NeuroToxicology 22 (2001) 491502

Serum Dioxin and Cognitive Functioning among Veterans of Operation Ranch Hand
Drue H. Barrett1, Robert D. Morris2, Fatema Z. Akhtar3, Joel E. Michalek4,*
1 Division of Environmental Hazards and Health Effects, National Center for Environmental Health, Centers for Disease Control and Prevention, Atlanta, GA, USA 2 Department of Psychology, Georgia State University, Atlanta, GA, USA 3 SpecPro Inc., San Antonio, TX, USA 4 Air Force Research Laboratory, Brooks Air Force Base, TX, USA

Received 20 July 2000; received in revised form 26 February 2001; accepted 28 June 2001

Abstract
We used the HalsteadReitan neuropsychological test battery, the Wechsler adult intelligence scale-revised, the Wechsler memory scale, and the wide range achievement test to assess cognitive functioning among Air Force veterans exposed to Agent Orange and its contaminant, 2,3,7,8-tetrachlorodibenzo-p-dioxin (dioxin), during the Vietnam war. The index subjects were veterans of Operation Ranch Hand (N 937), the unit responsible for aerial herbicide spraying in Vietnam from 1962 to 1971. A comparison group of other Air Force veterans (N 1052), who served in Southeast Asia during the same period but were not involved with spraying herbicides served as referents. Cognitive functioning was assessed in 1982, and dioxin levels were measured in 1987 and 1992. We assigned each Ranch Hand veteran to the background, low, or high dioxin exposure category on the basis of a measurement of dioxin body burden. Although we found no global effect of dioxin exposure on cognitive functioning, we did nd that several measures of memory functioning were decreased among veterans with the highest dioxin exposure. These results became more distinct when we restricted the analysis to enlisted personnel, the subgroup with the highest dioxin levels. An analysis based on dioxin quintiles in the combined cohort produced consistent results, with veterans in the fth quintile exhibiting reduced verbal memory function. Although statistically signicant, these differences were relatively small and of uncertain clinical signicance. Published by Elsevier Science Inc.

Keywords: Dioxin; Epidemiology; Herbicide; Cognitive function; Memory; TCDD

INTRODUCTION To address the concerns of veterans, Congress, and the public about the consequences of exposure to Agent Orange and its contaminant 2,3,7,8-tetrachlorodibenzo-p-dioxin (dioxin), the Air Force has been conducting a 20-year prospective study of veterans of Operation Ranch Hand, the unit responsible for handling and aerial spraying of herbicides in Vietnam from 1962 to 1971. These men were exposed to herbicides while loading spray tanks, maintaining the aircraft and spray equipment, and in ight. Agent Orange
Corresponding author. Tel.: 1-210-536-3441. E-mail address: joel.michalek@brooks.af.mil (J.E. Michalek). 0161-813X/01/$ see front matter. Published by Elsevier Science Inc. PII: S 0 1 6 1 - 8 1 3 X ( 0 1 ) 0 0 0 5 1 - 1
*

comprised approximately 60% of the 19 million gallons of herbicides sprayed in Vietnam, was a 1:1 mixture of 2,4-dichlorophenoxyacetic acid and 2,4,5trichlorophenoxyacetic acid; and was contaminated, from less than 0.05 to almost 50 ppm, with 2,3,7,8tetrachlorodibenzo-p-dioxin (Institute of Medicine, 1994). Participants of the Air Force Health Study were administered physical and psychological examinations in 1982, 1985, 1987, 1992 and 1997. An additional examination is planned for 2002. In addition to concerns about physical health problems, veterans and their family members have expressed concern about the impact of Agent Orange exposure on cognitive functioning, such as attention and concentration skills, monitoring and regulation of

