5

Analysis and Interpretation of Exposures Data

This is the second of three chapters that describe the methods and results of the committee’s analysis of the initial months of Airborne Hazards and Open Burn Pit (AH&OBP) Registry data. It summarizes information on respondents’ exposure to burn pits and other important service-related airborne exposures; discusses the registry’s limited capacity to provide reliable estimates of the magnitude, duration, and frequency of the exposures; and explicates the committee’s approach to using the exposure-related information that is available. An approach to analyzing the exposure information collected by the registry is then presented. The chapter closes with a summary and interpretation of this information. In presenting this, the committee wishes to emphasize the limitations of these data and of the results of analyses using it—the work described here is intended to respond to the Department of Veterans Affairs’ (VA’s) request for guidance on how to categorize the self-reported exposures collected by the AH&OBP Registry and not as an endorsement of the registry as a means for gathering exposure data.

INTRODUCTION

Human exposure is defined as contact between a chemical, physical, or biological agent and the outer boundary of a human organism. Exposure is quantified as the amount of an agent available at the exchange boundaries of the organism (for example, the skin, lungs, or gut) (EPA, 2011), and can include a time component as well. It is related to dose, which for a chemical agent is the amount taken into the body by a specific route of entry (such as dermal, inhalation, or ingestion). The pollutant concentration of a substance in a medium (that is, air, water, food, soil and the like) affects the dose taken into or in contact with the body. Substances can cause effects at the point of entry (the respiratory tract, gastrointestinal tract, skin, . . .) or lead to effects elsewhere in the body after absorption. Individuals differ in their absorption, metabolism to more or less toxic chemical intermediates, distribution in the body, and excretion. All of these factors give rise to a substantial degree of variability in the response, and potential health effects, of individuals’ exposure to the same levels of an agent.

In health outcome studies, exposure is ideally measured at the individual level throughout the duration of exposure (by, say, a personal monitor). In absence of such data—which is typically the case—other measures can be used to approximate the relative degree of exposure, such as the amount of time spent in areas with elevated concentrations of airborne substances, the proximity of an individual’s work or residence to an exposure source, or the type of job duties and their associated exposures. In characterizing exposure to burn pits, information on meteorological conditions, the substances and quantities burned at different times, satellite data, and self-reported

information on exposure proximity and frequency might provide additional information on the relative magnitude and duration of exposure. The accuracy of the assessment of exposure, however, depends on the quality of the data and the extent to which surrogate information is representative of individual exposure. All of these measures have the potential for inaccuracies, bias, and confounding which affect the ability to detect or observe a true association between an exposure and resulting health effects.

Sources and Nature of Exposures

The duration and frequency of exposures to environmental contaminants vary greatly for service members, and the exposures occur in the presence of other physiological stressors. Because many U.S. personnel involved in the Southwest Asia theater of operations worked and lived at fixed sites, those who were in the presence of burn pits may have experienced exposures to burn pit emissions during the majority of their deployment to those sites, although the magnitude and nature of the exposures would have varied. Occupational duties, for example, may put some service members in close proximity to burn pits and other sources of airborne pollutants. Some exposures would be expected to be of relatively short duration: Joint Base Balad had a large burn pit, but the base functioned as a transit stop, resulting in a short-duration exposure for many service members passing through (IOM, 2011). Moreover, meteorological conditions varied, affecting whether pollutants were transported toward or away from individuals and affecting the rate of dissipation of the pollutant. Service members thus experienced large variations in the duration, frequency, and magnitude of exposures which are not easily characterized.

Burn Pits

From the beginning of the Southwest Asia conflicts, the uncontrolled open-air burning of waste in Iraq and Afghanistan has been the primary solid-waste management solution in those theaters of operations. Waste included that which would be expected in any community, such as food remains and latrine waste, but it also included plastic bottles, electronics, waste from medical facilities, weapons and munitions, paint, petroleum and lubricant products, plastics and Styrofoam, and rubber tires (VA, 2016). The usual waste management systems of land-filling, recycling, and incinerating these items were often not feasible, instead necessitating the use of open burning. However, open burning of the items generates more byproducts of incomplete combustion, thereby increasing potential health risks.

In 2009 the U.S. military’s use of burn pits in Iraq and Afghanistan was restricted by law, and by the end of 2010 their use in Iraq had gradually been phased out, but it did continue in Afghanistan, where 197 burn pits were operating as of January 2011 (IOM, 2011). The use of burn pits varies depending on the size of the base, its activities, and its population. In Iraq, as of November 2009, burn pits were operating at 14 out of the 41 existing small military sites (defined as housing less than 100 U.S. service members), 30 of the 49 medium-size sites (between 100 and 1,000 service members), and 19 of the 25 large sites (more than 1,000 service members); however, data were not available for all sites (DoD, 2011). The number of burn pits used in Iraq declined in response to the 2009 law, and a 2010 Government Accountability Office (GAO) study of open-air pit burning in Iraq and Afghanistan listed only 22 in use in Iraq in August 2010 (GAO, 2010). Their use in Afghanistan continued, however, and in January 2011, 126 out of the 137 small sites, 64 of the 87 medium-size sites, and 7 of the 18 large military installation sites in Afghanistan still had operating burn pits (DoD, 2011).

In 2011, the Department of Defense (DoD) estimated that an average of 8 to 10 pounds of waste was generated each day by each person in theater (DoD, 2011). Joint Base Balad—the largest base, serving up to 25,000 people at a time—burned perhaps 100–200 tons of waste a day in 2007. By 2009, three incinerators were operational on the base, but the burn pit was still in use, burning approximately 10 tons of waste daily until it ceased operation in October 2009. A 2010 Army Institute of Public Health study of burn pits in Iraq and Afghanistan reported that large bases burned waste that consisted generally of 5–6% plastics, 6–7% wood, 3–4% miscellaneous noncombustibles, 1–2% metals, and 81–84% combustible materials (APHC, 2010; IOM, 2011).

Other Airborne Exposures

Air quality in the theaters of operation is affected by winds, temperature, humidity, meteorological events, and dust storms as well as by anthropogenic (civilian and military) sources such as power plants, industrial facilities, trash burning, agriculture, the Al-Mishraq (Iraq) sulfur plant fire (Baird et al., 2012), combat dusts from mortar fire and improvised explosive devices (IEDs), and—in the case of the 1990–1991 Gulf War—soot from oil-well fires. Additional emissions resulting from activities on military bases, including combustion products from vehicles, aircraft, and generators, and the evaporation of volatile compounds in fuel and occupational settings, such as vehicle maintenance, are also important considerations (IOM, 2011). Personal activities such as smoking also contribute to airborne exposures. Elevated exposures can thus occur on base, in the field, or in urban areas.

Composition of Air Pollutants

Not only are the emissions released by burn pits a complex mixture of various chemicals and particulates that depend on factors such as the composition of the trash burned, accelerant used, temperature, ventilation, and the burn rate (Woodall et al., 2012), but the composition and magnitude of air pollutants on military bases in theaters of operation are also affected by a variety of other anthropogenic and natural toxicants.

