National Academies Press: OpenBook

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

Chapter: 3 Assessment of the Model and Its Capacity to Produce Useful Exposure Metrics

« Previous: 2 The Herbicide Exposure Assessment Model
Suggested Citation:"3 Assessment of the Model and Its Capacity to Produce Useful Exposure Metrics." Institute of Medicine. 2008. The Utility of Proximity-Based Herbicide Exposure Assessment in Epidemiologic Studies of Vietnam Veterans. Washington, DC: The National Academies Press. doi: 10.17226/12059.
×
Page 35
Suggested Citation:"3 Assessment of the Model and Its Capacity to Produce Useful Exposure Metrics." Institute of Medicine. 2008. The Utility of Proximity-Based Herbicide Exposure Assessment in Epidemiologic Studies of Vietnam Veterans. Washington, DC: The National Academies Press. doi: 10.17226/12059.
×
Page 36
Suggested Citation:"3 Assessment of the Model and Its Capacity to Produce Useful Exposure Metrics." Institute of Medicine. 2008. The Utility of Proximity-Based Herbicide Exposure Assessment in Epidemiologic Studies of Vietnam Veterans. Washington, DC: The National Academies Press. doi: 10.17226/12059.
×
Page 37
Suggested Citation:"3 Assessment of the Model and Its Capacity to Produce Useful Exposure Metrics." Institute of Medicine. 2008. The Utility of Proximity-Based Herbicide Exposure Assessment in Epidemiologic Studies of Vietnam Veterans. Washington, DC: The National Academies Press. doi: 10.17226/12059.
×
Page 38
Suggested Citation:"3 Assessment of the Model and Its Capacity to Produce Useful Exposure Metrics." Institute of Medicine. 2008. The Utility of Proximity-Based Herbicide Exposure Assessment in Epidemiologic Studies of Vietnam Veterans. Washington, DC: The National Academies Press. doi: 10.17226/12059.
×
Page 39
Suggested Citation:"3 Assessment of the Model and Its Capacity to Produce Useful Exposure Metrics." Institute of Medicine. 2008. The Utility of Proximity-Based Herbicide Exposure Assessment in Epidemiologic Studies of Vietnam Veterans. Washington, DC: The National Academies Press. doi: 10.17226/12059.
×
Page 40
Suggested Citation:"3 Assessment of the Model and Its Capacity to Produce Useful Exposure Metrics." Institute of Medicine. 2008. The Utility of Proximity-Based Herbicide Exposure Assessment in Epidemiologic Studies of Vietnam Veterans. Washington, DC: The National Academies Press. doi: 10.17226/12059.
×
Page 41
Suggested Citation:"3 Assessment of the Model and Its Capacity to Produce Useful Exposure Metrics." Institute of Medicine. 2008. The Utility of Proximity-Based Herbicide Exposure Assessment in Epidemiologic Studies of Vietnam Veterans. Washington, DC: The National Academies Press. doi: 10.17226/12059.
×
Page 42
Suggested Citation:"3 Assessment of the Model and Its Capacity to Produce Useful Exposure Metrics." Institute of Medicine. 2008. The Utility of Proximity-Based Herbicide Exposure Assessment in Epidemiologic Studies of Vietnam Veterans. Washington, DC: The National Academies Press. doi: 10.17226/12059.
×
Page 43
Suggested Citation:"3 Assessment of the Model and Its Capacity to Produce Useful Exposure Metrics." Institute of Medicine. 2008. The Utility of Proximity-Based Herbicide Exposure Assessment in Epidemiologic Studies of Vietnam Veterans. Washington, DC: The National Academies Press. doi: 10.17226/12059.
×
Page 44
Suggested Citation:"3 Assessment of the Model and Its Capacity to Produce Useful Exposure Metrics." Institute of Medicine. 2008. The Utility of Proximity-Based Herbicide Exposure Assessment in Epidemiologic Studies of Vietnam Veterans. Washington, DC: The National Academies Press. doi: 10.17226/12059.
×
Page 45
Suggested Citation:"3 Assessment of the Model and Its Capacity to Produce Useful Exposure Metrics." Institute of Medicine. 2008. The Utility of Proximity-Based Herbicide Exposure Assessment in Epidemiologic Studies of Vietnam Veterans. Washington, DC: The National Academies Press. doi: 10.17226/12059.
×
Page 46
Suggested Citation:"3 Assessment of the Model and Its Capacity to Produce Useful Exposure Metrics." Institute of Medicine. 2008. The Utility of Proximity-Based Herbicide Exposure Assessment in Epidemiologic Studies of Vietnam Veterans. Washington, DC: The National Academies Press. doi: 10.17226/12059.
×
Page 47
Suggested Citation:"3 Assessment of the Model and Its Capacity to Produce Useful Exposure Metrics." Institute of Medicine. 2008. The Utility of Proximity-Based Herbicide Exposure Assessment in Epidemiologic Studies of Vietnam Veterans. Washington, DC: The National Academies Press. doi: 10.17226/12059.
×
Page 48
Suggested Citation:"3 Assessment of the Model and Its Capacity to Produce Useful Exposure Metrics." Institute of Medicine. 2008. The Utility of Proximity-Based Herbicide Exposure Assessment in Epidemiologic Studies of Vietnam Veterans. Washington, DC: The National Academies Press. doi: 10.17226/12059.
×
Page 49
Suggested Citation:"3 Assessment of the Model and Its Capacity to Produce Useful Exposure Metrics." Institute of Medicine. 2008. The Utility of Proximity-Based Herbicide Exposure Assessment in Epidemiologic Studies of Vietnam Veterans. Washington, DC: The National Academies Press. doi: 10.17226/12059.
×
Page 50
Suggested Citation:"3 Assessment of the Model and Its Capacity to Produce Useful Exposure Metrics." Institute of Medicine. 2008. The Utility of Proximity-Based Herbicide Exposure Assessment in Epidemiologic Studies of Vietnam Veterans. Washington, DC: The National Academies Press. doi: 10.17226/12059.
×
Page 51
Suggested Citation:"3 Assessment of the Model and Its Capacity to Produce Useful Exposure Metrics." Institute of Medicine. 2008. The Utility of Proximity-Based Herbicide Exposure Assessment in Epidemiologic Studies of Vietnam Veterans. Washington, DC: The National Academies Press. doi: 10.17226/12059.
×
Page 52
Suggested Citation:"3 Assessment of the Model and Its Capacity to Produce Useful Exposure Metrics." Institute of Medicine. 2008. The Utility of Proximity-Based Herbicide Exposure Assessment in Epidemiologic Studies of Vietnam Veterans. Washington, DC: The National Academies Press. doi: 10.17226/12059.
×
Page 53
Suggested Citation:"3 Assessment of the Model and Its Capacity to Produce Useful Exposure Metrics." Institute of Medicine. 2008. The Utility of Proximity-Based Herbicide Exposure Assessment in Epidemiologic Studies of Vietnam Veterans. Washington, DC: The National Academies Press. doi: 10.17226/12059.
×
Page 54
Suggested Citation:"3 Assessment of the Model and Its Capacity to Produce Useful Exposure Metrics." Institute of Medicine. 2008. The Utility of Proximity-Based Herbicide Exposure Assessment in Epidemiologic Studies of Vietnam Veterans. Washington, DC: The National Academies Press. doi: 10.17226/12059.
×
Page 55
Suggested Citation:"3 Assessment of the Model and Its Capacity to Produce Useful Exposure Metrics." Institute of Medicine. 2008. The Utility of Proximity-Based Herbicide Exposure Assessment in Epidemiologic Studies of Vietnam Veterans. Washington, DC: The National Academies Press. doi: 10.17226/12059.
×
Page 56
Suggested Citation:"3 Assessment of the Model and Its Capacity to Produce Useful Exposure Metrics." Institute of Medicine. 2008. The Utility of Proximity-Based Herbicide Exposure Assessment in Epidemiologic Studies of Vietnam Veterans. Washington, DC: The National Academies Press. doi: 10.17226/12059.
×
Page 57
Suggested Citation:"3 Assessment of the Model and Its Capacity to Produce Useful Exposure Metrics." Institute of Medicine. 2008. The Utility of Proximity-Based Herbicide Exposure Assessment in Epidemiologic Studies of Vietnam Veterans. Washington, DC: The National Academies Press. doi: 10.17226/12059.
×
Page 58
Suggested Citation:"3 Assessment of the Model and Its Capacity to Produce Useful Exposure Metrics." Institute of Medicine. 2008. The Utility of Proximity-Based Herbicide Exposure Assessment in Epidemiologic Studies of Vietnam Veterans. Washington, DC: The National Academies Press. doi: 10.17226/12059.
×
Page 59
Suggested Citation:"3 Assessment of the Model and Its Capacity to Produce Useful Exposure Metrics." Institute of Medicine. 2008. The Utility of Proximity-Based Herbicide Exposure Assessment in Epidemiologic Studies of Vietnam Veterans. Washington, DC: The National Academies Press. doi: 10.17226/12059.
×
Page 60
Suggested Citation:"3 Assessment of the Model and Its Capacity to Produce Useful Exposure Metrics." Institute of Medicine. 2008. The Utility of Proximity-Based Herbicide Exposure Assessment in Epidemiologic Studies of Vietnam Veterans. Washington, DC: The National Academies Press. doi: 10.17226/12059.
×
Page 61
Suggested Citation:"3 Assessment of the Model and Its Capacity to Produce Useful Exposure Metrics." Institute of Medicine. 2008. The Utility of Proximity-Based Herbicide Exposure Assessment in Epidemiologic Studies of Vietnam Veterans. Washington, DC: The National Academies Press. doi: 10.17226/12059.
×
Page 62
Suggested Citation:"3 Assessment of the Model and Its Capacity to Produce Useful Exposure Metrics." Institute of Medicine. 2008. The Utility of Proximity-Based Herbicide Exposure Assessment in Epidemiologic Studies of Vietnam Veterans. Washington, DC: The National Academies Press. doi: 10.17226/12059.
×
Page 63
Suggested Citation:"3 Assessment of the Model and Its Capacity to Produce Useful Exposure Metrics." Institute of Medicine. 2008. The Utility of Proximity-Based Herbicide Exposure Assessment in Epidemiologic Studies of Vietnam Veterans. Washington, DC: The National Academies Press. doi: 10.17226/12059.
×
Page 64

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.

