The Utility of Proximity-Based Herbicide Exposure Assessment in Epidemiologic Studies of Vietnam Veterans (2008)

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

Summary ABSTRACT  In past studies evaluating whether health problems experienced by Vietnam veterans might be linked to wartime use of Agent Orange or other herbicides, a fundamental challenge has been a lack of information about the veterans’ level of exposure to these herbicides. To address that problem, researchers devel- oped a model to assess the opportunity for herbicide exposure among these veterans. Using a geographic information system and a computerized database engine, the model makes it possible to link georeferenced data on herbicide spray paths with information about troop locations and calculate two different proximity-based exposure opportunity metrics. This report presents the conclusions and recommendations of an Institute of Medicine committee that was convened to provide guidance to the Department of Veterans Affairs (VA) about the best use of this herbicide exposure assess- ment model. The committee concluded that the model’s approach of using an exposure surrogate based on individuals’ or military units’ proximity in space and time to herbicide spray paths is a reason- able one. It is an important improvement over the cruder yes/no exposure classification based on service in Vietnam that has been used in many past epidemiologic studies of the health of Vietnam veterans. However, the proximity-based exposure metrics inevita- bly have some unknown amount of misclassification and must be used and interpreted with caution. Sensitivity analyses should be done to determine the effect that the model’s assumptions have on  

 PROXIMITY-BASED HERBICIDE EXPOSURE ASSESSMENT the exposure assignments it generates, and other proximity-based approaches with the potential of estimating exposure more accu- rately should be explored. Moving beyond proximity-based expo- sure measures would require additional data on herbicide fate and transport, individual behavior, and pharmacokinetics, and some of this information is not likely to be available. To conduct epidemiologic studies using the exposure assess- ment model requires data on when and where each veteran served in Vietnam and on the veteran’s health outcomes. It is generally possible to obtain useful data on individuals’ unit assignments and unit locations, but the processes of gaining permission for access to relevant military records and of collecting data for individuals are likely to be administratively difficult for many researchers as well as time consuming and costly. Despite the shortcomings of the exposure assessment model in its current form and the inherent limitations in the approach, the committee agreed that the model holds promise for supporting informative epidemiologic studies of herbicides and health among Vietnam veterans and that it should be used to conduct studies. The committee offers criteria that VA should draw on as a basis for developing a request for proposals, and it recommends that VA work with the Department of Defense and the National Archives and Records Administration to facilitate access to and interpreta- tion of military records for use in the studies. B etween 1962 and 1971, several herbicides—most notably the p ­ roduct known as Agent Orange—were used in Vietnam by U.S. forces and their allies for defoliation of forest areas, destruction of crops, and control of vegetation around the perimeters of troop encamp- ments. Since then, many studies have been conducted to examine whether health problems experienced by some Vietnam veterans might be linked to their wartime exposure to any of these herbicides or to a particular c ­ ontaminant—2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD)—that was present in some of them. A fundamental and persisting challenge in these studies has been to determine the amount of herbicide in the environment in Vietnam, identify military personnel who were exposed to the herbicides, distinguish them from personnel who were not exposed, and estimate the herbicide or dioxin dose that exposed individuals received. In an effort to improve exposure assessment for Vietnam veterans, a group of academic researchers has Throughout this report the term “herbicide” encompasses the TCDD contaminant unless specifically stated otherwise. 

