Most Americans find it difficult to reconcile their notions of social justice with the fact that some 35 million people, most of whom are employed or are dependents of someone employed, lack basic health insurance coverage. We hear that government agencies, physicians, and public and private hospitals are straining to keep pace with requests for services amid budget constraints. Not only are some people losing their insurance coverage, but the size of vulnerable populations is growing. Progress against several health problems is blocked because of poor access to health care. The news media frequently depict the plight of poor unwed pregnant adolescents, children in low-income families, the homeless, minority groups, residents of rural areas, and refugees. When the focus shifts to whether we are making headway against specific diseases, the disparity in health status between vulnerable groups and the general populace becomes apparent. With respect to HIV/AIDS, drug abuse, cancer, and infant mortality, less favorable outcomes are often attributed to the inability of some segments of society to gain timely access to essential health services.
In addition to concerns over the less fortunate in our society, there is a growing uneasiness that the delivery system and insurance infrastructure are not meeting the needs of middle- and upper-income Americans as well as they once were. Many besides the poor may have difficulty getting access to health care. Those with preexisting disease conditions fear that they will lose insurance coverage if they change jobs. Middle-income people fear that shrinking insurance benefits will force them to pay more and more of
the costs of health care, increasing their reluctance to seek care when they need it. Spurred in part by the malpractice insurance crisis, obstetrical services are reported to be in short supply in some areas. Residents of this country's rural areas are fighting an uphill battle to keep hospitals open so that, at a minimum, they have access to emergency services and essential primary care. To avoid having to admit emergency room patients who may not have insurance, growing numbers of hospitals have ceased offering emergency services.
There is no shortage of stories depicting these and related problems. What policymakers and the public lack, however, is a systematic way of looking at the barriers people face in getting needed health care and the impact those barriers have on society. Data describing barriers to health care are incomplete, scattered, underanalyzed, or outdated. The information that has been gathered is not organized in ways that promote systematic thinking about how access to health care has changed over time, nor can differences among affected groups be compared. For these and other reasons, the desirable objective of improving access to health care remains elusive.
The IOM Access Monitoring Project was designed to develop a way to monitor access to health care that will be useful to health care policymakers. The charge to the IOM committee was to develop a rationale and framework for gauging how well or how poorly the nation provides access to personal health care. The focus of this effort was to be the development of a limited set of indicators that could reliably sense the direction and extent of change at the national level in access problems and at the same time give clues about what factors might be driving that change. In addition to clarifying perceptions about the status of health care access in the United States, it was hoped that these indicators would serve as a general guide to decisionmaking.
Unfortunately, the data and methods available to devise a reliable system for monitoring access to health care are incomplete in some areas. Thus, in addition to developing indicators, the committee was instructed to recommend strategies for improving state-of-the-art access monitoring. This included identifying ways to enhance ongoing data collection activities and refine measurement techniques as well as encouraging research on access problems in areas in which there are currently insufficient data or in which the appropriate indicator for measuring access is unclear.
Unlike many of IOM's activities, this project was self-initiated—neither mandated by federal legislation nor directly requested by a public or private agency. The call for such a study of health care access came from
the IOM membership, and the specific direction taken was one that emerged from meetings of IOM's Board on Health Care Services.
The board concluded that, because of both societal changes and shifts in public policy, better mechanisms are required for monitoring changes in access. The nation needs, but currently lacks, an entity to continuously monitor the numerous types of utilization and health status problems arising from insurance inadequacies, cultural impediments, geographic barriers, or other factors and place these problems in the broader context of national health policies. The 14-member IOM access monitoring committee was constituted in February 1990 as a first step toward this goal.
Although the mandate for this specific project was generated by the interest of IOM members, the idea to undertake a monitoring project had its intellectual roots in a 1985 workshop sponsored by IOM. In fact, many of the indicators discussed at that meeting, as well as the workshop's conclusion that access monitoring should focus on health care utilization, were incorporated into the present committee's work.
The 1985 workshop was the result of discussions among IOM, the Robert Wood Johnson Foundation, and the federal Health Resources and Services Administration, which was seeking better methods for targeting its dollars to the medically underserved. In its 1985 annual report the foundation's then-president David Rogers expressed the frustrations that led to the workshop:
Each day a quick glance at the newspaper tells us the relative standing of our baseball, football, and ice hockey teams. Likewise, a myriad of data is available daily on the status of American business or the economy in our major cities. Yet in health care, even fairly simple statistics are not regularly collected, or when they are, they are processed so slowly that they are not available until two to five years after the fact. (Rogers, 1985)
DEVELOPING A SET OF INDICATORS
The process of developing social indicators involves in principle at least three broad elements (DeNeufville, 1975). First, the concepts that underlie the phenomenon of interest—access, in this case—need to be made clear with the help of existing models and attempts at problem definition. Second, the constraints on and possibilities for quantifying the concepts with existing or new data need to be considered. Third, it must be determined how best to organize and summarize the data so that they are of most use to policymakers.
