of these costs—the cost of the drivers and the cost of the truck—would be borne by the trucking companies. The remaining inventory costs would be borne by the shippers and recipients of the goods shipped. TABLE 6-3 TOTAL ENFORCEMENT COSTS ALLOCATED BY PAYMENT SOURCE AT 0.10 PERCENT BAC Borne by Borne by Item Public Sector Private Sector Total Comprehensive commercial safety enforcement ($ millions) 16.0 5.9 21.9 Post-crash testing ($ millions) 10.3 0.6 10.9 Combined approach ($ millions) 23.3 6.5 29.8 Share of total cost (%) 78 22 NOTE: The combined approach does not equal the sum of its components because certain shared costs, such as publicity, could be reduce" in a combined program. Summary Drinking-driving enforcement efforts in the public sector are generally keyed to the passenger vehicle population. If commercial vehicle drivers are to be successfully deterred from drinking and driving, strategies are needed to sustain a high level of enforcement activity more directly targeted on the commercial vehicle driver and to ensure efficient detection of positive BACs in this target population. The strategy most likely to accomplish these objectives is to combine DUI enforcement with other commercial safety enforcement efforts, such as safety inspections and truck weight enforcement. This comprehensive commercial safety enforcement strategy approaches an all-out enforcement blitz, because it provides nearly 2 million opportunities for contact with commercial vehicle drivers each year, or two out of every five commercial vehicle drivers on the

COSTS AND BENEFITS 131 road. Increasing opportunities for contact with commercial vehicle drivers would be combined with a vigorous public information campaign and swift sanctioning through license revocation. Enforcement costs would be mini- mized by piggybacking them onto existing safety pmgrams directed toward the commercial vehicle driver. The option of testing after serious crashes involving commercial vehicle drivers was also considered. Post-crash testing would have a specific deterrent effect; however, gauging the effect of this enforcement alternative as a general deterrent to drinking and driving by commercial vehicle drivers is difficult and its efficiency, in terms of cost per driver tested, is less than that of the comprehensive commercial safety enforcement alternative. If both approaches are combined in an enforcement program, certain econo- mies of scale can be achieved. Implementing a combined program at a 0.10 percent BAC standard would cost $30 million above current DUT enforcement activities. These costs assume widespread use of passive sensors and portable breath-testing devices. If use of these testing devices were ruled unconstitu- tional, then total costs would be lower, reflecting equipment savings, but benefits would also be substantially reduced. The costs are most sensitive to assumptions about the BACs of commercial vehicle drivers tested. PRIVATE ENFORCEMENT Opportunities for controlling drinking-driving behavior also exist at the pri- vate or company level. Current private DUT enforcement practices affecting commercial vehicle drivers are reviewed next, followed by discussion of the impact of more stringent standards on these practices and examination of alternative enforcement strategies. With the exception of public transit, the major industries affected by the proposed change in standards—the trucking and intercity bus industries—are already subject to federal regulations regarding alcohol use. Specifically, the regulations prohibit consumption of an intoxicating beverage within 4 hr before going on duty (known as the "four-hour rule") and while on duty [49 CFR Section 392.5 (1986)]. Moreover, they prohibit possession of an intox- icating beverage while on duty or when operating a commercial vehicle. At a minimum, private companies observe these regulations. However, some com- panies have developed more specific alcohol policies and enforcement prac- tices as discussed in the following section.

132 ZERO ALCOHOL AND OTHER OPTIONS Current Enforcement Practices Trucking Industry Drivers in the trucking industry represent the largest pool of workers to be affected by the proposed change in alcohol standards. The alcohol policies of large trucking companies appear to be based on the federal regulations described earlier, and testing is generally confined to those incidents for which probable cause of. alcohol impairment can be established. The International Brotherhood of Teamsters, many of whose 200,000 mem- bers work for large trucking operations, instituted a national drug and alcohol testing and rehabilitation program in 1984. Members can be tested during their recurrent U.S. Department of Transportation physical examinations12 and when probable cause of an infraction can be established from the appearance, behavior, speech, or breath odor of an employee (IBT, 4). Any employee found to be under the influence of alcohol while on duty is subject to discharge; however, an employee is permitted to take a leave of absence to undergo treatment. The large companies and unions, however, represent only a fraction of all commercial vehicle drivers. As indicated in Chapter 2, companies of sufficient size to mount their own alcohol and drug abuse programs are responsible for less than one-third of the heavy trucks using the highways. The small owner- operator independent truckers generally have no formal company policy with respect to alcohol use. To the extent that the independents contract with the larger companies, however, they fall under the company's policy. The federal commercial motor vehicle safety regulations also apply to these drivers, but enforcement is likely to be limited. A recent survey by the Owner Operator Independent Drivers Association (OOIDA) on the desirability of more strin- gent alcohol standards found that its membership was evenly split between remaining at the current 0.10 percent BAC standard and reducing the standard to 0.04 percent. Only 11 percent of owner-operators favored zero BAC. Intercity Bus Industry The two largest intercity operators, Trailways and Greyhound, which repre- sent more than half the industry, report strict alcohol policies and enforcement practices. Alcohol consumption is prohibited 6 hr before reporting to work and company policy is to fire an employee caught drinking on the job. In fact, the companies would prefer to see a more stringent zero BAC standard than the 0.04 limit being considered. Concern over liability insurance because of the number of passengers at risk in intercity bus operations is the driving force

COSTS AND BENEFITS 133 behind these strict policies. Enforcement, however, is still limited to testing for cause at the discretion of a company supervisor. Public Transit A recent (1987) survey of the membership of the American Public Transit Association reported that 76 percent of the 192 responding transit agencies had some form of written rules, procedures, policies, or directives addressing drug and alcohol abuse in the workplace (APTA 1987, 1). Fifty-seven percent said they tested for alcohol and drug use at employment interviews; 43 percent reported that testing was required as part of an accident investigation (APTA 1987, 1). However, only a small fraction (12 percent) of the systems indicated that their collective bargaining agreements specify the conditions under which management may request or require an employee to take a blood or urine test as a check for drug or alcohol use (APTA 1987, 4). The Southern California Rapid Transit District (SCRTD) provides a good example of a transit system with a comprehensive and strict alcohol and drug abuse policy. SCRTD's revised policy, which became effective in December 1986, prohibits possession or use of alcohol or drugs while on duty. Manda- tory testing is required following an accident, an altercation (between two or more employees), or extended instances of tardiness or absenteeism (SCRTD 1986, 28-29).' Testing is also conducted at biennial physicals for operators and may be requested at the discretion of supervisors and managers for cause, including disruptive or unusual behavior by an employee. If an employee tests positive, he will be permitted to enter an Employee Assistance Program on a mandatory refenal or will be subject to discharge. If the former option is selected, the employee must agree to mandatory random testing for a period of 2 years after return to duty. This, however, is the only circumstance in which random testing is allowed (SCRTD 1986, 30-32). Private Enforcement Issues Two aspects of the commercial vehicle operator work environment may be directly affected by the proposed lowering of legal BAC limits for these drivers. First, the difficulty of detecting alcohol-impaired drivers is likely to increase. Detecting alcohol impairment is already a problem for many seg- ments of the commercial vehicle driver population because of the generally unsupervised nature of the work environment. Although the large trucking firms have their own terminals and supervisors, the majority of commercial truck drivers operate under limited supervision. Even in the intercity bus industry where most drivers are company employees, Trailways reports that nearly one-third of its drivers are dispatched by phone from a distant location.

134 ZERO ALCOHOL AND OTHER OPTIONS Bus operators in large public transit systems may begin their shift at a designated street location rather than at a terminal, thus limiting opportunity for supervisor contact. These problems will be exacerbated with a lower BAC limit at which detecting impairment, even under the best of circumstances, is difficult. A second effect of lowering the BAC legal limit may be to worsen the adverse impacts of scheduling uncertainties in the affected industries. The current rule prohibits the driver from drinking 4 hr before reporting for duty. If a driver is recalled without much advance warning, reducing the BAC legal limit may increase the likelihood of his having to refuse the recall notice or risk the possibility of arriving for work "under the influence" according to the new standard. Fortunately, this problem may not be widespread. Many of the large trucking operations, for example, have a 2-hr recall policy, but report that their drivers have a good indication of when they will be notified. According to the OOIDA, the long-haul independent drivers have considerable flexibility in determining their schedules. The majority of drivers in the intercity bus and public transit industries have regular hours.14 The major problem is with short-haul (i.e., less than 500-mi) loads in the trucking industry, for which there is less certainty about schedules. An estimated 25 to 30 percent of the owner-operators are involved in short-haul transport (personal communica- tion, OOIDA). Addressing the problems both of detecting impairment and of scheduling uncertainties may require new enforcement approaches. Alternative Enforcement Strategies One strategy to facilitate detection of impaired drivers at lower BACs is to expand testing for alcohol impairment from an incident-based to a random approach where probable cause need not be established. This controversial approach has been proposed for drug testing to cover thousands of airline and railroad workers as well as truck and bus drivers. The bill (S. 1041), which has passed the Senate Commerce Committee but must still pass the full House and Senate, would require preemployment, probable-cause, and random testing as well as testing after a serious or fatal accident. This same policy could be applied to testing for alcohol impairment. Industry support for this option is mixed. The American Trucking Associa- tions (ATA), which represents many of the largest for-hire trucking firms, supports the concept of random testing at the option of motor carriers, but not as a mandatory requirement. ATA has endorsed a pilot project to determine whether random testing is practical and cost-effective and whether it can be

COSTS AND BENEFITS 135 conducted without violating a driver's constitutional rights (Transport Topics 1987). Others oppose the concept outright. Some trucking firms believe that alcohol abuse is a relatively small problem among commercial vehicle drivers. Thus, an extensive program of testing would be costly and of limited benefit. Union opposition to random testing is strong; the unions argue that testing is an invasion of an individual's constitutional rights. Random testing may also present some practical problems. The unsuper- vised nature of the commercial vehicle driving environment may limit the opportunities for testing, and in those cases in which the drivers are in business for themselves, testing may simply not be possible. For these rea- sons, pilot studies are being recommended to test the feasibility of random testing in the trucking industry, the results of which will not be available by the time this report is published. Companies could also institute programs for employee self-testing. Upon reporting to work, employees could take a preliminary breath test and opt for a no-fault absence for that day or shift if they test at or above the legal BAC limit. Because the purpose of a self-testing program is to protect the em- ployee, many private companies may leave the decision to take the test up to the discretion of the individual employee. Addressing adverse impacts on scheduling uncertainties from a change in the BAC legal limit may require companies to review their recall policies. In addition, testing for alcohol use could be required of any driver who must report for duty with less than 4 hr notice. If the driver tested positive, he would be excused from duty without prejudice. This provision could be mandated by federal regulation. Properly implemented, this procedure would provide a barrier to driving by anyone with a BAC over the limit who was called in on short notice. It should have a positive deterrence effect, because more frequent testing is likely to create greater awareness of the issue of drinking and driving by the affected commercial vehicle drivers. However, testing under these circumstances is likely to reach only a small fraction of them. Trucks and buses could also be instrumented to require the driver, using a unique identification code, to key in simple functions related to driving. If these could not be performed, the vehicle would not start. Unless such instrumentation is mandated, however, implementing this option will require a significant effort to convince the trucking and bus industries of the desirability and cost-effectiveness of instrumented vehicles. Short of random testing, then, opportunities for expanding testing for suspected alcohol use by private employers are few. Because of limited opportunities for supervision in the private work environment, the major locus of enforcement activity is likely to remain in the public domain.

136 ZERO ALCOHOL AND OTHER ONIONS Summary By federal regulation most of the industries affected by the proposed change to a more stringent BAC standard, with the exception of public transit, already prohibit consumption of alcohol on the job and within 4 hr before going on duty. Many of the larger firms and public agencies have explicit alcohol and drug abuse programs that require testing of employees when probable cause for alcohol or drug impairment can be established. The main problem with these programs is the difficulty of enforcement in work environments where supervision is limited. Lowering the legal BAC limit will increase the difficulty of detecting alcohol impairment. It may also exacerbate the adverse impacts of scheduling uncertainties and short recalls. One solution to the problem of detecting alcohol abuse is to introduce random testing of employees. Industry support for this option is mixed, however, and union opposition is strong. Pilot tests in the trucking industry are being conducted, but the results will not be available by the time this report is published. Scheduling problems can be addressed by reexamining company recall policies and requiring alcohol testing for all drivers who must report for duty in less than 4 hr. Because of the generally unsupervised nature of the private work environ- ment, however, the primarycenter of enforcement activity is likely to remain in the public sector. BENEFITS AND COSTS OF ALTERNATIVE BAC STANDARDS In this section, a summary of the costs and benefits of the public commercial enforcement strategies discussed earlier is provided. Total social costs are compared with total social benefits, quantified where possible, for three alternatives: maintaining the status quo (i.e., 0.10 percent BAC), reducing the legal limit to the 0.04 percent standard of the rail and airline industries, or further reducing the legal limit to zero BAC (no alcohol). Included is an analysis of the incremental benefits and costs of extending the standards to cover the drivers of medium trucks as well as heavy trucks and buses. In practice, if the current standard is changed, the experience with enforcement costs and benefits during the first years after the change has been enacted will provide the basis for more definitive judgments about the appropriateness of the sanctions, the vehicles to be covered, and the need for further changes in the standard. Table 64 gives the benefits and costs of an intensive alcohol enforcement program at three alternative BAC standards for drivers of heavy and medium

TABLE 6-4 BENEFITS AND COSTS AT ALTERNATIVE BACs AND TRUCK WEIGHTS ASSUMING USE OF PASSIVE SENSORS AND PBTs BAC Legal Limit (%) Benefits No. of Lives Saved No. of Injuries Prevented Savings ($ thousands) Property Injury' Damageb Delay Total Enforcement Costs ($ thousands) 0.10 Heavy trucks 120 1,390 8,700 6,700 3,200 18,600 21,000 Medium trucks 10 340 2,100 1,200 800 4,100 8,800 All 130 1,730 10,800 7,900 4,000 22,700 29,800 0.04 Heavy trucks 170 1,970 12,300 9,400 4,600 26,300 29,400 Medium trucks 20 490 3,100 1,800 1,100 6,000 12,000 All 190 2,460 15,400 11,200 5,700 32,300 41,400 000 Heavy trucks 220 2,540 15,900 12,200 6,000 34,100 37,800 Medium trucks 30 630 4,000 2,200 1,400 7,600 16,300 All 250 3,170 19,900 14,400 7,400 41,700 54,100 a Injury savings estimates include medical costs, foregone taxes, legal and court fees, and other administrative costs. b Property damage savings estimates include the cost of damages primarily to the vehicles involved in crashes of all severity types plus a small amount of administrative costs for property-damage-only crashes.

