that is, the veterans were asked to report physician-diagnosed major health problems or impairments during the first five years of their children's lives. About half of the health conditions reported were respiratory disease (mostly asthma and pneumonia) and otitis media. For most of the conditions, the veterans reported more health conditions than non-Vietnam veterans (all conditions, OR = 1.3, CI 1.2-1.4). After excluding children with a serious health condition or either a birth defect or cancer, the overall crude OR was 1.2 (CI 1.1-1.3). Elevated crude odds ratios were found for anemias (OR = 2.0, CI 1.2-3.3), diseases of the skin (OR = 1.5, CI 1.1-1.9), rash (OR = 2.3, CI 1.1-4.9), and allergies (OR = 1.6, CI 1.2-2.1). Without medical records that validate for many of these types of common conditions and health problems, recall bias may be an explanation for many of these findings.

The CDC (1989) did conduct two substudies using hospital records to identify birth defects among the veterans' offspring. The first, the General Birth Defects Study (GBDS), compared the occurrence of birth defects recorded on hospital records for the children of Vietnam and of non-Vietnam veterans (130 cases and 112 cases, respectively) who participated in the medical examination component of the VES. For a variety of characteristics, there were no apparent differences between the group of men who participated in the exam and the total interview group. There was no difference in the prevalence of birth defects between the two groups of children (crude OR = 1.0, CI 0.8-1.3). There was a slight but nonsignificant excess for major birth defects (OR = 1.2, CI 0.8-1.9). When analyzed by organ system, only digestive system defects appeared to be elevated (OR = 2.0, CI 0.9-4.6), although the small number of defects precluded the analysis of several broad categories. The number of defects was also too small for the analysis of specific individual defects. An analysis by race did indicate an elevated odds ratio (3.4, CI 1.5-7.6) for black Vietnam veterans. An examination of the specific defects listed on hospital records for children of black veterans did not reveal any particular pattern. A comparison of interview and hospital records was also conducted to evaluate the extent of potential misclassification of veteran responses. In general, interview responses were not predictive of the presence of a defect for either veteran group. The agreement between interview and hospital records was slightly poorer for Vietnam veterans. For example, the positive predictive value of the interview response for the presence of a defect in the hospital record was 24.8 percent among Vietnam veterans and 32.9 percent among non-Vietnam veterans. Sensitivity was 27.1 percent among Vietnam veterans and 30.3 percent among non-Vietnam veterans. The kappa measure of agreement was also lower (20.9 percent versus 27.6 percent) among Vietnam veterans.

The second substudy, the Cerebrospinal Malformation (CSM) Study, involved the analysis of medical records for all cases of cerebrospinal malformations (spina bifida, anencephalus, hydrocephalus) and stillbirths reported by veterans in the interview study. The substudy found 26 cerebrospinal malformations (live and stillbirths) among children of Vietnam veterans and 12 among children