SUMMARY  developed an herbicide exposure assessment model. This report presents the conclusions and recommendations of the Institute of Medicine (IOM) study committee that was convened to provide guidance to the Department of Veterans Affairs (VA) on the use of this model. The committee found that assignment of exposure based on proximity to herbicide spraying offers the possibility of important improvement in the classification of exposure over most earlier approaches, although there are inherent limitations in the model. Using the model to assess herbicide exposure in epidemiologic studies may permit observation of associations between herbicide exposure and health effects in the Vietnam veteran popu- lation that were not identifiable in previous studies. The committee recom- mends further sensitivity analyses to provide a better understanding of the model’s strengths, limitations, and uncertainties and also recommends exploration of opportunities to refine the model’s estimation of exposure. STUDY BACKGROUND U.S. military personnel were present in Vietnam throughout the 1962– 1971 period when herbicides were in use for defoliation and crop destruc- tion. Among the several different herbicides used, Agent Orange and Agent White were the principal defoliants and Agent Blue was widely used for crop destruction. Available data indicate that 95 percent of the herbicide used in Vietnam was applied by fixed-wing aircraft as part of Operation Ranch Hand. Herbicides were also applied by helicopters and with ground- spraying apparatus. Records for Operation Ranch Hand are considered to be relatively more complete than those for other herbicide use. In 1991, Congress requested that IOM committees periodically review the scientific evidence to assess whether associations may exist between exposure to the herbicides (or the TCDD contaminant) used in Vietnam and health outcomes. Congress originally sought from the IOM reviews an assessment of the increased risk that Vietnam veterans would have for conditions found to be associated with herbicide exposure. None of the review committees have been able to make quantitative assessments, in part, because credible direct or proxy measurements of Vietnam veterans’ herbicide exposure have not been available. The first of the IOM’s biennial reviews included a recommendation that this problem be addressed by an attempt to develop an exposure reconstruction model that could be considered valid for use in epidemio- logic studies of veterans. VA responded to this recommendation by com-   Agent Orange was a combination of 2,4-dichlorophenoxyacetic acid (2,4-D) and 2,4,5- t ­ richlorophenoxyacetic acid (2,4,5-T). Agent White was a combination of 2,4-D and ­picloram. Agent Blue was cacodylic acid and sodium cacodylate. TCDD was a contaminant of 2,4,5-T. 

 PROXIMITY-BASED HERBICIDE EXPOSURE ASSESSMENT missioning the IOM to assess the scientific issues that a research proposal would need to address, to oversee the selection of a contractor to conduct the work, to monitor progress during the course of the contract, and to evaluate the product produced by the contractor. The contract for the work was awarded in 1998 to researchers at the Columbia University Mailman School of Public Health (Jeanne Mager Stellman, Ph.D., principal investiga- tor), who produced the proximity-based exposure assessment model that is the focus of this report. STUDY CHARGE The Committee on Making Best Use of the Agent Orange Exposure Reconstruction Model was convened in 2007 to advise VA on the best ways to employ the herbicide exposure assessment model in the evaluation of the long-term health effects of veterans’ wartime exposure to herbicides. VA requested that the committee consider several factors: the relevant IOM recommendations for evaluating such a model; approaches for evaluating the exposure model using existing data on health outcomes associated with herbicide or dioxin exposure among Vietnam veterans; the availability, quality, and usefulness of existing information on Vietnam veterans, includ- ing data on troop locations and health outcomes for diseases commonly associated with herbicide exposure as well as those not currently linked to such exposures; and how such information might be used in epidemiologic studies using the exposure assessment model. The committee was also asked to consider the role of epidemiologic studies of Vietnam veterans conducted using the new model in informing the IOM’s biennial evaluation of evidence on the association between herbicide exposure and health outcomes. VA’s interests included advice on planning future research using the model, guid- ing researchers to the potentially most fruitful areas of study, and alerting researchers to the challenges in doing studies using the model. Despite the name of the committee, it is important to note that the herbicide exposure assessment model that was reviewed is not an exposure reconstruction model. To the committee, “exposure reconstruction” sug- gests the possibility of arriving at a retrospective estimate of the quantity of herbicide that individuals or groups were exposed to, or even the dose they might have received. Instead, the model produces metrics based on proximity to herbicide spraying that are only surrogates for exposure. Further­more, the committee was not charged with conducting analyses using the exposure assessment model or with conducting an assessment of the scientific evidence on associations between any specific health effects and exposure to herbicides used in Vietnam. 