The development of IOM's access monitoring framework and set of specific indicators was an interactive process. The committee initially identified an extensive list of indicators that it believed to be both broadly representative of access problems and for which there was a possibility of
regularly available information. Each meeting of the committee was followed by study of the methodological issues and assessment of data availability related to individual indicators. The indicator list and the conceptual framework were then refined at each subsequent meeting. When gaps in the information base were found, the committee recommended ways to fill those gaps. Many of the recommendations were directed at public and private organizations that collect information intended to enhance the nation's ability to monitor access to personal health care services. Taken together, the recommendations constitute a research agenda.
The committee also identified a second set of potential indicators representative of important access problems but for which a lack of data or consensus about how they should be measured limited the degree to which they could be studied. Descriptions of these access problems are included in Chapter 3. It is the committee's belief that further studies in these areas are an important component of a research agenda. To begin the process of moving toward developing these indicators, the committee commissioned three background papers on AIDS, substance abuse, and migrants and people who are homeless (Appendixes A-C). The papers identify what is needed to develop useful indicators and other approaches that may be used for tracking access problems.
RELATIONSHIP OF ACCESS MONITORING TO OTHER RELEVANT ACTIVITIES
The aim of the IOM Access Monitoring Project was to improve the state of the art in measuring access problems by determining what indicators can be identified now, what important access problems do not have indicators, and what steps need to be taken over the long term to institutionalize a monitoring system based on valid, reliable data. The purpose of these activities is to create a data base that will allow decisionmakers to determine how the nation is faring with respect to access over time. The data base should allow both general assessments of access to health care in the United States and specific analyses of particular services or populations,
It is expected that evidence of time-dependent changes in access, and differences in access among population groups, will stimulate focused discussion about policies that can promote more appropriate utilization and better health outcomes. Although the indicators may suggest how well or how poorly current policies are working, the IOM monitoring project is not an evaluation study per se. A specific change in Medicaid eligibility, for example, cannot be detected by and ascribed to changes in one or more indicators because many other policies and forces are operating at the same time. In the future, however, evaluators may want to use the indicators as part of studies that take into account other factors that may explain changes in access.
Although the monitoring project is not a research effort, the committee has relied heavily on the studies of others. Particular attention has been devoted to research that examines services known to affect health status, the independent contribution of barriers to use, and health outcomes. This report includes some original analyses of survey and hospital discharge data to demonstrate the utility of specific types of analyses; it was not feasible, however, to pursue inquiries into all the lines of interest. It is hoped that this report will spark interest in focused research in a variety of areas.
The IOM Access Monitoring Project overlaps with several other efforts to develop health indicators, most notably the Year 2000 Health Objectives for the Nation, outlined in Healthy People (U.S. Department of Health and Human Services, 1991). These objectives focus principally on health promotion and protection and on preventive health services. Several of the IOM indicators fall within these categories. Where there is overlap, the IOM committee has attempted to be as consistent as possible with the Year 2000 Health Objectives, which serve as useful standards to assess how far the nation might be from access goals. The main focus of the IOM project, however, is personal health services, including curative medical services. Other factors that can affect health and that are addressed in Healthy People, such as traffic accidents, misuse of firearms, environmental controls, and safe work environments, are not dealt with by the IOM project. In contrast to the great breadth of Healthy People, other monitoring projects have focused on particular subpopulations, particularly children and pregnant women.
Additionally, the World Health Organization (1981) has developed a set of indicators as part of its Health for All Project. Although the WHO literature offers some useful ideas (particularly in the area of avoidable mortality), many of the indicators are broad social and economic measures (e.g., income distribution, adult literacy rate) rather than specifically health-related gauges or are not germane to the U.S. context (e.g., distance to protected clean water supply, malaria prophylaxis).
MAJOR DATA SOURCES: THEIR USES AND LIMITATIONS
Box 1-1 shows the major sources of data on which the committee drew to support the monitoring of access indicators. The sources cover a wide array of activities designed to measure different types of health services use and health states. Each data base has inherent strengths and weaknesses, and, although they hold great promise for access measurement, none were established with this as a principal or even secondary aim in mind. Therefore, the major challenge for the future will be to capitalize on their strengths while making the necessary changes to minimize the weaknesses that limit our ability to draw conclusions from available data about access problems. These changes must be judicious so as not to interfere with the original
BOX 1-1 Major Data Sources for Monitoring Access to Personal Health Care
purposes of the data collection activities. How to approach this challenge is a major focus of the committee's recommendations on ways to improve the state of the art of access monitoring.