492

D.H. Barrett et al. / NeuroToxicology 22 (2001) 491502 Table 1 Sample size reduction by group among US Air Force veterans who served in Southeast Asia from 1962 to 1971 Ranch Hand Examined in 1982 Missing dioxina Non-quantifiable dioxin No dioxin resultb Dioxin >10 ppt Invalid cognitive resultsc Included in study 1046 93 3 12 1 937 Comparison 1223 111 24 12 24 0 1052 Total 2269 204 27 24 24 1 1989

memory information, and processing complex information. Studies have found conicting results regarding an association between dioxin and cognitive abilities (Klawans, 1987; Stehr-Green et al., 1987; Peper et al., 1993). In this analysis, we examine the association between cognitive functioning and Agent Orange exposure by examining neuropsychological test results of Vietnam veterans participating in the Air Force Health Study. MATERIALS AND METHODS Study design and subject selection procedures for the Air Force Health Study are published elsewhere (Wolfe et al., 1990). The goal of the study is to determine whether veterans of Operation Ranch Hand have experienced adverse health effects due to exposure to herbicides or their dioxin contaminant. Health indices and the cumulative mortality experience of Ranch Hand veterans have been contrasted with those of a comparison group of Air Force veterans who served in Southeast Asia during the same period as Ranch Hand veterans but were not involved with spraying herbicides. Comparison veterans were matched to Ranch Hand veterans on age, race, and military occupation. All Ranch Hand and comparison veterans are male. Dioxin Assessment During the 1987 examination, blood from willing participants was collected and assayed for dioxin (Grubbs et al., 1996). Veterans with no quantiable dioxin result in 1987, those who refused dioxin testing in 1987, and those new to the study in 1992 were also asked to give blood for the assay at the 1992 examination. Participation was voluntary and written informed consent was obtained at the examination site after a complete description of the study and a full explanation of all procedures. We conducted a primary and an alternative analysis. The primary analysis excluded veterans with no dioxin measurement, those with a non-quantiable dioxin result, and comparison veterans with a dioxin result greater than 10 ppt in lipid, the value, we regard as a threshold for background dioxin exposure (Michalek et al., 1998). One veteran's cognitive test results were considered invalid and were excluded from this study because he had been diagnosed with epilepsy prior to the 1982 physical examination. This resulted in inclusion of 937 Ranch Hand and 1052 comparison

a Refused the dioxin assay or non-compliant to the 1987 and 1992 physical examinations. b Serum drawn but no result due to a failure of one or more laboratory quality control checks and insufficient sample to repeat the assay. c Diagnosed as epileptic prior to the administration of cognitive function tests.

veterans. Table 1 shows study size reductions by group. We estimated the initial dioxin dose at the end of the last tour of duty in Vietnam in Ranch Hand veterans having current dioxin levels greater than 10 ppt, using a rst-order model with a constant half-life of 8.7 years (Michalek et al., 1996a). We assigned each veteran to one of four exposure categories: ``comparison,'' ``background,'' ``low,'' and ``high.'' The ``comparison'' category comprised comparison veterans who had a 1987 or 1992 dioxin measurement less than or equal to 10 ppt. The ``background'' category comprised Ranch Hand veterans with a 1987 or 1992 dioxin measurement of less than or equal to 10 ppt. Ranch Hand veterans with a 1987 or 1992 dioxin measurement greater than 10 ppt were assigned to ``low'' or ``high'' dioxin categories if their initial dioxin levels were less than or equal to 94 or greater than 94 ppt, respectively. The cut point separating the ``low'' and ``high'' categories (94 ppt) was the median initial dioxin level among all Ranch Hand veterans having 1987 or 1992 dioxin levels greater than 10 ppt. On the basis of these categorizations, we included 1052 comparison veterans, 388 Ranch Hand veterans with background dioxin levels, 274 Ranch Hand veterans with low dioxin levels, and 275 Ranch Hand veterans with high dioxin levels. Results from the primary analysis are presented in tables and discussed in the text. In the alternative analysis, we included all veterans with a dioxin level (Ranch Hand: n 940, comparison: n 1100) and stratied the combined cohort by quintile of the dioxin distribution. Veterans with non-quantiable dioxin levels were assigned a level equal to one half of their quantitation limit. This approach accounted for dioxin exposure from any