3 Assessment of the Model and Its Capacity to Produce Useful Exposure Metrics T he central question before the committee was how the Stellman team’s model for generating herbicide exposure opportunity metrics might best be used in epidemiologic studies of the health of Vietnam veterans. Before addressing that question, the committee judged it appropri- ate to first assess the strengths and weaknesses of the data and the calcula- tions that are the basis for the Stellman team’s exposure metrics and the infrastructure of their Herbicide Exposure Assessment–Vietnam (HEA-V) software tool. In this chapter, the committee elaborates on the concept of an exposure assessment hierarchy, as introduced in Chapter 2, to serve as the context for its assessment of the Stellman team’s model. The chapter then focuses on (1) the data on geography and herbicide spraying that are the basic infrastructure of the model, (2) the exposure metrics—“hits” and an exposure opportunity index (EOI)—that the HEA-V helps to calculate, and (3) potential refinements of the model. The chapter also addresses concerns that have been raised about some aspects of the data and assump- tions on herbicide spraying and the environmental fate and transport of the herbicides. Use of the model in epidemiologic studies will require researchers to supply data on troop location histories and veterans’ health outcomes. The committee’s examination of these types of data and their acquisition is discussed in Chapter 4. 35

36 PROXIMITY-BASED HERBICIDE EXPOSURE ASSESSMENT AN EXPOSURE Assessment HIERARCHY FOR VIETNAM VETERANS The Stellman team has developed a geographic information system (GIS) for use in estimating herbicide exposure in Vietnam, and they have also developed the HEA-V, a computerized database engine that facilitates the linking of disparate georeferenced data and the calculation of exposure metrics (HEA-V, 2003). To understand the strengths and limitations of this approach for measuring the opportunity for herbicide exposure, the commit- tee viewed the model in the context of an exposure assessment hierarchy. Placing the Stellman Team’s Model in an Exposure Assessment Hierarchy As noted in Chapter 2, an exposure assessment hierarchy can help illustrate both the relationship between an environmental exposure and a health outcome and the levels at which “exposure” might be measured with greater or lesser accuracy. More specifically, Figure 3-1 illustrates the exposure assessment hierarchy that the committee used to guide its think- ing on herbicide spraying in Vietnam and the level at which the Stellman team’s model operates. The simplest marker of exposure opportunity is presence or absence of a veteran in Vietnam (level 1). Level 2 uses information on proximity to spraying in space and time for military units or individuals. This is primar- ily the level at which the Stellman team’s model operates. The exposure opportunity metrics of hits and EOIs are calculated for geographic loca- tions, and those results can be combined with user-supplied information on the location histories of military units or personnel to calculate hits and EOIs at the unit or person level. A unit-level measurement would in effect assign the same EOI to all individuals within the unit. Hits and EOIs can be considered to serve as proximity-based surrogates of exposure. The proximity-based exposure metrics might be refined by the incorpo- ration of fate and transport models that provide estimates of the concentra- tion of an herbicide in various environmental media (level 3). For example, a spray drift model or estimates of the proportion of the sprayed herbicide that reaches ground level might be used instead of proximity alone. The next level of refinement in estimating exposure (level 4) 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 most highly refined level (level 5), information on pharmaco­kinetics— which relates to the body’s absorption, distribution, metabolism, and elimina-

ASSESSMENT OF THE MODEL 37 1. Presence in Vietnam or not Inputs to Data on location/time of the Model military units and spraying • Spray dates • Flight path 2. Proximity to spraying in space Potential locations and time Refinements • Type of herbicide (exposure opportunity) • Primary and • Volume sprayed secondary drift • Location history Fate and transport • Foliage density for military units model for herbicide • Penetration of or personnel herbicide through 3. Group-level ambient exposure foliage of military units or personnel • Photodegradation characteristics Individual-level behavioral data 4. Exposure of individuals by various routes Pharmacokinetics 5. Individual absorbed doses FIGURE 3-1  An exposure assessment hierarchy showing levels at which herbicide exposure in Vietnam can be assessed. The box on the left shows the inputs to the Stellman team’s model, and the box on the right shows some potential inputs if a revised model were to incorporate fate and transport phenomena. fig 3-1 Type is enlarged from 6.2 points tion of chemicals—would be needed to estimate the doses of a toxic compound to 7 points that individuals receive. Thus advancing through the hierarchy moves closer to measures of a truly biologically relevant dose. TCDD (2,3,7,8-tetrachloro- dibenzo-p-dioxin) levels in serum or tissue have been used as biomarkers of exposure to the TCDD contaminant in Agent Orange and some of the other herbicides used in Vietnam, but comparable biomarkers are not available for any of the herbicides per se, and the usefulness of TCDD levels has receded as the time since exposure in Vietnam has increased. Proximity-Based Surrogates of Exposure in Vietnam The exposure assessment hierarchy is instructive in comparing the exposure metrics generated by the Stellman team’s model and the exposure assessments used in previous epidemiologic studies of Vietnam veterans.

38 PROXIMITY-BASED HERBICIDE EXPOSURE ASSESSMENT (See Appendix B for a table summarizing the exposure assessments used in studies of U.S. Vietnam veterans.) In most cases, military personnel were considered “exposed,” not necessarily to herbicides but to the “war experi- ence in total,” if they were present in Vietnam during designated years or were part of specific military units at the division or corps level. In several studies, veterans’ self-reports of herbicide exposure were used as the expo- sure metric. A few studies have included an effort to assess the location of study participants in relation to recorded herbicide spray locations. More detailed exposure assessments, including serum TCDD measurements, have been made for those military personnel who were part of the Operation Ranch Hand units or the Army Chemical Corps, both of which applied herbicides in Vietnam. Proximity-based exposure metrics similar to those developed by the Stellman team were used in a 1986 validation study conducted in con- junction with the planning for a large Agent Orange Exposure Study to be done by the Centers for Disease Control and Prevention (CDC, 1989). The CDC “hits” metric defined an exposure hit as a unit’s presence within 2 kilometers of spraying that occurred within the previous 6 days. CDC also developed a weighted hit score for which the weight was related to the environmental half-life of TCDD. The result was that troops within 2 kilometers of a spray path 1 day after spraying were assigned a higher exposure score than troops in the same location 5 days after spraying. Finally, CDC developed an “area score,” which was based on the number of days a company was in one of five large, heavily sprayed regions of the III Corps Tactical Zone during 1967–1968 (CDC, 1989). Other Uses of Proximity to Source as a Surrogate for Exposure In environmental health studies, it is often necessary to make a retro- spective assessment of a study population’s exposures to an agent of inter- est. When stronger data, such as biomarker measurements for individuals or ambient environmental levels of an agent are unavailable, proximity to the agent has been used as an exposure surrogate. Such studies provide some insight into the usefulness of proximity as an exposure metric for herbicide use in Vietnam. Seveso Studies of the health effects of TCDD exposure from the 1976 indus- trial accident in Seveso, Italy, offer one example of a proximity-based approach to exposure assessment. Subjects were assigned to exposure zones based on the location of their homes, and these exposure zones were defined

ASSESSMENT OF THE MODEL 39 by proximity to the plant and measurements of surface soil contamination (Caramaschi et al., 1981). These environmental measurement-based expo- sure zones were associated with rates of chloracne, diabetes, and lymphatic and hematopoetic cancers (Bertazzi et al., 2001). They have also been corre- lated with serum TCDD levels on the basis of blood samples collected after the accident in 1976 and, later, in the 1990s, with serum TCDD concentra- tions being much higher on average in residents of the more highly exposed areas (e.g., Bertazzi et al., 1998; Eskenazi et al., 2004). Agricultural Studies Proximity to the locations of pesticide applications has also been explored as a potential surrogate for exposure in several U.S. studies. In an area of orchard cultivation in Washington State, organophosphate insecti- cide levels in carpet dust and metabolites in urine of children in agricultural families increased with self-reported proximity of homes to crop fields (Lu et al., 2000). Another study of children residing in a similar area of the state found that concentrations of organophosphate insecticide metabolites in urine were not related to proximity to fields but increased during the pesticide application season compared with other times of the year (Koch et al., 2002). In Iowa, over 90 percent of crop acreage (primarily corn and soybeans) is treated with one or more herbicides. Detections of agricultural herbicides and herbicide concentrations in house dust samples increased significantly with increasing acreages of corn or soybean fields within 750 meters of homes (Ward et al., 2006). However, the location of crop acreage within specific buffer distances of 100–500 meters did not explain significantly more of the variation in pesticide level than total acreage within 750 meters (Ward et al., 2006). Another study in Iowa (Curwin et al., 2005) found no relationship between agricultural herbicide and insecticide concentrations in house dust and self-reported proximity to crop fields in nonfarm house- holds; distance was classified in quarter-mile increments, ranging from less than 0.25 miles to more than 1 mile. INFRASTRUCTURE OF THE STELLMAN TEAM’S MODEL In view of the merit of the Stellman team’s proximity-based approach as a reasonable step toward more accurate herbicide exposure assessment, the committee reviewed the components that are the infrastructure of the GIS and the HEA-V software. Three integral databases store basic inputs on dates (filename: DATES), geography (GridPoints), and the data on herbicide spraying (HERBS).