Strengths and Weaknesses
Vital statistics are derived from birth and death certificates. The birth certificate is the major source of information about the use of prenatal care and low birthweight. In general, physician records about a woman's visits during pregnancy are transferred to the hospital for inclusion on the certificate. However, information on a birth record about the content of prenatal visits is limited. Data problems arise either when women have no or multiple providers or when recall of service use is required after delivery. Thus, threats to the validity of the data come from those most likely to have poor pregnancy outcomes—the poor, the young, minorities, noncitizens illegally residing in the United States, and the poorly educated.
Death records provide mortality statistics and when linked to the birth record offer insight into the important correlates of infant mortality related to prenatal care. The cause-of-death information recorded on a death certificate can be used to compare the mortality experience of different subpopulations and to assess its relationship to access barriers. Concerns have been expressed about the accuracy of cause-of-death information because it is based on the judgment of the certifying physician.
For the purposes of monitoring access for all age groups, the major limitation of the death certificate is its lack of relevant information on the possible extent of access barriers as a result of lack of insurance, low-income, or other such impediments. This limitation can be overcome in part through special followback surveys of a patient's closest relative.
Large household surveys such as the National Health Interview Survey (NHIS) provide a wealth of information that allows analysts to relate the use of health services and self-reports of health status to characteristics of individuals and families. Key strengths of the NHIS are its periodicity (it is conducted annually), its large and carefully constructed sample (about 120,000 respondents), and its well-tested questionnaire items.
Like most health care surveys, the NHIS suffers from its reliance on respondents' recall of when and how most services were used and the imprecision of self-reports of health status when compared with health examination data or medical record abstracts. The NHIS has questionnaire items
on physician utilization—one with a two-week recall and one with a 12-month recall. This represents a tradeoff between response error and the stability of the estimate.
Despite its large sample size, the ability of the NHIS to disaggregate population subgroups of interest (e.g., certain minorities, the chronically ill) and other areas of concern is limited. For example, key groups with access problems, such as the homeless and migrant farmworkers, are not well captured. Despite its periodicity, some key questionnaire items are not routinely included—for example, items on health insurance coverage and those that appear in various topical supplements (e.g., cancer screening, dental services) to the main survey. Despite its thoughtful design, specific questions relevant to access monitoring have not been worded to elicit the most in-depth information about access barriers, in contrast to a survey that has been specifically formulated to probe access issues. Recognizing that these limitations are not all resolvable, the committee presents recommendations in the detailed discussions (Chapter 3) of the indicators about modest changes that could maximize the utility of the survey.
In some cases the committee undertook original analyses of the NHIS1 Following the convention of the National Center for Health Statistics, however, estimates of persons or events are not reported in the tables in this report if the relative standard error is greater than 30 percent. Differences between estimates are discussed only in cases in which a difference was significant at at least the 5 percent level. The report also notes findings from other national surveys—specifically, the National Medical Care Expenditure Survey, the Medicare Beneficiary Survey, the National Maternal and Infant Health Survey, the Robert Wood Johnson Foundation access surveys, and the behavioral risk factor surveys of the Centers for Disease Control. These and other instruments are extremely helpful in delineating the underlying relationships that are the root causes of access barriers. They complement the routinely available NHIS data, which are the mainstay of access monitoring.
In particular, the 1987 National Medical Expenditure Survey (NMES), undertaken by the Agency for Health Care Policy and Research, explores a variety of issues that are not amenable to study under NHIS's format. Because the design involves NMES multiple interviews with the same respondent over the course of a year, an individual's changing insurance status and expenses can be tracked. The survey also provides detailed information about a person's insurance policy and physician records, which permits assessment of, for example, the effects of underinsurance on utilization.
Hospital Discharge Data
Computerized abstracts of patient discharge records organized into state data bases are increasingly available to health services researchers interested in measuring a wide range of hospital quality and cost phenomena. The committee has capitalized on recent innovations in using such data; it employed them for the comparatively new purpose of monitoring the effect access barriers can have on increasing admission rates for those who appear to be lacking the ambulatory care that might prevent hospitalization for certain conditions. The technique can also be used to detect inequities in the distribution of high-cost discretionary procedures.
This promising analytic strategy for monitoring access requires additional research and development. The lack of income data on a discharge abstract poses a methodological obstacle that is addressed by using patient zip code information. To avoid what methodologists call the ''ecological fallacy," it is necessary to limit the unit of analysis to neighborhoods (characterized according to income levels) rather than patients. This means, however, that the technique cannot adequately capture access problems in localities in which the poor are dispersed throughout the population. In contrast to income information, insurance status is included on the discharge abstract under "expected source of payment." Unfortunately, this information is difficult to interpret because of the lack of good population-based data on insurance coverage. These data are needed as the basis for tracking whether admission rates per capita are related to the expected payment source. Beyond these specific problems, future studies must focus on a better understanding of the underlying reasons for delayed or inadequate care together with further scrutiny of the admission diagnoses used in the analysis.
