Click for next page ( 12


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 11
Characterizing Exposure ... Final Report As it became apparent that some tasks were larger and more time-consuming than originally thought or it became obvious that some records or databases could not be used for the intended purposes of the research, the proposed projects were modified, augmented, or partly abandoned. This report chronicles some of the activities undertaken by the Columbia University researchers as they met the various challenges posed by the study and in response to continuing communication with the TOM committee. PROJECT 1: MILITARY UNIT AND HERBICIDE SPRAYING DATABASES, AND EXPOSURE ASSESSMENT MODEL DEVELOPMENT The goal of Project ~ the primary task in the research effort was to construct a transportable system for evaluating herbicide exposure of veterans who served in Vietnam. The purpose of the system was to provide data for use in epidemiologic studies. The Columbia University researchers outlined the following specific aims in their proposal: T. Iclentify and fill in remaining gaps in the Military Unit Database-Vietnam (MUD-V) by retrieving, evaluating, and abstracting primary source materials. 2. Develop additional mathematical models for use as exposure opportunity indices (EOls). 3. Carry out sensitivity analyses of models to characterize the robustness of exposure indices to inaccuracies in the locations of the troops. 4. Where inaccuracies or inconsistencies are found, attempt to obtain and incorporate additional or alternative troop-location data from primary sources. 5. Create a final database of troop locations that contains alternative exposure estimates from a variety of models and their reliabilities. 11

OCR for page 11
Characterizing Exposure ... Final Report 6. Incorporate the final database into user-friendly software that will permit future investigators to use the models to assign exposures or to propose their own models. Interview Division Chemical Officers who served in Vietnam and served as principal advisers to division commanders on the use of all chemicals to complete to the extent possible the record! of known spraying missions. Thus, the central goal of the project was to develop a comprehensive database that contained all known information on the military herbicide spraying that had been carried out under Air Force Operation Ranch Hand; by the US Army for perimeter defense and other smaller localized purposes; and in other, unintentional releases. An additional database was planned to contain locations and dates of "residence" of US military units stationed in Vietnam. Consolidation, Quality Control, and Standarcl~zation of Databases Before the initiation of the contract, a set of individual geographic locations of military units assigned to Vietnam were collected by Columbia University investigators Jeanne Mager Stellman and Steven D. Stellman for use in the Agent Orange Veterans Payment Program (AOVPP), in collaboration with Lt. Col. Richarc! Christian (ret.~. The Drs. Stellman were consultants to the special master presiding over this program, which resulted Dom the Agent Orange Product Liability Litigation a class-action lawsuit brought by Vietnam veterans and their families regarding injuries allegedly incurred as a result of the veterans' exposure to chemical herbicides cluring the Vietnam war. In that effort, the Columbia University researchers created a military-unit database for claim evaluation. The database contained about 500,000 records, each of which provided an exposure opportunity indexi for one military unit during a discrete period. The exposure of any individual claimant was calculated by summing the tabled exposures for his or her unit(s) during service in Vietnam. The database, the ``Military Unit 'I This exposure opportunity index (EOI) is an earlier formulation of the E4 EOI that was developed under the contract. In general, an EOI may be defined as an estimate of the possibility that a person will come into contact with a toxic chemical without regard to route of entry or later metabolism. 12

OCR for page 11
Characterizing Exposure ... Final Report Database-Vietnam (MUD-V)," although not intended for epidemiologic purposes, was viewed as a prototype for the present study. The Columbia University researchers undertook to examine the extent of coverage of the MUD-V database, evaluate gaps in the data, assess the implications of misclassification of exposure, andgiven all these considerationsascertain how much usable information remained for epidemiologic investigations. One aim of Project 1 was to carry out a sensitivity analysis of the data in MUD-V to determine the extent to which it could satisfactorily produce a rank-ordered exposure rating for the military units. On the basis of their previous experiences, the Columbia University researchers concluded that many so-called gaps or inconsistencies in the secondary data sources were minor and would not seriously affect the accuracy of epidemiologic studies. Indeed, they had developed methods in the AOVPP for imputing reasonable locations and exposure scores where data were missing, such as substituting average battalion locations for companies or using the highest computed exposure among companies with known locations for a company whose location during a particular period was not known. Nonetheless, one of the immediate tasks undertaken in this contract was to re-examine the database and remove all imputed data. To accomplish that, the researchers returned to the original troop-movement data that had been collected by the US Army and Joint Services Environmental Services Group (ESG), now known as the US Joint Services Center for Research of Unit Records. Restoring the Troop-Movement Database The original troop-movement database was stored on 9-in. magnetic reels in virtual address extension (VAX) backup format, which is no longer manufactured. Recovery entailed first locating a working Digital Equipment Corporation VAX and obtaining licenses for appropriate operating systems and other software. Eventually, the original data and directory structure were restored, original research notebooks were located, and data were made compatible with current 13