40 PROXIMITY-BASED HERBICIDE EXPOSURE ASSESSMENT Time and Geography Time references are essential in the calculation of the exposure metrics. Each date from January 1, 1961, through December 31, 1975, has been assigned four unique integer values corresponding to its day, week, month, and year within that 15-year period. The integer values facilitate the date- related calculations. As noted in Chapter 2, the model uses a GIS that is based on a grid that covers all of South Vietnam (as well as sprayed areas in Cambodia and Laos). The grid uses fixed intervals of 0.01° of latitude and longitude, which results in 176,060 cells of approximately 1.2 square kilometers each ( ­ Stellman and Stellman, 2003). The coordinates of the southwest corner of a cell serve as the reference point, and each cell is assigned a unique iden- tification number in the GridPoints database. Calculations of distance are made from the centroid of a cell, which is reported as being no more than 800 meters from any of the cell’s corners (Stellman and Stellman, 2003). These two databases are straightforward tools and are subject to little potential error or uncertainty. The committee notes, however, that the loca- tion data in military records from the Vietnam era, such as those for spray missions or sites of military encampments, were recorded with map coor- dinates in a version of the Universal Transverse Mercator (UTM) system used by the military, rather than in coordinates of latitude and longitude. The UTM values were converted to latitude and longitude using software from what is now the National Geospatial-Intelligence Agency (Stellman, 2007). Herbicide Spraying The third integral data component in the GIS is the information that the Stellman team has assembled on herbicide spraying. The data elements of the HEA-V HERBS file include information on when and where each her- bicide mission took place, how the mission was conducted (e.g., fixed-wing aircraft or other means), what herbicide was used, and how much herbicide was used. Of particular interest to the committee in thinking about generat- ing exposure opportunity metrics were the nature and quality of the data on the location of spraying and the amount of herbicide applied. Stellman and colleagues (2003a) have described assembling this data- base by cleaning, combining, and reconciling data on spraying from several sources, including records on Operation Ranch Hand missions and U.S. Army helicopter and ground spraying activities (sources known as the HERBS and Services HERBS tapes); newly identified data from the National Archives; and data relating to aborted missions, emergency dumps, leaks, crashes, and other herbicide releases that were not part of standard spray-

ASSESSMENT OF THE MODEL 41 ing missions. They report having examined and reconciled multiple versions of previously compiled data on Ranch Hand missions, original U.S. Air Force records that included Daily Air Activities Reports (DAARs), and the contents of Air Force “project folders,” which could include maps, after- action reports, and other documentation of groups of spraying missions (Stellman and Stellman, 2003). Mission Records The HEA-V HERBS file contains information on 9,141 separate spray- ing missions, of which 5,957 (65 percent) are recorded as Ranch Hand flights by C-123 fixed-wing aircraft. Records of these missions are consid- ered relatively complete, especially for 1965–1971, in part because of the formal, high-level approval process for those missions (U.S. Army, 1985; Stellman et al., 2003a; Young et al., 2004a). Also included in the HEA-V file is what is recognized as incomplete information on U.S. Army helicopter and ground spraying activities (U.S. Army, 1985; Stanton, 1989; IOM, 1994). A review of the file showed that it has records for 2,108 helicopter missions and 446 missions that used ground spraying equipment. (The delivery method is not specified for the remaining 630 missions, 70 percent of which were recorded as being for perimeter spraying around base camps, fire bases, air bases, and other fixed military camps.) Records of helicopter missions were kept with Ranch Hand records beginning in 1968, but ground spraying was not tracked as part of a permanent record system (Stanton, 1989). Information on herbicide use by the U.S. Navy, U.S. Marine Corps, Vietnamese, and other allied forces is not known to be available. The Stellman team’s comparisons of spraying records with procurement records show disparities in both ­directions—in some cases (e.g., Agent Pink) it would appear that more herbicide was pro- cured than documentation shows was sprayed, while in others (e.g., Agent Purple) it appears that more was sprayed than surviving records would indicate was procured (Stellman and Stellman, 2004). Location of Herbicide Spraying The location data in the HEA-V HERBS database identify the mission’s region (the Corps Tactical Zone, e.g., III Corps) and sometimes the ­province in which the mission originated. A review of the file showed that for approximately 99 percent of the fixed-wing missions, UTM coordinates are available for the starting and ending points of the mission and for interme- diate points at which the flight path changed or the spray was turned on or off. About 50 percent of helicopter missions and 60 percent of ground spraying missions are represented by a single UTM coordinate.

42 PROXIMITY-BASED HERBICIDE EXPOSURE ASSESSMENT The HEA-V system represents the flight path for a mission by straight lines between the known UTM points (Stellman and Stellman, 2003), but actual routes may have curved to follow the path of a target such as a road or waterway. The Stellman team reported plans to use information on the nature of target areas to impute more realistic flight paths (Stellman and Stellman, 2004), but this work has not yet been completed (Stellman, 2007). The records also include the number of fixed-wing planes flown in a given mission: 4.6 percent of these missions were flown with one aircraft, 21 percent with two, 44 percent with three, 11 percent with four, and 19 percent with five or more planes. The number of planes flown on a mis- sion is a factor in the size of the area sprayed. Although the current version of the HEA-V does not account for differences among missions in the total width of their flight paths, the Stellman team has explored ways in which the information might be incorporated (Stellman, 2007). Herbicide Agents and Volume Estimates The HEA-V HERBS database identifies the herbicide used in a mission and the amount dispensed (the “gallonage”). “Incidents” that would have affected the dispersal of herbicide (e.g., an aborted flight, leaks) are identified as well. The Stellman team (2003b) reported that Agent Orange accounted for 62 percent of the total documented volume of herbicide used (approxi- mately 12 million out of 19.5 million gallons), with 28 percent being Agent White, 6 percent being Agent Blue, 3 percent being other known herbicides, and 1 percent unknown agents. Although Stellman and colleagues (2003a) discuss the amount and variability of the TCDD contamination that may have been present in some of these herbicides (discussed later in this chap- ter), the HEA-V data do not incorporate any explicit estimates of TCDD levels for individual spray missions. The available records show that 95 percent of the herbicide used was applied via the missions flown by fixed-wing aircraft as part of Operation Ranch Hand (Stellman et al., 2003a). The Stellman team’s database shows that, overall, data on the amount of herbicide used are missing for 781 (8.5 percent) of the missions. The data are missing for only 0.9 percent of fixed-wing missions but for 33 percent of ground spraying missions. As with the data on the location of spraying, the information on the herbicide agents and volumes sprayed appears to reflect the Stellman team’s review of multiple sources, including attempts to reconcile herbicide pro- curement records with records of use and destruction of remaining stocks of these products (Stellman et al., 2003a). The magnitude of a separate set of volume estimates that were based on procurement and disposition records for 1965–1971 (17.4 million gallons of agents Orange, White, and Blue combined) (see Young et al., 1978) is similar to the amount in the Stellman

ASSESSMENT OF THE MODEL 43 team’s data drawn from Ranch Hand files for the same period (17.5 million gallons). The Stellman team’s data include an additional 1.6 million gallons in records covering the period before 1965 and the use of other agents. Evaluation of the Infrastructure of the Stellman TEAM’S Model In evaluating the infrastructure of the Stellman team’s model, the issues of principal concern to the committee included the general completeness of the data, the completeness of data on herbicide spraying conducted sepa- rately from Ranch Hand flights, the potential for errors in the location of spraying arising from errors or imprecision in UTM coordinates or from the representation of flight paths as straight lines, and the appropriateness of assumptions about the extent of the area considered exposed to herbicide by a given mission. The committee included in its considerations limitations of the model noted by the Stellman team (e.g., Stellman and Stellman, 2004) as well as concerns about aspects of the model that have been raised by others (e.g., Young and Newton, 2004; Young et al., 2004a,b; Ross and Ginevan, 2007; Young, 2007; Ginevan et al., 2008). Completeness of Data Efforts have been made since the early 1970s to compile information about herbicide use in Vietnam. Records from the Vietnam War are known to vary in their quality and completeness (Shaughnessy, 1991; Young et al., 2004a; Boylan, 2007). Much of the data on which the HEA-V HERBS file is based were originally recorded by field units and forwarded to the Chemical Operations Division of the central military command in Vietnam. A 1971 audit of an early version of the Ranch Hand spraying data characterized the statistical quality of the data as good; but it found that 2 percent of the records had missing data, 6 percent had “serious” tran- scription or measurement errors, and 23 percent had errors in the length of the track sprayed (Heizer, 1971). An assessment of the Ranch Hand records by a National Academy of Sciences committee (NRC, 1974) concluded that the data as a whole were reliable despite inaccuracies in some records. That committee’s comparisons of flight path coordinates from a sample of records with aerial photographs suggested good agreement for defoliation missions. The Stellman team has described a substantial review of various types of original records as part of its 1998–2003 work to develop its GIS and exposure opportunity model (Stellman and Stellman, 2003). The work was done in consultation with the Army unit now known as the Joint Services Records Research Center (JSRRC). The Stellman team used information

44 PROXIMITY-BASED HERBICIDE EXPOSURE ASSESSMENT they collected to fill in missing data on spraying records and to identify and correct errors when possible. In comparisons between the compiled records on spray missions and separate archival records on 60 percent of Ranch Hand target areas, spray paths of related missions were found to generally fall within these identified target areas (Stellman et al., 2003b). Because the Air Force target data and the mission flight path database are independent of each other and generally corroborate each other, the Stellman team sees them as providing some validation of the spray location data (Stellman et al., 2003b). The committee is persuaded that the Stellman team’s HERBS database is as complete a record of herbicide spraying as currently exists. However, the data appear to be more complete for the Air Force Ranch Hand mis- sions than for spraying conducted by other services or by other means, such as helicopter or ground spraying. Because formal reporting for Army heli- copter spraying is described as having started in 1968 (U.S. Army, 1985), the records on helicopter spraying are presumed to be more complete for the 1968–1971 period than for earlier years. To the committee’s knowledge, factors influencing the availability of non-Ranch Hand spraying records have not been identified in any systematic way; it is therefore not ­possible to judge whether the available data are relatively representative or whether they may under-represent spraying activities in certain areas or time ­periods. The IOM committee that oversaw the Stellman team’s work for VA reached a similar conclusion (IOM, 2003). Accuracy of Flight Path Data The Stellman team notes (Stellman and Stellman, 2004) that flight paths are represented with straight-line segments but that this assumption may not always hold. Young and colleagues (2004a) point to DAARs for reports that aircraft may have adopted zigzag flight patterns in response to enemy fire. The straight-line assumption may also not hold when the flight path followed features such as a river or a highway, with variations of as much as a kilometer or more from recorded locations suggested (Young et al., 2004a). Anecdotal evidence suggests that aircraft crews navigated by a combination of visual orientation and maps that were precise to no better than 120–240 meters (Young et al., 2004a). The committee heard concerns that the Stellman team’s EOI calcula- tions take into account an excessively wide area (up to 5 kilometers) on either side of a flight path (Ross and Ginevan, 2007; Ginevan et al., 2008). This issue is discussed again later in the chapter, but the committee notes here that considering the wider area when assessing exposure opportunity would seem to address, at least to some extent, the concern that true flight paths may have deviated from the straight lines used in the model. Missions