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



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

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

OCR for page 289
--> that is, the veterans were asked to report physician-diagnosed major health problems or impairments during the first five years of their children's lives. About half of the health conditions reported were respiratory disease (mostly asthma and pneumonia) and otitis media. For most of the conditions, the veterans reported more health conditions than non-Vietnam veterans (all conditions, OR = 1.3, CI 1.2-1.4). After excluding children with a serious health condition or either a birth defect or cancer, the overall crude OR was 1.2 (CI 1.1-1.3). Elevated crude odds ratios were found for anemias (OR = 2.0, CI 1.2-3.3), diseases of the skin (OR = 1.5, CI 1.1-1.9), rash (OR = 2.3, CI 1.1-4.9), and allergies (OR = 1.6, CI 1.2-2.1). Without medical records that validate for many of these types of common conditions and health problems, recall bias may be an explanation for many of these findings. The CDC (1989) did conduct two substudies using hospital records to identify birth defects among the veterans' offspring. The first, the General Birth Defects Study (GBDS), compared the occurrence of birth defects recorded on hospital records for the children of Vietnam and of non-Vietnam veterans (130 cases and 112 cases, respectively) who participated in the medical examination component of the VES. For a variety of characteristics, there were no apparent differences between the group of men who participated in the exam and the total interview group. There was no difference in the prevalence of birth defects between the two groups of children (crude OR = 1.0, CI 0.8-1.3). There was a slight but nonsignificant excess for major birth defects (OR = 1.2, CI 0.8-1.9). When analyzed by organ system, only digestive system defects appeared to be elevated (OR = 2.0, CI 0.9-4.6), although the small number of defects precluded the analysis of several broad categories. The number of defects was also too small for the analysis of specific individual defects. An analysis by race did indicate an elevated odds ratio (3.4, CI 1.5-7.6) for black Vietnam veterans. An examination of the specific defects listed on hospital records for children of black veterans did not reveal any particular pattern. A comparison of interview and hospital records was also conducted to evaluate the extent of potential misclassification of veteran responses. In general, interview responses were not predictive of the presence of a defect for either veteran group. The agreement between interview and hospital records was slightly poorer for Vietnam veterans. For example, the positive predictive value of the interview response for the presence of a defect in the hospital record was 24.8 percent among Vietnam veterans and 32.9 percent among non-Vietnam veterans. Sensitivity was 27.1 percent among Vietnam veterans and 30.3 percent among non-Vietnam veterans. The kappa measure of agreement was also lower (20.9 percent versus 27.6 percent) among Vietnam veterans. The second substudy, the Cerebrospinal Malformation (CSM) Study, involved the analysis of medical records for all cases of cerebrospinal malformations (spina bifida, anencephalus, hydrocephalus) and stillbirths reported by veterans in the interview study. The substudy found 26 cerebrospinal malformations (live and stillbirths) among children of Vietnam veterans and 12 among children

OCR for page 289
--> of non-Vietnam veterans. No formal analysis of the difference in malformations between the veteran groups was conducted, because negative responses (i.e., children without a reported malformation) were not verified and the participation rates differed between groups (7.8 percent of Vietnam veterans and 22.1 percent of non-Vietnam veterans refused to participate). The VES did find suggestive associations for birth defects. It is interesting to note that some potential associations were found for birth defects considered by the investigators to be "relatively common, easily diagnosed, and observable at birth" (CDC, 1989). These include hydrocephalus (OR = 5.1, CI 1.1-23.1) and hypospadias (OR = 3.1, CI 0.9-11.3). The GBDS did not replicate these findings, but this sample had limited power for the analysis of specific defects. Although associations were not found for all conditions, there was clearly a general pattern of a greater prevalence of birth defects in the offspring of Vietnam veterans, according to self-reports. The authors properly note the potential for recall bias as an explanation for the pattern of excess risk. As an attempt to evaluate recall bias, two record validation studies of birth defects were conducted. Overall, the GBDS did not find any association with an increased risk of birth defects among offspring of Vietnam veterans. However, this validation study had limited power to detect an increased risk for specific birth defects. The second validation substudy, the CSM review, was flawed by the differentially poor response rate among the non-Vietnam veteran group. This result and the fact that negative responses were not pursued discouraged the investigators from estimating the relative risk for cerebrospinal malformations. Another important study of Vietnam veterans was the CDC Birth Defects Study (Erickson et al., 1984a,b). In this study, children with birth defects among 428 fathers who were reported to have been Vietnam veterans were compared to children with birth defects among 268 control fathers who were non-Vietnam veterans. The odds ratio for Vietnam veteran status in relation to any major birth defect among offspring was 1.0 (CI 0.8-1.1). Analysis of the Agent Orange exposure opportunity index (EOI; see VAO Chapter 6 for details) based on both military records and self-reports did not indicate a statistically significant trend of increasing risk of all types of birth defects (combined) with increasing levels of Agent Orange exposure. No association was noted between Vietnam veteran status or self-reported Agent Orange exposure and risk of fathering a child with multiple birth defects (OR = 1.1, CI 0.7-1.7). The odds ratios for Vietnam veteran status, self-reported Agent Orange exposure, and logistic regression coefficients for EOI based on self-report and military records for most of the 95 birth defect groups were not significantly elevated. Although the odds ratio for spina bifida was not elevated with Vietnam veteran status (OR = 1.1), the EOI indices showed a pattern of increasing risk. For example, the odds ratios for the EOI based on information obtained during the interview for low to high levels of exposure (levels 1 to 5) were 1.2 (CI 1.0-1.4), 1.5 (CI 1.1-2.1), 1.8 (CI 1.1-3.0), 2.2 (CI 1.2-4.3), 2.7 (CI 1.2-6.2). A similar pattern was found for cleft lip with/without