OCR for page 11
Characterizing Exposure ... Final Report software systems. The data library was documented and written to CD-ROM. All programs and databases are now VAX-independent. The AOVPP database exposures were calculated by using {-month periods. However, the original data sources contain considerably finer time detail for many military units; in many instances, biweekly data exist; and in a few cases, daily coordinates were abstracted. The researchers went back to the original tapes to make the more detailed data available. All data were converted from the Universal Transverse Mercator (UTM) coordinate system used by the US military to a Cartesian" system more amenable to the required modeling. The researchers also appended the Department of Defenses (DOD) uniform unit-identification system designation, the Unit Identification Code (UTC), to the information in the files. Identifjir~g Units That Served in Vietnam In addition to reconstructing the existing troop-movement databases and determining where gaps or inconsistencies existed, the researchers updated the UIC LIST, a compilation of all the unit-identification numbers developed by ESG for military units with service in Vietnam. The UIC LIST had not been designed to be a comprehensive catalog of the units but rather was developed by ESG for its work in support of VA and the Centers for Disease Controls (CDC), as a recordkeeping system. It thus represented, to a large extent, units to which a VA or CDC study subject may have been assigned and units that had been identified for other study-specific purposes. The Columbia University researchers systematically examined the UIC LIST, compared and combined it with other data sources on military units, and created a master list. The master list is the first easily accessible and cross-referenced comprehensive list of all Army units that were stationed in Vietnam and the numbers of troops assigned to them in Vietnam. Where possible, the database also includes the identification of the next-higher command to which a unit was assigned. The next-higher command provides important " The Cartesian system expresses coordinates in terms of latitude and longitude. |2 Now called the Centers for Disease Control and Prevention. 14

OCR for page 11
Characterizing Exposure ... Final Report information about unit location and enables researchers to locate military records more easily at the National Archives and in other reference materials. Classifying and Modeling the Mobility of Military Units The Columbia University researchers developed a new concept for studying troop movement for use in reconstructing herbicide-exposure histories based on a concept they called the mobility factor. The mobility factor is a three-part classification system for characterizing the location and movement of military units in Vietnam. It comprises a mobility designation (stable, S; mobile, M; or elements mobile, E), a distance designation (usually in a range of kilometers) to indicate how far the unit might travel in a clay, and a notation of the modes of travel available to the unit (air; groundtruck, tank, or armored personnel carrier; or water). They then assigned a mobility factor to every unit that served in Vietnam. The mobility-factor concept simplifies the task of characterizing exposure of military personnel to hazardous substances and conditions during the course of military conflict. The mission of the organization had to be considered in conjunction with the organizational structure when mobility factors were being assigned. For example, if the mission was transportation, the mobility of the unit would vary with the command. In some cases, the mobility factor was determined in whole or in part by the type of installation to which it was assigned (for example, an airfield or a fire-support base). Not all stable units remained in the same location throughout the war. The researchers wrote a program that provides a list of all stable units that "moved", according to the database. In assessing the data, they found that in most cases the "move" was reala unit was reassigned to a different location. In some cases, however, typographic errors were responsible for the ostensible movement. In other cases, units were reclassified because some elements were, indeed, mobile. Because the mobility factor was a new concept, the researchers assembled a panel of military experts to review the concept itself and to examine the designations given to the military units stationed in Vietnam. In general, the concept was strongly endorsed by the panel, and the mobility assignments given to particular units were approved. 15