SUMMARY  AN EXPOSURE Assessment HIERARCHY The committee viewed the Stellman team’s exposure assessment model in the context of an exposure assessment hierarchy for herbicide spraying in Vietnam (see Figure 3-1). At the simplest level, “exposure” is defined based on a veteran’s presence or absence in Vietnam during the period of herbicide spraying. Measures of exposure at the second level are based on information on the location, timing, and volume of herbicide spraying com- bined with information on the location in space and time of ­individuals or military units. At the third level, proximity-based exposure metrics might be refined by the incorporation of more detailed data or models for the fate and transport of herbicides in the environment, such as spray drift models, estimates of the proportion of the sprayed herbicide that reached the ground, or consideration of secondary transport of the herbicides or the TCDD contaminant in the environment. The fourth level of the hierarchy would require data on individual-level interactions with the environment (e.g., dermal exposure to soil, consumption of local food) to better estimate personal exposures and permit examination of differences among units or individuals present at the same places and times. At the fifth and most highly refined level, information on pharmacokinetics would be needed to estimate the doses of a toxic compound that individuals receive. The Stellman team’s model operates primarily at the second level of this hierarchy, relating the location of military units in time and space to the ­ timing, location, and volume of herbicide spraying. Understand- ing this aspect of the model is necessary in order to accurately evaluate its strengths and weaknesses and to advise future researchers about its appropriate use. A TOOL FOR HERBICIDE EXPOSURE ASSESSMENT The Stellman team developed a geographic information system (GIS) and the Herbicide Exposure Assessment–Vietnam (HEA-V), which is a computerized database engine that facilitates the linking of disparate geo­ referenced data as well as the calculation of two exposure opportunity metrics at the spatial scale of approximately 1 square kilometer. The “hits” metric is based on an individual’s or a unit’s presence at a given location within a specified distance from a spray path at the time of a spray mission. Hits are calculated for each cell in the GIS grid that lies within 0.5, 1, 2, or 5 kilometers of a spray path. The other metric is the Exposure Opportunity Index (EOI), which factors in both direct exposure and an estimate of indirect exposure to residual herbicide from spray- ing that occurred before an individual’s or a unit’s entry into a location. The EOI calculations incorporate data on the quantity of herbicide sprayed, 

 PROXIMITY-BASED HERBICIDE EXPOSURE ASSESSMENT the distance from the spray path, the time since spraying, and an environ­ mental half-life. Hits and EOIs can be calculated for a specific location or for military units or individual military personnel when their location his- tories are provided. CONSIDERATIONS IN ASSESSING HERBICIDE EXPOSURE IN VIETNAM FOR EPIDEMIOLOGIC STUDIES The committee’s assessment was guided by four primary considerations. First, it was essential to be clear about the nature of the exposure assess- ment model and what it does and does not claim to do. The committee approached the model as a means of generating a quantitative representa- tion of opportunity for herbicide exposure, acknowledging that it could not provide sophisticated estimates of individual dose or exposure levels. Second, the committee gave careful consideration to the information needed to use the model. This included gaining an understanding of the strengths and limitations of data on herbicide spraying, troop locations, and health outcomes. Issues of access to and usability of these data were also important. Although attention in the past has often focused on TCDD, the spraying database includes information on all herbicides for which records were available, and the committee made its assessment from this broader perspective. Third, the committee considered whether the model “works,” that is, whether it locates spraying and troops accurately (opportunity for expo- sure) and whether doing so is related to actual exposure (through com- parisons with other sources of information on exposure, such as blood levels or environmental samples). Because environmental epidemiology often advances through successive approximations of exposure and not necessarily by applying the standard of absolute accuracy, the utility of the model is defined in part by the methods it improves upon. Fourth, with the nature of the model and its metrics clearly in mind, the committee considered the potential contributions and pitfalls of using it in epidemiologic studies. Of particular interest in these deliberations were the potential to study Vietnam veterans directly, the degree to which expo- sure classification might be improved if the model were to be used, and the appropriate interpretation of the results of any such studies. ASSESSMENT OF THE MODEL AND ITS EXPOSURE METRICS The committee considered it necessary to assess the strengths and weaknesses of the Stellman team’s approach to herbicide exposure assess- ment before­ commenting on the best use of the model. This initial assessment­ 