OCR for page 289
--> cleft palate—namely, EOI-1 (OR = 1.2, CI 1.0-1.4), EOI-2 (OR = 1.4, CI 1.0-1.9), EOI-3 (OR = 1.6, CI 1.0-2.6), EOI-4 (OR = 1.9, CI 1.0-3.6), and EOI-5 (OR = 2.2, CI 1.0-4.9). The category ''specified anomalies of nails" had an increased odds ratio for Vietnam veteran status and elevated coefficients (not statistically significant) for the two exposure indices. The category "other neoplasms" was related to the EOI based on the father's self-reported Agent Orange exposure. This group included a variety of congenital neoplasms such as cysts, teratomas, and benign tumors. In an attempt to search for a Vietnam veteran birth defect "syndrome," pairs and triplets of defects were examined for combinations that yielded significant differences in the distribution among Vietnam veterans and controls. According to the authors, these analyses did not produce any important associations or patterns among defect combinations. The results of this study were generally negative; that is, there was not a general pattern of increased risk for birth defects among the offspring of Vietnam veterans. However, the analysis of the Agent Orange EOIs based on military records found a significant trend for increased risk for spina bifida with increased exposure. As the authors note, this finding must be viewed with caution, because a related defect, anencephalus, was not found to be associated with a significant EOI trend. Another positive association was noted for cleft lip without cleft palate, where a significant regression coefficient was found for the EOI index based on the father's interview. No association was found for the EOI from military records. The CDC Birth Defects Study has many strengths, including the use of a population-based registry system with careful classification of birth defects for analysis. The statistical power of the study was excellent for many major birth defects. Use of the Agent Orange EOIs is an attempt to refine exposure assessment procedures compared to measures used in most other studies. The study did have several important limitations. First, the response rates among cases and controls were problematic, with approximately 56 percent of eligible case and control fathers interviewed. Examination of the nonparticipation group revealed lower participation among persons classified as "nonwhite." The analyses by race did not find important differences, but the potential for bias should not be overlooked. Another problem relates to the fact that case births occurred from 1968 through 1980, but interviews took place during 1982 and 1983, up to 14 years after the birth. To minimize the potential recall bias induced by this long lag period, controls were matched on year of birth. Aschengrau and Monson (1990) studied late adverse pregnancy outcomes among 14, 130 obstetric patients who delivered at Boston Hospital for Women from August 1977 to March 1980. History of the fathers' military service in Vietnam was determined from Massachusetts and national military records by using the husbands' names and Social Security numbers. The likelihood of combat experience, based on branch of service and military occupation, was used to estimate potential herbicide exposure. The analyses compared the risk of