OCR for page 11
Characterizing Exposure ... Final Report Stable Units An important distinction made by the researchers was that the great majority of military units assigned to Vietnam were "stable"; that is, their missions did not require routine movement around Vietnam, and they were assigned to a specific base-camp location. Thus, the military-occupancy probability assigned to these units would be 100% for the grids that they occupied. Initially, the Columbia University researchers made a rough working estimate of the exposure-opportunity scores for the stable units. Most units were in the low-exposure category for aerial applications because the strategy underlying the herbicide projects was to defoliate or destroy crops in areas away from the main support-troop locations. However, some stable units were in heavily sprayed areas, and the perimeters of the base camps were also subject to backpack and other, more poorly documented spraying. In investigating the issue further, the researchers identified a previously unrecognized source of supplementary data in the National Archives. From those data, they decluced that each time an installation or base camp was built or occupied by American forces, a formal land-transfer agreement was executed between the governments of South Vietnam and the United States. These agreements contained specific maps and descriptions of, for example, base camps and power stations. The researchers reviewed those documents and extracted extensive quantitative data, including complete dimensions of about 200 base camps, locations of airfields, water supplies, and hospitals. In addition, they were able to identify the precise locations of 36 military units that were stationed in those installations. They also obtained coordinates for a large number of perimeters of U.S. installations. The data were used for additional quality control of the stable-units database. The stable-units database of base-camp locations and dates of residence created by the researchers covers about 80/O of the troops stationed in Vietnam. Mobile-Troop :Location Modeling The Columbia University researchers developed and tested algorithms and programs for describing and analyzing the movement of mobile battalions and their elements to further the goal of developing models of troop movement and unit dispersion. The ultimate aim was to assign military-occupancy probabilities to specific grids in the map of 16

OCR for page 11
Characterizing Exposure ... Final Report Vietnam for every military unit, down to the company level, for the duration of its assignment in South Vietnam. The primary problem in characterizing mobile units' locations was incomplete data. A critical issue was to determine whether data on a unit for particular dates were missing simply because the unit did not move and therefore did not report its continued residence at one location or whether the gaps are missing data. To develop statistical methods that could be used to evaluate the data in the entire database, the researchers studied patterns of movement of units on which substantial data already existed. That exercise yielded criteria for identifying data that indicated "short-term location stability" (where there may not have been consistent recording of location but there were not data gaps), which in turn helped to identify units for which "true" gaps existed. The researchers could then more reliably impute missing locations for these units. A preliminary cleaning and analysis of a large dataset of troop locations of Army combat battalions assigned to Military Region III~3 suggested that data sources available in the National Archives (including Daily Journals, After-Action Reports, and Operation Reports-Lessons Learned) could be used to resolve most of the data problems encountered. That indicated to the researchers that it should be possible to assemble location databases for other mobile unitsan important finding because such units are likely to have been among the most heavily exposed to herbicides. Herbicide Dispersion Data The Columbia University researchers determined that a major aspect of linking a military location with an exposure opportunity rests in the computerized records of herbicide application commonly known as the HERBS files. Records indicate that 95% of all herbicide used during the war were dispersed under Operation Ranch Hand, the US Air Force aerial spraying program (Stellman IM et al., 2003). Those records, although incomplete, are by far the most important i3 Military Region III, also known as the III Corps Tactical Zone (III CTZ), comprises a large area in the south of Vietnam surrounding Ho Chi Minh City (formerly Saigon). It was an area of heavy combat and wartime spraying. 17