D.H. Barrett et al. / NeuroToxicology 22 (2001) 491502

493

source, including comparisons with levels greater than 10 ppt, and those with non-quantiable levels, who were presumably only lightly exposed. The dioxin quintiles, in parts per trillion units (and sample sizes) were (1) 0 dioxin 3:13 (n 411), (2) 3:13 < dioxin 4:63 (n 409), (3) 4:63 < dioxin 6:95 (n 405), (4) 6:95 < dioxin 15:70 (n 407), and (5) 15:70 < dioxin 617:75 (n 408). Cognitive Assessment To assess cognitive functioning, we administered the HalsteadReitan (HR) neuropsychological test battery (Jarvis and Barth, 1984), a widely used assessment of a variety of cognitive and behavioral functions; the Wechsler adult intelligence scale-revised (WAIS-R) (Wechsler, 1981), a standard measure of intelligence; the Wechsler memory scale (WMS) Form I (Wechsler, 1945), a standard measure of memory functions; and the reading subtest of the wide range achievement test (WRAT) (Jastak and Jastak, 1965). The HR test battery included the following subtests and outcome measures. 1. The category test an assessment of new problem solving, judgment, abstract reasoning, and concept formation (outcome measure: the total number of errors). 2. The tactual performance test an assessment of tactile perception and memory, and visuospatial performance (outcome measures: the time to completion in minutes using the dominant hand, the non-dominant hand, and both hands; the total time for the three trials; the number of shapes reproduced from memory (tactual memory); and the total number of blocks drawn in proper relationship to other blocks on memory trial (localization)). 3. The seashore rhythm test an assessment of auditory rhythmic pattern discrimination, attention, concentration, and coordination among ear, eyes, and hand (outcome measure: the number of correct responses). 4. The speech-sounds perception test an assessment of auditory verbal discrimination, attention, and concentration (outcome measure: the number of errors). 5. The finger-tapping test an assessment of motor speed and coordination (outcome measures: the average number of taps over five consecutive 10-s trials on the dominant and the non-dominant hand).

6. Grip strength an assessment of motor strength using a hand dynamometer (outcome measure: the strength in kilograms on the dominant and the nondominant hand). 7. Trail-making tests A and B an assessment of attention, perceptual ability, problem solving, motor speed, and coordination (outcome measure: the time to completion in seconds). We calculated the Halstead impairment index based on scores from the category test, the tactile performance test (total time, tactual memory, and localization), the seashore rhythm test, the speech-sounds perception test, and the nger-tapping test (dominant hand) (Jarvis and Barth, 1984). The impairment index was calculated by dividing the number of scores that were in the impaired range as established by Reitan (1955, 1959) by the number of scores that contributed to the impairment index (seven scores). Thus, the impairment index score ranged from 0 to 1.0. In addition, we examined established impairment index score categories (Jarvis and Barth, 1984). Impairment index scores between 0 and 0.2 were categorized as ``normal'', scores between 0.3 and 0.4 as ``mild'', scores between 0.5 and 0.7 as ``moderate'', and scores between 0.8 and 1.0 as ``severe.'' We analyzed the impairment index both in its continuous and discrete forms. For the WAIS-R, we examined age-adjusted scores on the information, digit span, vocabulary, arithmetic, comprehension, similarities, picture completion, picture arrangement, block design, object assembly, and digit symbol subscales. We also examined the verbal intelligence quotient (IQ), performance IQ, and fullscale IQ scores. For the WMS, we analyzed scores from the logical memory (immediate and delayed recall), visual reproduction (immediate and delayed recall), and associated learning subtests. These WMS subtests assess recall of logical verbal material, visual memory, and verbal retention, respectively. For the WRAT, we examined the raw reading score. All cognitive function tests were administered in 1982. We assessed combat exposure with a combat index designed for this study. The combat index was computed as a weighted sum of indicators of positive responses to 15 questions, with positive (yes) responses indicated by 1 or 2 and negative (no) responses indicated by 0. The questions (and weights) were: Did you receive combat pay? (1), Crash land or bail out or were you shot down? (1), Receive sniper or sapper re in or around the base? (1), Move killed or wounded personnel? (2), Serve as a forward air controller?