138 ZERO ALCOHOL AND OTHER OPTIONS trucks under a scenario in which passive sensors and portable breath-testing devices (PBTs) are used. The primary benefits are, lives saved, injuries pre- vented, property damage savings, and delay savings. The numbers of prevent- able fatalities and injuries were drawn from the estimates presented in Chapter 5. (These figures will be further rounded in the final summary chapter so as not to give a false impression of the precision of the estimates. They were left unrounded here so that a more accurate assessment of the savings associated with these benefit numbers could be derived.) Injury and property damage costs averted were drawn from data developed by Kragh et al. (1986, 19), 'adjusted for the higher property damage costs attributable to accidents involv- ing a truck" and updated to 1986 dollars using the implicit price deflator. Quantifying the delay savings proved more difficult. Although crashes involving a commercial vehicle can create significant delays,16 few systematic studies of delay by crash type, time of day, or type of road are available. A recent study by the California Department of Transportation of truck accidents on freeways in the Los Angeles area found that truck crashes on average took 3 hr 35 min to clear and investigate and that there was an average loss of 2,006 vehicle-hr per incident (Golob 1985,7-20). At an average value of travel time lost of $8.31 per vehicle hour, the traffic delay cost per accident was estimated at $16,670.'' This estimate was then applied to the number of urban freeway alcohol-involved truck crashes (approximately 9 percent of all alcohol-in- volved truck crashes) that could be avoided through an effective commercial safety enforcement program. Crashes in rural areas were assumed not to create delays. Table 6-4 also shows the companion costs of enforcement for each BAC alternative. The costs were drawn from Table 6-1 and assume an intensive enforcement strategy that combines testing for alcohol use with truck safety inspections and truck weight enforcement as well as testing after serious crashes. Table 6-5 presents the results of a scenario in which testing devices are ruled unconstitutional. Although enforcement costs are $2 million to $5 million less because of equipment savings, relying on behavioral testing only will substantially reduce expected benefits, particularly at low BACs. Without the benefit of passive sensors and PBTs, enforcement efforts are likely to yield only two-thirds of expected savings at the current 0.10 percent standard. As the standard is reduced to 0.04 percent and to 0.00 percent, benefit levels would be reduced accordingly to 62 percent and 55 percent, respectively, of initial savings estimates. These declining benefits reflect the greater difficulty of detecting alcohol-impaired drivers at low BACs without the use of testing devices. They are based on an experiment conducted in Charlottesville, Virginia, in 1984 (discussed in Chapter 5), which compared the percentage of drivers detained at sobriety checkpoints with and without the use of passive sensors by the police (Jones and Lund 1986).

TABLE 6-5 BENEFITS AND COSTS AT ALTERNATIVE BACs AND TRUCK WEIGHTS ASSUMING NO USE OF PASSIVE SENSORS AND PBTs BAC Legal Limit (%) Benefits No. of Lives Saved No. of Injuries Prevented Savings ($ thousands) Property Injury Damage Delay Total Enforcement Costs ($ thousands) 0.10 Heavy trucks 80 915 5,700 4,400 2,100 12,200 19,000 Medium trucks 5 225 1,400 800 500 2,700 8,400 All 85 1,140 7,100 5,200 2,600 14,900 27,400 0.04 Heavy trucks 105 1,220 7,600 5,800 2,800 16,200 25,100 Medium trucks 10 305 1,900 1,100 700 3,700 11,400 All 115 1,525 9,500 6,900 3,500 19,900 36,500 0.00 Heavy trucks 120 1,400 8,700 6,700 3,300 18,700 33,600 Medium trucks 15 345 2,200 1,200 800 4,200 15,600 All 135 1,745 10,900 7,900 4,100 22,900 49,200

140 zo ALCOHOL AND OTHER OPTIONS TABLE 6-6 TOTAL NET ENFORCEMENT COST PER CASUALTY AVERTED AT ALTERNATIVE BACs AND TRUCK WEIGHTS ASSUMING USE OF PASSIVE SENSORS AND PBTs Cost by Truck Type ($) BAC Legal Heavy Trucks Limit (%) and Buses Medium Trucks All 0.10 1,600 13,400 3,800 0.04 1,400 11,800 3,400 0.00 1,300 13,200 3,600 NoTe: These data are drawn from Table 6-4 and represent a ratio of total enforcement costs less all benefits, short of the value of life, to the number of casualties (fatalities and injuries) averted at each BAC limit and truck weight category. TABLE 6-7 TOTAL NET ENFORCEMENT COST PER CASUALTY AVERTED AT ALTERNATIVE BACs AND TRUCK WEIGHTS ASSUMING NO USE OF PASSIVE SENSORS AND PBTs Cost by Truck Type ($) BAC Legal Heavy Trucks Limit (%) and Buses Medium Trucks All 0.10 6,800 24,800 10,200 0.04 6,700 24,400 10,100 0.00 9,800 31,700 14,000 NOTE: These data are drawn from Table 6-5 and represent a ratio of total enforcement costs less all benefits, short of the value of life, to the number of casualties (fatalities and injuries) averted at each BAC and truck weight category. Tables 6-6 and 6-7 compare costs and benefits at alternative BACs and truck weights. Benefits, short of the value of life, are subtracted from enforce- ment costs, and the residual is divided by the number of casualties (fatalities and injuries) averted to provide an estimate of the relative efficiency of each enforcement scenario. In an enforcement environment in which use of passive sensors and PBTs is widespread, the net costs per casualty averted remain relatively constant, reflecting the assumptions made to estimate costs and benefits (see Table 6-6). The benefits assume a relatively linear increase from 0.10 percent to 0.00 percent, whereas the costs increase proportionately less, reflecting the high level of fixed costs at each enforcement standard. This same effect is evident if the incremental, or marginal, net cost of reducing the legal BAC limit is examined.

COSTS AND BFJS'EFITS 141 In a legal environment in which passive sensors and PBTs are ruled unconstitutional, net cost of enforcement per casualty averted is substantially higher at each BAC, reflecting benefit reductions that are not offset by cost savings. As enforcement levels become more stringent, total enforcement costs per casualty averted increase by 40 percent from approximately $10,000 at 0.10 percent and 0.04 percent BAC to $14,000 at 0.00 percent BAC (see Table 6-7). The marginal net cost of reducing the legal BAC limit from 0.10 percent or 0.04 percent to 0.00 percent slightly more than doubles. When enforcement costs per casualty are compared by truck size, the costs are significantly less if the policy is restricted to heavy trucks and buses alone (see Tables 6-6 and 6-7). This finding reflects the proportionately larger number of fatalities that could be averted by preventing heavy-truck accidents. Clearly the most significant benefit in determining the effectiveness of a safety policy is the value of lives saved. Placing a value on human life, however, is highly controversial. Values from studies using the willingness-to- pay approach, when converted to a uniform base, range from $800,000 to $2.7 million (Miller 1986, 15). Estimates of the value of life used by various units of government range much higher. An alternative way of comparing benefits and costs is to quantify all the benefits without placing a value on human life. There is considerably less debate over the valuation of benefits such as averted injuries or pmperty damage saved. Then estimated savings can be compared with program costs and the additional cost needed to save a life can be calculated. If this approach is applied here to each enforcement scenario and the savings from prevented injuries, averted property damage, and avoided delay are all quantified and compared with enforcement costs, the value of a human life would range from $48,000 to $195,000, well below the values previously mentioned. Although this example would suggest that a policy of reducing the legal BAC limit can be supported in benefit-cost terms, policy choices are rarely made on economic grounds alone. Many other nonquantiflable issues raised earlier in this report may influence the decision. These are summarized in Chapter 7. SUMMARY To maximize deterrence effects at a cost that taxpayers and private industry are willing to pay, alcohol enforcement efforts should be targeted directly on the commercial vehicle driver. One option is to expand opportunities for contact with such drivers on the highway by combining DUT enforcement with other commercial safety activities, such as vehicle safety inspections and truck weight enforcement. Another alternative would be to adopt the approach

142 ZERO ALCOHOL AND OTHER OPTIONS used in other commercial environments and require testing after serious crashes. The general deterrence effect of the latter option, however, is less well defined. If the two approaches are combined and the 0.10 standard were maintained, total enforcement costs would reach $30 million above current DUT enforce- ment activities. The costs cover testing for alcohol impairment, sanctioning, publicity, and the economic losses of shipment delays. If a 0.04 percent or 0.00 percent standard were selected, enforcement costs would increase by approximately 40 percent and 80 percent, respectively, over the 0.10 percent baseline. These increased costs can be atiributed to the need for more objec- tive testing devices at lower BACs and hence higher equipment and training costs, and to the larger numbers of drivers that will test positive as the legal BAC limit is lowered. Seventy percent of the costs can be allocated to buses and heavy trucks (i.e., those weighing more than 26,000 lb), because many of the enforcement methods are targeted on heavy trucks. Finally, public law enforcement agen- cies are likely to pay approximately four-fifths of the costs. The remaining, largely economic, costs would be borne by private trucking firms and shippers. These costs are predicated on an environment in which the use of passive sensors and portable breath-testing devices is widespread. If these testing devices were ruled unconstitutional, then enforcement costs would be reduced by the savings in equipment, but the efficacy of enforcement efforts would be sharply reduced as well. The cost estimates are most sensitive to estimates concerning the alcohol levels of the commercial vehicle drivers tested. Private companies represent another avenue for enforcement efforts and many firms already have alcohol policies. However, because of the nature of the work environment, in which supervision of commercial vehicle drivers is limited, opportunities for expanding private enforcement efforts, particularly at low BACs, will be difficult. Random testing represents one option, but it is contnwersial and will require pilot testing to demonstrate its feasibility for commercial trucking operations. Companies may also reexamine recall pol- icies if legal BAC limits are lowered and testing may be required of drivers reporting to work with less than 4 hr notice. It is likely, however, for the foreseeable future that the primary enforcement role will be left to public law enforcement officials. When the costs of public enforcement are compared with the benefits, short of any quantification of the value of human life, the costs appear reasonable at the three alternative BAC standards analyzed-0.10, 0.04, and 0.00 percent. The net cost per casualty averted at the most stringent zero BAC standard ranges from $3,600 to $14,000 in a legal environment that constrains the use of passive sensors and preliminary breath tests. When the same calculation is

COSTS AND BENEFITS 143 made by truck size, the net cost per casualty averted for medium trucks is substantially higher than the cost for heavy trucks and buses, reflecting the significantly larger number of fatalities and injuries that could be avoided by preventing heavy-truck accidents. Casualties comprise the sum of fatalities and injuries. The savings from avoiding such casualties include the direct economic benefits from reduced medical costs, averted property damage, and prevented traffic delays. Apply- ing only a modest value to the societal benefit of saving a life, well below those values typically used in cost-benefit studies, would result in benefits that significantly outweigh the costs even at the most stringent BAC standard. NOTES The program provides funds on a matching basis: 80 percent federal and 20 percent local. According to the Truck Inventory and Use Survey, there are 3.6 million trucks that weigh more than 10,000 lb (Census Bureau 1985, 121-123). The truck is towed to a protected location; it is the responsibility of the driver to pay the towing costs and arrange for deliveiy of the load. Weigh-in-motion scales are being introduced that can weigh the trucks without stopping them as they pass over equipment on the pavement suiface. However, this type of equipment was used for only 6 percent of all vehicles weighed in 1984 (FHWA 1985, 17). Only a small percentage of the medium-weight trucks (between 10,000 and 26,000 lb) are likely to exceed the maximum weight-per-axle standard of 20,000 lb. A more extensive evaluation will involve a walk-around inspection and weighing on portable scales. Even these benefits in terms of future accident avoidance may be lessened in those crashes in which the driver of the passenger vehicle, who is not tested for alcohol use, is at fault. These same percentages are evident at 0.04 percent BAC and 0.00 percent BAC. Medium trucks (between 10,000 and 26,000 lb) represent 54 percent and large trucks (greater than 26,000 lb) and buses represent 46 percent of the total inventoty of trucks and buses. Medium trucks account for 18 percent of all truck- and bus-involved crashes, whereas heavy trucks and buses account for 82 percent of this total. Again these percentages remain relatively constant at 0.04 percent BAC and at 0.00 percent BAC. The difference here is the length of the license reinstatement period. If the commercial vehicle driver loses his license and thus his source of income for 1 year for the first infraction and for life for the second, he is unlikely to have the resources to pay the administrative costs or, in the second instance, the need for license reinstatement. The employee, however, is to be given written notice of the test 30 to 60 calendar days before the test (IBT, 3). The district also has the right to search persons, personal property, lockers, and vehicles located on all property owned, leased, or operated by the district (SCRTD 1986, 22). Only the "extra-board" drivers in the intercity bus industiy are subject to a 2-hr recall, but the majority of drivers know where they are on the recall list. The property damage figures were adjusted for fatal crashes drawing on data developed by the Bureau of Motor Carrier Safety in a special computer tabulation of property damage costs by accident severity for truck-involved crashes.

144 ZERO ALCOHOL AND OThER OPTIONS For crashes involving trucks cariying hazardous materials, delays can last up to several hours and evacuation of area residents may be required. See Appendix G for a more detailed explanation of the methodology used to derive this estimate. REFERENCES ABBREVIATIONS APTA American Public Transit Association IBT International Brotherhood of Teamsters SCRTD Southern California Rapid Transit District APTA. 1987. Survey of Rules, Policies and Activities Relating to Drug and Alcohol Abuse by Transit System Employees. Washington, D.C. Census Bureau. 1985. Truck Inventory and Use Survey, 1982 Census of Transporta- tion. U.S. Department of Commerce. FHWA. 1985. Overweight Vehicles—Penalties and Permits: Report to Congress from the Secretary of Transportation. U.S. Department of Transportation. Golob, T. F. 1985. A Preliminary Analysis of Major Freeway Accidents in Los Angeles, Orange, and Ventura Counties Involving Large Trucks, 1983-1984. Institute of Transportation Studies, University of California, Irvine. lET. National Freight Industry Negotiating Committee. Drug and Alcohol Abuse Program (undated). Jones, I. S., and A. K. Lund. 1986. Detection of Alcohol-Impaired Drivers Using a Passive Alcohol Sensor. Journal of Policy Science and Administration, Vol. 14, No. 2, pp. 153-160. Kragh, B. C., et al. 1986. Accident Costs for Highway Safety Decisionmaking. Public Roads, Vol. 50, No. 1, pp. 15-20. Latchaw, J. 1986. Do's and Don'ts on Administrative License Suspensions, and Other Issues. In Reducing Highway Crashes Through Administrative License Revocation, NHTSA, U.S. Department of Transportation, pp. 7-13. Miller, T. 1986. Benefit-Cost Analyses of Health and Safety: Conceptual and Empirical Issues. Working Paper. The Urban Institute, Washington, D.C., 27 pp. Oates, J. F. 1974. Factors Influencing Arrests for Alcohol-Related Traffic Violations. Report 801-230. NHTSA, U.S. Department of Transportation. SCRTD. 1986. Comprehensive Alcohol and Drug Abuse Policy. Los Angeles, Calif. Transport Topics. 1987. AFA Will Support Random Drug Testing of Industiy Drivers. Mar. 30, P. 1.