OCR for page 11
Characterizing Exposure ... Final Report records available for exposure-assessment purposes. In 1974, the National Research Council organized data into a file, known as the HERBS tape, that contained coordinate points for more than 6,200 missions flown by fixed-wing aircraft and included date, type of herbicide, gallons applied (gallonage), and purpose of mission (NAS, 1974). The data in the HERBS file are given in UTM coordinates; a mission is recorded as a series of turning points in the flight path. Algorithms were written to translate from UTM to Cartesian coordinates so that spray-mission data could be integrated with other geographic information. Other programs transformed the discrete turning points into continuous flight paths so that exposure potential could be evaluated over complete spray routes. Although herbicide operations in Vietnam began in August 196T, the HERBS file contains data only from 1965 on. During the course of their work, the researchers found additional information on pre-HERBS-file spraying in a variety of sources, including two early Air Force documents (Buckingham, 1982; CHECO, 1967), the Air Force Herbicide Project folders housed at the National Archives (which contain detailed information on the planning and execution of most Air Force herbicide missions), publications produced by the US Military Assistance Command, and information retrieved from the government of South Vietnam "202 Tasks Realized" report. In the early 1980s, ESG assembled a tape to supplement HERBS the Services-HERBS tape consisting of helicopter, backpack, truck, and other smaller-scale spraying data. The database also included some information on aborted spray missions, leaks, and other unintentional releases. The combined data from those two sources comprised over 8,800 individual military spray missions. However, several discrepancies existed between the two files. Many could be resolved by a careful comparison between consecutive legs with cross- referencing to the map of Vietnam or by reference to the Daily Air Activity Reports (DAARS). When discrepancies could not be resolved, a panel of military experts assembled by the contractor reviewed the missions and determined whether it was clear that the error was typographic and whether it was possible to resolve the differences. if it was impossible to resolve 18

OCR for page 11
Characterizing Exposure ... Final Report a difference, the mission was flagged as questionable in the database so that it could be included or excluded at the discretion of the user. As a result of that exercise, the researchers developed a composite HERBS file that contains . . the date of spraying, coordinates of the spray mission, type of herbicide, number of gallons, mode of application, type of aircraft (if appropriate), and where available area sprayed. For missions flown with aircraft, the file also contains connectivity indicators that enable one to trace the path of spraying. During the researchers' reinspection of the DAARS, it became clear that individual missions were associated with particular targets and that identifying numbers (project numbers) were assigned to the targets. An earlier NAS report (1974) described how herbicide operations were organized into projects to be approved by various committees. In re-examining the original HERBS file obtained from NARA and ESG, the researchers found a field that could be related to the project numbers on the DAARS. When the HERBS file was grouped according to this field, all the Ranch Hand missions fell into specific projects. That was a major breakthrough in understanding the herbicide-spraying program because it allowed several thousand spray missions to be rationalized into a few hundred projects for analysis purposes. It is described in greater detail in a paper by the researchers that was featured on the cover of Nature on April 17, 2003 (StelIman IM et al., 20034. Combining and validating data from all those sources produced a final composite spray- mission database that comprised 9,141 missions (19,977 sorties~4) that dispersed 19,491,090 gal of herbicide in ~ 961-1971 (Steliman and Stellman, 2003~. Depending on the assumptions used, the researchers estimate that the herbicides contained 487-807 lb (221-336 kg) of TCDD (Stellman IM et al., 2003~. |4 A mission was executed by one to four aircraft. Each aircraft's flight was considered a sortie. 19

OCR for page 11
Characterizing Exposure ... Final Report Approach to Exposure Assessment The Columbia University researchers proposed the EO] as a useful alternative to such traditional toxicology-based measures as blood or adipose-tissue concentrations. It was not meant to be a substitute; neither method is perfect, and both can yield valuable information about potential exposure. The EOI method used in this project was originally developed and published by the researchers during the 1980s (Stellman and Stellman, 1986). Its central feature is the comparison of the geographic location of a potentially-exposed military unit with all known locations of herbicide release. Total exposure opportunity for the unit is the sum of the EOI estimates for all temporally-appropriate!: data in the database. Four models to quantitatively assess exposure opportunity were developed by the Columbia University researchers in the course of this and previous work. They incorporate increasingly realistic (and more complex) exposure concepts of distance and time of potential exposure to herbicide application. The simplest, E1 (the "hit" model), simply counts the instances in which a person was within a specified distance of a known spray. The second, E2, also counts hits but makes close hits count more by weighting each hit according to inverse distance from the spray. The third model, E3, begins with distance-weighted hits and factors in the total time during which the person is considered to have been exposed. E1 and E2 can be regarded as representing acute or direct exposures, since no allowance is made for exposure engendered by entering a sprayed area after the spraying has occulted or for the length of time spent in the sprayed area. E3 is analogous to acute followed by chronic exposure. Time is an essential characteristic of the current, E4 EOI model. Any person or entity that is present on the day of spray would be considered to have "direct exposure". Those entering a sprayed location after that time and those remaining in the location after having been directly exposed would be considered to have "indirect exposure", that is, exposure to any residual Is The exposure assessment software that implements the model allows the user to set the time period over which potential exposure should be factored. This may be relatively short if the user is examining herbicide ingredients that break down in sunlight or extremely long for a chemically-stable compound like dioxin. 20