494

D.H. Barrett et al. / NeuroToxicology 22 (2001) 491502

(1), Fly the same aircraft when a fellow crew member was wounded or killed? (2), Fly in the same formation or the same sortie when a fellow crew member was wounded or killed? (1), Fly an aircraft that received battle damage? (1), Receive incoming artillery or rocket re at your home base or camp? (1), Encounter mines or booby traps? (1), Engage VC or NVA in a reght? (2), Did you kill VC or NVA in strang or bombing runs? (2), Were you wounded? (2), Captured by the enemy? (2), Was a close friend killed in action? (2). Each veteran was assigned to one of four strata depending on whether his sum fell into the ranges 02, 35, 68, or 9 or greater, the approximate quartiles of the distribution. We collected information on psychiatric diagnoses and psychotropic medication use during the 1982 physical examination. If the veteran reported that he had experienced mental or emotional illnesses, his written consent was obtained to access medical records in order to verify the diagnosis. We classied psychiatric diagnoses according to International Classication of Disease (ICD) codes following the ICD-9-Clinical Modication convention (Puckett, 1995). Four dichotomous (yes, no) variables for each of the following categories of psychiatric diagnoses were determined by medical record review: (1) organic psychotic conditions, including dementia, alcoholic withdrawal delirium, drug withdrawal syndrome, and acute delirium (ICD codes 290294), (2) other psychoses (ICD codes 295299), (3) neurotic personality and other non-psychotic disorders (ICD codes 306316, 300302), and (4) substance abuse (ICD codes 303305). We dened an indicator for the use of psychotropic medications (yes, no) using American Hospital Formulary Service codes (Board of the American Society of HealthSystem Pharmacists, 2000) and by reviewing medical records; a veteran was assigned the value ``yes'' for this variable if he was taking psychotherapeutic agents (formulary code 28:16), anti-manic agents (formulary code 28:28), or anxiolytics, sedatives and hypnotics (formulary code 28:24) at any time during the postservice period beginning with departure from Southeast Asia and ending with cognitive testing in 1982. We dened a drink-year as drinking one 2 oz shot of 80 proof whiskey, 12 oz of beer, or 5 oz of wine per day for 1 year. Military occupation (ofcer, enlisted yer, enlisted ground crew) was determined from military records and served as a surrogate for both education and training. Most ofcers were college educated, and most enlisted personnel were high school educated. Enlisted yers were trained for ight status while enlisted ground crew were not.

Data Analysis In the primary analysis, we contrasted mean scores on the cognitive function scales in the Ranch Hand background, low and high dioxin exposure categories with those in the comparison category, adjusting for military occupation (ofcer, enlisted yer, enlisted ground crew), age (in years at the time of the 1982 physical examination), race (black, non-black), drinking history (drink-years), marital status (married, unmarried), combat exposure quartile, the four psychiatric diagnoses indicators (yes, no), and the psychotropic medication use indicator (yes, no). In the discrete analysis of the Halstead impairment index, we limited the adjusted variables to military occupation, age, and race to reduce the number of empty cells. We assessed the signicance of contrasts on continuously distributed outcomes using differences of leastsquare means and the mean-squared error arising from linear models. We used logistic regression models to analyze the categorized impairment index. In an alternative analysis, we contrasted the second, third, fourth and fth dioxin quintile with the rst quintile with regard to the same outcome measures using the same adjustment variables. All analyses were conducted using SASTM software (SAS/STATTM, 1997). We used main effects models with no stepwise reduction throughout (Kleinbaum et al., 1982). To further examine the differences in memory functioning, we re-examined the WMS logical memory scores and the HR tactual memory score, restricting the analyses to enlisted personnel. Enlisted Ranch Hand personnel have the highest dioxin levels (Wolfe et al., 1990) and, based on a questionnaire results, their dioxin measurements correlate with reported herbicide skin exposure in Vietnam (Michalek et al., 1995). Of those who responded to the questionnaire, 517 had a measured dioxin level and participated in the 1982 physical examination. We restricted the analysis to those enlisted Ranch Hand veterans who reported at least some skin exposure (N 460) with reference to comparison enlisted personnel (N 659). RESULTS Dioxin and Demographic Characteristics The distribution of military occupation and demographics are presented in Table 2. Ranch Hand veterans in the high dioxin category were younger on average than those in the low, background, or comparison