7 Summary Assessment Driving a heavy truck is hazardous work. The occupational death rate of heavy-truck drivers is five times greater than the average for all workers in the United States. Because the commercial vehicle driver's work takes place on public highways, other motorists share in the hazards and their share is substantial. Each year, the drivers of medium and heavy trucks and buses are involved in about 5,000 fatal crashes, the majority of which are collisions between heavy trucks (gross weight in excess of 13 tons) and other vehicles and pedestrians. About 80 percent of the 5,750 persons killed in those fatal crashes are passenger vehicle occupants, cyclists, or pedestrians rather than the commercial vehicle drivers themselves. The extent to which alcohol- impaired driving contributes to the frequency of commercial vehicle crashes is a matter of serious concern. At the request of Congress and the Secretary of Transportation, the Trans- portation Research Board appointed a study committee to examine the avail- able evidence on the alcohol problem in commercial vehicle driving and to weigh the benefits and costs of alternative blood alcohol concentration (BAC) limits. FINDINGS Alcohol and Commercial Vehicle Crashes Commercial vehicle drivers are less likely to drink and drive and are less frequently involved in crashes when under the influence of alcohol than drivers of noncommercial vehicles. About 45 percent of all drivers involved in fatal motor vehicle crashes have a measurable amount of alcohol in their bloodstreams. Alcohol is present in the blood of about 24 percent of medium- 145

146 ZERO ALCOHOL AND OThER OPTIONS truck drivers and in about 14 percent of heavy-truck drivers involved in fatal crashes. Crashes involving alcohol-impaired drivers of passenger buses are rare. Commercial vehicle drivers who had been drinking are involved in about 750 fatal crashes, 7,700 injury crashes, and about 4,750 property damage crashes each year. Risk of Crashes at Alternative BACs Ethanol, the psychoactive ingredient in alcoholic beverages, interferes with cognitive and motor responses. These effects are assumed to be increased by the amount of alcohol in the brain, which can be approximated by determining the BAC in capillary blood. Because of the variability in human size, experi- ence with alcohol, and the rate at which alcohol is metabolized, individuals reach different BACs and eliminate alcohol at different rates when given equivalent amounts. The best studies control for this variability. Although most studies have examined BACs of 0.08 percent or more (0.10 percent BAC is the legal limit in most states), some studies have demonstrated decreased performance in some driving-related physical or mental abilities at much lower BACs. Credible studies testing skills at low BACs indicate that performance on some cognitive tasks decreases significantly at BACs as low as 0.015 percent. Although individual reactions to alcohol vary depending on such factors as fatigue, motivation, previously acquired skill at the task being measured, and experience as a drinker, sensory and cognitive performance is significantly reduced at or below 0.04 percent BAC. Large-scale studies of actual automobile crashes confirm the results of laboratory tests that show performance decrements at low BACs. These studies indicate that, across broad populations of drivers, BACs exceeding about 0.04 to 0.05 percent clearly increase the probability of causing a crash. Moreover, thorough analysis of these studies indicates that when the driver's age and experience with alcohol are controlled for statistically, the risk of crash involvement increases at any recorded BAC above zero. Although alert and highly skilled professional drivers may be less affected at low BACs than other drivers, this potential advantage is offset by the demands and conditions of commercial vehicle driving. Large commercial vehicles are much more demanding to operate than automobiles; they have longer stopping distances, accelerate more slowly, require wider turning radii, and have more controls and displays. The largest vehicles on the road, such as twin trailer trucks, operate close to the design limits of the highway. The drivers of commercial vehicles typically work long hours, and alcohol contrib- utes to drowsiness. For all these reasons the effects of alcohol at low BACs on

SUMMARY ASSESSMFRT 147 drivers of trucks and buses are more hazardous than they are for drivers of automobiles. Legal and Technical Ability To Test for Low BACs Legal Ability To enforce a much lower BAC for commercial vehicle drivers, officers would need to rely more heavily on hand-held breath-screening devices to detect and measure low BACs. Use of these devices must fit within the current enforce- ment framework established in case law. Judicial interpretation of the Fourth Amendment to the Constitution—the protection against unreasonable search and seizure—has set definite guidelines for enforcing laws against driving under the influence of alcohol (DUT). Specifically, officers cannot test drivers (test is equivalent to a search) for alcohol unless they (1) have a reasonable ("articulable") suspicion that the driver is impaired and (2) can establish probable cause for arrest. As the first step in establishing reasonable suspicion, traffic officers look for behavioral evidence of driver impairment. This evidence may be manifested by erratic driving, slurred speech, or the odor of alcohol. Behavioral cues, however, are not easily detected at low BACs. Passive alcohol sensors, hand- held devices that detect the presence of alcohol in the exhaled breath of a motorist, could greatly assist investigating officers in establishing reasonable suspicion of alcohol use. Enforcement agencies in the United States, however, have been reluctant to use passive sensors without a definitive court ruling on their legality. At issue is whether the use of the sensor as an aid to establishing reasonable suspicion is actually a test that would be interpreted as a search. If it is, officers would have to rely on behavioral evidence to establish suspicion. in contrast to the passive alcohol sensor, which detects the presence of alcohol in exhaled breath, the portable breath tester provides an approximation of BAC based on a sample of deep lung air. Although not accurate enough in field use to be relied on for evidential purposes, these devices could be used to establish probable cause. A positive reading in the low BAC range would provide officers with sufficient evidence to arrest the driver and give him a formal, evidential test. About half of the states now allow use of portable breath testers in the screening phase of DIJI enforcement. Although lower courts have upheld their use, the constitutionality of portable breath testers has not been adjudicated at a definitive level. Opinions are divided regarding the constitutionality of using passive sen- sors and portable breath testers in the screening phase of DUT enforcement. The constitutional standard is clear. Officers must have a reasonable suspicion

148 ZERO ALCOHOL AND OTHER OPTIONS and a probable cause before arresting a suspect. The uncertainty regarding enforcing a low BAC standard is whether passive sensors can be used to assist officers in establishing reasonable suspicion and whether portable breath testers can be used in establishing probable cause. Without these devices the ability to enforce a lower BAC standard would be much diminished. One line of argument holds that whenever a driver is required to cooperate with a preliminary breath test, he is being searched and that without probable cause the officer cannot require this cooperation. Another line of argument holds that consent to preliminary breath testing would be viewed by the courts as a relatively minor intrusion when balanced against the government's interest in protecting the public against the risks of alcohol-impaired driving of heavy commercial vehicles. These different views cannot be reconciled in the ab- stract. The issue will only be resolved through adjudication. Assuming that the balancing argument prevails and officers are able to use breath-screening devices to establish reasonable suspicion and probable cause, drivers would still need to consent to alcohol screening. Otherwise their refusal to cooperate would stymie enforcement. In some states, Minnesota, for example, implied consent laws require drivers to cooperate with authorities as a condition of licensure. Refusal to cooperate may be penalized by license revocation. Technical Ability Although breath-screening devices are subject to some legal uncertainties, they can detect low BACs. Passive sensors can indicate the presence of alcohol and help officers decide whether further examination of a driver is proper under state and federal law. Portable breath-testing devices have been greatly refined over the decades and, when properly calibrated and used, can provide a preliminary measure of BAC. The potential accuracy of passive alcohol sensors and portable breath testers can be reduced by field conditions. Wind, adverse weather, operator inexperience, use of breath fresheners containing alcohol, and lack of cooper- ation from drivers suspected of driving under the influence can reduce the accuracy of these devices considerably. Proper handing and calibration can overcome many of these shortcomings. Refusal of the suspect to cooperate, however, can only be handled by punishing refusal to cooperate as in Min- nesota's implied consent law. Evidential breath-testing equipment must meet federal guidelines that allow for measurement error of plus or minus 0.005 percent when alcohol vapor concentration is measured under controlled conditions. In actual practice, evidential equipment tends to underestimate the BAC of the subject by 10

SUMMARY ASSESSMENF 149 percent or more, in part because of the ratio used to approximate BAC from the breath alcohol concentration. Even so, reliable measurements of breath alcohol concentrations with evidential equipment require careful adherence to the scientific procedures developed for forensic alcohol analysis. Probable Benefits of Deterrence Recent efforts at deterring driving under the influence of alcohol, which combine extensive publicity campaigns with sustained increases in enforce- ment, have demonstrated considerable and persistent reductions in alcohol- related crashes. In contrast, short-lived enforcement blitzes or statutory changes not accompanied by enhanced enforcement in the United States and abroad have sometimes shown dramatic prompt effects, but in almost all cases these effects have proved transient. Whenever threats of higher penalties or greater risk of apprehension for drinking and driving are not backed up with highly visible enforcement, drivers soon learn to ignore them. The most notable instance of increased deterrence on a large scale occurred recently in New South Wales, Australia. In this case the government widely publicized its intentions to use new procedures to apprehend impaired drivers and new procedures to punish them. The government delivered on its threat by con- ducting checkpoint-based random breath tests at an annual rate of one in three licensed drivers and by fining the driver and suspending his license for virtually every instance of a BAC of 0.05 percent or more. This program has reduced traffic fatalities in the region (which has a population of 5 million) by about one-fourth. The benefits of this effort are still too recent to classify as permanent. Nonetheless, the benefits have persisted longer than has pre- viously been seen. Some smaller-scale efforts in the United States have also produced notable benefits, although none quite so definitive as those observed in New South Wales. In successful deterrence efforts the particular BAC limits enforced are less important than the degree of vigor applied to their enforcement. The recurring element appears to be creating a widespread impression among drivers that the responsible authorities are committed to reducing the drinking-driving problem and are bringing to bear much greater enforcement resources in a circumstance where there had earlier been an impression of relative tolerance or inefficiency in detecting and sanctioning violations. Enforcement targeted at commercial vehicle drivers and patterned on the successful deterrence efforts could improve traffic safety. Estimates of the benefits that might result from alternative BAC limits for commercial drivers require extrapolating from a small and imperfect data base. The benefits estimated in this report, therefore, should be interpreted as approximations

150 ZERO ALCOHOL AND OThER OPTIONS TABLE 7-1 RANGE OF BENEFITS AND COSTS AT ALTERNATIVE BAC STANDARDS Benefits BAC Reduced Increased Public Legal Nonfatal Involvements and Private Costs Limit Lives Injuries in Property of Enforcement (%) Saved Averted Damage Crashes ($ millions) 0.10 80-140 1,100-1,800 700-1,200 27-30 0.04 110-190 1,500-2,500 1,000-1,800 37-41 Zero 130-250 1,700-3,300 1,200-2,200 49-54 and not analytical findings. Vigorously enforcing the prevailing BAC limit of 0.10 percent could avert about 80 to 140 fatalities, about 1,100 to 1,800 nonfatal injuries, and about 700 to 1,200 truck involvements in property damage crashes annually (Table 7-1). At a BAC limit of 0.04 percent, vigorous enforcement could avert roughly 110 to 190 fatalities, 1,500 to 2,500 nonfatal injuries, and 1,000 to 1,800 truck involvements in property damage crashes. The benefits of a complete prohibition against alcohol in the blood- stream of drivers (zero BAC) can only be speculated about because of the complete lack of empirical evidence for such a standard. Assuming that a zero BAC standard, vigorously enforced, might increase the deterrent effect at a 0.04 percent BAC standard by one-fourth to one-third, about 130 to 250 fatalities might be averted annually. Such an effort might also avert 1,700 to 3,300 nonfatal injuries and roughly 1,200 to 2,200 truck involvements in property damage crashes. These benefit estimates are based on the few programs that have suc- cessfully deterred drinking and driving. The high end of the range of benefits depends on the assumption that the authorities would be able to rely on breath- screening devices to enforce low BAC standards. Without these devices, enforcement would be less efficient, particularly at the lowest BACs. To account for this reduced efficiency without breath-screening devices, the minimum benefits (represented by the low end of the range) are estimated at 0.10 percent to be roughly 65 percent of the maximum benefits; at 0.04 percent, 60 percent; and at zero BAC, 55 percent (Table 7-1). Strategy and Costs of Enforcing Alternative BACs A strategy for deterring drinking and driving by commercial vehicle drivers requires three basic elements: (1) opportunities to screen large numbers of drivers, (2) extensive publicity, and (3) a means of meting out punishment with dispatch. Driver screening could take place as part of ongoing vehicle

SUMMARY ASSESSMEWF 151 safety inspections and at vehicle weigh stations. Sufficient opportunities for contact with drivers exist in these situations to be within the range of driver contacts achieved in successful deterrence efforts. These efforts could be supplemented with more vigorous testing for alcohol following crashes in- volving commercial vehicle drivers. A large-scale public information cam- paign could communicate to drivers the increased effort to detect and punish driving while under the influence of alcohol. Finally, the threat could be delivered on with reasonable speed by adopting an administrative license revocation process handled by the state Department of Motor Vehicles rather than having the cases handled by civil or criminal courts. With administrative license revocation, already adopted by many states as part of ongoing DUI enforcement, the driver's license is revoked by the Department of Motor Vehicles upon notice that a driver has been arrested and has tested higher than the existing state BAC limit. Although the framework for a successful program exists, past efforts at deterring drinking and driving by the general public indicate that it is neither inexpensive nor easy to keep people from driving after consuming alcohol. In order to deter driving under the influence of alcohol by commercial vehicle drivers under the prevailing BAC limit of 0.10 percent, the overall cost would have to increase by an additional $30 million annually (Table 7-1). The largest share of increased cost (51 percent) results from the need for additional personnel with adequate training and equipment to carry out their charge. An additional 21 percent of the increased costs results from the administrative review, hearings, and expected appeals process. An ongoing public informa- tion campaign would cost about $5 million annually. About 11 percent of the total costs are economic costs associated with driver delays. Lowering the BAC limit to 0.04 percent would increase the total cost to $41 million. This estimate assumes that passive sensors and portable breath testers could be used in the screening phase of enforcement. In addition to increasing the deterrence effect, use of these devices compared with reliance on be- havioral tests lowers the time required by the investigating officer for each screening. The increased equipment needs, however, more than offset these time savings. Without passive sensors and portable breath testers the cost would be roughly $3 million less at each BAC standard. Reducing the BAC limit to zero increases the total cost by $54 million. The costs of enforcing lower BACs do not increase much over the cost of enforcing 0.10 percent BAC because simply gearing up for enforcement—allocating personnel to the task and training and equipping them—makes up the bulk of the cost. Most of the increase at lower BACs is attributable to the increased number of drivers who would test positive for alcohol and then be handled through the admin- istrative system and, when cases are appealed, through the legal system.

152 ZERO ALCOHOL AND OTHER OPTIONS Comparison of Alternative BACs The benefits of enforcing lower BACs appear justified when total social costs are compared with total social benefits. For example, the net cost of each BAC option per casualty averted ranges from a low of roughly $4,000, assuming maximum benefits, to a high of $14,000, assuming the minimum benefit at zero BAC (Table 7-2). Casualties averted is simply the sum of lives saved and injuries averted as estimated in Table 7-1. Total direct economic benefits include the savings from reduced medical costs, reduced property damage, and reduced motorist delay. The public and private costs associated with enforcing the alternative BAC limits are almost offset by directly quantifiable benefits alone. The estimate of net costs per casualty averted results from dividing the difference between public and private costs and economic bene- fits by the casualties averted. At any BAC limit the direct costs to society are low. Applying the lowest estimates of the value of life and the value of reduced pain and suffering to the numbers of casualties averted would make the benefits greatly outweigh the costs. TABLE 7-2 COST PER CASUALTY AVERTED AT ALTERNATIVE BAC STANDARDS BAC Net Costa Legal Total Economic Public and per Casualty Limit Casualties Benefits Private Costs Averted (%) Averted ($ millions) ($ millions) ($ thousands) 0.10 1,200-1,900 15-23 27-30 4-10 0.04 1,600-2,700 20-32 37-41 4-10 Zero 1,800-3,400 23-42 49-54 4-14 a Public and private costs less total economic benefits divided by casualties averted The decision about the appropriate BAC standard, however, turns on more than a narrow comparison of quantifiable benefits and costs. A number of nonquantifiable issues influence the ultimate policy decision, including the potential limiting of constitutional rights, the severity of the mandated penalty, the introduction of a dual BAC standard for highway driving, the effects of a lower standard on industry work schedules, and the commitment of state and local authorities to enforcing the new BAC limit. The uncertainties of future legal rulings on hand-held breath testers could be avoided by retaining the prevailing BAC limit in state law, but vigorously enforcing it. This policy could be enfoited by relying on behavioral cues without potentially narrowing the constitutional rights of commercial vehicle drivers. As an additional advantage, maintaining the status quo (but actually enforcing it) would retain the simplicity of a single BAC limit for all drivers rather than a dual system.