OCR for page 11
Characterizing Exposure ... Final Report herbicide or dioxin that is present. The E4 EOI may be expressed mathematically as a product of three factors: Concentration of the toxicologically active substance in the herbicide mixture. A person's distance from spray paths. Residence time at an exposed location. Thus, E4 EOI = concentration factor x distance factor x time factor. The E4 EOI for a given mission is calculated as the sum of the component E4 values for all its legs. The researchers expanded the scope of assessment for the models in three important ways: by giving greater consideration to the questions of dispersion of individual troops than to reported locations of their units, using nonlinear error methods; by carrying out sensitivity analyses to account for spray-coordinate errors introduced by deviation of aircraft from flight paths and wind dispersion of herbicides; and by investigating other exposure models as extensions of or alternatives to the above. Two elements needed improvement for exposure-opportunity models: incorporation of gallonage of herbicide and consideration of connectivity of fixed-wing aircraft flight paths. The intent of this work was thus to attempt to improve the modeling of herbicide dispersal by creating a mathematical means of estimating the herbicide release along the entirety of flight paths rather than at points along flight paths. Modeling Herbicide Exposure As the project evolved, the Columbia University researchers refined their approach to calculating the EOI. Rather than estimating it directly for a military unit on the basis of its location, they chose to calculate an exposure score for a series of contiguous grids that cover Vietnam and a military-unit occupancy probability, which represents the likelihood that an individual military unit will occupy any specific grid. These grids are 0.01 x 0.01, which is 21

OCR for page 11
Characterizing Exposure ... Final Report equal to ~1.2 km2 on average. The "ridding concept permits use of statistical methods to evaluate exposure of units whose locations are known only probabilistically. It also makes it possible to superimpose contours of exposure on maps of Vietnam to visualize regions of especially high and low exposure. Finally, it permits more rigorous treatment of the various geophysical characteristics that may affect exposure to and bioavailability of herbicides (soil type, presence of water, and the like). The geographic information system (GIS) developed by the researchers transforms exposure estimation into a more efficient process. The researchers digitized an existing soil map of the former Republic of Vietnam that was incorporated into the GIS to extend exposure modeling to account for differential environmental decay of herbicides (or dioxin) that may depend on soil typology. The map was derived from a 1961 field survey that was carried out by the EN Food and Agricultural Organization (FAO) for the Vietnam Ministry of Agriculture (Moormann, 19614. The original FAO map was scanned, ant! each of the 925 contours was traced manually. A point-in-polygon algorithm was used to assign to each of the GIS grids an integer corresponding to the soil contour in which it fell. This allows the EOf for a grid to be computed by using a half-life appropriate to the soil type. To make the GIS more easily accessible to researchers, the contractors developed a user- friendly software interface called "Herbicide Exposure Assessment-Vietnam" (HEA-V). The MEA-V employs information from various components of the database highs paths of aerial spray missions, number of gallons sprayed (gallonage), and chemical agents; documented spray- mission targets; herbicide storage, transport, and unplanned-dispersal information; military-unit identification codes; locations of military units, bases, structures, air Relets, and lancling zones; movements of combat troops; land features ~ 6; soil typology; and locations of civilian populations to calculate exposure opportunity. A paper published in the March 2003 edition of the journal Environmental Health Perspectives details the development and application of the GTS ant! associated software system (Steliman SD et al., 20034. i6 Including coordinates of a variety of: elevations and land contours, rivers and streams, mountains and highlands, coastal areas and mangrove forests, bays and estuaries, and such structures as roadways and utilities. 22