D.H. Barrett et al. / NeuroToxicology 22 (2001) 491502 Table 2 Distribution of dioxin and demographic characteristics in 1982 by dioxin exposure category Characteristic Dioxina Median (range) Age (years) Mean (S.D.) Race Black (%) Military occupation Officer (%) Enlisted flyer (%) Enlisted ground crew (%) Comparison Ranch Hand Background 4.0 (010) 43.9 (7.6) 5.3 37.4 16.6 46.0 5.7 (010) 44.9 (7.3) 5.2 60.6 12.4 27.1 Low 53.2 (2794) 45.1 (7.5) 8.0 38.7 21.9 39.4 High

495

194.9 (943290) 40.9 (7.3) 5.1 2.9 21.5 75.6

a Measured (in parts per trillion) in 1987 or 1992 in comparison and Ranch Hand background categories, extrapolated to end of service in Vietnam in Ranch Hand low and high categories.

categories. Most of the veterans with high dioxin levels were enlisted ground crew, and most with background levels were ofcers. Cognitive Functioning We found few signicant differences in measures of cognitive functioning among veterans in the four dioxin categories (Tables 3, 4 and 5). Although not signicant, there was a decreasing trend in the mean HR tactual memory score among Ranch Hand veterans (Table 3) with those with high dioxin levels scoring the lowest. We found statistically signicant differences with comparisons on the HR nger-tapping and grip strength tasks (Table 3). Ranch Hand veterans in the low dioxin category scored lower than comparison veterans on the nger-tapping test with the both the dominant and non-dominant hands, and Ranch Hand veterans in the background dioxin category scored lower than comparison veterans on non-dominant hand grip strength. Table 4 shows the distribution of Halstead impairment index categories by dioxin categories and the results of logistic regression analyses. Ranch Hand veterans in the low dioxin category were more likely than other veterans to be categorized as severely impaired. When veterans with severe impairment ratings were compared with all other veterans, those in the low dioxin category were almost two times more likely to be rated as severely impaired. No signicant or meaningful decrements were found in any Ranch Hand dioxin category with regard to the WAIS-R or WRAT results. There was some indication that memory function may be affected among the

veterans in the high dioxin category (Table 5). Ranch Hand veterans in the high category scored signicantly lower than comparison veterans on both the immediate and delayed recall trials of the WMS logical memory test. Analyses restricted to enlisted personnel supported the ndings for memory function from the entire cohort (Table 6). We found signicant decrements when memory scores of enlisted Ranch Hand veterans who reported skin exposure were compared with scores of enlisted comparison veterans, with respect to both immediate and delayed recall WMS logical memory tasks and the HR tactual memory score. The results of the alternative analyses, using dioxin quintiles, were consistent with the results using dioxin categories. Veterans with dioxin levels in the fth quintile had the lowest mean scores on the WMS logical memory tasks (immediate and delayed recall), while none of the fth versus rst quintile contrasts with regard to the HR and WAIS subscales were signicant. The mean difference between the fth and rst quintiles (fth minus rst) was signicantly different from zero for both WMS logical memory tasks (immediate recall: mean difference 0:47, 95% CI 0:88 to 0.05, P-value 0:03; delayed recall: mean difference 0:45, 95% CI 0:84 to 0.06, P-value 0:02). Time to completion for the HR tactual performance test using both hands was greatest for veterans in the fth dioxin quintile but the contrast with the rst quintile fell short of the conventional level of statistical signicance (mean difference 0:20, 95% CI 0:01 to 0.41, P 0:06). Some contrasts between the fourth and rst and between the third and rst quintiles with regard to