OCR for page 289
--> malformations among children of 107 Vietnam veterans to that for children of 1,432 men without known military service; the risk in 313 non-Vietnam veterans compared to the men without military service; and the risk in the Vietnam veterans compared with the non-Vietnam veterans. There was a slight, nonsignificant increase in the odds ratio for all congenital anomalies for Vietnam veterans compared to men without known military service (OR = 1.3, CI 0.9-1.9) and for Vietnam veterans compared with non-Vietnam veterans (OR = 1.2, CI 0.8-1.9). For major malformations, the odds ratio was elevated for Vietnam veterans compared with men without military service (OR = 1.8, CI 1.0-3.1), but the ratio decreased for Vietnam veterans compared with non-Vietnam veterans (OR = 1.3, CI 0.7-2.4). Only slight increases were found for the analysis of minor malformations and "only normal variants." Although based on small numbers, the analyses of 12 malformation groups found that children of Vietnam veterans, compared to children of men with no known military service, had an increased risk of malformations of the nervous system, cardiovascular system, genital organs, urinary tract, and musculoskeletal system. Confidence intervals were not presented with the odds ratio estimates, but it was noted that they included 1.0, so elevated risks were not significantly increased. Further examination of specific anomaly diagnoses for the 18 infants of Vietnam veterans with major malformations did not reveal any pattern of association with potential herbicide exposure. Although the study did find a positive association between paternal military service in Vietnam and the risk of major malformations in offspring, the authors suggest cautious interpretation of their findings, given the small number of subjects in many of the comparisons involving specific groups of birth defects. Additionally, it was noted that some of the malformations observed can also be due to maternal and delivery factors (endocrine condition and fetal presentation). An important problem relates to misclassification of herbicide exposure due to equating exposure to service in Vietnam. Two state health surveys of veterans (Iowa and Hawaii) did not indicate an increased prevalence of birth defects (Rellahan, 1985; Wendt, 1985), but a survey in Maine did report an increased risk of birth defects among veterans (Deprez et al., 1991). The limitations of these general survey studies affect their usefulness in this evaluation. As part of the National Vietnam Veterans Birth Defects/Learning Disabilities Registry and database, a joint project of the Association of Birth Defect Children and the New Jersey Agent Orange Commission, a self-administered questionnaire was sent to Vietnam veterans to inquire about birth defects and a variety of conditions and disabilities in the children of Vietnam veterans and non-Vietnam veterans (Lewis and Mekdeci, 1993). A preliminary analysis indicated no differences in birth defects between the two groups; however, for a variety of conditions, including allergies, frequent infections, benign tumors, cysts, and chronic skin disorders, the veterans showed a higher frequency. The possibility of recall bias and the self-selected nature of the registry are of concern. Nonetheless,

OCR for page 289
--> a carefully designed and comprehensive epidemiology study with review of medical records could address the possibility of an association with some of these childhood health conditions. A study of birth defects among offspring of Australian Vietnam veterans was conducted using a total of 8,517 matched case-control pairs, with 127 infant cases and 123 infant controls having a father who served in Vietnam (Donovan et al., 1984). There were 202 cases and 205 controls whose fathers were in the Army but did not serve in Vietnam. The adjusted odds ratio for birth defects among children of Vietnam veterans versus all other men was 1.02 (CI 0.8-1.3). Analysis of subgroups based on the type of Army veteran (Australian Regular Army enlistees, National Service draftees) did not detect any increased odds ratios for these comparisons. There was a slight, statistically nonsignificant increase in the odds ratio for National Service Vietnam veterans versus those who did not serve in Vietnam (OR = 1.3, CI 0.9-2.0). The risk was independent of the length of Vietnam service and the time between service and conception. Analyses by diagnostic group (central nervous system, cardiovascular, oral clefts, hypospadias, musculoskeletal, dislocation of hip, chromosomal anomalies) did not show an excess risk for Vietnam veterans. However, two defects had odds ratios above 1.5 (statistically nonsignificant)—ventricular septal defects (OR = 1.8) and Down's syndrome (OR = 1.7). Overall, this study was negative; that is, there was no evidence of an increased risk of fathering a child with a congenital anomaly for Australian Army veterans who served in Vietnam. As indicated by the upper confidence limit (1.3), this study had adequate power to rule out an odds ratio greater than 1.3 for congenital anomalies. Assessment of potential Agent Orange exposure in this study is limited, because "history of service" in Vietnam was used as the primary "exposure" variable. This uncertainty is further compounded by potential differences in the location and nature of service of Australian veterans in Vietnam and their resultant herbicide exposure. The Australian study of veterans living in Tasmania reported more congenital anomalies among the 357 Vietnam veterans than among the comparison families (Field and Kerr, 1988). The authors suggested that the results indicated a pattern of association with congenital heart disease and anomalies of the central nervous system. As described earlier in the section on spontaneous abortion, there are several notable problems with this study, including inadequate presentation of results, potential selection bias, self-reported health outcomes, and using service in Vietnam as a surrogate for herbicide exposure. Ranch Hand Study The latest report from the Air Force Health Study (AFHS) of Operation Ranch Hand veterans ("Ranch Hands") and their children was published in 1995 (Wolfe et al., 1995). The Air Force released a first report on the analysis of reproductive effects in 1982, and this report was reviewed in VAO (AFHS, 1992). The original study cohort comprised 1,098 Ranch Hands who