ASSESSMENT OF THE MODEL 45 flown with multiple aircraft also would have contributed to variation in the width of the area where herbicide was applied. Accuracy in locating herbicide spraying is essential for effective assess- ment of exposure, but the concerns that have been raised do not appear to point to major misrepresentations of locations in the spraying database. Other Issues It has been noted that troops on the ground may have mistaken the frequent aerial spraying of insecticide (e.g., malathion) for herbicide spray- ing (Young et al., 2004a; Cecil and Young, 2008). Young and colleagues (2004a) point to reports that between 1966 and 1972 more than 3.5 million liters of malathion were sprayed over approximately 6 million hectares of South Vietnam and that by 1970 malathion was being sprayed at 9-day intervals. The committee did not attempt to determine whether records exist that document the flight paths of insecticide spray missions. If they do, it would be appropriate to consider adding that data to the Stellman team’s GIS. THE STELLMAN TEAM’S EXPOSURE OPPORTUNITY METRICS As previously described, the Stellman team’s model produces two expo- sure metrics that are based on proximity to herbicide spraying: hits and the EOI. The hits metric represents direct exposure, and the EOI incorporates consideration of indirect exposure from previous spraying. The model uses a two-stage approach to calculate the exposure values. The first stage relies on the datasets that the committee has described as the infrastructure of the model. The data on the location and date of each spraying mission are used to calculate a hit and an EOI value for each indi- vidual cell in the GIS grid that falls within 5 kilometers of that mission’s spray path. These geographically based exposure calculations are stored, along with essential information about the associated spray mission, as individual records for each cell exposed to spraying during that mission. The database containing all this information (Exposure_Master) contains approximately 1.45 million records and is an integral part of the HEA-V tool. At the second stage, this geographic exposure database is used in combination with user-supplied information on the location histories of military units, individual military personnel, or other study subjects to calculate exposure scores for the period that the units or the individuals spent in Vietnam.

46 PROXIMITY-BASED HERBICIDE EXPOSURE ASSESSMENT Direct Exposure The hits metric is generated for four specific distances from a spray path: 0.5, 1, 2, and 5 kilometers. A cell that fell within 0.5 kilometer of a given mission’s spray path would have a hit score of 1 for each of the four distance possibilities. A cell that fell more than 2 kilometers but less than 5 kilometers from the flight path would have a hit score of 1 for only the 5-kilometer option. When information on military units or personnel is submitted for analysis, a GIS cell’s hit scores for a given date are assigned to each individual or unit whose location history puts them in that cell on that date. The hit scores for a given distance option can be summed across spray missions to generate a cumulative score for a specified time period. For example, an HEA-V analysis found that during March 1969, 72 of 278 (15 percent) companies in Army combat battalions in III Corps were within 0.5 kilometer of a spray path (Stellman and Stellman, 2003). Of these units, 27 had cumulative scores for that month ranging from 2 to 19 for hits within the 0.5-kilometer distance. Indirect Exposure Opportunity The EOI calculation for each mission includes three main components: the amount of herbicide sprayed, a GIS cell’s distance from the spray path, and an herbicide decay rate (Stellman and Stellman, 2003). The amount of herbicide is used as a concentration factor. Exposure is treated as inversely proportional to distance from the flight path (referred to as a 1/D factor), with calculations restricted to cells that are within 5 kilometers of the spray path. Herbicide decay is represented by a first-order decay model, with a 30-day half-life specified for the calculations at the geographic level. Users are given the option to specify an alternative half-life when calculations are made for military units or personnel. For any given date, then, the EOI for a GIS cell reflects the combined effects of these three factors for any spraying that occurred on that date, plus the appropriate residual effect of previous spraying that impinged on that cell. In principle, these cell-specific EOI scores are cumulative for units or individuals in a manner similar to that for the hits score. Evaluation of The Stellman TEAM’S Exposure Opportunity Metrics Much of the criticism of the model has focused on the exposure metrics, generally with the argument that the metrics are inaccurate because the model fails to take into account other factors that influence the exposure of troops. The committee agrees that a number of potentially important

ASSESSMENT OF THE MODEL 47 factors­ have not been included in the Stellman team’s model, and that many of these have the potential to cause misclassification of exposure that, if non-differential with respect to disease, will often tend to bias results toward the null. In other words, if the error in measuring exposure does not depend on disease, then the ability to associate exposure and disease will be reduced. The committee found, however, that several of the criticisms reflect expectations of the model that extend beyond the capabilities of a ­proximity- based exposure opportunity approach. For example, concerns that the model does not take into account the chemical properties of the herbicides or the herbicide contaminant TCDD would be warranted for consideration in a model that purported to approximate the ambient levels of dioxin or herbicides in which troops operated in Vietnam (level 3 in Figure 3-1). As designed, however, the model provides only for the rough exposure classi- fication permitted by assessing proximity to spray paths (level 2). With that in mind, the committee offers its assessment of the exposure opportunity metrics, including consideration of several criticisms. Direct Exposure The hits metric is calculated for distances from a spray path that range from 0.5 to 5 kilometers (i.e., a maximum swath 10 kilometers wide). This offers users the opportunity to assess the effect of different assumptions about proximity to spraying. In principle, modifications to the HEA-V would make it possible to use smaller distance factors, but the 1.2-square- kilometer resolution of the GIS grid imposes a limit on the precision with which small distances can be represented. The committee heard concern that the model allows for attribution of exposure over an area that is much larger than the swath of 0.08– 0.1 kilometer­ for a single C-123 fixed-wing aircraft (Ross and Ginevan, 2007; Ginevan et al., 2008). The area is also larger than the area that would be affected by spray drift as estimated from primary drift ­ models (Ross and Ginevan, 2007) or that was affected in Air Force test flights (Young et al., 2004b). It was noted that even though spray missions typi- cally involved multiple aircraft, the 80 percent that were flown with four or fewer planes would have generally have had a swath of no more than about 0.5 kilometer. However, there is support for use of distances much greater than the nominal width of a plane’s spray swath for an exposure opportunity metric. A 1969 report on herbicide use in South Vietnam included calculations that, under unfavorable but acceptable operating conditions, spray drift damage to broadleaf crops could occur at distances up to 2 kilometers (Darrow et al., 1969). The report also noted that unintended crop damage occurred on

48 PROXIMITY-BASED HERBICIDE EXPOSURE ASSESSMENT defoliation missions because of malfunctions of the spray equipment. An Army field manual on the use of herbicides indicated that crops should be safe from drift at a distance of 7 kilometers from spraying conducted with the equipment typically used on the C-123 Ranch Hand aircraft at the rec- ommended altitude (150 feet) and close to the maximum acceptable ground wind speed (8 knots) (Department of the Army, 1971). Spray drift appeared evident in aerial photographs reviewed for a National Academy of Sciences study of the effects of herbicides in Vietnam (NRC, 1974). Whereas many spray swaths were reflected in bands with clearly defined boundaries, others showed a diffuse edge indicative of drift. Indirect Exposure Opportunity Concerns have been raised about some of the components of the EOI calculation: the herbicide decay rate, distance from the spray path as an exposure modifier, and the amount of herbicide sprayed. The decay rate is currently modifiable within the user interface of the software; therefore researchers who can support other half-life choices are free to use them for calculating the EOI. The model represents the effect on exposure of increas- ing distance from the spray path with an inverse distance (1/D) factor, which cannot be modified in the software’s user interface. It would be useful if this were a changeable function in the software so that a researcher who could justify other drop-off rates could use them to calculate the EOI. An alternative to a simple drop-off rate in the EOI model would be to expand the EOI estimation to incorporate more advanced spray drift models (e.g., AgDRIFT®). This potential expansion of the model is discussed later in the chapter. The committee also noted that expanding the model to consider secondary drift might be useful in further addressing the effects of distance on exposure. Another component of the Stellman team’s EOI model is the amount of herbicide sprayed during a mission (gallonage), which is used to weight the EOI. Because herbicides were sprayed at a relatively constant rate over the flight path, herbicide gallonage is an indirect way to account for multi-plane missions along a single flight path. Since the HEA-V databases contain information for most fixed-wing missions on gallonage and the number of planes flying together, the model might benefit from incorpora- tion of a better representation of the implicit width of the flight path. As noted previously, the Stellman team has explored methods to do this but has not incorporated this element into the current version of the HEA-V (Stellman, 2007). Given the availability of the data on gallonage and num- bers of planes, other users of the model might be able to consider means of incorporating flight path width into EOI calculations.