SUMMARY  focused on (1) the basic data inputs on geography and herbicide spraying, (2) the exposure metrics of hits and EOI, and (3) the assumptions regarding the effects of distance from spraying in time and space that are incorporated into the calculation of the EOI. Infrastructure for Proximity-Based Exposure Measures The committee concluded that the Stellman team’s approach of using an exposure surrogate based on individuals’ or military units’ proximity in space and time to herbicide spray paths is a reasonable exposure assess- ment strategy. This approach is an important improvement over the cruder exposure classification based on service in Vietnam that has been used in many past studies of the health of Vietnam veterans. The databases and GIS were found to provide a useful infrastruc- ture for estimating proximity-based surrogates of exposure to herbicides in Vietnam. However, data on spraying by fixed-wing aircraft are more complete than the data for spraying by ground equipment or helicopters. As a result, the model is currently better suited to examining proximity to fixed-wing spraying. Even so, it is important for the potential contribution of herbicide exposures from helicopter and ground spraying to be taken into consideration in planning and interpreting studies, recognizing that sources of exposure other than from fixed-wing aircraft could introduce misclassification in the rank-order of exposure assignment. In addition, given the significant uncertainties in the levels of TCDD contamination in the herbicides used in Vietnam, proximity-based exposure models may be better suited to studies of the health effects of herbicides for which the active ingredients were consistent over time, such as 2,4-D and 2,4,5-T, rather than TCDD. Should researchers want to distinguish between proximity to Agent Orange and proximity to Agent White or Agent Blue, for example, it is possible to generate separate exposure opportunity values for each agent using the Stellman team’s model. The committee concluded that the proximity-based exposure metrics of hits and EOI have value in that they move in a favorable direction along the exposure assessment hierarchy described above. However, the methods by which the hits and EOI scores are calculated have the potential for exposure misclassification of unknown magnitude, and so these metrics must be used with caution. Other proximity-based approaches to estimating exposure may be more accurate and should be explored using the existing GIS. Moving beyond proximity-based measures would require additional data on meteorologic conditions, herbicide fate and transport, individual behavior, and pharmacokinetics. 

 PROXIMITY-BASED HERBICIDE EXPOSURE ASSESSMENT Testing and Refining Herbicide Exposure Assessment The Committee concluded that factors influencing the environmental fate and transport of the herbicides that were not incorporated into the cur- rent version of the Stellman team’s model (e.g., width of the spray swath, concentration of contaminants, primary and secondary drift, soil condi- tions, initial and remaining canopy, and photodegradation) are likely to have affected exposure to herbicides and their contaminants. Incorporating these phenomena into an exposure model could possibly reduce exposure misclassification, but it would require additional data that may or may not be available. Furthermore, the resolution of the Universal Transverse Mercator (UTM) system used in the military records and the approxi- mately 1-square-kilometer resolution of the Stellman team’s GIS grid map of Vietnam limits to some extent the benefits of adding fine-scale fate and transport modeling. The committee emphasizes that, regardless of the exposure model used, sensitivity analyses are necessary to determine the impact of the model’s assumptions on the exposure assignments it generates. In addition, the committee concluded that it is not feasible to validate the exposure scores produced by the Stellman team’s model—or any other proximity-based model—by comparisons with biomarkers or soil samples because of the passage of time and the unavailability of archived environmental or bio- logical samples. AVAILABILITY AND ACCESSIBILITY OF DATA ON VETERANS The Stellman team’s herbicide exposure assessment model starts from a data infrastructure that focuses on the timing, location, and content of ­ herbicide spraying in Vietnam and the software tools for calculating exposure metrics. To generate exposure metrics that can be used in epide- miologic studies, this infrastructure must be supplemented with data on when and where military personnel served in Vietnam and on their health outcomes. Working with the Joint Services Records Research Center at the Depart- ment of Defense (DoD) and with records held by the National Archives and Records Administration (NARA), the Stellman team has assembled loca- tion data for many of the military units that moved infrequently during the time they served in Vietnam (i.e., stable units) as well as for some combat units, which moved frequently. Conducting a person-based study requires the further step of identifying the individuals who served in these units and gaining access to their military records. However, access to this information is constrained by privacy laws and is especially challenging for researchers who are not affiliated with DoD or VA. 