OCR for page 11
Characterizing Exposure ... Final Report While the research effort was directed at characterizing the exposures of US veterans of Vietnam, the databases and model created by the Columbia University researchers could be adapted for use in studies of other countries' veterans and of Vietnam residents. Expan(ling Perimeter Spraying Database Seven Army divisions served in Vietnam during the years in which herbicide spraying took place (1965-1970~. A Division Chemical Officer, who generally held the rank of lieutenant colonel or major, served in each division. Each officer served for a 1-year tour of duty, although some may have served for shorter periods and some for more than one tour. An objective of this Project 1 task was to attempt to learn how much perimeter spraying and local spraying took place during the Vietnam War for which no records exist on the Services-HERBS tape. The researchers intended to identify Division Chemical Officers from available military records, such as the Daily Journals and the Morning Reports, and interview them. They also anticipated retrieving further data on people in the Chemical Corps from the National Personnel Records Center (NPRC) in St. Louis, which holds all military records for the armed services ant! Coast Guard. An exchange of correspondence occurred among the investigators, NPRC, representatives of the individual branches of the services, the DOD Privacy Board, the Secretary of Defense, and the Defense Manpower Data Center (DMDC) (which maintains the Vietnam file) to secure more information. The records requestedunit assignments, military occupational specialties, and awards won are releasable public records, in contrast with medical records and Social Security numbers, which are considered private records. The armed services have set up "routine-use" mechanisms by which private records may be released to federal contractors but have not established routine-use mechanisms for the release of releasable records. After negotiations among VA, NPRC, and DOD, it was agreed that a contract would be drawn between VA and NPRC for finding Social Security numbers of the chemical officers. The results of the search would be given to NAS, the VA contractor, and then to Columbia University, the NAS subcontractor. 23

OCR for page 11
Characterizing Exposure ... Final Report A text file containing 983 records of Chemical Corps officers and enlisted personnel was given to NAS to transfer to VA for transfer to NPRC. Each record contained either Army serial number or Social Security number. Of the 983 records, NPRC was able to match AS, 160 of them officers. Social Security numbersthe most straightforward means of tracing the veterans were available for just 29 of these officers. Given the additional challenges of locating the surviving veterans in this truncated cohort and the fact that only some of them would have been involved in herbicide spraying operations, the researchers concluded that this approach was unlikely to yield enough data to warrant its pursuit. In consultation with the committee, the task was abandoned. Continuing Work The researchers have indicated that they hope to pursue further information-gathering and analysis of veterans' exposure to herbicides, using funding from other sources. Among the materials submitted in fulfillment of the contract with NAS (listed in Appendix A) is a draft Web site and its associated documentation. The intent of the Web site is to gather voluntarily submitted information on the locations of military units that served in Vietnam directly from individuals or organizations. That would be used to fill gaps in data on the highly mobile combat units that server! in Vietnam. Accomplishments . Expansion and cleaning of an archive of previously tracked locations of combat battalions. Development of an approach to classifying military units so that they can be broken down by the degree to which their mission required frequent changes in location. The approach has permitted the development of a database of locations of about 80/O of all Army troops, most Air Force personnel, and Navy personnel assigned to construction battalions or permanent installations and calculation of exposure opportunity and hit scores for them. 24

OCR for page 11
Characterizing Exposure ... Final Report . . . Expansion and cleaning of herbicide-spray data (the HERBS and Service-HERBS files) that had been used extensively by the researchers and others in the past. That work has led to a substantially expanded inventory of spraying and to a higher estimate of the amount of dioxin likely to have been deposited in the Republic of Vietnam during the war. Refinement of the computational approach of a previously developed EO! and refinement of the model itself to account for gallonage and direct-hit exposures better. Development of a GIS for Vietnam into which were placed extensive databases, such as the HERBS file of spraying missions, an exposure table of hits and exposure-opportunity scores, military-unit identification codes, and military locations. Design ant! development of a unique user-friendly software systemthe Herbicide Exposure Assessment-Vietnam that implements the GTS and may serve as an archetype for other epidemiologic software for GIS-based analyses. Exploiting the National Archives data resulted in a revision of both the tasks and the timetable for the Columbia University researchers' work. The committee, who were consulted on the changes, felt that they were appropriate and desirable. It should be noted that although these data have substantially expanded knowledge about spray activities, they do not constitute a complete accounting of all herbicide releases. Indeed, it is not possible to document the myriad opportunities for in-country exposure. The best that any database of wartime herbicide exposures can do is to provide a basis for better-informed epidemiologic studies of veterans. 25