ASSESSMENT OF THE MODEL 49 Troop Presence Although official policies have been described as ensuring that Ranch Hand spray targets were kept clear of U.S. troops during spray missions (Young et al., 2004a), the committee did not find that to be sufficient evidence that troops would not have been close to spray missions. In fact, GAO (1979) compared location data for Marine Corps infantry ­battalions for 1966 through 1969 with the location of Ranch Hand flights. Of 218,000 marines who served during that period, 5,900 were estimated to have been within 0.5 kilometer of a spray path on the day of the spraying, and 17,400 were within 2.5 kilometers. To the extent that documentation of troop loca- tions near spraying exists, it must be taken as the best available evidence. Proximity to Perimeter Spraying As noted, the model calculates direct hits for a series of distances of from 0.5 to 5.0 kilometers from a fixed-wing flight path or the location of helicopter or ground spraying. It is possible, however, that drift from ground or helicopter spraying operations would have been less than that for spraying by fixed-wing aircraft, resulting in less exposure at a distance but more exposure near the spray path. Those operations may have been substantial contributors to exposure opportunity for at least some types of military units and their personnel. In the analyses done in conjunction with the planning for the CDC Agent Orange Study, most hits (i.e., presence within 2 kilometers of spray- ing within the previous 6 days) were found to be from helicopter and ground spraying, including perimeter spraying of fire bases (CDC, 1989). Likewise, when the Stellman team’s GIS was used to investigate exposures of stable units in March 1969, it was found that 2 percent (36) of the 1,982 unit locations had hits within 0.5 kilometer from fixed-wing spraying, and 7 percent (141) had hits within 0.5 kilometer from helicopter and ground spraying (Stellman and Stellman, 2003). If the hit zone is widened to 5 kilometers,­ 10 percent of the locations had hits from fixed-wing spraying and 18 percent had hits from perimeter spraying during the month. For the subset of operations that involve spraying the perimeter of a base, it may be possible to match the base coordinates with the coordinates in the HERBS files to assign “hits” to bases during that spray event. Spray Penetration It was suggested to the committee that little of the herbicide sprayed actually made it to the forest floor where ground troops could be exposed (Ross and Ginevan, 2007; Young, 2007). Some of the herbicide applied

50 PROXIMITY-BASED HERBICIDE EXPOSURE ASSESSMENT would be absorbed into the waxy layer of the plants and most of this absorbed herbicide would undergo photodegradation by ultraviolet radia- tion in a few hours. TCDD has been shown to have a photodegradation half-life of 6 to 10 hours under some conditions (Young et al., 2004b). A report describes experiments by the U.S. Department of Agriculture in Puerto Rico and Texas finding that on average 21 percent of the spray penetrated the upper canopy and 6 percent penetrated to ground level (Young et al., 2004a). Young and colleagues (2004a) also note that a “leaf area index” for the Vietnamese jungle predicts that only 1 percent to 6 percent of an aerial spray would have reached the lowest levels. However, this does not appear to account for spraying in areas where the amount of protective foliage was more sparse or may have diminished over time as a result of previous spraying. As noted, calculations using the Stellman team’s HERBS database show that approximately 95 percent of the herbicide known to have been used in Vietnam was applied via fixed-wing aircraft. The remainder was dispersed via helicopters or ground-based spraying. The assumptions made about the proportion of herbicide reaching the ground from fixed-wing spray- ing in comparison with other application methods would be important in judging the contribution of each source to a more refined estimate of exposure opportunity. This sort of refinement would require more detailed data regarding factors such as canopy cover and remaining canopy for previously sprayed areas, which would move beyond the proximity-based approach adopted by the Stellman team. The committee also notes that while such adjustments offer the potential to provide more local detail to measures of exposure opportunity, the geographic size of the grid cells underlying the GIS presents a practical limit to the ultimate spatial resolu- tion of any such adjustments. TCDD Exposure In its current form, the Stellman team’s model does not offer means of generating exposure scores linked specifically to TCDD. Although it is possible to calculate scores specifically for exposure to Agent Orange and the other herbicides that contained 2,4,5-T, the level of TCDD contamina- tion in these herbicides varied over time and by several orders of magnitude (from less than 0.05 ppm to 50 ppm; IOM, 1994). Because the model incor- porates no adjustments for varying levels of TCDD, the potential exists for misclassification in estimates of exposure to TCDD. This implies that the model will generally be better suited for examining exposure to herbicides than for examining exposure to TCDD.

ASSESSMENT OF THE MODEL 51 Does the EOI Produce a Range of Potential Exposure Values? For exposure measures to be useful in epidemiology, they must show a range of exposure in the population under study. To investigate the range of potential exposures among Vietnam veterans, the Stellman team identi- fied 1,957 “stable” Army units in Vietnam in June 1969 and the 2,095 cells in the GIS grid that these units occupied during that month (Stellman et al., 2003b). Of the 2,095 occupied cells, 56 percent of the locations had been sprayed and were occupied by 1,045 units. Personnel counts could be estimated for 815 of these units and totaled 142,583 soldiers. The resulting distribution of EOI scores for these soldiers was lognormal and spanned several orders of magnitude (see Figure 3-2). However, unpublished calculations provided to the committee (now published as Ginevan et al., 2008) gave results showing that in some instances the EOI scores of locations directly under a flight path and loca- tions as much as 4 kilometers away from the flight path are not significantly different. Although the implications of these differences are unclear, the 40,000 30,000 Frequency 20,000 10,000 0 0.0 1.0 2.0 3.0 4.0 5.0 6.0 Logarithm of EOI Score FIGURE 3-2  Frequency distribution of the logarithm of the non-zero EOI scores for 142,583 soldiers in 815 units designated as “stable” or “stable with mobile elements,” June 1969. SOURCE: Stellman et al., 2003b. Reproduced with permission from Environmental Health Perspectives. 3-2 The figure has been reduced by from 4.25 inches wide to 3.5 inches

52 PROXIMITY-BASED HERBICIDE EXPOSURE ASSESSMENT committee urges, as discussed below, that more sensitivity testing be done to gain a better understanding of the dynamics of the model. The committee also emphasizes that the EOI data do not describe the magnitude of exposure in units that can be used to make comparisons with exposures measured in other populations. Because EOI is a ­proximity measure that does not account for fate and transport, personal activity pat- terns, or biological uptake characteristics, it seems unlikely that the EOI will be proportional to exposure. Instead, the measure is more likely to be a reasonable means of ranking potential exposure. Comparison of the Hit Score and EOI The committee sees strengths and weaknesses in both of the ­Stellman team’s proximity-based exposure opportunity measures. The hit score has the virtue of being very simple. However, when used as an exposure s ­ urrogate, it implicitly makes two assumptions that are likely to cause exposure misclassification: (1) residual herbicide does not contribute to future exposure (i.e., as if it has a short half-life in the environment), and (2) distribution of herbicide is uniform within the assigned radius (i.e., ignoring drop-off with distance from the flight path). The EOI takes into account both the presence of residual herbicide (via the decay rate) and distance, but it does this in ways that may not be accurate. For example, it was not clear to the committee that the current procedure of integration of the 1/D function along the flight path is the best approach. Both hits and the EOI have merits, but both are likely to cause some level of exposure misclassification. Sensitivity Analyses Needed Given the nature of the proximity-based exposure opportunity mea- sures and the associated assumptions regarding decreases in herbicide con- centration with distance and time, the committee stresses the importance of quantitative assessments of the sensitivity of these measures to the model assumptions. Such assessments are often based on recording the amount of variation in model outputs resulting from systematic variation of model inputs (e.g., distance from the flight path in the “hits” approach, or the rate of distance-decay in the EOI). In the case of the HEA-V software, gen- eral sensitivity analyses could involve summaries of exposure opportunity scores associated with particular units or individuals, or, more broadly, involve local summaries (ranges, distributions, or percentiles) of exposure opportunity scores mapped for each of a number of grid cells. An initial effort in this direction is reflected by the Stellman team’s evaluation of the impact of small shifts in the coordinates, which suggested minimal impact

ASSESSMENT OF THE MODEL 53 on exposure assignment (Stellman and Stellman, 2003). It is important to note that such analyses can be conducted even in the absence of “gold standard” exposure measures. They provide information on the stability of the proposed measures under uncertainty around model inputs and assump- tions, and they provide important context for any epidemiologic study in which the measures are used. Additional directions researchers might pursue include more detailed statistical modeling of data uncertainties, including models of measurement error. However, in the absence of a gold standard by which to evaluate error, researchers face a challenge. If they are able to estimate the likely distribution of errors in the study population and thus quantify the range in uncertainty, this should be taken into account in sensitivity analyses and power calculations. POTENTIAL FOR CORROBORATION OF EXPOSURE MEASURES In large retrospective occupational or environmental studies, it is often impossible to obtain detailed exposure information for all subjects. When a surrogate measure on the exposure assessment hierarchy (see Figure 3‑1) is used instead, it is then desirable to compare the surrogate with other exposure indicators. In the case of the Stellman team’s exposure oppor- tunity model, correlation of the EOI with variations in TCDD levels in independent environmental data or in tissue samples could theoretically increase confidence in the model’s measures of exposure opportunity for the herbicides that contained the TCDD contaminant. Unfortunately, neither approach appears likely to be sufficiently informative to be worthwhile. Utility of TCDD Measurements in Soil, Water, and Sediments It is estimated that the use of herbicides in Vietnam resulted in the deposition of between 170 and 680 kilograms of TCCD over the southern part of the country (Dwernychuk et al., 2002). Although TCDD persists in the environment, the committee sees little opportunity to use information on its current levels to provide more than broad qualitative support for the Stellman team’s model. TCDD is part of a family of polychlorinated dibenzo-p-dioxins (PCDDs) which are often found together. The mix of PCDDs varies depending on their source. They are byproducts of low-temperature com- bustion processes, such as the open burning of waste material, and they are also generated during the production or use of some chlorinated chemical products, including phenoxy herbicides such as Agent Orange (ATSDR, 1998). TCDD was present as a PCDD contaminant of special concern in the 2,4,5-T herbicide used in Agent Orange and in some of the other

54 PROXIMITY-BASED HERBICIDE EXPOSURE ASSESSMENT military herbicides used in Vietnam, but it is not prominent in PCDDs generated by combustion. TCDD and related compounds have very low water solubility and are lipophilic. Following aerial spraying, waste incineration, or other atmospheric releases, the compounds may be deposited on plants, on soil, and in bodies of water and their sediments. They can also enter water and sediments from such sources as effluent discharge and soil runoff. Although TCDD can be degraded by sunlight, it is otherwise highly persistent in the environment and can accumulate in the fatty tissue of some fish and mammals. Studies carried out at Eglin Air Force Base in Florida indicated that even though 99 percent of the TCDD applied was photodegraded, TCDD was still detectable in test plots several years after application (Young and Newton, 2004). The half-life of TCDD has been estimated at 9–15 years in surface soil and 25–100 years in subsurface soil (Paustenbach et al., 1992). In sediments, TCDD has been found to persist over decades (Bopp et al., 1991). An investigation in one area of Vietnam found that TCDD levels were somewhat elevated in soil samples from locations where aerial spraying had occurred 30 years earlier (Dwernychuk et al., 2002). The same investiga- tion also found that soil samples from Special Forces bases in the area had significantly higher concentrations of TCDD than the soil samples collected from flight paths. The higher levels in the base areas were attributed to her- bicide spillage and disposal. A separate analysis of soil and sediment from areas at and around the site of the Da Nang Airbase, which was a base used for Operation Ranch Hand missions, found very high levels of TCDD (365,000 ppt) in soil samples from specific sites where herbicides had been stored or airplanes’ spray tanks had been loaded (Hatfield Consultants and Office of the National Steering Committee 33, MNRE, 2007). Because of the persistence of TCDD and other PCDDs, samples of undisturbed soil or sediment can provide a historical record of their deposi- tion (e.g., Czuczwa and Hites, 1984, 1986; Baker and Hites, 2000). Com- parisons of TCDD and PCDD levels found in areas that were sprayed and areas that were not may offer some qualitative insight into past exposure sources. However, several factors hinder the use of data on current TCDD concentrations to validate the Stellman team’s model. With sediment cores from a lake or delta, the historical profile of the total levels of TCDD depos- ited over the watershed does not adequately indicate localized variations in soil deposition. In addition, the precision with which TCDD deposition in sediments can be dated is limited and may be no better than 2 to 5 years (e.g., Frignani et al., 2004). Therefore sediment analysis could only be used to validate the estimates of exposure opportunity generated by the Stellman team’s model on very broad geographic and temporal scales. With soil analysis, the range of 9 to 15 years for the estimated half-life of TCDD in surface soil and the use of herbicides extending over a 10-year