DoD conducted environmental monitoring and health studies at Joint Base Balad in Iraq starting in 2004. The base operated a large burn pit and was a central logistics hub for U.S. forces deployed in support of Operation Iraqi Freedom (OIF)/Operation Enduring Freedom (OEF)/Operation New Dawn (OND). In response to personnel complaints of odor, poor visibility, and health effects attributed to burn pit emissions, the U.S. Army Public Health Command and the Air Force Institute for Operational Health conducted a series of ambient-air sampling and screening health-risk assessments of burn pit exposures there in 2007 and again in 2009. The assessments were designed to measure the concentration of airborne pollutants released by burning at several sites on base and to detect potentially harmful inhalation exposures for personnel (APHC, 2010; CHPPM and AFIOH, 2009; IOM, 2011; Taylor et al., 2008). Even though these efforts were limited by their inability to contribute to individual exposure assessment as well as by their inability to distinguish the contributions from particular sources (combustion engines, burn pits, dust storms, and the like), they do yield some information about the constituents and ambient levels of airborne toxicants that may have been present on bases with burn pits.

A 2011 review of air monitoring efforts at Joint Base Balad conducted by a committee of the Institute of Medicine (IOM, 2011) found that

• Particulate matter (PM) concentrations in ambient air were on average higher than U.S. pollution standards. PM was most likely a result of local sources (vehicle traffic, aircraft emissions) and regional sources (long-range anthropogenic sources, dust storms), although the burn pit likely made some contribution.
• Polychlorinated dibenzo-p-dioxins and dibenzo-p-furans (PCDD/Fs) were detected at low concentrations. Although species associated with greater toxicity were higher than generally found in the United States or urban environments worldwide, they were lower than levels associated with some individual sources.
• Concentrations of volatile organic compounds (VOCs) and polycyclic aromatic hydrocarbons (PAHs) were similar to those reported in major urban areas outside the United States with major sources being regional background, ground transportation, stationary power generation, and the airport at Joint Base Balad (IOM, 2011).

Subsequent studies have also noted the contribution of the Joint Base Balad burn pit to PCDD/Fs on base as well as the important role of other sources of emissions including the airfield as the primary source of PAHs. Other important contributors to PAH levels were aircraft, vehicle emissions, space heaters, and diesel generators (Masiol et al., 2016a,b).

These observations were limited to the pollutants that were targeted by DoD and the conditions (meteorological, waste stream composition, operating conditions) present at the time of the measurements. Several criteria pollutants commonly monitored in the United States and likely released by burn pits were omitted by DoD’s

sampling, including sulfur dioxide, ozone, nitrogen dioxide, and carbon monoxide. Other pollutants not included in the sampling included those known to be associated with the burning of household waste (EPA, 1997, 2001; Lemieux et al., 2003, 2004), geologic material, carbon from combustion sources, metals from regional smelting activities, and other gaseous pollutants produces by combustion engines (Engelbrecht et al., 2009; IOM, 2011). Thus, the available monitoring data provide information on exposures to the major types of constituents from burn pit emissions, but they lack information on other chemicals that were likely present as well as exposure variability over time (IOM, 2011).

Other data collected as part of a monitoring program for a solid waste disposal facility at the Bagram Airfield in Afghanistan emphasize the variability of exposures associated with burn pits (Blasch et al., 2016). The facility operated a burn pit from 2005 to 2012. The investigators collected breathing zone samples, unlike the case with Joint Base Balad, but only PM and VOCs were studied. Sampling was conducted at four security locations (up to 125 meters from the burn pit) and a control location (4 km from the burn pit) during 30 12-hour shifts. Among the VOCs detected, only Acrolein exceeded the 1-year military exposure guideline but benzene was detected in all samples. The range of PM concentrations varied considerably in association with airfield activity (vegetation removal, demining, road construction, vehicle traffic, industrial activity, air traffic). The highest recorded concentrations of environmental PM_2.5¹ (0.615 mg/m³) occurred at the solid waste disposal facility where the burn pit and incinerators were located. High PM_2.5 and PM₁₀ concentrations were also noted at the bazaar, a highly populated site with unpaved roads and considerable vehicular traffic. The investigators concluded that “[t]he diversity of results support the concept of a complex environment with multiple polluting sources and changing meteorological and operational conditions” (Blasch et al., 2016, p. S38).

Limitations of Exposure Information

Emissions from burn pits were only one of the many potential exposures experienced by military personnel deployed to the Southwest Asia theater of operations. Other exposures included agents used as preventive measures (such as vaccines, pesticides, and insecticides), hazards of the ambient environment (such as sand, insects, air pollution, and endemic diseases), job-specific agents (such as paints, solvents, and diesel fumes), war-related agents (such as smoke from oil-well fires and depleted uranium), and hazards associated with cleanup operations in the 1990–1991 Gulf War (such as sarin and cyclosarin) (IOM, 2010). Neither DoD nor VA records nor other sources contain detailed information on all the agents to which military personnel might have been exposed, at what doses, or for what amount of time. The number and combination of these sources make it difficult to examine whether any agent or combination of agents may have caused or exacerbated health problems in a deployed population. Further complicating such research is the fact that other physical and psychological stressors that may have been experienced in addition to the airborne chemical and particulate exposures mentioned here could also have an effect on some health outcomes.

Determining whether veterans and service members deployed to the Southwest Asia theater of operations face an increased risk of illness because of their exposures during deployment requires information about specific agents, durations of exposure, routes of entry, internal dose, and documentation of adverse reactions. DoD initiated an environmental monitoring effort following the 1990–1991 Gulf War primarily because of concerns related to smoke from oil-well fires and possible exposure to sarin and cyclosarin, and it focused its information-gathering accordingly (IOM, 2010). When the military engaged in OEF in Afghanistan in 2001 and OIF in Iraq in 2003, it sampled the air, water, and soil to characterize the deployment environment. Using this baseline exposure information, DoD—through the U.S. Army’s Public Health Command (formerly the Center for Health Promotion and Preventive Medicine)—designed and implemented its Enhanced Particulate Matter Surveillance Program to characterize and quantify particulate matter in the ambient environment at 15 sites in the Persian Gulf region (NRC, 2010). This was one of the first attempts to measure and characterize exposures to PM that could be used for studies of the health effects of service members and veterans deployed to those areas. The National Research Council (NRC) found, however, that this effort had several flaws in its methodology, and that its study design

___________________

¹ PM_2.5 is particulate matter with a diameter of 2.5 microns or smaller.

limited its usefulness. Accordingly, the NRC recommended that the methodology be improved before undertaking future monitoring efforts. Limitations included inappropriate design or analyses to address objectives that were not clearly defined, a lack of validation of the sampling equipment for high levels of PM concentrations recorded at the sites, the failure to collect sufficient particle mass and composition data on a consistent basis to be useful for quality assurance and for health-effects studies, and the use of different filter media, each analyzed with different techniques, which limited comparability among results and precluded source apportionment and mass-balance assessments (NRC, 2010).

Nevertheless, results from this PM exposure assessment effort clearly showed that service members deployed to the Southwest Asia theater of operations were exposed to high concentrations of PM and that the particle composition varied considerably over time and location. The NRC committee further concluded that to appropriately investigate health effects resulting from exposure to a complex mixture of pollutants, the monitoring strategy needs to be tailored to the specific goals and hypotheses that future health-surveillance and research studies are designed to address (NRC, 2010). In other words, study design should be based on a priori hypotheses that encompass basic analysis plans as opposed to attempting to design a study to fit the data after it has been collected, sometimes for a different purpose.