SUMMARY  The committee concluded that it is generally possible to obtain useful data on individuals’ unit assignments and unit locations. Locations for most stationary units have already been catalogued, and it appears feasible to gather adequate location information for mobile troops. However, the processes of gaining permission for access to relevant military records and of collecting data for individuals are likely to be administratively difficult for many researchers, as well as time consuming and costly. Assistance from experts in the location and interpretation of Vietnam-era military records is likely to be essential for the effective collection of data from these sources. Epidemiologic studies will also require assembling mortality or mor- bidity data for Vietnam veterans. With appropriate identifying informa- tion, mortality data for Vietnam veterans are readily and reliably available through the National Death Index and the Social Security Administration’s Death Master File. Mortality data can also be obtained from VA’s Benefi- ciary Identification and Record Locator Subsystem (BIRLS), but access to this database is more limited. Obtaining morbidity data presents greater challenges. VA has several databases that could contribute information about the health status of some veterans, but only about 20 percent of Vietnam veterans are receiving care from the VA, and access to VA data is typically restricted to VA researchers. However, as more Vietnam veterans reach age 65 and become eligible for Medicare, it will be possible to use Medicare records to conduct morbidity studies. Researchers could also explore other resources such as state cancer registries and hospital discharge datasets. The committee notes that the Air Force Ranch Hand personnel and Army Chemical Corps personnel are two groups of veterans that are not suitable study populations when the Stellman team’s model is to be used, even though their exposures to herbicides are likely to be among the high- est of all veterans. Most of their herbicide exposures were a direct result of duties that required handling or applying herbicides. By contrast, the model is designed to assess the exposure opportunity that would result from unintended proximity to herbicide spraying. EPIDEMIOLOGIC STUDIES USING THE HERBICIDE EXPOSURE ASSESSMENT MODEL Despite recognizing shortcomings in the exposure assessment model in its current form and inherent limitations in the proximity-based approach, the committee concluded that the assessment model holds promise for contributing to informative epidemiologic studies of herbicides and health among Vietnam veterans and that it should be used to conduct such studies. 

10 PROXIMITY-BASED HERBICIDE EXPOSURE ASSESSMENT Two key considerations led to this conclusion. First, the exposure assessment model is applicable to the population of ultimate interest, namely Vietnam veterans. No other group has the confluence of exposures and exposure circumstances experienced by the Vietnam veterans, so given an adequate model, there is inherent value in asking the question in this group. Second, many previous studies of this population have been severely limited with respect to exposure assessment. A more accurate, if still imper- fect, method should increase the specificity of exposure classification and may permit observation of associations between herbicide exposure and health effects in the Vietnam veteran population that were not identifiable in previous studies. The committee also concluded that the ongoing work by VA investiga- tors is constructive in characterizing the logistical challenges and magnitude of effort needed to apply the Stellman team’s herbicide exposure assessment model. However, the committee views this work as too limited and insuf- ficiently accessible to the broader research community to constitute, in isolation, the best use of the model. Two types of study design were judged to be the most promising for application of the model. One approach is a cohort study that would start from military units and identify individual veterans who served in those units. Units might be selectively sampled based on their exposure potential. The other approach suggested by the committee is to build on large cohorts already assembled and pursue nested case-control studies of outcomes of interest within those cohorts. Either approach would require assembling location histories for individual study subjects. Efforts to validate the model solely by examining existing data on health outcomes associated with herbicide exposure are of limited value. If positive, such studies would add support to the model’s potential value, but if negative, the model’s value is not disproved because the levels of exposure may be lower than would cause adverse health effects or the study’s power may be insufficient to address the question adequately. Studies of veterans based on the exposure assessment model should include analyses to assess how sensitive an estimated association between the exposure opportunity metric and specific health outcomes is to different parameters and sources of uncertainty in the exposure assessment measure. Sensitivity analyses that summarize the variability in the model’s exposure metrics under different assumptions underlying the exposure opportunity model will provide ranges and distributions of exposure opportunity ­metrics (hits or EOI) that may be included in statistical estimates, such as rate ratios and logistic regression parameters. In this way, researchers can offer a range or distribution of the estimated risks associated with the exposure and so illuminate the underlying uncertainty in the assumptions of the model. 