ASSESSMENT OF THE MODEL 55 period mean that present-day TCDD levels may reflect levels that were anywhere from 3.4 to 15 times higher during the war. There are also likely to have been other sources of TCDD and PCDD deposition over the more than 45 years since herbicide spraying began, including contributions from open burning of waste during and after the war, burning of materials con- taining polychlorinated biphenyls, and long-range atmospheric transport of TCDD and PCDDs. As a result, it may be difficult to isolate the contribu- tions of TCDD from war-era herbicides except in areas where they were extensively used. For the herbicides themselves that were used in Vietnam, virtually no opportunity exists to test for residues. The principal components of Agent Orange, the military herbicide used most extensively, were 2,4-D and 2,4,5-T in various formulations. Unlike TCDD, these compounds are fairly water soluble and not persistent. For example, 2,4-T has a half-life in soil of approximately 6 days and in aerobic aquatic environments of 15 days (EPA, 2005). Similarly, malathion, which was sprayed to control mosquitoes and potentially confused with herbicide spraying, has a half life of 11 days or less in soil and up to 2 weeks in an aerobic aquatic environ- ment (EPA, 2006). As a result, no major accumulation of these compounds in soil and sediment is likely. Utility of Serum and Adipose TCDD Measurements In humans, an initial rapid rate of elimination of TCDD is followed by a slower rate that results in an estimated half-life of 7–10 years (e.g., Michalek et al., 2002). As it has been more than 35 years since any U.S. military personnel were exposed to TCDD in Vietnam, studies conducted now may not be able to reliably distinguish TCDD exposure in Vietnam from background exposure in the United States from other sources (e.g., combustion or food). The IOM committee that oversaw the Stellman team’s work for VA reached a similar conclusion (IOM, 2003). Some previous studies have explored the correlation between measures of exposure and TCDD levels in tissue samples. These studies have used various approaches to exposure measurement and have had varying results. CDC Study In 1987, CDC (1988, 1989) measured TCDD levels in serum samples from 646 enlisted Vietnam veterans with a pay grade between E1 and E5 and only one tour of duty in Vietnam (average 320 days) who served in one of five combat battalions in III Corps during 1967–1968. Comparison samples were drawn from 97 U.S. Army veterans of the same era who did not serve in Vietnam (CDC, 1988). The Vietnam veterans were selected to

56 PROXIMITY-BASED HERBICIDE EXPOSURE ASSESSMENT represent low, medium, and high categories of exposure, which were defined on the basis of the number of hits. (As described above, a hit was defined as a person’s company being located within 2 kilometers of a recorded Agent Orange spraying mission within 6 days after the spraying.) Subjects in the high exposure category had five or more hits. Fifty-four percent of the Vietnam veterans enrolled in the study had at least one hit, and 30 percent had five or more. The analysis used this exposure measure as well as two others, all of which were based on proximity in space and time to Agent Orange spraying, as the Stellman team’s measures are. Although a range of TCDD levels was found in this study, the serum levels of those who served in Vietnam, which ranged from non-­detectable levels to 45 ng/g lipid, were not significantly different from the levels among veterans with no Vietnam service (range: non-detectable levels to 25 ng/g lipid) (CDC, 1988). Only 4 percent of the Vietnam veterans had TCDD ­ levels above 8 ng/g lipid. There was also no significant trend in serum TCDD levels by exposure score category. The lack of relationship between serum TCDD levels and exposure scores was an important factor in the cancellation of the larger CDC study of the long-term health effects of exposure to Agent Orange (CDC, 1989). There are several reasons why the CDC study may have failed to detect a difference in TCDD serum concentrations between the veterans who served in Vietnam and those who did not. The TCDD concentrations in Agent Orange used during the exposure period studied may have been relatively low. Contamination levels could vary by production run and manufacturer and were found to range from 0.05 to 50 ppm, averaging 1.98 and 2.99 in two sets of samples (NRC, 1974; Young et al., 1978). It is also possible that the study selection criterion requiring a single tour of duty may have undersampled subjects with high exposure opportu- nity, limiting the power of the study to detect differences. Only 30 percent of the Vietnam veterans tested had hit scores of 5 or greater (i.e., being in locations within 2 kilometers of spraying that had occurred in the previous 6 days five or more times). Differences in pharmacokinetics between veterans due to differences in metabolism, body composition, or weight change, could have influenced the rate at which TCDD was eliminated. Furthermore, the pharmacokinetics of TCDD is more complicated than the simple first-order models assumed in the CDC study as the basis for power calculations (e.g., Michalek et al., 2002; Emond et al., 2005). However, elevated serum and adipose tissue levels of TCDD have been used elsewhere to document higher exposure (e.g., Flesch-Janys et al., 1998; Michalek et al., 2002). Another possibility is that most American military personnel who served in Vietnam did not have high TCDD exposure. The range of expo- sure opportunity scores seen for most Vietnam veterans may not represent

ASSESSMENT OF THE MODEL 57 significant elevations in TCDD exposure. The lack of correlation between exposure score and TCDD serum levels could also mean that the exposure score is a poor surrogate for TCDD exposure. Residents of Vietnam Comparisons have also been made between TCDD levels and EOIs for two groups of Vietnamese (Verger et al., 1994; Kramarova et al., 1998; Stellman and Stellman, 2003). In one comparison, adipose tissue samples from 25 participants in a cancer case-control study conducted in Ho Chi Minh City were analyzed for TCDD and other PCDDs and PCDFs ( ­ Stellman and Stellman, 2003). These subjects were recruited in 1993–1997 for a study supervised by the International Agency for Research on ­Cancer (IARC). Of the analyzed samples (from a mix of cases and controls), 11 had detectable TCDD levels, ranging from 1.0 to 4.2 pg/g lipid. The S ­ tellman team (Stellman and Stellman, 2003) computed EOI values for the study subjects on the basis of geocoded residential histories. Seven subjects had an EOI of zero; scores for the others ranged from 295 to 7.7 × 105. The Pearson correlation coefficient between the TCDD levels and the EOI (both log-transformed) was 0.23 and not significantly different from zero, but no sample had both a high TCDD level and a low EOI (Stellman and Stellman, 2003). A second study examined 27 Vietnamese men admitted for abdominal surgery in Ho Chi Minh City in 1989; all patients were born before 1953 (Verger et al., 1994). TCDD concentrations in adipose tissue samples (range: non-detectable levels to 49.6 pg/g lipid) were compared with EOI scores (range: 0 to 9,868) computed based on residential history and an earlier version of the Stellman team’s model. The Spearman correlation ­coefficient for all samples was 0.32, p = 0.10 (with log transformation: Pearson r = 0.36, p = 0.07). Restricting the analysis to the 22 subjects with non-zero EOIs increased the correlation (Spearman = 0.44, p = 0.04; ­Pearson = 0.50, p = 0.02) (Verger et al., 1994; Stellman and Stellman, 2003). The committee sees the data on Vietnamese residents as providing weak evidence that the EOI can serve as a predictor of TCDD concentrations in people. It is unclear whether these results can be generalized to U.S. vet- erans because the Vietnamese study subjects could have been exposed to TCDD for long periods of time, including extended exposure via the food chain, a route less likely for American military personnel. TESTING AND REFINING THE STELLMAN TEAM’S MODEL The Stellman team’s model for herbicide exposure in Vietnam counts direct exposure events and also produces a quantitative representation of

58 PROXIMITY-BASED HERBICIDE EXPOSURE ASSESSMENT indirect exposure that takes account of the quantity of herbicide sprayed, the distance from the spray path, and, by using an environmental decay factor for the herbicides, the time since spraying. The committee notes first that the exposure assessment model and metrics developed by the Stellman team are one approach but not necessarily the only approach that might be followed for a proximity-based assessment of exposure opportunity. Other approaches might be explored, possibly without requiring collection of additional data. There are also several possible expansions of the model and the GIS databases, each of which has advantages and disadvantages. For exam- ple, the Stellman team (Stellman et al., 2003b) has described developing other geocoded databases that might be used to refine exposure estimates. Examples of these databases include locations of inhabited places during the period of the Vietnam War, roadways and other constructed features, locations of military sites, and soil types. Only the database on soil types was available to the committee for examination. Aerial spray drift dispersion models might also be useful. For example, the AgDRIFT (Teske et al., 2002) and AGDISP models (Bilanin et al., 1989) use inputs such as the type of aircraft and spray nozzle, characteristics of the spray material, spraying height, swath width, wind speed and direc- tion, and temperature to estimate the percentage of the spray that deposits at various distances from the flight path. For application to spraying in Vietnam, inputs such as weather conditions are likely to be available from the DAARs, but they would have varied relatively little if official operating procedures were followed. Those procedures specified that spraying flights take place in clear weather, with wind speeds of no more than 10 miles per hour, and that herbicide be dispensed at an altitude of 150 feet at a constant delivery rate of 3 gallons/acre (MACV, 1969; Stellman and Stellman, 2003; Young et al., 2004a). Other factors such as the types of vegetation, characteristics of the ini- tial and remaining canopy, and meteorological parameters that could affect the ground-level deposition, photodegradation rate, and the availability of herbicide in the topsoil could also be incorporated into a more detailed exposure model that might use the Stellman team’s EOI or the spraying location data in the GIS as a starting point. Consideration of secondary drift (e.g., through evaporation from treated plant materials or transport of aerosolized particles) might further improve estimation of exposure. However, the committee is not aware of any currently existing secondary drift models that could be directly applied. Rather, such models would need to be developed. Although use of spray drift models or incorporation of other factors could potentially result in improved quantification of herbicide deposition, it is unclear whether they would result in changes in the relative ranking of