OCR for page 289
--> regularly handled and sprayed herbicides in Southeast Asia from 1962 to 1971 ("exposed cohort") and a comparison group of 1,549 Air Force veterans who were in Southeast Asia at the same time but presumably were not exposed to herbicides. In 1987, 995 Ranch Hands (91 percent of original study group) and 1,299 comparison veterans (84 percent of original group) participated in a physical exam and agreed to provide serum samples for the dioxin assay. A total of 872 Ranch Hands (79 percent of original cohort, 88 percent of 1987 cohort) and 1,036 comparison subjects (67 percent of original group, 80 percent of 1987 cohort) were available for analysis, after exclusion of samples that were unreliable because of laboratory error or that had dioxin levels below the level of detection or above an upper threshold for background (10 parts per trillion [ppt]) for comparison subjects. Of the 872 Ranch Hands, 454 had 1,006 self-reported conceptions and 419 fathered 792 liveborn infants during their service in Vietnam or until January 1990. Of the 1,036 comparison veterans, 570 had 1,235 conceptions and 531 fathered 981 liveborn infants during this period. The initial dioxin level was estimated from the current level using a first-order decay rate model with a fixed 7.1-year half-life estimate. The referent group for the Ranch Hands included the conceptions and offspring of comparison men with "background" levels (≤10 ppt, N = 570, mean = 3.9ppt). Ranch Hands with levels at background were analyzed as a separate stratum (N = 179), since the authors felt this group included a mixture of exposed and unexposed veterans, given their mean level of 6.1 ppt and uncertainties in dioxin elimination. The other strata used in the analysis included Ranch Hand "low" (current ≤10 ppt and initial ≤110 ppt, N = 119) and Ranch Hand "high" (current ≤10 ppt and initial >110 ppt, N = 156). The 110 ppt level was chosen because it is the median estimated initial dioxin level at the time of conception of the Ranch Hands with levels greater than 10 ppt. As the authors point out, this cutoff is arbitrary, with no assumed biologic meaning. Reproductive outcomes of comparison veterans with a current dioxin levels of greater than 10 ppt were not analyzed, because the investigators suspected that these may reflect dioxin exposure after service in Vietnam. The reproductive and developmental outcomes included in the analyses included spontaneous abortion (miscarriage, fetal death less than 20 weeks gestation), stillbirth (fetal death 20 weeks or greater gestation), and birth defects. All conceptions reported by the men, their wives, or their partners were verified through medical records and vital statistics review. The proportion of adverse outcomes verified by specific sources was not stated. This may be important, given the known limitations of vital statistics records for the identification and classification of certain pregnancy outcomes. Stratified analyses were performed, adjusting for six covariates, including father's race, mother's smoking and drinking during pregnancy, mother's and father's age at birth or conception, and father's military occupation (officer, enlisted flyer, enlisted nonflyer). In addition, adjustment was made for history of