496

D.H. Barrett et al. / NeuroToxicology 22 (2001) 491502

Table 3 Adjusteda HalsteadReitan neuropsychological test battery results by dioxin category Comparison Category test Mean Differenceb 95% CIc P-value Tactual performance test Localization Mean Difference 95% CI P-value Tactual memory Mean Difference 95% CI P-value Dominant hand Mean Difference 95% CI P-value Non-dominant hand Mean Difference 95% CI P-value Both hands Mean Difference 95% CI P-value Total time Mean Difference 95% CI P-value Seashore rhythm Mean Difference 95% CI P-value Speech-sounds Mean Difference 95% CI P-value Finger-tapping Dominant hand Mean Difference 95% CI P-value Ranch Hand Background 37.29 0 36.03 1.26 3.33, 0.81 0.23 Low 38.32 1.03 1.3, 3.36 0.39 High 38.06 0.77 1.66, 3.19 0.53

2.59 0

2.57 0.03 0.28, 0.23 0.84 6.21 0.05 0.13, 0.23 0.58 7.95 0.02 0.28, 0.23 0.85 6.97 0.09 0.33, 0.15 0.46 3.87 0.03 0.15, 0.2 0.76 18.77 0.1 0.66, 0.47 0.74 26.15 0.12 0.43, 0.20 0.47 7.47 0.01 0.39, 0.38 0.96

2.49 0.1 0.38, 0.19 0.50 6.05 0.12 0.31, 0.08 0.24 8.11 0.13 0.15, 0.41 0.35 7.25 0.2 0.07, 0.46 0.15 3.98 0.13 0.06, 0.33 0.19 19.32 0.46 0.17, 1.08 0.15 26.14 0.13 0.48, 0.23 0.48 7.62 0.14 0.29, 0.58 0.51

2.6 0 0.29, 0.3 0.98 5.98 0.18 0.39, 0.02 0.08 7.9 0.07 0.37, 0.22 0.64 7.27 0.22 0.06, 0.5 0.13 4.00 0.16 0.05, 0.36 0.14 19.02 0.15 0.51, 0.81 0.65 26.38 0.11 0.26, 0.48 0.56 7.12 0.35 0.81, 0.1 0.13

6.16 0

7.98 0

7.06 0

3.85 0

18.87 0

26.27 0

7.48 0

50.18 0

50.41 0.22 0.51, 0.96 0.55

49.15 1.03d 1.87, 0.19 0.02

49.9 0.28 1.15, 0.6 0.54

D.H. Barrett et al. / NeuroToxicology 22 (2001) 491502 Table 3 (Continued ) Comparison Ranch Hand Background Non-dominant hand Mean Difference 95% CI P-value Grip strength Dominant Hand Mean Difference 95% CI P-value Non-dominant hand Mean Difference 95% CI P-value Trail-making Test A Mean Difference 95% CI P-value Test B Mean Difference 95% CI P-value Halstead impairment index Mean Difference 95% CI P-value
a

497

Low 44.46 0.83 1.63, 0.04 0.04

High 45.51 0.21 0.62, 1.04 0.61

45.3 0

45.72 0.42 0.29, 1.12 0.25

51.17 0

50.38 0.79 1.77, 0.2 0.12 47.14 1.04 1.96, 0.11 0.03

51.32 0.15 0.95, 1.26 0.79 48.21 0.03 1, 1.06 0.95

51.1 0.07 1.23, 1.09 0.91 47.99 0.19 1.26, 0.88 0.73

48.18 0

26.05 0

26.55 0.5 0.44, 1.45 0.30 66.6 0.27 2.98, 2.44 0.85 0.40 0.01 0.02, 0.03 0.58

25.71 0.34 1.4, 0.72 0.53 67.46 0.59 2.42, 3.6 0.70 0.41 0.02 0.01, 0.05 0.22

25.63 0.42 1.52, 0.6 0.45 65.78 1.08 4.21, 2.04 0.50 0.38 0.01 0.04, 0.02 0.58

66.87 0

0.39 0

Adjusted for military occupation, age, race, drink-years, marital status, combat exposure, organic psychotic conditions, other psychoses, neurotic personality and other non-psychotic disorders, substance abuse, and use of psychotropic medications. b Difference of means. c Confidence interval. d Values in bold indicate statistical significance.