SUMMARY ASSESSMENT 153 Retaining 0.10 percent BAC, however, ignores the considerable risk to the public of tolerating driving by commercial vehicle drivers while under the influence of this concentration of alcohol. In actual practice, the BAC limits enforced are higher than those specified in law. Officers want to ensure that any case they bring to court will result in conviction. As a result they typically only arrest drivers with BACs at least 0.01 to 0.04 over the legal limit. Lowering the BAC limit to 0.04 percent reduces the risk to the public and would make U.S. Department of Transportation BAC regulations for airline crews, railroad engineers, and commercial vehicle drivers uniform. But this lower BAC may introduce other problems. A much reduced BAC may complicate industry work scheduling. Many drivers in commercial truck and bus transportation are covered by a 2-hr recall provision. A 150-lb man who had consumed three beers just before being recalled could technically be subject to loss of his license for one year if he climbs into a cab within 2 hr. Although most drivers have a reasonable idea of when they might be recalled, not all do, and some companies do not look favorably on a driver who does not report when called. One way to solve this problem is to allow drivers to take a self-administered breath test upon reporting to work and to prohibit companies from penalizing them for electing not to drive. Given the tolerance built into enforcement of DUI, a 0.04 percent BAC standard might result in an enforced standard of 0.06 to 0.08 percent. From a safety perspective such a standard would still expose the public to consider- able risk. A zero BAC limit eliminates the image that the consumption of even small amounts of alcohol is acceptable for drivers of large vehicles and recognizes the impairing effects of hangovers. A driver on the road after an evening of heavy drinking might have a low BAC but still be significantly impaired. Although a zero BAC standard raises the risk that drivers with trace amounts of alcohol in their blood from medicines or a single drink taken an hour or so before driving might be punished, as a practical matter such an outcome is unlikely. Most evidential tests are not taken until about 1 hr after the driver has been stopped, at which time trace amounts of alcohol would have been metabolized. In addition, given the tolerance built into enforcement of any BAC limit to account for such factors as measurement error, the zero BAC limit would probably result in an enforced limit of 0.02 to 0.04 percent. Officers are probably no more likely to arrest drivers with preliminary BAC readings just above zero than they are to arrest drivers just above the current limits. Along with the advantages, a zero BAC introduces the same potential problems as the 0.04 percent BAC standard. Driver work scheduling would become more complicated. The need for hand-held screening devices would

154 ZERO ALCOHOL AND OTHER OPTIONS be even greater. The primary question about adopting a zero BAC is whether the enforcement and legal systems would support it given that the risk of a crash at very low BACs is low and the consequences for the driver are quite severe. RECOMMENDATIONS The Commercial Motor Vehicle Safety Act of 1986 requires the Secretary of Transportation to establish a BAC standard for commercial vehicle drivers and mandates a penalty for violating the newly established BAC limit as license revocation for one year on the first offense and permanent loss of the commercial license on the second offense. The study committee believes that any consumption of alcohol on the job by commercial vehicle drivers is inappropriate for the workplace and incom- patible with traffic safety. The majority of the committee (three-fourths) recommends that the penalties required by the Commercial Motor Vehicle Safety Act be applied to violations of 0.04 percent BAC. Moreover, consistent with the principle that alcohol consumption is inappropriate for the workplace, the majority favors reducing the BAC limit for commercial vehicle drivers to zero, but also recommends that a lesser penalty than that required by the Commercial Motor Vehicle Safety Act be applied to DUT violations below 0.04 percent. A penalty of license revocation for 24 hr to 30 days is recom- mended for drivers detected with BACs greater than 0.01 but less than 0.04 percent (use of 0.01 for the lower end of the range would account for measurement error). The driver should also be referred to a competent au- thority to determine whether he has an alcohol problem, and if so, should receive treatment. In addition, the violation should be placed on the driver's record. This recommendation parallels the current federal regulation that prohibits interstate commercial vehicle drivers from drinking on the job and within 4 hr of reporting to work. The penalty for violating the current federal regulation is for the driver to be placed out of service for 24 hr. In contrast to this penalty, the majority of the committee recommends that some action on the license be taken to ensure that a record is made of the violation and that the driver is referred for screening and subsequent treatment. On the second and subsequent offense of driving with a BAC greater than zero but less than 0.04 percent, the period for loss of license should range from 30 days to one year. Through the broad regulatory power of the office, the Secretary of Transporta- tion could set the BAC standard at zero with these lower penalties and still require the states to adopt the penalties mandated by the Commercial Motor Vehicle Safety Act for violations of 0.04 percent BAC. The majority of the committee favors setting an explicit policy of zero BAC because it provides an unequivocally clear message to drivers that alcohol in

SUMMARY ASSFSSMFRT 155 the bloodstream—whether from a beer with lunch or because of a hangover— is incompatible with the safe operation of commercial vehicles. A zero limit enforced by the public sector would also strengthen the policies of many private companies that strictly prohibit alcohol consumption on the job or before reporting to work. A minority of the committee members believes that the existing state per se limits (perhaps reduced to 0.08 percent BAC) are appropriate when the penalties associated with violation are those mandated by the Commercial Motor Vehicle Safety Act. These members also believe that it will be very difficult to establish probable cause at low BACs. The steps needed to establish probable cause at low BACs, facilitated by use of portable breath testers to screen drivers, could be interpreted as a search under the Fourth Amendment and could therefore be ruled unconstitutional. Without the use of portable breath testers and passive sensors, few drivers with BACs below 0.08 would be detected and the deterrent effect would be small. The majority of the committee, however, believes that the courts will recognize the government's interest in protecting innocent drivers and permit use of portable breath testers for screening. The entire committee recommends that the BAC standard decided on by the Secretary of Transportation apply to drivers of commercial vehicles weighing 26,000 lb or more, drivers of passenger buses, and drivers of vehicles hauling hazardous materials. The estimated benefits and costs of the zero BAC policy for drivers of buses and trucks weighing 26,000 lb or more are summarized in Table 7-3. The Federal Highway Administration has already ruled that the uniform license provisions of the Commercial Motor Vehicle Safety Act will not apply to drivers of commercial vehicles weighing between 10,000 and 26,000 lb. The federal government and the states have a considerable task ahead of them to develop and coordinate the application of these new regula- tions to drivers of heavy trucks and passenger buses. Once that process is well under way, the committee recommends reducing the BAC limit, as suggested earlier, for drivers of commercial vehicles weighing less than 26,000 lb. If the Secretary adopts a lower BAC standard, some additional steps are necessary to ensure a successful policy. The U.S. Department of Transporta- tion should develop support for enforcing a lower BAC limit from state and local authorities and develop a public information campaign to communicate the new policy to commercial vehicle drivers. To assist in deterrence, the states should handle violations of the adopted BAC limit by revoking driver licenses through an administrative process with appropriate protection of due process. If the Secretary adopts a BAC limit below 0.05 percent, the states will also need to repeal that section of their presumptive laws under which a person testing below 0.05 percent BAC is presumed not to be under the influence of alcohol. Those states that have not already adopted the language

156 ZERO ALCOHOL AND OThER OPTIONS TABLE 7-3 ESTIMATED BENEFITS AND COSTS OF ZERO BAC STANDARD (EXCLUDING MEDIUM TRUCKS) Category Annual Effect Lives saved 120-220 Nonfatal injuries averted 1,400-2,500 Medical savings ($ thousands) 9,000-16,000 Property damage savings ($ thousands) 7,000-12,000 Delay savings ($ thousands) 3,000-6,000 Public and private enforcement costs ($ thousands) 34,000-38,000 recommended in the Uniform Vehicle Code that defines per se limits in terms of both blood alcohol concentration and breath alcohol concentration should be urged to do so. The Secretary should also consider adopting additional steps to facilitate enforcement. For example, it could be required, by regulation, that as a condition of licensure, drivers consent to being screened with hand-held breath-testing devices to assist in establishing reasonable suspicion and proba- ble cause. Such a requirement would assist officers in those cases when a driver refuses to cooperate in the screening phase. The refusal of a driver to submit to screening would be permitted as evidence in establishing probable cause. Drivers might also be required to consent to an evidential test once probable cause has been established. In addition, in order to maximize the deterrence benefits, the Secretary could urge the 25 states not currently allowing the use of portable breath testers in the screening phase to pass enabling legislation. Adoption of lower BAC limits and experimentation with different enforce- ment strategies and sanctions should be carefully evaluated. This evaluation should be carried out by independent researchers. To improve the quality of data for such an evaluation, the Department of Transportation should continue to emphasize the importance of reporting BACs of commercial vehicle drivers involved in crashes and should support research to determine the incidence of drinking and driving by commercial vehicle drivers. Experience with the costs of alternative BAC limits and estimates of their deterrent effects will provide the best basis for deciding whether the BAC limits and sanctioning patterns recommended in this report are appropriate or should be adjusted.

Appendix A Alcohol Involvement in Fatal Truck Crashes Past efforts to estimate alcohol involvement in fatal crashes have relied on the 15 states reporting the highest percentage of BACs. The approach of using data from states with good alcohol reporting is reasonable, yet may introduce reporting biases. Within the "good" states the fatally injured drivers who were tested for alcohol do not necessarily represent a random sample of drivers within those states. That is, alcohol testing is more likely to occur for TABLE A-I STATES WITH MOST COMPLETE REPORTING OF ALCOHOL USE BY FATALLY INJURED DRIVERS, 1984 (NHTSA 1986) No. of Percent with Known State Fatally Injured Drivers BAC Test Result California 2,856 89.5 Colorado 358 87.7 Delaware 75 90.7 District of Columbia 23 95.7 Hawaii 73 91.4 Illinois 863 82.9 Maine 138 89.1 Minnesota 382 87.4 Montana 154 80.5 Nevada 126 92.8 New Jersey 471 84.7 New Mexico 256 87.9 North Carolina 813 80.9 Oregon 325 91.7 Rhode Island 46 93.4 Vermont 68 91.2 Virginia 584 86.5 Washington 448 91.5 West Virginia 280 81.1 Wisconsin 504 82.5 Wyoming 103 83.5 Total 8,946 157

158 those types of accidents where previous experience indicates possible alcohol use. For example, rural, nighttime, and single-vehicle accidents were over- represented in the 15 states with good reporting in the 1982 FARS data (Voas 1984). The net result of this bias is overestimation of the alcohol involvement across all accident types. For the purpose of this report, 1984 FARS data were examined to determine whether biases were present. Twenty states and the District of Columbia were found to have relatively complete reporting (Table A-i), a substantial increase over the 15 such states in 1982 (NHTSA 1986). To identify biases in report- ing, the relative number of accidents by subgroup for each of several known BAC correlates—accident characteristics known to be related to positive BAC—were compared (Table A-2). For the general population of drivers several observations may be made. The good states are similar to the poor TABLE A-2 COMPARISON OF BAC CORRELATES AMONG SUBGROUPS OF DRIVERS (NHTSA 1986) Percent Correlate (%) Group and No. of of Single Subgroup Drivers Total Night Vehicle Weekend Rural Male Young° All drivers Good reporting states (21) 20,056 35 40N 42 52N 53 77N 62 Poor reporting states (30) 37,506 65 40 40 52 58 78 61 Fatally injured drivers 25,631 45 44 50 52 63 79 62N Surviving drivers 31,931 55 37 34 51 51 77 61 Drivers with known BAC 21,870 38 49 49 55 60 81 67 Drivers with unknown BAC 35,692 62 35 36 50 54 76 58 Truck driversb Good reporting states 1,658 31 27N 25N 34N 66N 96N 49N Poor reporting states 3,696 69 28 23 35 68 97 48 Fatally injured drivers 2,112 39 28N 67 34N 73 97N 46 Surviving drivers 3,242 61 27 14 35 64 97 49 Drivers with known BAC 1,099 21 33 43 36N 75 95N 48N Drivers with unknown BAC 4,255 79 26 18 35 66 97 48 NOTE: N = not statistically different at the 0.01 confidence level using the chi-square test. All other pairs are statistically different. "Young" drivers are under the age of 35. b Trucks are defined as all medium and heavy trucks.

159 TABLE A-3 COMPARISON OF BAC CORRELATES FOR PREDICTOR SUBGROUPS AND TARGET SUBGROUPS (NHTSA 1986) Percent Correlate (%) No. of of Single Subgroup Drivers Total Night Vehicle Weekend Rural Male Younga Predictor sample: fatally injured drivers with known BAC Good states (21) 7,679 13 46 52 54 59 79 65 All states 16,011 28 48 52 54 62 80 64 Extrapolation sample: known and unknown BACs Good states (21) Fatally injured drivers 8,855 15 44 51 53 59 79 63 All drivers 20,056 35 40 40 52 58 78 61 All states Fatally injured drivers 25,631 45 44 50 52 63 79 62 All drivers 57,562 100 40 41 52 56 78 61 a "Young" drivers are under the age of 35. ones with the exception that rural accidents are more prevalent in the poor states. Fatally injured drivers vary significantly from surviving drivers on all correlates except age. Likewise, drivers with known BACs are more likely to have experienced the types of accidents specified by the conditions. For truck drivers, the variations for fatally injured versus surviving and known versus unknown BAC are fewer in number than those for the general population. The data in Table A-3 carry the analysis a step further. Two samples have been identified that are useful for predicting unknown BACs in fatal acci- dents: fatally injured drivers with known BACs for the 21 good states and those for all states. By comparing these first two rows with the others, it is possible to see the direction and magnitude of oversampling on the basis of the BAC correlates. Although both predictor samples are reasonably close, they are overrepresentative of the entire population of drivers (row 6). For example, 40 percent of all fatal accidents occurred at night compared with 46 percent in the sample of fatally injured drivers with known BACs from 21 good states. Thus, predictions from these two samples will result in an overestimation of alcohol involvement in fatal crashes unless corrected for in some manner.