Other efforts to understand individual military personnel exposures are also fraught with challenges. One major issue is that reconstructing past exposure events is difficult and prone to problems. For example, one commonly used method to collect exposure information is to survey subjects about their perceived exposures to various agents, but this is limited by error in recollection and recall bias, which only increase with time. Individual responses have rarely been verified by in situ measurements or records (IOM, 2010).

Various exposure assessment tools and methods have been used in research to attempt to fill gaps in exposure information. Models specific to certain chemical exposures have been proposed (such as for in-theater exposure to sarin; GAO, 2004), but are of questionable reliability because of the difficulty in incorporating meteorological data, transport and dispersion data, and service member-unit location information—types of information that may not be recorded accurately, if at all, or easily available. For example, unit locations are rarely accurate, and are not reliable for reflecting the actual location of individual service members because members of the same unit may be deployed to different forward operating bases (GAO, 2004; IOM, 2006, 2013) These limitations on individual locations and experiences related to exposure greatly limit assessments of associations with, or the likelihood of increased risk for, diseases, symptoms, or other adverse health effects that are due specifically to airborne hazards for service personnel.

This committee concurs with the previous NRC (2010) committee that concluded that to appropriately investigate health effects resulting from exposure to a complex mixture of pollutants, the monitoring strategy needs to be tailored to the specific goals and hypotheses that future health-surveillance and research studies are designed to address (see Box 5-1). That includes matching the monitoring period with the deployment period of the military personnel being studied. In particular, different types of exposure monitoring may be required for the study of exposure events or conditions leading to potential persistent effects, such as asthma and chronic obstructive pulmonary disease, compared with the study of acute effects, such as day-to-day variability in respiratory or cardiac responses (NRC, 2010). Future monitoring studies should include other ambient pollutants that military personnel may be exposed to in the field and that may be relevant to human health, such as ozone, hazardous air pollutants, and other materials such as diesel exhaust. Consideration should also be given to the toxicity of mixtures. Although the health effects of mixtures are complex and incompletely known, the characterization of all relevant constituents is a necessary first step in understanding which combinations or concentrations might have combined effects for certain health endpoints. In addition, more repeated sampling with a consistent filter type would provide a greater library of gravimetric and chemical-specific data and thus increase statistical power. Finally, increasing the sampling frequency would make it possible to estimate more accurate annual-average concentrations of particle mass and chemical components (NRC, 2010).

EXPOSURE INFORMATION COLLECTED BY THE AH&OBP REGISTRY

The exposure information collected by the AH&OBP Registry questionnaire consists of self-reports of deployment exposures, occupational activities, and personal habits. The result is qualitative information on individual-level exposures or exposure potential. This section briefly summarizes the questionnaire’s collection of exposure information and discusses its limitations. Questions are discussed in the order in which they are asked in the questionnaire, which is reproduced in Appendix C: deployment-related exposures (Section 1.2); military occupational exposures (Section 1.3); and environmental exposures and regional air pollution (Section 1.4). Exposures not related to military service, including nonmilitary occupational exposures (Sections 5.1–5.5) and residential and hobby-related exposures (Section 6), are collected by the questionnaire but are not discussed here given the low response rates and quality issues with the questions themselves. Exposures to tobacco smoke (Sections 2.5 and 2.6) collected by the questionnaire are discussed as potential confounders of health effects in Chapter 6. Assessment of the questions themselves is found in Chapter 3.

Location-Specific Deployment Exposures

Section 1.2 asks respondents about exposures related to each deployment that the respondent verifies from the VA Defense Information Repository Database (referred to as a “deployment segment”). Thus, a person with multiple deployments will answer these questions multiple times. For each deployment, respondents are asked about three sources of emissions— oil-well fire smoke, burn pits, and sewage ponds—and about where the person spent the majority of his or her time with relation to these exposures.

The first of these questions (1.2.A), “Were you exposed to soot, ash, smoke, or fumes from the Gulf War oil well fires?” was asked only of individuals who had deployed during or directly after the 1990–1991 Gulf War. A yes or no response is required.² Two follow-up questions (1.2.B and 1.2.C) ask about deployment location(s) in the 1990–1992 time period. Respondents are asked to identify the base where they spent most of their time using a drop-down list or to type in an answer. They are then asked for the location where they spent the second most amount of time. Because the committee was only granted basic deployment information (country and year), these responses were not made available to the committee and thus were not part of the analysis.

The next four questions collect information about exposure to burn pits during each deployment. For service members who answer “yes,” indicating that they were “near” a burn pit during a deployment (1.2.D), follow-up questions elicit who ran that burn pit (1.2.E), if their duties included direct work with the burn pit (1.2.F), and the numbers of hours they were exposed in a typical day (1.2.G). The committee focused on those later two questions, which characterized exposure to burn pits:

• Did your duties during these dates include the burn pit (examples include trash burning, hauling trash to the burn pits, burn pit security, trash sorting at the burn pit)? [Yes or No]
• On a typical day, how many hours did smoke or fumes from the burn pit enter your work site or housing? [Never or Enter 1–24 hours]

Two additional questions ask about the number of hours that a service member was outside or in an open tent or shelter (1.2.H) or near a sewage pond (1.2.I) on a typical day. For both questions, respondents could answer “never” or enter a number of hours. These questions had relatively high rates of nonresponse: 18.4% and 38.4% for Questions 1.2.G and 1.2.I, respectively. Additionally, the committee deemed sewage ponds to be a relatively small source of airborne exposure.

General Military Occupational Exposures

In Section 1.3, questions collect information about seven potential exposures related to occupational duties. These questions are asked only once of any respondent and apply to any deployment. The first question in the section asks the respondent to answer yes or no to “Were you ever close enough to feel the blast from an IED (improvised explosive device) or other explosive device?” All of the other questions are formatted in a similar manner, asking the number of days of exposure in a typical month. Those exposures include being near heavy smoke from weapons; being in a convoy; performing refueling duties; performing aircraft, generator, or other large engine maintenance; performing construction duties; or performing pesticide duties. All of these exposures were considered in the committee’s analyses except for pesticide duties (Question 1.3.G). This question appears to have been included to generate information on confounding exposures, but it is too vague and open to interpretation to yield useful information.

Environmental Exposures and Regional Air Pollution

A series of questions in Section 1.4 collects information about air quality, dust storms, and their suspected impact on some symptoms. The first question asks “Did you do anything differently during your deployment(s)

___________________

² Almost all of the questions discussed in this chapter include options for “don’t know” or “I don’t wish to answer.” These are characterized as nonresponses in the committee’s analyses and discussed in Chapter 4.

when you thought or were informed air quality was bad (for example, during dust storms or heavy pollution days)?” Respondents could indicate “yes,” “no,” “never thought of this,” or “I was not informed or aware of bad air quality” (1.4.A). Those who answered “yes” were asked what measures they took (1.4.B).

The next question (1.4.C) asks directly about dust storms as a specific exposure: “In a typical month during your deployment(s), how many days did you experience dust storms?” Respondents could answer “never” or enter a number of days from 1 to 31.

Three questions ask if respondents experienced “wheezing, difficulty breathing, an itchy or irritated nose, eyes or throat” that they “thought was the result of poor air quality” (1.4.D) and, if so, how many days in an average month were they affected (1.4.E). It also asks if they sought care for those symptoms (1.4.F).