Because early case finding through medical screening tests is a critical factor in whether people with certain types of cancer will survive, one dimension of access can be measured by the prevalence of cancers found at a late stage. The major source of this information is the tumor registry. These state/local registries record information about the pathology of tumors and other useful data such as a patient's race, age, sex, and geographic location. Like hospital discharge data bases, however, tumor registries are not established in all states and localities—which limits the ability to generalize from them to the nation as a whole. Tumor registries also lack good data on income and insurance.
Some of the access indicators in this report rely on the tracking of certain communicable diseases. Physicians are expected to report occurrences of these diseases to their local health departments. These agencies in turn are charged with reporting them to the CDC, which routinely publishes the data. Major problems with the system are underreporting and misreporting, which occur for several reasons. Physicians' lack of understanding of the importance of sustained attention to reporting is sometimes cited as the broad reason for underreporting. Also contributing to the problem, however, are physician concerns about patient privacy, changing definitions and reporting guidelines, and the difficulties of recognizing diseases with relatively low incidences.
Information reported to third-party payers for the purpose of paying claims could potentially be a source of information for access monitoring. Health insurance claims contain data about utilization, health status, and costs, but they also have several major drawbacks to their use in monitoring. The major problem is an obvious one: claims data bases do not contain information about the uninsured. In addition, they are not uniform and are thus expensive to analyze. There are numerous ongoing efforts to make claims data more usable, efforts that will be considerably furthered by the movement toward widespread use of computerized medical records. In the meantime, the Health Care Financing Administration could investigate the potential contribution of the Medicaid and Medicare data bases to access monitoring. A few studies reported in the discussion of the indicators rely on analyses of these data.
Most of the data bases described above are subject to two prominent concerns about obtaining information for monitoring access. One is methodological—how to obtain valid information about racial and ethnic subgroups—and the other procedural—the timely public release of data.
The access problems of racial and ethnic minorities have been a consistent focus of concern among health care policymakers. Inconsistent classification or misclassification of these population subgroups has frustrated both health service researchers and spokespersons for the subgroups themselves
for decades. Recent results from the Hispanic Health and Nutrition Examination Survey highlight the analytic problems of lumping together Americans of Cuban, Puerto Rican, and Mexican origin who have quite different health care experiences and circumstances. Similar problems exist when Asian population subgroups are aggregated. Samples need to be of sufficient size to disentangle the effects of poverty from cultural or other characteristics specific to a group that may constitute access barriers. Oversampling in existing surveys or conducting special studies is one answer, albeit one that raises methodological and logistical issues.
An important rationale for access monitoring is that it allows periodic updating of indicators to identify policy and environmental changes influencing access. For this reason, making information available in a timely fashion is important, but doing so involves tradeoffs among cost, accuracy, and the response time of the indicators to changing conditions.
Data sources frequently have multiple purposes and are controlled by a variety of organizations. Therefore, the considerations involved in how the tradeoffs should be made are complex. Nevertheless, the committee's concern about lag time is real. To this end, one of the committee's recommendations addresses the need for a federal locus of authority for oversight of the monitoring process. Among other functions, such an organization would make more visible the need for and utility of data and thereby act as an advocate for the availability of timely information.
Poor race/ethnicity data and lack of timeliness are major weaknesses common to many data bases and threaten the validity inherent in the monitoring process. How the data are manipulated also demands attention. Some examples of the need for more research and analysis to further the evolution of measuring access can be cited from the committee's work. Future capacity to monitor access would be improved by creating better surrogate measures of such concepts as socioeconomic and insurance status. Further research is required to operationalize concepts like chronic disease follow-up care and excess mortality. Indices that combine data in useful ways, as in measurement of the use of prenatal care, must also be improved. The general aim over the long term should be to move toward acceptance and standardization of approaches to monitoring access to health care services.
DeNeufville, J. N. 1975. Social Indicators and Public Policy: Interactive Processes of Design and Application. New York: Elsevier.
National Center for Health Statistics. 1989. Current Estimates from the National Health
Interview Survey. Series 10, No. 176. Hyattsville, Md.: National Center for Health Statistics.
Rogers, D. E. 1985. Examining the Adequacy of a Community's Health Care. Annual Report. Princeton, N.J.: The Robert Wood Johnson Foundation.
U.S. Department of Health and Human Services. 1991. Healthy People 2000: National Health Promotion and Disease Prevention Objectives . Washington, D.C.: U.S. Government Printing Office.
World Health Organization. 1981. Development of Indicators for Monitoring Progress Towards Health for All by the Year 2000. Geneva: WHO.