ASSESSMENT OF THE MODEL 59 exposures among military personnel or units. It is also unclear how much of the historical data needed to use these more advanced spray deposition models would be available in the military records. No “gold standard” exists for use in testing the accuracy of retro- spectively estimated exposure of veterans to herbicides in Vietnam using the Stellman team’s hits or EOI score, alternative simple proximity-based measures (e.g., incorporating other distance functions), or approaches that use the Stellman team’s infrastructure as a foundation for more elaborate fate and transport modeling (e.g., primary and secondary drift). Thus the committee sees it as essential that sensitivity analyses be done to compare various approaches to estimating exposure. The exposure measures they produce should be compared with each other to see how assigned exposures are changed, particularly rank orderings. The impact of assumptions (e.g., distance functions, decay rates) should be examined in the same way. CONCLUSIONS Based on its review of the Stellman team’s herbicide exposure assess- ment model, the committee reached several conclusions. 1. Using a surrogate of exposure that is based on individuals’ or military units’ proximity in space and time to herbicide spray paths is a reasonable exposure assessment strategy. This approach is a clear improve- ment over the cruder measures of exposure or opportunity for exposure, such as those based on service in Vietnam, that have been used in some past studies of the potential effects of herbicide exposure on the health of Vietnam veterans. Such proximity-based surrogates are similar to exposure measures commonly used in occupational health studies (e.g., job title) and in environmental studies of proximity to sources of exposure. 2. The Stellman team’s databases and GIS provide a useful basis for estimating proximity-based surrogates of exposure to herbicides in Viet- nam. Because of the availability of relatively more complete data on spray- ing by fixed-wing aircraft, the model is currently better suited to examining proximity to that type of spraying than to spraying from ground equipment or helicopters. The uncertainty about the completeness of the data on heli- copter and ground spraying should be taken into consideration, especially when studying stable units, which may have had limited exposure to fixed- wing spraying. 3. The Stellman team’s hits and EOI scores have value in that they move further along the exposure assessment hierarchy than exposure assess- ment based only on presence in Vietnam. However, the methods by which the hits and EOI scores are calculated have the potential for significant exposure misclassification, and so these metrics must be used with caution.

60 PROXIMITY-BASED HERBICIDE EXPOSURE ASSESSMENT Other proximity-based approaches with the potential for estimating exposure scores more accurately should be explored. Moving from exposure metrics based on spray and troop location data to more accurate exposure and dose metrics would require the incorporation of additional data, such as herbicide fate and transport, individual behavior, and pharmacokinetics. 4. Fate and transport processes not incorporated into the current version of the Stellman team’s model (e.g., width of the spray swath, con- centration of contaminants, primary and secondary drift, soil conditions, initial and remaining canopy, and photodegradation) will affect estimates of exposure to herbicides and their contaminants. Incorporating these p ­ henomena into an exposure model could possibly reduce exposure mis- classification but would require additional data that may or may not be available. However, the relatively coarse resolution of the military UTM system used in military records and the Stellman team’s GIS grid map of Vietnam may limit, to some extent, the benefits of adding fine-scale fate and transport modeling. 5. Regardless of the exposure model used, sensitivity analyses are nec- essary to determine the impact of model assumptions regarding decreases in herbicide concentration with distance and time on the exposure assignments generated. Such studies provide important information on the stability of the proposed exposure opportunity measures. 6. Given the significant uncertainties about the levels of TCDD con- tamination over time and from different lots of the herbicides used in Viet- nam, proximity-based exposure models may be better suited to studies of the health effects of herbicides in general rather than TCDD specifically. 7. It is not feasible to validate the exposure scores produced by the Stellman team’s model, or any other proximity-based model, by compari- sons with biomarker or soil samples because of the passage of time and the unavailability of archived environmental or biological samples. RESEARCH OPPORTUNITIES From its review of the Stellman team’s model, the committee identified two areas where it urges further investigation. 1. 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 ­parameters that account for more aspects of herbicide fate and transport in the environ- ment. Further development of the model will require an assessment of the additional data needed and of the availability of these data. 2. The sensitivity of the Stellman team’s model’s results to changes in parameter values should be assessed systematically. The committee specifi-

ASSESSMENT OF THE MODEL 61 cally urges attention to 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. REFERENCES ATSDR (Agency for Toxic Substances and Disease Registry). 1998. Toxicological profile for chlorinated dibenzo-p-dioxins. Atlanta, GA: U.S. Department of Health and Human Services. Baker, J. I., and R. A. Hites. 2000. Siskiwit Lake revisited: Time trend of polychlorinated dibenzo-p-dioxins and dibenzofuran deposited at Isle Royale, Michigan. Environmental Science and Technology 34:2887–2891. Bertazzi, P. A., I. Bernucci, G. Brambilla, D. Consonni, and A. C. Pesatori. 1998. The Seveso studies on early and long-term effects of dioxin exposure: A review. Environmental Health Perspectives 106(S2):625–633. Bertazzi, P. A., D. Consonni, S. Bachetti, M. Rubagotti, A. Baccarelli, C. Zocchetti, and A. C. Pesatori. 2001. Health effects of dioxin exposure: A 20-year mortality study. American Journal of Epidemiology 153(11):1031–1044. Bilanin, A. J., M. E. Teske, J. W. Barry, and R. B. Ekblad. 1989. AGDISP: The aircraft spray dispersion model, code development and experimental validation. Transactions of the American Society of Agricultural Engineers 32:327–334. Bopp, R. F., M. L. Gross, H. Tong, H. J. Simpson, S. J. Monson, B. L. Deck, and F. C. Moser. 1991. A major incident of dioxin contamination: Sediments of New Jersey estuaries. Environmental Science and Technology 25(5):951–956. Boylan, R. 2007. Accessing military unit records at the College Park Archives. Oral presenta- tion to the IOM Committee on Making Best Use of the Agent Orange Reconstruction Model, Meeting 2, April 30–May 1, Washington, DC. Caramaschi, F., G. del Corno, C. Favaretti, S. E. Giambelluca, E. Montesarchio, and G. M. Fara. 1981. Chloracne following environmental contamination by TCDD in Seveso, Italy. Inter­ national Journal of Epidemiology 10(2):135–143. CDC (Centers for Disease Control and Prevention). 1988. Serum 2,3,7,8-tetrachlorodibenzo- p-dioxin levels in U.S. Army Vietnam-era veterans. Journal of the American Medical Association 260(9):1249–1254. CDC. 1989. Comparison of serum levels of 2,3,7,8-tetrachlorodibenzo-p-dioxin with indirect estimates of Agent Orange exposure among Vietnam veterans: Final report. Atlanta, GA: Agent Orange Projects, Center for Environmental Health and Injury Control. Cecil, P. F., Sr., and A. L. Young. 2008. Operation FLYSWATTER: A war within a war. Envi- ronmental Science and Pollution Research 15(1):3–7. Curwin, B. D., M. J. Hein, W. T. Sanderson, M. G. Nishioka, S. J. Reynolds, E. M. Ward, and M. C. Alavanja. 2005. Pesticide contamination inside farm and nonfarm homes. Journal of Occupational and Environmental Hygiene 2(7):357–367.

62 PROXIMITY-BASED HERBICIDE EXPOSURE ASSESSMENT Czuczwa, J. M., and R. A. Hites. 1984. Environmental fate of combustion-generated poly­ chlorinated dioxins and furans. Environmental Science and Technology 18(6):444–450. Czuczwa, J. M., and R. A. Hites. 1986. Airborne dioxins and dibenzofurans: Sources and fates. Environmental Science and Technology 20(2):195–200. Darrow, R. A., K. R. Irish, and C E. Hinarik. 1969. Herbicides used in Southeast Asia. Technical Report SAOQ-TR-69-11078. Fort Detrick, MD: U.S. Army Plant Sciences Laboratories. Department of the Army. 1971. Tactical employment of herbicides. Field Manual FM 3-3. Washington, DC: Department of the Army. Dwernychuk, L. W., H. D. Cau, C. T. Hatfield, T. G. Boivin, T. M. Hung, P. T. Dung, and N. D. Thai. 2002. Dioxin reservoirs in southern Viet Nam: A legacy of Agent Orange. Chemosphere 47:117–137. Emond, E., J. E. Michalek, L. S. Birnbaum, and M. J. DeVito. 2005. Comparison of the use of a physiology based pharmacokinetic model and a classical pharmacokinetic model for dioxin exposure assessments. Environmental Health Perspectives 113(12):1666–1674. EPA (Environmental Protection Agency). 2005. 2,4-D RED facts. EPA-738F-05-002. http:// www.epa.gov/oppsrrd1/REDs/factsheets/24d_fs.htm (accessed December 4, 2007). EPA. 2006. Reregistration eligibility decision (RED) for malathion. EPA 738-R-06-030. http://www.epa.gov/pesticides/reregistration/REDs/malathion_red.pdf (accessed Decem- ber 4, 2007). Eskenazi, B., P. Mocarelli, M. Warner, L. Needham, D. G. Patterson, Jr., S. Samuels, W. Turner, P. M. Gerthoux, and P. Brambilla. 2004. Relationship of serum TCDD concentrations and age at exposure of female residents of Seveso, Italy. Environmental Health Perspec- tives 112(1):22–27. Flesch-Janys, D., K. Steindorf, P. Gurn, and H. Becher. 1998. Estimation of the cumulated exposure to polychlorinated dibenzo-p-dioxins/furans and standardized mortality ratio analysis of cancer mortality by dose in an occupationally exposed cohort. Environmental Health Perspectives 106(S2):655–662. Frignani, M., R. Piazza, L. G. Bellucci, C. N. Huu, R. Zangrando, S. Albertazzi, and I. Moret. 2004. Polychlorinated biphenyls in sediments of the Tam Giang-Cau Hai Lagoon ­(Central Vietnam): First results. Organohalogen Compounds 66:3657–3663. GAO (General Accounting Office). 1979. U.S. ground troops in South Vietnam were in areas sprayed with Herbicide Orange. FPCD-80-23. Washington, DC: U.S. Government Print- ing Office. Ginevan, M. E., J. H. Ross, and D. K. Watkins. 2008. Assessing exposure to allied ground troops in the Vietnam War: A comparison of AgDRIFT and exposure opportunity index models. Journal of Exposure Science and Environmental Epidemiology advance online publication, March 12 (DOI:10.1038/sj.jes.2008.12). Hatfield Consultants and Office of the National Steering Committee 33, MNRE (Ministry of Natural Resources and Environment, Vietnam). 2007. Assessment of dioxin contamina- tion in the environment and human population in the vicinity of Da Nang Airbase, Viet Nam: Final report. West Vancouver, British Columbia, Canada. HEA-V (Herbicide Exposure Assessment–Vietnam). 2003. CD-ROM, version 1.0.2. Software and accompanying electronic documentation. New York: Columbia University. Heizer, J. R. 1971. Data quality analysis of the HERB 01 data file. MITRE Technical Report, MTR-5105. Prepared for the Defense Communications Agency. McLean, VA: MITRE. IOM (Institute of Medicine). 1994. Veterans and Agent Orange: Health effects of herbicides used in Vietnam. Washington, DC: National Academy Press. IOM. 2003. Characterizing exposure of veterans to Agent Orange and other herbicides used in Vietnam: Interim findings and recommendations. Washington, DC: The National Academies Press.