Limitations of AH&OBP Registry Questionnaire Exposure Information

The information collected by the AH&OBP Registry questionnaire has a number of limitations, many of which are a consequence of it being self-reported. As noted in Chapter 2, self-reported data are by nature subject to recall and reporting biases. The ability of service members to accurately recall their exposures and activities—including duration and frequency, sometimes over multiple deployments as long as 25 years ago—is open to question. Another problem is the tendency for some individuals to over- or underreport exposures (reporting bias). This can result in exposure misclassification—that is, individuals who are actually exposed do not report relevant exposures, or individuals who were not or less exposed report that they were more exposed. Too much misclassification in one direction or the other can result in attenuated associations or spurious conclusions.

In considering the quality of the information provided for each deployment (Section 1.2), the potentially many responses required of respondents—particularly for those with multiple deployments (and thereby the potential for a longer duration of exposure)—may have resulted in less accurate information for those with more exposure because of the difficulty in remembering each individual deployment, or because of response fatigue (Bosnjak and Tuten, 2001). The amount of information or attention given to responding to each question may have decreased with each round of deployment-related queries in the section. Gasper and Kawata, in a 2015 analysis of respondents, reported that

And if those with a greater potential for exposure were more motivated to complete the questionnaire if they had symptoms, this could result in reporting bias.

The accuracy of self-reported data can be improved if there is a way to confirm its validity. However, in the case of the exposure information collected by the AH&OBP Registry questionnaire, few data were available to verify self-reported exposures, although databases can confirm some deployment information, including dates, locations, and job categories. One way to validate self-reported exposure to burn pits would be to compare self-reported information to data on which bases had active burn pits and when incinerators went into use at various bases, thus defining periods of no or reduced burn pit exposure. The committee did not have access to data that would allow them to make such an evaluation, but Gasper and Kawata (2015) found

While some data were collected on particular predictors of exposure, such as job assignment and deployment locations, without additional information on the numbers of individuals who served in those jobs or at specific locations to serve as denominators in analyses, no inferences can be made about patterns in the prevalence of exposure. For example, if a large number of individuals who report a particular job duty participate in the registry and report a specific exposure, there is no way of differentiating among several possibilities—whether that reflects a large number of individuals who have those job duties, whether there was a greater motivation on the part of those with certain job duties to enroll in the registry, or whether a greater proportion of individuals with those job duties were exposed.

The IOM’s 2011 report Long-Term Health Consequences of Exposure to Burn Pits in Iraq and Afghanistan highlights additional potential issues with the use of the registry data, beyond those identified earlier that are associated with using such data in general. While the levels of pollutants found at Joint Base Balad were elevated, urban areas around the globe have elevated levels of many of the pollutants observed as well, so individuals’ exposures can be elevated not only when they are in the field, but also during duties and activities related to being in urban environments. Furthermore, it is important to note that similar exposures can be experienced at other sites that are not related to deployment. Such exposures may also contribute to, or exacerbate, the development of health conditions related to burn pit exposure.

The questions discussed above do not provide information on the intensity of exposure beyond a binary yes/no for exposure, and this is a major shortcoming. Intensity is a central component of exposure characterization. Information regarding it could be obtained from direct questions (a self-reported rating of the intensity of the smoke, for example), or indirect questions (How far from the burn pit did you live or work?), or by asking additional questions about burn pit duties or more frequent or longer exposure to burn pits. Other studies have used assessments of proximity to the source in studies of burn pit exposures (AFHSC et al., 2010; Jones et al., 2012; Powell et al., 2012; Smith et al., 2012), but such information on individual locations was not available to the committee.

The lack of good quantitative data for exposure assessment purposes also includes information that might allow one to characterize acute or chronic exposure or to identify the constituents and chemical species present, both of which are important to understanding potential health effects. For example, the health effects known to be associated with acute exposure to PM (such as increased mortality secondary to respiratory or cardiac diseases) are different from those associated with chronic PM exposure (such as lower respiratory symptoms and reduced lung function) (WHO, 2006). The AH&OBP Registry does not contain quantitative information on the level of exposure (beyond yes/no) and duration, so differentiating between acute or chronic exposures is not possible. Furthermore, there are several factors that may influence the toxicity of airborne PM, including bulk chemical composition, trace element content, strong acid or sulfate content, and particle size distribution (Harrison and Yin, 2000). Because there are no data available describing those aspects of exposure that can be linked to the registry, the ability of the data in the registry to be used to investigate health effects is further weakened.

THE COMMITTEE’S ANALYSIS OF AH&OBP EXPOSURE INFORMATION

The committee’s analysis of the exposure data collected by the registry questionnaire included a careful inspection of descriptive data, consideration of how to create exposure variables, and a look at potential cumulative exposure. The examination in this chapter is limited to descriptive statistics for the questionnaire items related to exposures and the creation of independent variables describing exposure for use in the analysis of potential associations between exposures and health effects (discussed in Chapter 6). All analyses were carried out under the committee’s direction by an external contractor using data supplied directly to the contractor by VA—neither committee members nor staff had access to the raw data.

The committee only examined exposure data as collected by the registry questionnaire. However, data on deployments (as described in Chapter 4) were extracted by VA from two DoD Defense Manpower Data Center

datasets, the Gulf War Oil Well Fire Smoke Registry and the Contingency Tracking System (CTS) database. Deployment information was limited to country and year of deployment.

While other sources of data could potentially serve as exposure proxies (such as distance from the specific source of concern), the quality of much of that data is poor and is plagued by many of the same limitations as the registry data. Other data pertaining to exposures that could be linked to the registry data were not available to the committee.

During its deliberations, the committee recognized the variety of limitations to using the registry data as discussed above, with the objective of identifying potential methods to use the exposure data appropriately in its further analyses and execution of its charge. The committee’s primary concerns included that exposures are multidimensional (time, intensity, chemical composition, and the like) and that data characterizing those dimensions were lacking, the high degree of correlation between responses to questionnaire questions, the potential for bias in those responses, and the high nonresponse rates for some questions. The text below describes the approaches the committee took to making the best use of the available data without over-interpreting it.

Descriptive Statistics

The committee focused its analysis on six potential sources or types of airborne emissions: burn pits, dust (including sandstorms and desert environment), diesel, exhaust, and fuel (including jet fuel, combat, construction, and soot [the last for 1990–1991 Gulf War respondents only]). All of these exposures are important contributors to PM, VOC, PAH, and PCDD/F exposures in theater, so it is inaccurate to solely focus on exposure to burn pits, and it is not reasonable to assume that individuals were exposed to these airborne substances as a result of burn pits alone.

Table 5-1 lists the exposures of interest, applicable questions in the questionnaire, and the type of response for those questions included in the committee’s analyses.

Responses to questions related to exposures are summarized in Table 5-2. The majority of respondents indicated that they were near burn pits during their deployments (62.6% of deployments, but 96% of individuals overall; Question 1.2.D) but only about one-third of respondents indicated that their duties were related to burn pits (32.0% of deployments; Question 1.2.F). On average, respondents reported 7.5 hours per day of smoke or fumes from burn pits at their work site or housing, but 16% reported 24-hour exposure; about 7% reported 12 hours of exposure per day; and 15% reported 6 hours of exposure or less per day. A large proportion (72%) also reported having been near an IED blast or other explosion. For several of the duration questions (1.2.G, 1.2.I, 1.3.B, 1.3.C, 1.3.D, 1.3.E), about 15% of respondents indicated continuous exposure (24 hours per day or 31 days per month).