TABLE A-4 ALLOCATION OF FATALLY INJURED TRUCK DRIVERS WITH UNKNOWN BAC LEVELS (NHTSA 1986) Percentage by Correlate Single Vehicle Multiple Vehicle Urban Rural Urban Rural Day Night Day Night Day Night Day Night BAC (N = 41) (N = 32) (N = 203) (D = 115) (N = 32) (N = II) (N = 88) (N = 43) Level (%) (U = 260) (U = 74) (U = 289) (U = 145) (U = 833) (U = 313) (U = 1,824) (U = 597) No alcohol 81 50 81 69 94 27 88 72 0.01-0.03 7 0 3 4 0 0 2 2 0.04-0.09 2 9 3 6 3 18 1 2 ~:0.10 10 41 13 21 4 55 9 24 NOTE: U = unknowns to be allocated (all seventies).

161 METHODOLOGY A simplified procedure for compensating for alcohol oversampling was de- veloped as part of this study. Because there is always some reporting of BACs in all states, even among surviving drivers, the approach was to accept all the known BAC values and to allocate the unknown values by comparing their accident characteristics against a predictor sample, which in this case was chosen to be fatally injured drivers with known BACs in all states. The decision to use all states instead of the 21 states with good alcohol coverage was based solely on sample size considerations. The predictor sample was divided into drivers of personal vehicles and of medium and heavy trucks. Each of these was further subdivided in a hierarchi- cal fashion based on selected alcohol correlates. Number of vehicles in crash, land use, time of day, sex, age, and day of week were used for personal vehicles, whereas number of vehicles, land use, and time of day were used for trucks. Within each of the final levels the distribution of BACs was calculated; these percentages form the basis for allocating the unknowns. To do so, the drivers with unknown BACs were subdivided similarly and the percentages were applied to the number of drivers in each of the levels (Table A-4). A potential drawback of this method is the sparseness of data in some of the predictor cells. For example, urban nighttime multiple-vehicle truck accidents were allocated on the basis of a total of 11 observations. However, sample size problems would plague other statistical approaches as well (e.g., logistic regression or log-linear models) and the current method has the advantage of simplicity. RESULTS Alcohol use by all drivers involved in fatal crashes has been decreasing since 1980 (Voas 1984; Fell 1985) and is summarized in Table A-5. The method used in this report verifies that alcohol use is substantially less than it was in TABLE A-5 PERCENTAGE OF ALCOHOL INVOLVEMENT FOR ALL DRIVERS IN FATAL CRASHES, 1980-1984 (Voas 1984; Fell 1985) "Good State" Extrapolationa Current Methodc BAC Level (%) 1980 1981 1982 1983 1984 1984L 1984 ~t0.0l 62 60 59 57 54 54 51 ~0.10 50 49 49 43 43 44 40 a As reported by Fell (1985) and in FARS data (NHTSA 1986). b Extrapolated from 21 states shown in Table A-i. C Includes drivers of all vehicles, not just personal and medium and heavy trucks.

162 TABLE A-6 PERCENTAGE OF ALCOHOL USE IN FATAL CRASHES BY DRIVERS OF PERSONAL VEHICLES AND TRUCKS, 1984 Personal Vehicles Medium and Heavy Trucks "Good State" Current "Good State" Current BAC Level (%) Extrapolation' Method Extrapolationa Method No alcohol 45.5 47.3 78.3 80.1 0.01-0.03 2.9 1 3.5 2.8 1 2.3 0.04-0.09 7.0 54.5 7.5 52.7 2.4 21.7 3.6 19.9 J ~ 0.10 4.4.7 J 41.7 J 16.5 j 14.0 a Based on extrapolation of results for fatally injured drivers in the 21 states with good alcohol coverage. TABLE A-7 PERCENTAGE OF ALCOHOL USE AMONG DRIVERS OF MEDIUM AND HEAVY TRUCKS IN FATAL CRASHES, 1984 BAC Level (%) Medium-Truck Drivers No. Percent Heavy-Truck Drivers No. Percent No alcohol 469 69.8 3,661 83.9 0.01-0.03 5 0.7 1 118 2.7 1 0.04-0.09 52 7.7 30.2 112 2.6 16.1 0.10 146 21.8 J 471 10.8 j Total . 672 4,362 NOTE: Medium trucks are those that weigh between 10,000 and 26,000 lb. Heavy trucks are those that weigh more than 26,000 lb and truck-tractors. Not included in this tabulation were 318 drivers of medium and heavy trucks of unknown size and weight. 1980 (51 percent versus 62 percent). Also, when compared with the approach of extrapolating data from good reporting states, the current method yields an estimate of alcohol use that is 3 percent lower. This difference is due to the oversampling effect previously discussed, which, however, is relatively small in magnitude and probably less than the error present in both methods. On the basis of the methodology developed for this report, the overall effect of alcohol oversampling is small, at least in the 1984 FARS data. Compensating for it produces only a minor decrease in the estimate of alcohol use from 54 percent to 51 percent for BACs ~ 0.01 percent. Further, when the data are segregated by vehicle type, the difference in the methods is on the order of 2 percent for both personal vehicles and medium and heavy trucks (Table A-6). Apparently, oversampling is greater for other types of vehicles (e.g., buses, motorcycles, recreational vehicles) than it is for personal vehicles and trucks. Application of the current methodology to the 1984 FARS data file also revealed that drivers of medium and heavy trucks in fatal crashes were far less likely to have used alcohol than drivers of personal vehicles; alcohol was

163 detected in an estimated 53 percent of personal vehicle drivers as compared with 20 percent for truck drivers (Table A-6). If the data are further segregated by truck type, it is revealed that drivers of heavy trucks are nearly half as likely to have used alcohol before a fatal crash than drivers of medium trucks. Although 30 percent of the 672 medium-truck drivers had positive BACs, only 16 percent of the 4,362 heavy-truck drivers did (Table A-7). The difference might be explained by the nature of the driving (long haul for heavy trucks, short haul for medium trucks) and the drivers (more experienced and better-trained drivers with better driving records). REFERENCES Fell, J. C. 1985. Tracking the Alcohol Involvement Problem in HighwayCrashes. In Alcohol and Highway Safety 1984: A Review of the State of the Knowledge, Report DOT-HS-806-598, U.S. Department of Transportation. NHTSA. 1986. Fatal Accident Reporting System: 1984. U.S. Department of Transportation. Was, R.B. 1984. Estimating Alcohol Involvement in Fatal Crashes: A Note on the Reporting of BAC in the FARS. Abstracts and Reviews in Alcohol and Driving, Vol. 5, No. 1.

Appendix B . FARS and Texas Data on Fatalities for Drivers of Medium and Heavy Trucks Olga J. Pendleton In Table B-i the peitentage of known BACs for fatally injured drivers of medium and heavy vehicles is given for each BAC limit (0, ~:0.04, 2t0.10). The percentages are based on the Fatal Accident Reporting System (FARS) data for 1982-1985. For the purposes of this study, medium vehicles were defined as BODYTYPE = 70, 71, or 75 and heavy vehicles were defined as BODYTYPE = 72, 74, or 76, according to FARS format. The total known BACs are listed in parentheses. Thus, in 1982 there were 61 known BACs for medium-vehicle driver fatalities and 68.9 percent of these drivers (or 42) had zero BAC, 26 percent had BAC 2: 0.04, and 21.3 percent had BAC ~t 0.10. The percentages of unknown data are specified in the last row of the table. Note that during the 4-year period, 83 percent of the 16,030 fatalities for medium and heavy vehicles have missing BAC values. Thus, percentages in Table B-i represent approximately 17 percent of such fatalities. The extent of the missing data makes statistical inference virtually impossible and is too great to allow for estimation of the missing values. Statistical models based on only 17 percent of the data have very little potential for providing reliable estimates of the missing data. The distribution of known BACs creates a further complication for statistical inference. Although the frequency distribution above about 0.04 percent BAC approaches a normal distribution, the greatest frequency occurs at 0.02 percent BAC (Figure B-i). This distribution suggests that the role of alcohol in commercial vehicle fatal crashes is quite different from the role that alcohol plays in noncommercial vehicle fatal crashes. 164

TABLE B-I PERCENTAGE OF KNOWN AND UNKNOWN BACs FOR FATALLY INJURED DRIVERS BY VEHICLE TYPE AND BAC CATEGORY Truck Type BAC Limit (%) 1982 Percent Total No. 1983 Percent Total No. 1984 Percent Total No. 1985 Percent Total No. All Years Percent Total No. Known BACs Medium trucks 0.0 68.9 1 64.5 1 68.4 1 71.1 1 67.6 1 0.04 26.0 61 34.0 62 30.5 79 27.6 90 30.2 292 0.10 21.3 J 29.0 j 26.6 j 22.2 J 25.1 Heavy trucks 0.0 73.3 83.6 1 83.3 1 88.3 1 84.6 1 0.04 12.6 590 11.3 706 13.0 825 9.2 989 11.4 3,110 0.10 10.0 j 10.6 J 9.6 j 7.8 j 9.8 Medium and heavy trucks 0.0 79.7 1 82.0 1 82.0 1 86.8 1 83.2 1 0.04 15.0 651 13.6 768 14.0 904 10.8 1,079 12.9 3,402 0.10 12.0 J 12.1 J 11.0 J 9.0 J 11.0 Unknown BACs 83.0 3,913 81.0 4,006 78.0 4,129 73.0 3,982 83.01 16,030

irc These data must therefore be used in a purely descriptive sense in recogni- tion of the limitations and potential bias of the estimates. The estimates tend to show a certain degree of consistency from year to year. It would appear that the proportion of driver fatalities with BAC ~! 0.10 is somewhat higher for 100 90 80 70 60 0 z w 50 U. 40 30 20 10 0 j.:.>:.:.i r:.:-Y.I i:.<I . 0 .02 .04 .06 .08 .10 .12 .14 .16 .18 .20 .22 .24 .26 .28 .30 .32 .34 .36 .38 .40 BLOOD ALCOHOL CONCENTRATION FIGURE B-i Frequency of fatalities for drivers of medium and heavy vehicles by BAC. medium-vehicle drivers (25 percent) than for heavy-vehicle drivers (10 per- cent). The proportion with BAC greater than 0.04 would appear to contribute a relatively small amount to this proportion (3.0 and ii percent for medium and heavy vehicles, respectively). If the BAC value is zero, the proportions are 32 and 16 percent, respectively. /

167 TABLE B-2 TEXAS DPS-REPORTED ALCOHOL INVOLVEMENT FOR DRIVERS OF MEDIUM AND HEAVY VEHICLES, 1980-1986 Vehicle Type Medium and Heavy All Others Crash Type Percent No. Percent No. Fatal 7.3 38 32.0 7,882 Incapacitating 6.0 2,015 24.0 34,815 Nonincapacitating 4.6 4,565 22.0 65,080 Possible injury 3.0 3,332 12.0 44,381 Noninjury 1.2 40,906 8.4 352,042 All 8.4 352,040 12.0 504,201 In Table B-2 the percentage of alcohol-involved accidents is listed by type of injury for medium- and heavy-vehicle drivers and all others for 1980-1986. These figures are based on accidents reported by the Texas Department of Public Safety (DPS) only. Alcohol involvement in driver fatalities is only 32 percent according to Table B-2. Thus, these estimates may be underestimates for all other categories as well. The 7.3 percent of alcohol involvement for medium and heavy vehicles is, in fact, lower than the ii percent based on the FA.RS data (Table B-i).

Appendix C Description and Costs of Testing Techniques Breath-screening devices of various types assist officers in detecting the presence of alcohol. Behavioral sobriety tests are more commonly used to determine whether a driver is sufficiently impaired by alcohol to be arrested. These two different types of field tests are discussed in the following section. The subsequent section gives the costs and training requirements of testing methods. Information on costs and training requirements was provided by Ron Engle of the National Highway Traffic Safety Administration and by Robert Voas. DESCRIPTION OF TESTING TECHNIQUES Behavioral Sobriety Tests When an officer thinks that a driver may be intoxicated but wants to test the person before making an arrest, a field sobriety test is the likely next step. Burns and Moskowitz (1977) established three tests as the most effective after evaluating six of the most commonly used behavioral sobriety tests. These three tests are horizontal gaze nystagmus, walk and turn, and the one-leg stand. Horizontal Gaze Nystagmus In this simple procedure, a suspect fixes his eyes on an object, such as a pencil, held in front of his face. His eyes will make an involuntary jerking movement 168

169 (called nystagmus) as they follow the side-to-side movement of the pencil. The angle of nystagmus for an intoxicated person is reduced from 55 degrees to 40 degrees. Besides alcohol, other drugs, generally central nervous system depressants, also cause this jerking motion at an earlier angle. The gaze nystagmus test is relatively nonintrusive. It can be performed while the driver remains seated in the car and only requires about 40 sec. Walk and Turn The walk-and-turn test is probably the most commonly used sobriety test. The suspect takes nine heel-to-toe steps along a line, turns on the heel of his foot, and returns to the starting point in the same manner. The suspect first assumes the heel-to-toe position, and the officer gives instructions once he has started to walk. This test provides a measure of divided attention, described in Chapter 3, because the suspect must concentrate on two activities, his walking and the officer's instructions. One-Leg Stand In the one-leg stand, the suspect stands on one foot, holding the other leg stiffly in front of him with his eyes focused on his raised foot and counts for 30 sec while maintaining this position. The walk-and-turn and the one-leg-stand tests measure a person's balance and muscular coordination. Both tests provide a measure of divided attention. Methods of Preliminary Breath Testing Breath testing by having a suspect inflate a plastic bag through a test tube has been used for many years. Alcohol in the breath causes chemicals coating the tube to change from yellow to green. This device is small, lightweight, and inexpensive. Borkenstein's breathalyzer uses the oxidation-photometric method. Alcohol exhaled in the breath passes through a chemical solution. Chemical changes caused by alcohol are measured by a photocell to indicate BAC. The electrochemical cell technique measures alcohol in exhaled breath as it oxidizes in a fuel cell. Electric current flowing from anode to cathode causes oxidation. Measuring this reaction indicates the amount of alcohol present. This method requires that the breath sample be deep lung (alveolar) air, which requires that the test be conducted properly.