ASSESSMENT OF THE MODEL 63 Koch, D., C. Lu, J. Fisker-Andersen, L. Jolley, and R. A. Fenske. 2002. Temporal associa- tion of children’s pesticide exposure and agricultural spraying: Report of a longitudinal b ­ iological monitoring study. Environmental Health Perspectives 110(8):829–833. Kramarova, E., M. Kogevinas, C. T. Anh, H. D. Cau, L. C. Dai, S. D. Stellman, and D. M. Parkin. 1998. Exposure to Agent Orange and occurrence of soft-tissue sarcomas or non-Hodgkin lymphomas: An ongoing study in Vietnam. Environmental Health Per- spectives 106(Suppl 2):671–678. Lu, C., R. A. Fenske, N. Simcox, and D. Kalman. 2000. Pesticide exposure of children in an agricultural community: Evidence of household proximity to farmland and take home exposure pathways. Environmental Research 84(3):290–302. MACV (Military Assistance Command, Vietnam). 1969. Military operations: Herbicide opera- tions. Directive 525-1. Springfield, VA: National Technical Information Service. Michalek, J. E., J. L. Pirkle, L. L. Needham, D. G. Patterson, Jr., S. P. Caudill, R. C. Tripathi, and P. Mocarelli. 2002. Pharmacokinetics of 2,3,7,8-tetrachlorodibenzo-p-dioxin in Seveso adults and veterans of operation Ranch Hand. Journal of Exposure Analysis and Environmental Epidemiology 12:44–53. NRC (National Research Council). 1974. The effects of herbicides in Vietnam. Part A— ­ ummary and conclusions. Washington, DC: National Academy of Sciences. S Paustenbach, D. J., R. J. Wenning, V. Lau, N. W. Harrington, D. K. Rennix, and A. H. Parsons. 1992. Recent developments on the hazards posed by 2,3,7,8-tetrachlorodibenzo-p-dioxin in soil: Implications for setting risk-based cleanup levels at residential and industrial sites. Journal of Toxicology and Environmental Health 36(2):103–149. Ross, J. H., and M. E. Ginevan. 2007. Points for the committee to consider when evaluating the Stellman model. PowerPoint presentation to the IOM Committee on Making Best Use of the Agent Orange Exposure Reconstruction Model, Meeting 2, April 30–May 1, Washington, DC. Shaughnessy, C. A. 1991. The Vietnam conflict: “America’s best documented war”? The His- tory Teacher 24(2):135–147. Stanton, S. L. 1989. Area-scoring methodology for estimating Agent Orange exposure status of U.S. Army personnel in the Republic of Vietnam. In Comparison of serum levels of 2,3,7,8-tetrachlorodibenzo-p-dioxin with indirect estimates of Agent Orange exposure among Vietnam veterans: Final report, by Centers for Disease Control and Preven- tion. Atlanta, GA: Agent Orange Projects, Center for Environmental Health and Injury Control. Stellman, J. M. 2007. Responses to IOM 091407. Unpublished document submitted to the IOM Committee on Making the Best Use of the Agent Orange Reconstruction Model, September 14. Stellman, J. M., and S. D. Stellman. 2003. Contractor’s final report: Characterizing exposure of veterans to Agent Orange and other herbicides in Vietnam. Submitted to the National Academy of Sciences, Institute of Medicine, in fulfillment of Subcontract VA-5124-98- 0019, June 30, 2003. Stellman, S. D., and J. M. Stellman. 2004. Exposure opportunity models for Agent Orange, dioxin, and other military herbicides used in Vietnam, 1961–1971. Journal of Exposure Analysis and Environmental Epidemiology 14(4):354–362. Stellman, J. M., S. D. Stellman, R. Christian, T. Weber, and C. Tomasallo. 2003a. The ex- tent and patterns of usage of Agent Orange and other herbicides in Vietnam. Nature 422(6933):681–687. Stellman, J. M., S. D. Stellman, T. Weber, C. Tomasallo, A. B. Stellman, and R. Christian, Jr. 2003b. A geographic information system for characterizing exposure to Agent Orange and other herbicides in Vietnam. Environmental Health Perspectives 111(3):321–328.

64 PROXIMITY-BASED HERBICIDE EXPOSURE ASSESSMENT Teske, M. E., S. L. Bird, D. M. Esterly, T. B. Curbishly, S. L. Ray, and S. G. Perry. 2002. AgDRIFT®: A model for estimating near-field spray drift from aerial applications. Envi- ronmental Toxicology and Chemistry 21(3):659–671. U.S. Army. 1985. Services HERBS Tape. Report No. AD-A160 563. Washington, DC: U.S. Army and Joint Services Environmental Support Group. Verger, P., S. Cordier, L. T. B. Thuy, D. Bard, L. C. Dai, P. H. Phiet, M. F. Gonnord, and L. Abenhaim. 1994. Correlation between dioxin levels in adipose tissue and estimated exposure to Agent Orange in South Vietnamese residents. Environmental Research 65:226–242. Ward, M. H., J. Lubin, J. Giglierano, J. S. Colt, C. Wolter, N. Bekiroglu, D. Camann, P. Hartge, and J. R. Nuckols. 2006. Proximity to crops and residential exposure to agri- cultural herbicides in Iowa. Environmental Health Perspectives 114(6):893–897. Young, A. L. 2007. Public statement to the Committee on Making Best Use of the Agent ­Orange Exposure Reconstruction Model, Meeting 2, April 30–May 1, Washington, DC. Young, A. L., and M. Newton. 2004. Long overlooked historical information on Agent O ­ range and TCDD following massive applications of 2,4,5-T-containing herbicides, Eglin Air Force Base, Florida. Environmental Science and Pollution Research 11(4):209–221. Young, A. L., J. A. Calcagni, C. E. Thalken, and J. W. Tremblay. 1978. The toxicology, envi- ronmental fate, and human risk of Herbicide Orange and its associated dioxin. OEHL TR-78-92, Final Report. Brooks Air Force Base, TX: U.S. Air Force Occupational and Environmental Health Laboratory. Young, A. L., P. F. Cecil, and J. F. Guilmartin, Jr. 2004a. Assessing possible exposure of ground troops to Agent Orange during the Vietnam War: The use of contemporary military records. Environmental Science and Pollution Research 11(6):349–358. Young, A. L., J. P. Giesy, P. D. Jones, and M. Newton. 2004b. Environmental fate and bioavail- ability of Agent Orange and its associated dioxin during the Vietnam War. Environmental Science and Pollution Research 11(6):359–370.

Next: 4 Data for Epidemiologic Studies of Vietnam Veterans »
The Utility of Proximity-Based Herbicide Exposure Assessment in Epidemiologic Studies of Vietnam Veterans Get This Book
×
Buy Paperback | $52.00 Buy Ebook | $41.99
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

A fundamental challenge in past studies evaluating whether health problems experienced by Vietnam veterans might be linked to wartime use of Agent Orange or other herbicides has been a lack of information about the veterans' level of exposure to these herbicides. To address that problem, researchers developed a model to assess the opportunity for herbicide exposure among these veterans.

The Utility of Proximity-Based Herbicide Exposure Assessment in Epidemiologic Studies of Vietnam Veterans presents the conclusions and recommendations of an Institute of Medicine committee (IOM) that was convened to provide guidance to the Department of Veterans Affairs (VA) about the best use of a model to characterize exposure to the troops based on their proximity to herbicide spraying in Vietnam. This book's assessment is guided by four primary considerations: to be clear about what the assessment model does and does not claim to do; to gain understanding of the strengths and limitations of data on herbicide spraying, troop locations, and health outcomes; to consider whether the model locates spraying and troops accurately to consider the potential contributions and pitfalls of using it in epidemiologic studies. Of particular interest in these deliberations were the degree to which exposure classification might be improved if the model were to be used, and the appropriate interpretation of the results of any such studies.

In light of the questions that remain concerning herbicide exposure and health among Vietnam veterans and the array of evidence that has thus far been brought to bear on that issue, The Utility of Proximity-Based Herbicide Exposure Assessment in Epidemiologic Studies of Vietnam Veterans concludes that the application of this model offers a constructive approach to extending knowledge about the effects of herbicides on the health of these veterans and merits the initial steps recommended in our report.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!