Not surprisingly, the reported exposure to dust storms was quite high, with a mean of 8.9 days per month. This, along with the relatively frequent convoy-related activities, would suggest that exposures to PM would be elevated. While it is true that PM resulting from dust storms and kicked up by convoys will be compositionally different than combustion-related PM (Cassee et al., 2013; Engelbrecht et al., 2008; Lyles et al. 2011), current standards do not differentiate risks based on PM composition (EPA, 2013).

Other than burn pit, dust storms, or Gulf War oil-well fire soot exposure, most respondents (68%) endorsed only one exposure related to their military service (Questions 1.3.A–G; heavy smoke from weapons, convoys, refueling duties, large engine maintenance, construction, and pesticides, Question 1.4.C not included). The numbers and percentages of respondents who reported up to 5 or more exposures are presented in Table 5-3.

The committee examined the distribution of responses to questions pertaining to the durations of exposures. Those distributions are depicted in Figures 5-1 and 5-2. In Figure 5-1 the duration of exposure to burn pits in hours per day, as reported for each deployment, is shown by bars that represent the number of deployments for which the duration of burn pit exposure was reported (1.2.G). Two-hour increments for exposure duration were chosen due to the tendency of respondents to endorse an even number of hours. Of the more than 206,000 eligible deployment segments, 37% skipped the question (reported no burn pit exposure on Question 1.2.D) and 12% responded “don’t know,” represented by the nonresponse bar to the right of the figure. Elevated frequencies for each duration occur at 24 hours (16% of deployments), 12 hours (7% of deployments), and 4 hours (4% of deployments).

TABLE 5-1 Exposures of Interest and Associated Questions

^a For questions that required input of time, “never” was the first response option and the range of hours or days was available as the second possible response to the question.

^b Question 1.3.C was used to examine dust, diesel, exhaust, and fuel, and combat exposures.

Figure 5-2 shows the distribution of days per month of exposure to other hazards. The questionnaire did not ask respondents to estimate their burn pit exposure in days per month. Responses were grouped into 5-day categories because a review of the raw data suggested that those categories were natural breaking points, with greater numbers of respondents endorsing 5, 10, 20, 25, or 31 days of exposure per month. Most respondents indicated 0 to 5 days of exposure; however, for Questions 1.3.B, 1.3.C, 1.3.D, and 1.3.E, the second most commonly reported duration was 26 to 31 days.

Respondents with No Exposure

A small proportion of respondents indicated that they were not exposed to airborne hazards or open burn pit emissions. With regard to burn pit exposure, fewer than 18% of deployment segments represented in the registry were not near burn pits (1.2.D), and 29% of deployment segments represented in the registry did not have duties that involved the burn pit, (1.2.F; see Table 5-2). Pertaining to exposures to blasts, smoke from weapons, convoys, refueling, large engine maintenance, construction, and pesticides (Questions 1.3A–G), 24% of respondents indicated that they were not exposed to any of these hazards. Only 1% of respondents indicated that they had never experienced a dust storm (1.4.C). Among respondents who had been deployed in the 1990–1991 Gulf War, 6% reported that they were never exposed to soot from Gulf War oil-well fires (1.2.A; see Table 5-2).

TABLE 5-2 Descriptive Statistics for Questions Related to Deployment, Environmental, and Occupational Exposures

NOTE: Med = median; SD = standard deviation.

^a Don’t know, missing, refused, or skipped.

^b Question asked only for Gulf War deployments.

^c Question asked for each deployment.

TABLE 5-3 Number of Exposures Other Than Burn Pits, Gulf War Oil-Well Fire Soot, or Dust Storms Reported by Respondents^*

* These exposures are blast, smoke from weapons, convoys, refueling operations, large engine maintenance, construction, or pesticides associated with military occupations (Questions 1.3.A–G).

All respondents reported exposure to at least two of the ten exposures that all respondents were asked about (soot was asked about only for Gulf War veterans): burn pits (1.2.D), sewage ponds (1.2.I), blasts (1.3.A), smoke from weapons (1.3.B), convoy (1.3.C), refueling operations (1.3.D), large engine maintenance (1.3.E), construction (1.3.F), pesticides (1.3.G), and dust storms (1.4.C). One individual reported exposure to just 3 of the 10 hazards, and only 4 reported exposure to only 4 of the 10 hazards. However, 2% of respondents did not answer all 10 questions (missing, don’t know, or refused responses).

Because the proportion of respondents who were not exposed is small, a better reference group for analyses would be those with levels of exposure believed to be relatively low.

Committee-Created Exposure Variables

Several approaches were used to explore the data because each approach was viewed as having significant limitations by itself, given the underlying issues with the data. By using several approaches, the committee aimed for a more complete picture of the exposures. Thus, the committee examined burn pit exposure based on responses to the individual questions related to burn pit exposure as well as with a committee-created variable that combined those responses to create an exposure potential score. The committee also developed a cumulative metric for use in multivariate analyses.

One method the committee developed resulted in a reduced set of exposure metrics to characterize exposures to sources; the metric included information on whether an exposure may have occurred and, when possible, the duration of the exposure, with the recognition that any metric was potentially flawed, so undue precision would not be assumed. Self-reported information on exposures was interpreted as providing indicators of potential exposure to those agents (the committee uses the term “exposure potential” in its analysis to reflect this). Having a reduced set of metrics that more broadly characterizes exposures also serves in the interpretation of the exposure-related questions. These metrics can be used themselves or combined to provide a distribution of overall exposures to multiple potential agents.

Because of the qualitative nature of the information collected, ordinal variables expressing low, middle, or high levels were created for each of six exposures of interest (burn pits, dust, diesel/exhaust/fuel, combat, construction, and Gulf War oil-well fire soot) to express the gradation of exposures. Categories of exposure were assigned based on a score that incorporated responses to one, two, or three questions that, collectively, characterized the exposure potential of each respondent. (Gulf War oil-well fire soot exposure is based on one question, and that exposure is specific to soot exposure in Kuwait during the Gulf War, as opposed to soot exposure to burn pits or other combustion sources. This exposure was viewed as being particularly unique, given the timing and source differences.) The magnitude of the score is used as an indicator of the magnitude of exposure, ranging from 0 to indicate never exposed or no exposure to 6 to indicate greater exposure. The integration of both the binary and

duration aspects of the questions captures both potential acute and chronic exposures, recognizing that individuals have varying levels of response and reflecting the limited information on exposure–response relationships for the agents of concern. A representation of how the exposure potential score was derived for each exposure (burn pits, dust, diesel/exhaust/fuel, combat, construction, and Gulf War oil-well fire soot) is presented in Box 5-2.

For burn pit exposure potential, the possible responses to Questions 1.2.F and 1.2.G (see Table 5-1) were combined. Because burn pit exposure was assessed for each deployment and the committee sought a representation of exposure for each individual, the created variable used responses to Question 1.2.F to indicate ever exposed (3) or never exposed (0) and the cumulative number of hours of burn pit exposure collected by Question 1.2.G across deployments. For persons with a cumulative number of hours of burn pit exposure in the highest tertile, 3 points were assigned, for the middle tertile, 2 points were assigned, and for the lowest tertile, 1 point was assigned. By adding the scores for each part (ever/never exposed + tertile of cumulative hours of burn pit exposure), the resulting exposure potential scores ranged from 0 (never/no exposure) to 6 (ever exposed/highest tertile of cumulative hours).