170 Semiconductor sensors use metal oxide as the conductor, which shows significant increases in conductivity in the presence of reducing gases, in this case alcohol. The sensor can be used with many repetitions without error but is sensitive to temperature and must be recalibrated weekly. Passive alcohol sensors test normally exhaled air in front of a person's mouth. These devices became available with the development of microchip technology. There are currently two types of passive devices, one based on semiconductor technology and the other on the fuel cell. The Insurance Institute for Highway Safety recently developed a passive device incorporat- ing an electrochemical fuel cell sensor inside a flashlight (Jones 1986). A pump built into the device draws air from in front of the person and passes the air over the sensor. Saliva-Alcohol Testing Saliva testing became unpopular many years ago due to the rise of breath techniques and the requirement of large quantities of saliva for testing. The test was time consuming and in some cases infeasible given the condition of the suspect. Current methods require much smaller quantities of saliva. It will soon become practical to use solidstate test strips for direct saliva-alcohol detection. "The era of 'touch and tell' has not quite arrived for strip tests for saliva-alcohol analysis, but it is near" (Dubowski 1986, 32). Evidential Methods of Testing Devices for collecting evidence are generally used in police stations or laboratories, but are portable if placed in vans equipped with a special power supply. Gas chromatography is the best method for alcohol analysis (Dubowski 1986). Although useful with a variety of biological specimens, most police stations have machines set up for breath analysis. Current models come as small as a desktop copier. In this technique a small breath sample is introduced into a stream of carrier gas and fed into a heated column. This column is filled with material that has different retention times for the different constituents of the gas sample (Schmidt 1980). Gas components are analyzed according to the time needed to move through the column. In the infrared absorption technique a breath sample is introduced into an optical cell and the transmission of an infrared light beam is measured. Changes in transmission are proportional to the concentration of alcohol in the

171 sample. After the optical cell has been purged with alcohol-free air, test cycles of less than 1 min are possible. Laboratory tests and devices used in police stations are quite accurate, even at low BACs. But for these devices to be used, an officer has to have some reason to arrest a suspect. COSTS AND TRAINING REQUIREMENTS A portable breath tester costs approximately $400, with discounts available when purchased in quantity. The equipment used at police stations costs about $4,000. Portable testers that can be used for evidential purposes cost about $1,200 with an attached printer. Passive alcohol sensors cost between $500 and $675. In principle, officers trained to use a portable breath tester should not need extra training to accurately measure lower BACs. In reality, refresher courses will likely be needed. The key to using the hand-held breath tester is the reaction time of the operator in pushing the "read" button on the sensor to get deep lung air. Some devices have a light to indicate when deep lung air enters the instrument, whereas others rely on the officer to push the read button after a certain number of seconds. Between one-half and one full day of training for the new user of the portable breath testers should be sufficient, and a brief refresher course for those already trained is adequate. Passive alcohol sensors require a minimum of one-half day of training plus field supervision for an officer to be prepared to use this device. Most officers tend to hold the device farther than 6 in. from the driver's mouth, as reported in field studies of the passive sensor. Field supervision can easily correct this tendency. Studies on field sobriety tests, especially the three-test battery, indicate that a minimum of 1 day and an average of 3 days of training is required before these techniques can be applied. The majority of this time is generally spent on the horizontal gaze nystagmus test, because it is new to most officers and has proven to be the most effective of the behavioral tests. The cost of training will depend on the enforcement strategy of a particular agency or department. Many departments have only one trained "Tiger Team," which consists of about 10 percent of the force. These officers are specially trained to staff roadside sobriety checks and patrol for impaired drivers during peak drinking hours, such as Friday and Saturday nights. REFERENCES Bums, M., and H. Moskowitz. 1977. Psychophysical Tests for DWI Arrests. U.S. Department of Transportation.

172 Dubowski, K. M. 1986. Recent Developments in Alcohol Analysis. In Alcohol, Drugs, and Driving: Abstracts and Reviews, Vol. 2, No. 2, pp. 13-46. Jones, I. S. 1986. The Development and Evaluation of a Passive Alcohol Sensor. In Proceedings of the 30th Annual Conference, American Association for Automotive Medicine, Montreal, Quebec. Schmidt, M. 1980. Comparison of Different Methods for the Determination of Breath Alcohol. In Alcohol, Drugs and Traffic Safety (L. Goldberg, ed.), Almqvist och Wiskell International, Stockholm, Sweden, Vol. 2, pp. 686-690.

Appendix D Estimating the Costs of a Comprehensive Commercial Safety Enforcement Option The methodology used to derive the cost estimate for the comprehensive commercial safety enforcement option is discussed in more detail in this appendix. There are four types of costs associated with a successful enforce- ment program: those related to alcohol testing, those of the sanctioning process, those of a publicity campaign, and economic costs associated with delay created by the enforcement process. A more complete explanation of the last type of cost is contained in Appendix E. ALCOHOL TESTING Costs are based on a total of 1,833,000 opportunities for contact with commer- cial vehicle drivers. This assumes that safety inspections would rise to 1,158,000 as projected by FHWA's Office of Motor Carrier Safety and that truck citations for overweight loads would remain at their current level, approximately 0.7 percent of all truck weighings. Although more than 100 million truck weighings are performed each year, an opportunity for contact with the driver occurs only when he is given a citation. Personnel Costs All of those commercial vehicle drivers with whom the police come into contact as part of a safety inspection or truck weighing would be screened for 173

174 alcohol use. This roadside interview of 1,833,000 drivers would take 3 mm per driver. Assuming a 240-day year, this would require 48 person-years at a cost of $3.3 million, using an officer's base salary of $27,500 plus a multiplier of 2.5 to cover fringe benefits and overhead items. Salary estimates are based on Justice Expenditure and Employment data for 1983, inflated to 1986 dollars by using the implicit price deflator for the gross national product, which applies to compensation of state and local employees. Of the 1.8 million commercial vehicle drivers observed, 1.8 percent, or 32,994, would show some presence of alcohol (i.e., >0.00 percent BAC). Figures on alcohol involvement of commercial vehicle drivers were derived in the following manner. First, it was assumed that the probability of crash risk with alcohol involvement would be the same as that for passenger cars, as shown in the case control studies discussed in Chapter 3. The curves shown in Figure 3-2 are essentially flat up to 0.08 percent BAC; that is, they show that the probability of being involved in a crash is roughly equal for drinking drivers compared with drivers who have not been drinking. Above 0.08 percent, and more dramatically above 0.10 percent, the incidence of alcohol involvement in crashes is about two to four times greater than the incidence of alcohol-impaired driving on the road. For simplicity's sake, 0.10 percent BAC rather than 0.08 percent BAC was selected as the cut-off point where the alcohol incidence in crashes rises above the alcohol incidence on the road. A multiple of 3 was used as the differential crash risk between drinking and nondrinking drivers above 0.10 percent BAC. Estimates of commercial vehicle driver alcohol involvement in crashes were then derived by using data from the state of Texas adjusted to lower BACs with the Fatal Accident Reporting System (FARS) figures on commer- cial vehicle driver alcohol involvement as discussed in Chapter 2. The data shown in Table D-1 resulted. These figures were then adjusted to reflect commercial vehicle driver alcohol involvement on the road by dividing the crash involvement number above 0.10 percent by 3 as follows: Distribution of Commercial Alcohol Involvement Alcohol Involvement Vehicle Drivers' BACs in Crashes (%) on the Road (%) 0.01-0.039 0.7 0.7 0.04-0.099 0.5 0.5 >0.10 1.7 0.6 Given these assumptions, it is evident that 0.6 percent of commercial vehicle drivers on the road have some alcohol in their blood at a 0.10 percent legal limit, 1.1 percent at a 0.04 percent legal limit, and 1.8 percent at a 0.00 percent standard. Further testing would be required of those 32,994 drivers testing positive at >0.00 percent BAC. If the standard were 0.10 percent BAC, preliminary field

175 TABLE D-1 COMMERCIAL VEHICLE DRIVER ALCOHOL INVOLVEMENT BY CRASH TYPE AND BAC STANDARD Alcohol Involvement by BAC Limit Type of No. of 0.10 0.04 0.00 Crash Drivers Percent No. Percent No. Percent No. Fatal 5,100 9.1 464 11.8 602 15.0 765 Injury 110,000 4.2 4,620 5.4 5,940 7.0 7,700 PDO 279,000 1.0 2,790 1.3 3,627 1.7 4,743 Bus 63,500 0.0 0 0.0 0 0.0 0 Total 457,600 1.7 7,874 2.2 10,169 2.9 13,208 testing using roadside sobriety tests and possibly a preliminary breath test would require nearly 1 hr per driver or 16 person-years at a cost of $1.1 million. If a lower legal BAC limit (0.04 or 0.00) were adopted, greater use of preliminary breath tests would be required to detect alcohol impairment, but the time required to test could be reduced from nearly 1 hr to 15 min per driver, reducing testing costs for this phase of the enforcement process. On the basis of the results of the preliminary tests, 0.6 percent of the drivers, or 10,998, would actually be arrested and test positive in an evidential test at a BAC standard of ~!0.10 percent. This figure was derived as described earlier. Arresting the driver, conducting the evidential test, and completing a report would require 2 additional hr per driver, or 11 person-years, at a cost of $788,000. The full range of testing activities—from the roadside interview to eviden- tial testing—would thus require 76 person-years at a cost of $5.2 million. Equipment Equipment needs would consist of a minimum of 156 portable breath testers (PBTs) (three per state) at a cost of $400 each. One-fourth of the nation's 500 truck stops would also be outfitted with evidential testing equipment at a cost of $4,000 each and a custom-designed van to hold the equipment at a cost of $14,000 each. Equipment costs were annualized assuming a 3-year equipment life. At a 0.04 percent or 0.00 percent legal BAC limit, officers would also have to be outfitted with passive sensors at a cost of $585 each to detect the presence of alcohol in initial roadside interviews. Using the same assumptions as for PBTs, this would require an annual investment of $30,000.

176 Training One day of training in the use of PBTs for each of the 156 devices would be required at a cost of $286 per day. Because officers would not have to be retrained in the use of the equipment each year, except for new recruits, average annual training costs would be lower than this initial figure. Assuming a 10 percent turnover rate in the on-duty state police force, annual average training costs should be about $4,500. At lower legal BAC limits, officers would also have to receive training in the use of passive sensors. Using the same assumptions as for PBTs, with the exception that only one-half day of training would be required, average annual training costs would be $2,230. Testing The cost of the evidential test for those 10,998 drivers who test positive at 0.10 percent BAC or greater would be $30 per test. This figure is based on state estimates of testing costs. It is assumed that evidential-quality breath- testing equipment is already available at police stations. Towing The trucks of all drivers testing at 0.10 percent BAC or greater would have to be towed at a cost of $250 per truck. SANCTIONING Each of those 10,998 commercial vehicle drivers who test positive at 0.10 percent BAC or greater would require an administrative review at a cost of $25 per review. Half of the cases would require a subsequent administrative hearing at a cost of $324 per hearing. (This latter cost assumes that a police officer would be present at the hearing for a minimum of 4 hr of overtime, requiring an additional 11 person-years.) Half of the administrative hearing findings would be contested and thus would require a judicial review at an average estimated cost of $1,025 per review. (Half of these judicial reviews would require 4 hr of police overtime, resulting in 3 additional person-years at $215 per review.) These figures are based on recent estimates of administra- tive per se programs in California, Maryland, Illinois, and Iowa.

177 PUBLICITY Costs are based on an intensive publicity effort involving a national campaign at a cost of approximately $1 million to develop a campaign theme, advertis- ing materials, and the like. Accompanying state programs of approximately $75,000 each would be needed to develop promotional materials, such as billboards and brochures, tailored to local audiences. Figures assume at least two major campaigns a year and were derived from estimates provided by NHTSA and by individual states of the costs of publicity campaigns for drinking-driving and seat-belt use. ECONOMIC COSTS OF DELAY The derivation of these costs is explained in Appendix E. REFERENCE Justice Expenditure and Employment, 1983. 1986. Bureau of Justice Statistics, U.S. Department of Justice.

Appendix E Estimating the Economic Costs of Shipment Delay A description is given of the methodology used to derive an estimate of the economic costs of shipment delay associated with stopping and testing com- mercial vehicle drivers for alcohol use. The estimate covers three types of cost. [A more complete discussion of the relevant costs may be found elsewhere (AASHTO 1977).] First, there are the time- and distance-related costs associated with paying the drivers. These include the cost of paying the driver who is stopped for alcohol testing for his time out of service. In those incidents in which the driver tests positive at the legal BAC or greater and is unable to continue driving, the costs also cover the time of another driver to pick up the load. Second, the estimate examines the capital costs associated with the truck itself. These include time-related depreciation of the truck and an opportunity loss on the investment capital tied up in the truck during the period it is not being productively utilized. Finally, the estimate addresses another type of opportunity loss experienced as a result of delay—the return that could have been earned on the value of the inventory carried by the truck. Each of these costs will be discussed in turn. DRIVER COSTS The primary source of information on costs was interviews with selected firms representing various segments of the trucking industry—for-hire carriers, private carriers, and owner-operators or companies that use owner-operators. Most companies pay their long-haul drivers on a per-mile basis. If a delay should occur, the method of payment switches to an hourly basis. The 178

179 exceptions here are the independent owner-operators, who are paid a percent- age of the expected revenue from a shipment. The policy of the firms interviewed, with the exception of those that use owner-operators, is to pay a driver on an hourly basis for time out of service even if the driver is at fault. The for-hire and private carriers pay an average of $13 per hour, including fringe benefits. Those that contract with owner- operators pay nothing if the driver is delayed, because the driver is paid a percentage of the total expected revenue from the shipment and thus is paid nothing if he cannot deliver the shipment. Using a weighted average of the miles traveled by each type of carrier (TRB 1986, 67) as a proxy for the relative probability that each of these different types of drivers will be on the road, an average figure was obtained for all carrier types of $11.33 per hour: Driver Hourly Total Truck Carrier Type Cost ($) Miles (%) For-hire 14 40.1 Private 12 47.7 Owner-operator 0 12.2 To calculate the cost of the second driver, two figures had to be estimated: the per-mile cost of the driver and the average distance he would have to travel to pick up the load. Using the weighting methodology just described, an average cost of 39 cents per mile was calculated, assuming a figure of 45 cents per mile (including fringe) for the for-hire carriers, 38 cents per mile for the private carriers, and 21 cents per mile for owner-operators contracting with a trucking firm. Calculating the average distance traveled to pick up a load required infor- mation about the average distance between shipping origin and destination points by type of carrier. The larger carriers tended to operate between terminals located about 200 and 500 mi apart. The owner-operators tended to travel the longest distances, generally averaging between 1,000 and 1,500 mi. For each of these different carrier types, the farthest distance traveled to pick up a load would be the midpoint between their origin and destination points. The average distance traveled to pick up a load then would be half of this median distance: Distance Traveled Between Origin and Destination (mi) Carrier Total Miles Type per Haul Longest Average For-hire 400 200 100 Private 300 150 75 Owner-operator 1,250 625 312

180 Weighting the latter figures according to the method described earlier resulted in an average iravel distance of 115 mi to pick up a load. TRUCK COSTS The capital costs associated with delay are depreciation and an opportunity loss on the capital invested in the truck during the period of delay. (Conser- vatively it was assumed that depreciation is a time-based as well as a mileage- based phenomenon; older trucks are worth less than newer vehicles.) Estima- ting these costs required several calculations. First, the purchase price of a truck was calculated by using $89,000, the average price of a tractor-semi- trailer (American Trucking Associations 1986a, 22). Second, the average useful economic life of a truck was calculated at 12 years, using data on survival rates for heavy trucks compiled by the Motor Vehicle Manufacturers Association. Finally, an appropriate industry return was calculated by using an after-tax return on assets computed from financial data provided by the American Trucking Associations (1986b, 1). Combining these figures—an $89,000 purchase price, an average useful life of 12 years, and a rate of return of 5.3 percent after taxes—resulted in an hourly capital cost of delay of $4.18. This figure was then applied to various estimates of delay time developed in the enforcement scenario. The length of delay took into account two factors: the length of time the truck was out of service during the alcohol testing process (which could be as long as 3 hr), and, for those drivers who tested positive at the legal BAC limit or greater, the time for another driver to pick up the load. This latter time assumed a 2-hr recall period and an additional 3 hr of travel at an average of 40 mph to cover the average 115-mi distance to pick up the load. The maximum delay, based on these assumptions, was 8 hr. INVENTORY COSTS The final cost calculated was a measure of the opportunity loss on the investment tied up in the value of a shipment during the period of the delay. Several other inventory-related costs were not measured because their effect was assumed to be small. For example, one could assume some cost of spoilage for perishable items being shipped. However, agricultural com- modities account for only 6 percent of the total long-haul truck miles traveled (Census Bureau 1985, Table 10, 74). Moreover, many of the shipments would only be delayed a few hours, reducing the potential for spoilage. Similarly, if delays were substantial, the inventory or safety stocks of the shippers and the recipients of goods shipped would have to be increased.