OCR for page 289
--> spontaneous abortion prior to service in Southeast Asia. The authors noted that the adjustment of father's military occupation was performed because it may serve as a proxy for education and occupation is associated with dioxin level. Adjustment for occupation may, in fact, lead to some degree of "overadjustment" owing to the high correlation between occupation and exposure potential. Comparison of the adjusted estimates with the unadjusted risk ratio estimates derived from the data provided in the paper showed little difference, indicating that the adjustment for military occupation did not materially affect the results. Summary The validation of self-reported birth defects in this study was systematic and of high-quality. Although the etiology of most birth defects remains unknown, the study accounted for an array of factors controlled for in most previous studies of birth defects. Considering all birth defects combined, there was a slightly higher proportion of defects among Ranch Hand children than among comparison children (22.3 percent versus 20.8 percent). No general pattern of increasing risk with increasing dioxin levels was found. A small increased RR of 1.3 (CI 1.0-1.6) was found for the low-dioxin category. There was a slightly higher prevalence of major birth defects among Ranch Hand children compared to comparison children (7.4 percent versus 5.7 percent). There was an elevated risk ratio for the low-level category (RR = 1.7; CI 1.1-2.7), although a dose-response gradient was not evident, with an RR of 1.1 for background (0.6-1.8) and 1.2 (CI 0.8-2.1) for the high-level category. The analysis of birth defect groups yielded a total of 11 increased and five decreased risk ratios for the lowand high-level comparisons with the referent category. For example, the analysis of circulatory system and heart defects found risk ratios of 2.3 for low and 0.9 for high levels. Genital defects had risk ratios of 1.8 for low and 1.2 for high; urinary system defects had risk ratios of 2.0 for low and 2.1 for high. Examination of specific defects included in this larger defect grouping did not show any specific associations or patterns. Interestingly, neural tube defects (spina bifida, anencephaly) were in excess among offspring of Ranch Hands, with four total (rate of five per 1,000), in contrast to none among the comparison infants (exact p = .04). The four cases were distributed as two spina bifida in the high-level category, one anencephaly and one spina bifida in the low-dioxin category. There was no clear pattern of association with developmental disabilities in terms of specific delays in development or hyperkinetic syndrome, although the low-level stratum for specific delays in development had a risk ratio of 1.5 (CI 1.0-2.3). The recently published results of the analysis of birth defects among the offspring of Ranch Hands suggest the possibility of an association between dioxin exposure and risk of neural tube defects. These findings require a consideration of the current evidence for an association between herbicides and neural tube defects and an increased risk among Vietnam veterans exposed to herbicides.

OCR for page 289
--> Table 9.1 includes a summary of the studies that have reported results specifically for neural tube defects (typically anencephaly and/or spina bifida), including studies in VAO and more recent publications. Unfortunately, some studies (e.g., Seveso), particularly the occupational and environmental studies, do not have results specific for individual birth defects, usually because of the small number of cases. A number of studies of veterans appear to show an elevated relative risk for either service in Vietnam or estimated exposure to herbicides or dioxin and neural tube defects (anencephaly and/or spina bifida) in their offspring. Many of the estimates are imprecise, and chance cannot be ruled out. Nonetheless, the pattern of association warrants further evaluation. The CDC Birth Defects Study (Centers for Disease Control, 1988), the CDC Vietnam Experience Study (Centers for Disease Control, 1989), and the Ranch Hand Study (Wolfe et al., 1992) are of the highest overall quality. The CDC VES cohort study found that more Vietnam veterans reported that their children had a central nervous system anomaly (OR = 2.3; 95 percent CI 1.2-4.5) than did non-Vietnam veterans (Centers for Disease Control, 1989). The odds ratio for spina bifida was 1.7 (CI 0.6-5.0). A substudy was conducted in an attempt to validate the reported cerebrospinal defects (spina bifida, anencephaly, hydrocephalus) by examination of hospital records. A difference was detected, but its interpretation was limited by differential participation between the veteran groups and failure to validate negative reported—i.e., the veterans not reporting their children having a birth defect. Thus, the issue of a recall bias remains a major concern with this study. The CDC Birth Defects Study utilized the population-based birth defects registry system in the metropolitan Atlanta area (Centers for Disease Control, 1988). There was no association between overall Vietnam veteran status and the risk of spina bifida (OR = 1.1, CI 0.6-1.7) or anencephaly (OR = 0.9, CI 0.5-1.7). However, the exposure opportunity index based on interview data was associated with an increased risk of spina bifida; for the highest estimated level of exposure (EOI-5), the OR was 2.7 (CI 1.2-6.2). There was no similar pattern of association for anencephaly. This study has a number of strengths, including the use of a population-based birth defects registry system and adjustment for a number of potentially confounding factors. Two study limitations include the relatively low response proportions among the case and control subjects (approximately 56 percent) and the lag between birth and interview for some cases and controls. Thus, all three epidemiologic studies (Ranch Hand, VES, CDC Birth Defects Study) suggest an association between herbicide exposure and an increased risk of spina bifida in offspring. Although the studies were judged to be of relatively high-quality, they suffer from methodologic limitations, including possible recall bias, nonresponse bias, small sample size, and misclassification of exposure. In addition, the failure to find a similar association with anencephaly, an embryologically related defect, is of concern.