The dust exposure potential scores were based on responses to Questions 1.4C, 1.3C, and 1.3F, which asked the respondent to report the number of days per month that the respondent experienced dust storms, was in a convoy, and performed construction duties. Scores were derived by adding the points assigned for each questions (highest tertile, 2 points; middle tertile 1 point; and lowest tertile 0 points). The resulting dust exposure potential score ranged from 0–6.

The diesel/exhaust/fuel exposure potential scores were based on responses to Questions 1.3C, 1.3.D, and 1.3.E, which elicited the number of days per month the respondent was in a convoy, performed refueling duties, and performed large engine maintenance. Scores were derived by adding the points assigned for each questions (highest tertile, 2 points; middle tertile, 1 point; and lowest tertile, 0 points). The resulting diesel/exhaust/fuel exposure potential score ranged from 0–6.

The combat exposure potential score was based on responses to three questions (1.3.A, 1.3.B, and 1.3.C). The possible responses to 1.3.B and 1.3.C were combined so that respondents with a high number of days exposed to combat-related smoke and a high number of days in a convoy (and thus were potentially exposed to combat-related stressors), were assigned a score of 4, whereas respondents reporting low exposure to both were assigned a score of 0. Responses indicating exposure to blast were incorporated by adding two points for a positive (yes) response to question 1.3.A. This resulted in a range of scores from 0 to 6 for combat exposure.

The construction exposure potential was based on one question that asked the respondent to report the number of days per month that he or she performed construction duties (1.3.F). Again, points were awarded based on tertiles, the lowest tertile was assigned a score of 0, the middle tertile was assigned a score of 3, and the highest tertile was assigned a score of 6 so that scores ranged from 0 to 6.

The Gulf War oil-well fire soot exposure potential score was based on the response to one question (1.2.A) with a yes or no response. A response endorsing exposure to Gulf War oil-well fire soot was assigned a score of 6, whereas 0 was assigned to responses reporting no exposure.

The exposure potential scores were transformed into categories of exposure potential reflecting assumed low, medium, and high potential for exposure so that each service member had one estimate of exposure potential for each exposure variable. Low, medium, or high exposure was meant to reflect the gradation of exposures but not in a quantitative way. Table 5-4 shows how scores for each variable were assigned to low, medium, or high categories of exposure potential.

The distributions for the exposure variables that the committee created for the six exposures of interest are shown in Figure 5-3. A medium level of exposure potential is most predominant for burn pits, dust, diesel/exhaust/fuel, and combat exposure. However, there were nearly as many respondents with a high exposure potential to burn pits as medium. Combat exposure also had a relatively high proportion of respondents in the high and medium exposure potential groups. Low-level exposure potential is dominant for construction. Gulf War oil-well fire soot exposure is different in that nearly all who had eligible Gulf War deployments indicated that they had exposure, which falls into the high exposure potential category. Only 146 Gulf War respondents indicated that they did not have exposure to oil-well fire soot.

TABLE 5-4 Categories of Exposure Potential Based on Exposure Potential Scores for Each Exposure Variable

Collinearity Among Exposure Categories

Potential collinearity among exposure potential scores (0 to 6) was examined using Pearson correlation coefficients. An assessment of collinearity is informative in interpreting the resulting health analyses, and it also reflects the tendencies of persons to report being consistently highly exposed to multiple agents, as discussed previously.

All six exposure potential variables were statistically significantly correlated, with one exception. Construction and Gulf War soot from oil fires were the only two not correlated (R = 0.02, p = 0.15). All other correlations ranged from 0.04 (Gulf War oil-well fire soot and burn pits, p = 0.01) to 0.71 (construction and dust, p <0.01). This indicates that respondents who reported one exposure of interest were likely to report other exposures. However, correlation between dust, diesel/exhaust/fuel, and combat exposure variables was high due to the fact that all three incorporate Question 1.3.C (days per month being in a convoy or other vehicle operations). The same occurs for dust and construction, which both incorporate Question 1.3.F (days performing construction duties).

Exposures Among Gulf War Service Members

Given the different context and activities of the 1990–1991 Gulf War compared with the later conflicts, the committee examined the exposure potential categories (low, medium, high) reported among the 5,595 service

members who had been deployed during the Gulf War time period.³ The proportions of Gulf War service members with low, medium, or high exposure potential to burn pits, dust, diesel/exhaust/fuel, combat, and construction were very similar to those of the complete cohort of respondents (see Figure 5-3). Generally, the difference between Gulf War service members and the larger group was less than 5% for each level and each variable. The greatest difference occurred between the percentage of Gulf War service members with medium exposure potential to burn pits (49.7%) and those of all respondents (41.8%).

Composite Exposure Potential

Because there were many sources of airborne emissions that contributed to a service member’s exposures to PM (dust storms, convoys, construction), and PAHs, VOCs, and PCDD/Fs (refueling operations, convoys, large engine maintenance) in addition to burn pits, and because there were insufficient data⁴ with which to determine which sources contributed the most or posed the most harm, the committee chose to weigh each potential exposure equally and to create a metric that places emphasis on the totality of exposures. This is similar to the U.S. National Ambient Air Quality Standards PM standard which is defined for all particles in a certain size range and does not discriminate by source (EPA, 2013). Given the construction of the variables developed, this approach acknowledges that multiple physiologic insults weigh on the health of the service member.

To examine the experience and potential health effects of highly exposed individuals, the committee devised an approach to identify composite exposure potential levels across all six exposure potential variables. For each individual, the number of low and high exposure potentials, up to six each, was determined. For example, a respondent who had a low exposure potential for two of the exposure potential variables, a medium exposure potential for two variables, and a high exposure potential for the last two variables. A composite exposure potential score to all variables was then created by assigning a 0 to the lowest exposure tier, 1 to the middle tier, and 2 to the most exposed tier from the self-reported exposures. The respondent in the example above would have a score of 6 [(2×0)+(2×1)+(2×2) = 6]. The possible scores range from 0 (low exposure potential to all six variables) to 12 (high exposure potential to all six variables). Respondents with equal numbers of high and low exposures, or many medium-level exposures would score about 6.

The committee-created composite exposure potential score generally followed a normal distribution (see Figure 5-4). The majority of scores (16.6%) were at the mid-point (6), and very few individuals reached either extreme (0.3% on the high end, and none on the low end). This is not unexpected given how the exposure metrics were created (e.g., using tertiles of the distributed variables), though it indicates that few individuals were at one extreme (low or high) for the majority of the exposure metrics. However, the proportion of respondents in the two most exposed groups (composite exposure potential scores 11 and 12) is higher compared with the two least exposed groups (composite exposure potential scores 0 and 1), indicating some skewness. As noted previously, there is a significant degree of correlation among the exposure metrics, and respondents who reported on one exposure of interest were likely to report other exposures.

EXPOSURE METRICS FOR MULTIVARIATE ANALYSES

A set of “cumulative” exposure metrics were developed to characterize burn pit exposure. The multivariate analyses were conducted only among post-9/11 respondents to control for different latency periods and because the types of exposures encountered were likely different for Gulf War and post-9/11 service members. These measures exploit the fact that burn pit exposure is asked separately for each deployment, allowing the construction of cumulative exposure measures by multiplying responses by the length of each deployment. These metrics focused on burn pit exposures and were calculated in three different ways:

___________________

³ This group includes service members who also deployed in the theater of operations in later time periods.