181 However, shippers already build a delay factor into the calculation of their inventory needs and an estimate of the delay involved from alcohol testing amounted to less than 1 percent of the total truck miles traveled.' This was not considered significant enough by industry experts to change the level of safety stocks carried. Several data elements were required to calculate the opportunity loss on the investment tied up in the value of the shipment during the time of delay. First, an average value of, a truck shipment was calculated at $41,300 based on industry figures for general freight of $1.45 per pound provided by the National Freight Claim and Security Council (1986, 9) and an average trailer weight of 28,500 lb. Second, an average delay time of 12.5 hr was computed based on estimates of the total miles per haul-500 mi on average—and a 40- mph traveling speed. Finally, the same after-tax return on assets of 5.3 percent was used as an appropriate rate of return. The actual cost was derived by computing the future value of a $41,300 shipment, assuming a 12.5-hr travel time, and an after-tax rate of return of 5.3 percent and subtracting this from the present value of the load. The resulting opportunity loss is $11, or $0.88 per hour. This figure was then applied to the actual delay times estimated in the enforcement scenario, using the same approach as that in calculating the truck costs. NOTE 1. The comprehensive commercial safety enforcement option would result in 199,980 hr of delay. Assuming an average haul distance of approximately 500 mi and an average travel time of 12.5 hr driving at 40mph, this would result in 15,998 delayed hauls for a total of 8.0 million mi per year. This represents 0.01 percent of the total combination -truck miles traveled. REFERENCES AASHTO. 1977. Manual on User Benefit Analysis of Highway and Bus-Transit. Washington, D.C., 189 pp. American Trucking Associations. 1986a. American Trucking Trends 1986. Alexandria, Va. American Trucking Associations. 1986b. 1985 Motor Carrier Annual Report. Alex- andria, Va. Census Bureau. 1985. Truck Inventory and Use Survey, 1982 Census of Transporta- tion. U.S. Department of Commerce. Motor Vehicle Manufacturers Association. Motor Vehicle Facts and Figures '86. Detroit, Mich., undated, p. 27. National Freight Claim and Security Council. 1986. Results of Cargo Value Per Pound Survey. American Trucking Associations, Alexandria, Va. TRB. 1986. Special Report 211: Twin Trailer Trucks. National Research Council, Washington, D.C. 1986.

Appendix F Estimating the Costs of Post-Crash Testing This appendix provides a more detailed explanation of the methodology used to derive the cost estimate for the post-crash testing option. There are three types of costs associated with a successful enforcement program: costs related to alcohol testing, costs of the sanctioning process, and costs of a publicity compaign. Because crashes will cause delays regardless of whether drivers are tested for alcohol, the economic costs associated with these delays were not attributed to alcohol enforcement efforts. A summary of all of the costs of this option is given in the accompanying text box. ALCOHOL TESTING This scenario assumes that all commercial vehicle drivers involved in crashes causing a fatality or requiring any person to be transported to a hospital would be tested for alcohol use. According to the estimates in Chapter 2 (Table 2-10), there was an average of 441,500 crashes annually between 1983 and 1985 involving medium and heavy trucks and buses. Of these, 5,177 crashes involved a fatality (Table 2-10) and 61,810 required transporting one or more of the persons involved in the accident to the hospital. This latter figure is based on a 1985 estimate from the National Accident Sampling System (NASS) that 14 percent of those involved in truck crashes had to be hospi- talized or transported to a hospital. (This same percentage was also applied to bus crashes.) Thus testing will be required in a total of 67,000 crashes, or 15 percent of the average annual total. 182

183 Post-Crash Testing Cost Estimate Based on 0.10 Percent BAC Alcohol testing Personnel $2,315,000 Equipment 2,453,000 Training 527,000 Testing 793,000 Towing 558,000 Subtotal $6,646,000 Sanctioning Administrative review $67,000 Administrative hearing 436,000 Judicial hearing 762,000 Subtotal $1,265,000 Publicity campaign $3,000,000 Total $10,911,000 According to Table 2-11, there are 457,500 commercial vehicle drivers (assuming that each vehicle has one driver) involved in 441,500 annual crashes or an average of 1.036 commercial vehicle drivers per crash. Applying this same ratio to the 67,000 crashes results in 69,400 commercial vehicle drivers who will have to be tested for alcohol. This was reduced to 68,300 to account for the 20 percent of the commercial vehicle drivers who will be killed in fatal crashes and therefore will not require testing. Personnel Costs Post-crash testing requires a dual enforcement strategy, as shown in Figure F-i. According to the 1985 NASS estimates, approximately 4 percent of all commercial vehicle drivers in a crash would require testing for alcohol at a hospital. (It was assumed conservatively that the majority of these occur- rences would take place in serious accidents.) Three hours of an officer's time

67,000 Crashes 68,300 Crashes 0.05 hr Roadside Jr Interviews 0.95 hr Field Hospital Tests for Evidential impairment Tests (3,360) (18,300) 2hr Poflce Station Evidential Tests (2,020) Tests Positive; Administrative Sanctions Required (2,690) Administrative Reviews (2,690) Administrative Hearings (1,345) Judicial Reviews (670) FIGURE F-i Enforcement scenario for post-crash testing option.

185 would be needed to see that the 18,300 injured persons (457,500 x 0.04) are escorted to the hospital, to obtain the test results, and to complete a report. Assuming a 240-day year, this would require 29 person-years. Using an officer's base salary of $27,500 (based on Justice Expenditure and Employ- ment figures for 1983 inflated through 1986) plus a 2.5 percent multiplier for fringe benefits and overhead, this would result in additional personnel costs of $1.9 million. The remaining 50,000 commercial vehicle drivers can be tested for alcohol using a three-stage enforcement process. A roadside interview of these drivers would take 3 min each. This would require 1 person-year at a cost of $90,000 and would identify an estimated 3,360 drivers with BACs >0.00 percent. This latter estimate is based on Texas data on alcohol involvement of commercial vehicle drivers in fatal and injury truck accidents at the 0.10 percent legal limit and was adjusted to lower BACs by using the Fatal Accident Reporting System (FARS) data on commercial vehicle driver alcohol involvement as discussed in Chapter 2. Alcohol involvement was assumed to be zero for bus drivers. These alcohol involvement figures were then applied to the percent- age of drivers estimated to fall in each category: truck drivers in fatal crashes, truck drivers in injury crashes, and bus drivers. The 3,360 drivers would be tested further to determine their BACs by using roadside sobriety tests if the standard were 0.10 percent BAC, and possibly a preliminary breath test. Preliminary field testing would require nearly 1 hr per driver, or 2 person-years, at a cost of $114,000. If a lower legal BAC limit (0.04 or 0.00) were adopted, greater use of preliminary breath tests would be required to detect alcohol impairment, but the time required to test on-the-road drivers could be reduced from nearly 1 hr to 15 mm, reducing testing costs for this phase of the enforcement process. Evidential testing of those 2,020 drivers who register a BAC of 0.10 percent or greater in field tests would require an additional 2 hr per driver to escort the individuals to the police station, conduct the test, and complete a report. This would require another 2 person-years at a cost of $145,000. Again, adjusted Texas data were used to estimate the number of drivers who would test positive at 0.10 percent BAC or greater. All of the personnel costs associated with the alcohol testing process in the hospital or at the roadside total $2.3 million. Equipment Costs are based on the assumption that all on-duty state law enforcement officers would require portable breath testers (PBTs) at a cost of $400 each. Assuming that 80 percent of the 69,000 state highway troopers (most recent

186 estimate from Justice Expenditure and Employment) are on active duty and that only one-third of the force is on duty at any one time (i.e., 69,000 divided by three 8-hr shifts), then 18,400 devices would be needed. Costs were annualized assuming a 3-year useful life for the equipment, resulting in a total annual investment of $2.5 million. At a 0.04 percent or 0.00 percent legal BAC limit, officers would also have to be outfitted with passive sensors at a cost of $585 each to detect the presence of alcohol in initial roadside interviews. Using the same assumptions as those for PBTs, this would require an annual investment of $3.6 million. Training One day of training would be required in the use of PBTs by 18,400 officers at a daily cost of $286 each. Assuming that 10 percent of the on-duty state police force turns, over each year, average annual training costs would be $527,000. At lower legal BAC limits, officers would also have to receive training in the use of passive sensors. Using the same assumptions as those for PBTs, with the exception that only one-half day of training would be required, average annual training costs would be $264,000. Testing Those receiving tests at the hospital would take a blood or urine test at an average cost of $40 per test. Those tested at the police station would take a breath, blood, or urine test at an average cost of $30. It is assumed that evidential-quality breath-testing equipment is already available at police stations. Towing According to the 1985 NASS data, 17 percent of the trucks involved in crashes are towed from the accident scene because of damage to the vehicle. Thus only 83 percent of the vehicles of those drivers testing at 0.10 percent BAC or greater would have to be towed, at a cost of $250 per truck, and the costs would be attributed to alcohol enforcement. SANCTIONING Of the commercial vehicle drivers involved in accidents, 2,690 are likely to test at or above a BAC of 0.10 percent and thus require sanctioning. These figures are based again on estimates of alcohol-related accidents involving commercial vehicle drivers using Texas data adjusted by FARS. These 2,690

187 drivers would require an administrative review at a cost of $25 per review. Half of the cases would require a subsequent administrative hearing at a cost of $324 per hearing. (This latter cost assumes that a police officer would be present at the hearing for a minimum of 4 hr of overtime, requiring an additional 3 person-years.) Half of the administrative hearing findings would be contested and thus would require a judicial review at an average estimated cost of $1,025 per review. (Half of these judicial reviews would require 4 hr of police overtime, resulting in an additional person year and a $215 cost per review.) These figures are based on recent estimates of administrative per se programs in California, Maryland, Illinois, and Iowa. PUBLICITY Because this enforcement strategy is more highly targeted on a certain seg- ment of the commercial vehicle driver population (i.e., drivers involved in crashes) than the comprehensive commercial safety option, a less extensive publicity campaign is required. Costs assume a $0.7 million national cam- paign to develop a campaign theme and advertising materials and a compan- ion effort of $45,000 in each state to develop promotional materials tailored to local audiences. REFERENCE Justice Expenditure and Employment, 1983. 1986. Bureau of Justice Statistics, U.S. Department of Justice.

Appendix G Cost of Traffic Delays Caused by Accidents A potentially significant item that is often omitted from estimates of accident costs is the value of time lost in traffic delays. The reduction in capacity and the resulting traffic delay that accompany most urban freeway accidents lead to losses of efficiency that cannot be measured directly, because the amount and value of lost time must be estimated for all motorists involved in the delay. Nevertheless, sufficient information is available from studies of acci- dent delays and the value of travel time to make a reasonable estimate of delay costs. VALUE OF TRAVEL TIME Presumably motorists attach a monetary value to travel time that can be identified. Numerous studies have attempted to value this time, most employ- ing the wage rate, adjusted for the type and purpose of travel, as a proxy. After a review of many such studies, Miller et al. (1985, App. G) recommend the use of a travel-time value of 55 and 100 percent of the wage rate for commuter and "on-the-clock" work travel, respectively.' In addition, Miller identifies the few studies that estimate the pecuniary value of leisure travel time; however, these studies are criticized in the literature for yielding unreliable and contradictory results. Consequently, for lack of a universally accepted estimate, Miller's recommendation for the value of work and commuter travel time are employed in these calculations, whereas the value of leisure travel time is assumed to be zero. On the basis of wage rates that are weighted for traffic composition by using the data and method shown in Table G-1, the estimated value of travel time is $8.31 per vehicle-hour. 188

TABLE 0-1 VALUE OF TRAVEL TIME Value of Time Value of Time Weighted Average Occupants per Hour per Hour Percentage Value of Time per Purpose of per - x per Occupant = per Vehicle of Vehicle Hour per Vehicle Vehicle Tnpa Vehicle' ($) ($) Tripsc ($) Commercial trucking 1.1 1639d 17.62e 22 3.88 Work related 1.3 8.931 11.61 3 0.35 Commuting to and from work 1.2 4.919 5.89 24 1.41 Personal travel (nonleisure) 1.6 4919 7.86 34 2.67 Leisure 1.7 N.A. N.A. 17 N.A. Total 8.31 NOTE; N.A. = not available. a The category "commercial trucking" includes all types of commercial trucks. "Work-related" travel includes all nontruck travel that is undertaken while on work time. "Personal travel" includes journeys to and from personal business, shopping, school, and church. "Leisure" travel includes vacation, pleasure, and social driving. '' Estimates of occupants per vehicle for commuter, personal, and work-related travel are from the Nationwide Personal Transportation Study (FHWA 1985a, Table 12, 15). The figure for commercial trucks was estimated by TRB staff. C The percentage for trucks is based on data from Highway Statistics (FHWA 1985b, 171). The percentages for commuter, personal, work- related, and leisure travel are based on daytime travel data from the Nationwide Personal Transportation Study (FHWA 1985a, Table 11, 13-14). d The value of trucking time per hour is based on the sum of the hourly wage rate and fringe benefits of the driver ($11.33) and hourly inventory and truck costs ($5.06) as explained in Appendix E of this report. e To avoid overcounting of truck and inventory costs, only the wage and fringe benefits components of the hourly time value are multiplied by the fractional portion of the occupant-per-vehicle estimate. Based on the average hourly wage rate for nonagricultural employees for May 1987 (provided by the Bureau of Labor Statistics, U.S. Department of Labor). g Based on the product of the wage rate markdown of 0.55 recommended by Miller (1985, App. 0) and the average hourly wage rate for nonagricultural employees for May 1987 ($8.93) (provided by the Bureau of Labor Statistics, U.S. Department of Labor).