SUMMARY 11 In order for other researchers to benefit from such analyses as well as any refinements or extensions of the model, the committee encourages those who conduct studies applying the exposure assessment model not only to address meaningful epidemiologic questions but also to follow the example set by the Stellman team in contributing publicly accessible building blocks for reanalyses and refinements by others. Work that ­investigators—including­ VA researchers—carry out to determine unit locations and calculations used to arrive at exposure indices should be documented and made available, ideally through the Internet, to enable others to repeat the analyses. Any extension or refinement of the model and associated sensitivity analyses should also be documented so that others can evaluate and build upon the work. RECOMMENDATIONS The committee’s conclusions from its consideration of the Stellman team’s herbicide exposure assessment model led it to make the following recommendations: 1. VA should sponsor epidemiologic studies of Vietnam veterans that take into account the criteria below regarding the appropriate characteristics of informative research on herbicide exposure and health outcomes in this population. VA should draw on the criteria as the basis for developing a request for proposals. S  pecifically, to make the best use of the exposure assessment model, epidemiologic studies of Vietnam veterans should have the follow- ing characteristics: a. The study population should be broadly representative of Viet- nam veterans, with care taken to include sufficient numbers of study participants with relatively higher exposure. b. A broad range of health outcomes should be considered, not just those that are suspected of being related to herbicide expo- sure. Where feasible, morbidity should be studied in addition to mortality. c. The health data should be as complete and up-to-date as possible. d. The study should have sufficient statistical power to address the range of health outcomes of concern. e. To isolate the effects of herbicide exposure, potential con- founding factors need to be carefully addressed in the study design or the analytic approach. 

12 PROXIMITY-BASED HERBICIDE EXPOSURE ASSESSMENT f. Analyses should be conducted to evaluate how sensitive the estimated associations between exposure opportunity and health outcomes are to the uncertainty in the exposure oppor- tunity metrics and to varying approaches to estimating her- bicide exposure, possibly including alternative approaches to exposure assignment as discussed in Chapter 3. g. Opportunities to conduct research using the exposure assess- ment model should be open to investigators beyond the VA sys- tem to allow for the benefits of engaging the broader research community and to enhance public acceptance and credibility. 2. In support of the recommended epidemiologic studies, VA should work with DoD and NARA to • facilitate health research uses of military records that are sub- ject to access barriers arising from privacy laws, and • arrange for assistance from DoD and NARA staff with appro- priate expertise to aid researchers in the location and interpre- tation of military records for health research uses. RESEARCH OPPORTUNITIES From its review of the Stellman team’s model, the committee also iden- tified two areas where it urges further investigation. First, efforts should be made to improve and refine the Stellman team’s model by exploring alternative formulations of the proximity-based expo- sure metrics and by incorporating alternative or additional model param- eters that account for more aspects of herbicide fate and transport in the environment. Further development of the model will require an assessment of the additional data needed and the availability of these data. Second, the sensitivity of the Stellman team’s model’s results to changes in parameter values should be assessed systematically. The committee spe- cifically urges attention to the effects of potential inaccuracies in the data on the location of herbicide application or troop presence. It is also important to investigate, especially with any attempt to add refinements to the existing model, the effect of assumptions on factors such as spray swath, the con- centration of the TCDD contamination, primary and secondary drift, soil conditions, initial and remaining canopy, and photodegradation of sprayed herbicide. Although the committee concluded, based on the information it reviewed, that direct validation of the accuracy of exposure assignment is not feasible, it encourages efforts to quantify the degree of accuracy and incorporate those estimates into the sensitivity analysis.