⁴ There are a few exceptions where environmental monitoring data are available for some sort of quantitative exposure assessment, such as at Joint Base Balad, but those data are quite limited and were not viewed as being sufficient for a meaningful exposure analysis (IOM, 2011). Furthermore, the committee did not have access to base-specific deployment information.

1. Cumulative days deployed near a burn pit, derived by summing the number of days of each deployment for which a respondent indicated that he or she was near a burn pit (based on Question 1.2.D) expressed as quartiles;
2. Cumulative days deployed with burn pit duty, derived by summing the number of days of each deployment for which a respondent indicated that he or she had duties that included the burn pit (based on Question 1.2.F) expressed as quartiles; and
3. Cumulative hours of exposure to burn pit smoke, derived as the product of the number of days of deployed times the hours per day that smoke or fumes from burn pits entered the work site or housing (based on Question 1.2.G), summed over all deployments and expressed as quartiles.

Given the importance of the other sources of the main pollutants released by the burn pits (PM, PAHs, VOCs, PCDD/Fs), the committee felt it was appropriate to create a variable that would characterize exposure to these pollutants. To do so, it used the composite exposure potential variable created to qualitatively express the potential exposure to all six main sources (burn pits, dust, diesel/exhaust/fuel, combat, construction, and Gulf War oil-well fire soot). In the multivariate analyses, the composite exposure potential score (0 to 12) was further consolidated and expressed as quartiles. For the three cumulative metrics of burn pit exposure and the composite exposure potential metric expressed as quartiles, the values that define each quartile are presented in Table 5-5.

The association between health outcomes and exposure potential was also examined using the committee’s exposure potential metric (0 to 6) for dust, diesel/exhaust/fuel, combat exposure, and construction. Soot from Gulf War oil-well fires was not examined for the multivariate analyses because the question about soot was only asked of the Gulf War respondents. Burn pit exposure as expressed by the committee’s exposure potential variable was included as a qualitative fourth proxy of burn pit exposure to be consistent with the presentation of the other exposures.

TABLE 5-5 Quartiles of Burn Pit and Composite Exposure Potential Metrics Used in Multivariate Analyses

Changes in Burn Pit Exposure Over Time

The committee examined changes in the patterns of reporting burn pit exposure over time by location. The analyses were limited to country and year of deployment, although others have been able to access and conduct similar analyses with more detailed data, including base location, compared with knowledge about when and where incinerators were in use (Gasper and Kawata, 2015). The number of deployments for which high exposure to burn pits as defined by the committee’s exposure potential variable was reported (rather than by the individual) by the country of deployment (Iraq, Afghanistan, or Kuwait) were examined by year of deployment, shown in Figure 5-5. The use of burn pits in Iraq and Afghanistan but not in Kuwait is well described (Liu et al., 2016).

As shown in Figure 5-5, reported burn pit exposure by year of deployment could be interpreted as an indicator of registry or deployment data validity because there were fewer reports of high burn pit exposures in Kuwait than

for Iraq and Afghanistan, as would be expected. On the other hand, the figure also raises some uncertainty about responses since if there were no burn pits in Kuwait, the exposures should presumably have been even lower than reported and shown. Perhaps the Kuwait data in part reflects the issue of multiple deployments and uncertainty as to where and when the burn pit exposures occurred (for example, military versus civilian trash-burning activities). Smaller-scale trash burning at bases without large burn pits may have been interpreted as burn pits.

As noted early in this chapter, the use of burn pits in Iraq and Afghanistan began to decline in 2009, and one would thus expect to see a dip in the number of deployments for which high burn pit exposure was reported after 2009. While the percentage of deployments to Iraq with high burn pit exposure potential shows a relatively steady decline since 2003, there is no notable deviation around 2009. Deployments to Afghanistan, however, show a rise in the percent of deployments with reported burn pit exposure in 2010 and a decrease after 2012.

SYNOPSIS AND CONCLUSIONS

Based on the information presented in this chapter, the committee has reached the following findings, conclusions, and recommendations regarding the analysis and interpretation of AH&OBP Registry exposure data.

Assessing exposure using self-reported registry data has a number of inherent limitations—even if the registry was well designed and there is a careful selection of items for analysis. The information collected by the AH&OBP Registry questionnaire consists of self-reports of

• location-specific deployment-related exposures (oil-well fires, burn pits, and sewage ponds);
• general military occupational exposures (being near heavy smoke from weapons; being in a convoy; performing refueling duties; performing aircraft, generator, or other large engine maintenance; performing construction duties; or performing pesticide duties);
• environmental exposures and regional air pollution (air quality, dust storms);
• exposures not related to military service, including nonmilitary occupational exposures (such as working as a fire fighter or in a dusty job);
• residential and hobby-related exposures (living near a farm or recreational woodworking, for example); and
• exposures to tobacco smoke and consumption of alcohol.

Other potentially problematic exposures such as endemic diseases, insects, depleted uranium, and hazards associated with cleanup operations in the 1990–1991 Gulf War were not included.

As detailed in Chapters 3 and 4, the committee identified several flaws with the way these data were collected and found that there was a particular problem with deployment-related exposure questions, which asked for specific information for each separate segment of a respondent’s time in theater. In addition, the information collected has a number of limitations, including the fact that self-reported data are subject to recall and reporting biases. The questions do not provide information on the intensity of exposure beyond a binary yes/no for exposure even though intensity is a central component of exposure characterization. While potential surrogates for the intensity of exposure to sources such as distance from a source are often used in analyses like this, such information was not available to the committee.

The data on burn pit exposures are limited by the lack of details on the chemicals and PM that comprised that exposure; other occupational and environmental sources of airborne pollutants, troop location, meteorological, satellite, or other data by which to conduct exposure assessments; and the absence of information that would allow for the consideration of acute versus chronic exposures. The analysis of the exposure data is complicated by the high fraction of registry participants reporting potential exposures to both burn pit emissions and dust, particularly dust storms but also convoys and construction. For many of the questions, there was a very high percentage of respondents indicating exposures, and there was a tendency for individuals reporting exposures to one type of source to report exposures to other sources as well. Some questions had high rates of non-response.

Thus, given the charge, and given the concern for over-interpreting the data at hand, the committee developed a reduced set of metrics to express exposure potential. Because there were many sources of airborne emissions that

contributed to a service member’s exposures to PM (dust storms, convoys, construction), and PAHs, VOCs, and PCDD/Fs (refueling operations, convoys, large engine maintenance) in addition to burn pits, and because there are insufficient data by which to determine which sources contributed the most or posed the most harm, the committee chose to weigh each potential exposure equally and to focus on the totality of exposures. Specifically, the metrics combined responses to multiple questions into single indicators of potential exposure for each of six exposures of interest: burn pits, dust, diesel/exhaust/fuel, construction, combat, and Gulf War oil-well fire soot. Combining responses that are related to a similar exposure can reduce the resulting number of variables (dimensionality) to be considered in health studies and can be used to construct an overall exposure to multiple stressors. The chapter text contains descriptive statistics related to the metrics.