190 TIME LOSS Few studies have estimated the loss of time of motorists involved in traffic delays caused by accidents on urban interstates and freeways. The most frequently cited study, developed from a daytime accident log kept by Houston police in 1968 and 1969 on a 6-mi section of the Gulf Freeway, found that the average freeway accident took 45 min to clear and investigate (Goolsby 1971,41). Similarly, a study conducted on Los Angeles freeways in 1974 yielded a duration estimate of 46 mm (Juge 1974). However, the estimates of both of these studies are based on accidents involving all types of vehicles. A more specific analysis of truck-related freeway incidents, con- ducted by the California Department of Transportation during 1983 and 1984, yielded a mean duration of 3 hr 35 min and an average loss of 2,006 vehicle-hr per incident (Golob 1985, 7-20). The latter estimate is the most relevant for this analysis. CALCULATION OF TRAFFIC DELAY COSTS The traffic delay cost per truck accident can be estimated by applying the previously derived time loss and value estimates as follows: Vehicle hours lost per accident = 2,006, Dollar value of time per vehicle-hour = $8.31, Traffic delay cost per accident = 1 x 2 = $16,670. To apply this delay cost to the preventable accident estimates derived in Chapter 5, it was necessary to determine the share of these crashes that occurs on urban Interstates.2 According to the 1984 FARS and NASS files, 13.2 percent, 8.6 percent, and 6.8 percent of heavy and medium truck-involved accidents, resulting in injuries, deaths, and property damage only, respec- tively, occur on urban Interstates. These percentages were then applied to the total preventable crash figures by accident severity category and multiplied by the delay cost per crash to determine the annual reduction in traffic delay costs as follows:

Traffic Delay Savings at Alternative BACs 191 BAC Limit Total No. of Pre- (%) ventable Accidents 0.10 2,390 0.04 3,400 0.00 4,380 No. of Prevent- able Accidents on Urban Inter- Savings states ($ millions) 240 4.0 340 5.7 440 7.4 NOTES "On-the-clock" work travel includes all work-related journeys (not including travel time to and from work). Commuter travel includes journeys to and from work, shopping, church, school, and other personal business. Leisure travel includes vacation, social, and pleasure driving. The estimates provided in Chaptei 5 were given in tcnns of truck involvements by crash severity. These figures were converted to crashes using the ratio of commercial vehicles to truck-involved crashes by crash severity provided by the 1984 PARS and NASS files. REFERENCES FHWA. 1985a. Survey Data Tabulations: Nationwide Personal Transportation Study 1983-1984. U.S. Department of Transportation. FHWA. 1985b. Highway Statistics. U.S. Department of Transportation. Golob, T. F. 1985. A Preliminary Analysis of Major Freeway Accidents in Los Angeles, Orange, and Ventura Counties Involving Large Trucks, 1983-1984. Institute of Transportation Studies, University of California, h-vine. Goolsby, M. E. 1971. Influence of Incidents on Freeway Quality of Service. In Highway Research Record 349, HRB, National Research Council, Washington. D.C., pp. 41-46. Juge, J., et al. 1974. Early Detection and Rapid Removal of Disabled Vehicles and Other Hazards from the Freeway. California Traffic Safety Program. Miller, T., et al. 1985. Development of a Value Criteria Methodology for Assessing Highway Systems Cost-Effectiveness. Report PB 85-219236/XAB. FHWA, U.S. Department of Transportation, 209 pp.

Study Committee Biographical Information M. W. PERRINE, Chairman, a psychologist, is Professor at the Schools of Medicine and Public Health at Boston University and Director of the latter's Alcohol Research Unit. Dr. Perrine received his B.A. from the University of Connecticut and his M.A. and Ph.D. from Princeton University. He was a lecturer in psychology and Director of the Research Center on Visual Percep- tion in Ulm, Germany, for 4 years, after which he became Professor of Psychology at the University of Vermont. He has been a consultant to IBM Corporation, the National Science Foundation, the German Ministry of Sci- ence, and the Office of the Director, National Institute on Alcohol Abuse and Alcoholism. Dr. Perrine was a coprincipal investigator for Project CRASH, the Vermont Alcohol Safety Action Program for the National Highway Traffic Safety Administration. His research areas include the influences of alcohol on perception, cognition, and driving behavior and behavioral aspects of alcohol and highway safety. Dr. Perrine has published a number of articles on the influence of alcohol on driving-related behavior. He is a member of the American Psychological Association, the Research Society on Alcoholism, and the National Safety Council's Committee on Alcohol and OLher Drugs. He also serves on the TRB Committee on Alcohol, Other Drugs, and Transportation. JAMES D. BEARD, a physiologist, has been Director of the Alcohol Research Center at the University of Tennessee College of Medicine and Memphis Mental Health Institute since 1967. He received his B.A. from DePauw University and his Ph.D. from the University of Tennessee, Memphis. Dr. 192

193 Beard was Assistant Professor of Physiology at the University of Tennessee College of Medicine in Memphis, then became an Associate Professor of Psychiatry and later Professor of Psychiatry there. He has published more than 80 articles, including 11 chapters dealing with the pathophysiology of acute and chronic alcoholism. Dr. Beard has been a consultant to the National Institute on Alcohol Abuse and Alcoholism. He has served as a principal consultant for the Fourth and Fifth Special Reports to the U.S. Congress on Alcohol and Health (and is on the Editorial Board for the Sixth Special Report), as a member and chairperson of the Alcohol Biomedical Research Review Committee, and as a member of the National Advisory Council on Alcohol Abuse and Alcoholism of the Alcohol, Drug Abuse and Mental Health Administration. He is a member of numerous scientific societies. Currently, Dr. Beard is a member of the Board of Directors of the Research Society on Alcoholism, on the Editorial Board of Alcoholism: Clinical and Experimental Research, and on the Scientific Advisory Board of the National Institute on Alcohol Abuse and Alcoholism. ROBERT J. FORMAN is Vice-President for Safety and Security of Greyhound Lines, Inc. A graduate of Michigan State University, Mr. Forman was District Director of the New England States for the National Safety Council before becoming the Vice-President for Safety of Greyhound Lines. He was with Trailways Lines, Inc. from 1979 to 1987. Mr. Forman is a past Vice-Chairman of the Department of Transportation's National Highway Safety Advisory Committee and past President of the Arizona Safety Association. He has been a member of the Board of Directors of the National Safety Council and of the Health and Safety Committee of the National Association of Manufacturers. He served on the TRB Committee for the Study of Safety Benefits and Costs of Using Citizens Band Radios on Intercity Buses. ROBERT B. FORNEY, a toxicologist, received his Ph.D. from Indiana Univer- sity. Dr. Fomey was an Assistant Professor and Associate Professor of Tox- icology at Indiana University for 14 years. During that time he became Director of the State Department of Toxicology at the time of its creation. Next he became Professor of Pharmacology and Toxicology at the Schools of Medicine of both Indiana University and Purdue University. In 1977 he was made a Distinguished Professor of Toxicology and Director of the State Toxicology Laboratory at the School of Medicine, Indiana University. He has done extensive research on the effects of alcohol and other drugs on perfor- mance. He is a member of the Traffic Safety Committee and the Committee on Alcohol and Drugs for the National Safety Council. Dr. Fomey is affiliated with the Society on Toxicology, American Academy of Forensic Science, American Society of Clinical Pharmacology and Therapeutics, and Sigma Xi.

194 GERALD J. FRIEDMAN, a physician, is Medical Director for United Parcel Service. He received his M.D. from New York University. After working at NYU's School of Medicine, he rose to the position of Physician-in-Charge of the Metabolism and Endocrine Clinic at Beth Israel Hospital. He has taught at NYU's School of Medicine since 1949, and is a Clinical Professor of Medi- cine at Mount Sinai School of Medicine. He has published more than 50 papers on internal medicine and is an expert on industrial policy on alcohol and drug abuse. Dr. Friedman is past president of the New York Chapter of the American Diabetes Association and a representative to the American Occupa- tional Medical Association. LARRY G. MAJERUS is currently the Administrator of the Montana Division of Motor Vehicles. After receiving a B.A. from Carroll College, Mr. Majerus was a budget analyst and auditor with the Montana Department of Highways and Highway Safety Management Specialist in the Denver Office of the National Highway Traffic Safety Administration. He has been President of Region IV of the American Association of Motor Vehicle Administrators and Chairman of the Driver License Compact Commission. KIMBALL I. MAULL, a physician, is Professor and Chairman of the Depart- ment of Surgery, University of Tennessee. He received his B.A. from the University of Virginia and his M.D. from Cornell University, after which he held the positions of Assistant Professor of Surgery at the University of Kentucky Medical Center and Associate Professor of Surgery of the Medical College of Virginia. Dr. Maull has written a number of papers on crash trauma and the relationship between alcohol abuse and vehicle crashes. He is the president of the American Association for Automotive Medicine and Editor- in-Chief of Advances in Trauma, an annual publication dealing with new developments in injury treatment and control. LAIMUTIS NARGELANAS, a law enforcement specialist, is currently the Super- intendent of the Division of State Troopers, Illinois State Police. He received his B.A. and M.A. from Sangamon State University and has been accepted for advanced candidacy to complete his Ph.D. at Southern Illinois University. Mr. Nargelanas served as an Illinois State Trooper for 5 years before joining the staff of the Illinois State Police Academy. He was Director of Curriculum Development for the Illinois Department of Law Enforcement Academy before becoming the Director of Training for the State Police Department of Law Enforcement Academy in 1979. Mr. Nargelenas has progressed through the ranks, achieving the: permanent rank of Major in 1983. He is a member of the International Association of Chiefs of Police, Highway Safety Committee, the Illinois Driving Under the Influence Task Force, Chairman of both the

195 Traffic Law Enforcement and Adjudication Committees for the National Safety Council, and the Fourth Vice-President of the Illinois Association of Chiefs of Police. OLGA J. PENDLETON, a statistician, is a Program Manager for the Statistical Analysis and Research Program at the Texas Transportation Institute. Dr. Pendleton received her B.S. from the University of South Alabama and her M.S. and Ph.D. from Emory University. She was an Assistant Professor in Computer Science and Statistics at Mississippi State University, an Assistant Professor in Statistics and Computer Science at the University of Georgia, and Assistant Biometrician with the Department of Biomathematics at the Univer- sity of Texas System Cancer Center. Dr. Pendleton became an Associate Research Statistician with the Texas Transportation Institute in 1981. She has written a number of articles relating to traffic fatalities, including the involve- ment of alcohol and blood alcohol concentration. ROBERT HARRY REEDER, a lawyer, serves as the General Counsel of the Traffic Institute at Northwestern University and as Executive Director of the National Committee on Uniform Traffic Laws and Ordinances. Mr. Reeder received his B.A. and J.D. from Washburn University. He was admitted to the Bar of the U.S. District Court of Kansas in 1960 and the U.S. Supreme Court in 1968. Before coming to the Traffic Institute as assistant counsel, he began his legal career as a research assistant with the Research Department of the Kansas Legislature. An expert in traffic law, Mr. Reeder is coauthor of Vehicle Traffic Law and The Evidence Handbook, and is author of Interpretation of Implied Consent by the Courts and Analytical Study of the Legal and Opera- tional Aspects of the Minnesota Law Entitled "Chemical Test for Intoxica- tion." Mr. Reeder has been Chairman of the National Safety Council and now serves on their Committee on Alcohol and Other Drugs. THOMAS II. ROCKWELL, an industrial engineer, is Professor of Industrial Engineering at Ohio State University. He received his B.S. from Stanford University and his M.S. and Ph.D. from Ohio State. Dr. Rockwell directed the Driver Performance Research Laboratory at Ohio State and was Associate Professor of Industrial Engineering before becoming a full professor in 1965. His research areas include human performance under impaired conditions. Dr. Rockwell has served on several TRB committees and chaired the Committee on Road User Characteristics. He is a fellow of the Human Factors Society. ALISON SMILEY, a human factors engineer, is President of Human Factors North, Inc. She received her B.S. from the University of Western Ontario and her M.A.Sc. and her Ph.D. in Systems Design Engineering from the Univer- sity of Waterloo. Dr. Smilcy was a researcher with the National Research

196 Council of Canada and with the Southern California Research Institute. She has done extensive work and written papers on the effects of drugs and alcohol on driving performance. Dr. Smiley is the president-elect of the Human Factors Society of Canada and a member of the Advisory Panel to the National Institute on Alcohol Abuse and Addiction and TRB 's Committee on Simulation and Measurement of Driving. Scorr D. SOLDON, an attorney, is a partner with Previant, Goldberg, Uelman, Gratz, Miller and Breuggeman. Mr. Soldon received his B.A. in Political Science from Marquette University and his J.D. from Northwestern Univer- sity School of Law. He was admitted to the State Bar of Wisconsin and the Bar of the U.S. District Courts for the Eastern and Western Districts of Wisconsin and the U.S. Courts of Appeals for the Fifth, Sixth, Seventh, Eighth, and the D.C. Circuits. Mr. Soldon is a specialist in labor and employment law. ROBERT B. VOAS is a psychologist with the Environmental Research Institute of Michigan. After graduating from the University of California at Los Angeles, Dr. Voas held research positions with the U.S. Navy and then became Assistant to the Director at the Manned Spacecraft Center of the National Aeronautics and Space Administration. He worked with Litton Industries and for the Peace Corps. For more than 10 years, Dr. Voas was Chairman of the Division of Evaluation for the National Highway Traffic Safety Administration. His areas of research include personality measure- ment, stress tolerance, and the relationship between alcohol countermeasures and safety.

The Transportation Research Board is a unit of the National Research Council, which serves the National Academy of Sciences and the National Academy of Engineering. The Board's purpose is to stimulate research concerning the nature and performance of transportation systems, to dis- seminate the information produced by the research, and to encourage the application of appropriate research findings. The Board's program is car- ried out by more than 270 committees, task forces, and panels composed of more than 3,300 administrators, engineers, social scientists, attorneys, educators, and others concerned with transportation; they serve without compensation. The program is supported by state transportation and high- way departments, the modal administrations of the U.S. Department of Transportation, the Association of American Railroads, the National High- way Traffic Safety Administration, and other organizations and individuals interested in the development of transportation. The National Academy of Sciences is a private, nonprofit, self-per- petuating society of distinguished scholars engaged in scientific and engi- neering research, dedicated to the furtherance of science and technology and to their use for the general welfare. Upon the authority of the charter granted to it by the Congress in 1863, the Academy has a mandate that requires it to advise the federal government on scientific and technical matters. Dr. Frank Press is president of the National Academy of Sciences. The National Academy of Engineering was established in 1964, under the charter of the National Academy of Sciences, as a parallel organization of outstanding engineers. It is autonomous in its administration and in the selection of its members, sharing with the National Academy of Sciences the responsibility for advising the federal government. The National Acad- emy of Engineering also sponsors engineering programs aimed at meeting national needs, encourages education and research, and recognizes the superior achievements of engineers. Dr. Robert M. White is president of the National Academy of Engineering. The Institute of Medicine was established in 1970 by the National Academy of Sciences to secure the services of eminent members of appropriate professions in the examination of policy matters pertaining to the health of the public. The Institute acts under the responsibility given to the National Academy of Sciences by its congressional charter to be an adviser to the federal government and, upon its own initiative, to identify issues of medical care, research, and education. Dr. Samuel 0. Thier is president of the Institute of Medicine. The National Research Council was organized by the National Academy of Sciences in 1916 to associate the broad community of science and technology with the Academy's purposes of furthering knowledge and advising the federal government. Functioning in accordance with general policies determined by the Academy, the Council has become the principal operating agency of both the National Academy of Sciences and the National Academy of Engineering in providing services to the government, the public, and the scientific and engineering communities. The Council is administered jointly by both the Academies and the Institute of Medicine. Dr. Frank Press and Dr. Robert M. White are chairman and vice chairman, respectively, of the National Research Council.

Transportation Research Board National Research Council 2101 Constitution Avenue, N.W. Washington, D.C. 20418 NON-PROFIT ORG. U.S. POSTAGE PAID WASHINGTON, D.C. PERMIT NO. 8970 ADDRESS CORRECTION REQUESTED .i1 N .0 :t LU 0 W N.: of 0 Cm LtJ ow o- 'o-i,: 00 o