On the basis of its evaluation, the committee concludes that the exposure data are of insufficient quality or reliability to make them useful in anything other than the most general evaluations of exposure potential. Under these limited circumstances, it believes that there may be some circumstances where supplementing these data with information from on-site environmental monitoring or meteorological, satellite, or other relevant measurements or observations might yield results that would better quantify the variation in exposures to specific constituents, thereby allowing more detailed assessments of health outcomes in particular populations.

The exposure potential metrics are also used in the health outcome assessment described in Chapter 6, individually and in combination, to indicate cumulative exposure potential in order to account for both the exposures to sources individually and multiple exposures to potentially harmful agents or stressors. The committee wishes to emphasize that this was done in the service of fulfilling the statement-of-task directive to address associations of self-reported exposures with self-reported health conditions and is not an endorsement of the data’s suitability for this task.

REFERENCES

EPA. 2001. Database of sources of environmental releases of dioxin-like compounds in the United States. EPA/600/C-01/012. http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=20797 (accessed September 23, 2010).

EPA. 2011. Community-based air pollution projects glossary. https://iaspub.epa.gov/sor_internet/registry/termreg/searchandretrieve/glossariesandkeywordlists/search.do?details=&vocabName=Community-Based%20Air%20Pollution&uid=1821538 (accessed September 15, 2016).

EPA. 2013. National ambient air quality standards for particulate matter. Federal Register 78 (10):3086–3287. https://www.gpo.gov/fdsys/pkg/FR-2013-01-15/pdf/2012-30946.pdf (accessed September 15, 2016).

GAO (Government Accountability Office). 2004. Gulf War illnesses: DOD’s conclusions about U.S. troops’ exposure cannot be adequately supported. Testimony before the Subcommittee on National Security, Emerging Threats, and International Relations, Committee on Government Reform, House of Representatives. GAO-04-821T. Washington, DC: GAO.

GAO. 2010. DOD should improve adherence to its guidance on open pit burning and solid waste management. http://www.gao.gov/assets/320/311365.pdf (accessed September 15, 2016).

Gasper, J., and J. Kawata. 2015. An evaluation of selection bias and data quality: Open burn pit and airborne hazards registry. Prepared for the Department of Veterans Affairs Office of Public Health, Post-Deployment Health Group by Westat. September.

Harrison, R. M., and J. Yin. 2000. Particulate matter in the atmosphere: Which particle properties are important for its effects on health? Science of the Total Environment 249(1–3):85–101.

IOM (Institute of Medicine). 2006. Gulf War and health, volume 4: Health effects of serving in the Gulf War. Washington, DC: The National Academies Press.

IOM. 2010. Gulf War and health, volume 8: Update of health effects of serving in the Gulf War. Washington, DC: The National Academies Press.

IOM. 2011. Long-term health consequences of exposure to burn pits in Iraq and Afghanistan. Washington, DC: The National Academies Press.

IOM. 2013. Returning home from Iraq and Afghanistan: Assessment of readjustment needs of veterans, service members, and their families. Washington, DC: The National Academies Press.

Jones, K. A., B. Smith, N. S. Granado, E. J. Boyko, G. D. Gackstetter, M. A. K. Ryan, C. J. Phillips, T. C. Smith, and the Millennium Cohort Study Team. 2012. Newly reported lupus and rheumatoid arthritis in relation to deployment within proximity to a documented open-air burn pit in Iraq. Journal of Occupational and Environmental Medicine 54(6):698–707.

Lemieux, P. M., B. K. Gullett, C. C. Lutes, C. K. Winterrowd, and D. L. Winters. 2003. Variables affecting emissions of PCDD/Fs from uncontrolled combustion of household waste in barrels. Journal of the Air & Waste Management Association 53(5):523–531.

Lemieux, P. M., C. C. Lutes, and D. A. Santoianni. 2004. Emissions of organic air toxics from open burning: A comprehensive review. Progress in Energy and Combustion Science 30(1):1–32.

Liu, J., N. Lezama, J. Gasper, J. Kawata, S. Morley, D. Helmer, and P. Ciminera. 2016. Burn pit emissions exposure and respiratory and cardiovascular conditions among airborne hazards and open burn pit registry participants. Journal of Occupational and Environmental Medicine 58(7):e249–e255.

Lyles, M. B., H. L. Fredrickson, A. J. Bednar, H. B. Fannin, D. W. Griffin, and T. M. Sobecki. 2011. Medical geology: Dust exposure and potential health risks in the Middle East. 34th International Symposium on Remote Sensing of Environment. Sydney, Australia. April 10–15, 2011. http://www.isprs.org/proceedings/2011/ISRSE-34 (accessed September 15, 2016).

Masiol, M., C. T. M. Mallon, K. M. Haines Jr, M. J. Utell, and P. K. Hopke. 2016a. Airborne dioxins, furans, and polycyclic aromatic hydrocarbons exposure to military personnel in Iraq. Journal of Occupational and Environmental Medicine 58(8S):S22–S30.

Masiol, M., C. T. M. Mallon, K. M. Haines Jr, M. J. Utell, and P. K. Hopke. 2016b. Source apportionment of airborne dioxins, furans, and polycyclic aromatic hydrocarbons at a United States forward operating air base during the Iraq war. Journal of Occupational and Environmental Medicine 58(8S):S31–S37.

NRC (National Research Council). 2010. Review of the Department of Defense Enhanced Particulate Matter Surveillance Program report. Washington, DC: The National Academies Press.

Powell, T. M., T. C. Smith, I. G. Jacobson, E. J. Boyko, T. I. Hooper, G. D. Gackstetter, C. J. Phillips, B. Smith, and the Millennium Cohort Study Team. 2012. Prospective assessment of chronic multisymptom illness reporting possibly associated with open-air burn pit smoke exposure in Iraq. Journal of Occupational and Environmental Medicine 54(6):682–688.

Smith, B., C. A. Wong, E. J. Boyko, C. J. Phillips, G. D. Gackstetter, M. A. K. Ryan, T. C. Smith, and the Millennium Cohort Study Team. 2012. The effects of exposure to documented open-air burn pits on respiratory health among deployers of the Millennium Cohort Study. Journal of Occupational and Environmental Medicine 54:708–716.

Taylor, G., V. Rush, A. Deck, and J. A. Vietas. 2008. Screening health risk assessment burn pit exposures, Balad Air Base, Iraq and addendum report. IOH-RS-BR-TR-2008-0001/USACHPPM 47-MA-08PV-08. Brooks City-Base, TX: Air Force Institute for Operational Health and U.S. Army Center for Health Promotion and Preventative Medicine. May.

VA (Department of Veterans Affairs). 2016. U.S. Department Of Veterans Affairs: 10 things to know about burn pits. http://www.publichealth.va.gov/docs/exposures/ten-things-to-know-fact-sheet.pdf (accessed September 15, 2016).

WHO (World Health Organization). 2006. Health risks of particulate matter from long-range transboundary air pollution. Bonn, Germany: European Centre for Environment and Health. http://www.euro.who.int/__data/assets/pdf_file/0006/78657/E88189.pdf (accessed September 16, 2016).

Woodall, B. D., D. P. Yamamoto, B. K. Gullett, and A. Touati. 2012. Emissions from small-scale burns of simulated deployed U.S. military waste. Environmental Science and Technology 46:10997–11003.