HR subscale means were signicant. Among veterans in the fourth quintile the mean grip strength of the dominant hand and among those in the third and fourth quintiles the mean grip strength in the non-dominant hand was signicantly increased. All other contrasts with the rst quintile were not signicant. The mean WAIS information score was signicantly decreased among veterans in the third quintile and the mean similarity score was signicantly increased among those in the fourth quintile; all other contrasts with the rst quintile were not signicant.

DISCUSSION The results of this study suggest that although there are not global decrements in cognitive functioning associated with Agent Orange exposure among Vietnam veterans, there may be specic decrements involving verbal and tactual memory functioning. Although our ndings of decreased memory functioning among Ranch Hand veterans with the highest dioxin exposure are somewhat small and of uncertain clinical signicance, they suggest that there is merit to continued

498

D.H. Barrett et al. / NeuroToxicology 22 (2001) 491502

Table 4 Adjusteda Halstead impairment index category by dioxin category Comparison Ranch Hand Background Normal Count (%) Mildly impaired Count (%) Moderately impaired Count (%) Severely impaired Count (%) Severe versus normal, mild, moderate Odds ratioa 95% CIb P-value Severe versus normal Odds Ratio 95% CI P-value
a

Low 92 (33.6) 118 (43.1) 48 (17.5) 16 (5.8) 1.9c 1.0, 3.5 0.05 1.8 0.8, 3.7 0.14

High 83 (30.2) 125 (45.5) 58 (21.1) 9 (3.3) 0.8 0.4, 1.8 0.62 0.7 0.3, 1.6 0.42

354 (33.65) 495 (47.1) 169 (16.1) 34 (3.2) Referent

137 (35.3) 193 (49.7) 53 (13.7) 5 (1.3) 0.5 0.2, 1.4 0.20 0.6 0.2, 1.8 0.40

Referent

Adjusted for military occupation, age, and race. Confidence interval. c Values in bold indicate statistical significance.
b

study of the long-term inuence of Agent Orange exposure on the cognitive functioning of these veterans, especially focusing on memory functions. Several studies have attempted to examine neuropsychological functioning and Agent Orange exposure among Vietnam veterans; however some of these studies have been hampered by small sample sizes, imprecise or subjective assessment of exposure, assessment of cognitive abilities long after potential exposure, or inadequate control of baseline functioning. Levy (1988) compared neuropsychological tests results of 6 Vietnam veterans who had current mild chloracne, a skin lesion associated with high exposure to dioxin, with results from a control group of 25 Vietnam veterans without chloracne who were matched for age, education, period of Vietnam service, and combat experience. In contrast to our ndings of limited cognitive decits, Levy (1988) found evidence to suggest that exposure to presumed high doses of dioxin are associated with a variety of decits in cognitive functioning. The exposed groups scored signicantly lower on measures of auditory memory, attention, and verbal uency. The exposed group also scored signicantly lower on the WAIS vocabulary subtest. Fiedler and Gochfeld (1992) examined the association between neurobehavioral test results and dioxin exposure in order to determine whether a marker of

exposure could be identied. In the Pointman I Project, Vietnam veterans who directly handled herbicides and, in the Pointman II Project, Navy, Marine, and Army combat ground troops were administered neuropsychological tests. As in our study, only limited differences in cognitive functioning were found. This included signicant differences on two measures of attention and immediate memory, the digit-span subscale of the WAIS-R and trail-making test B, with veterans who worked with herbicides showing poorer performances than those with no known herbicide exposure. Inconsistent results were found when ground combat troops were compared with control veterans. As part of a larger study of US Army Vietnam veterans and non-deployed peers, the Centers for Disease Control and Prevention examined the relation between neuropsychological functioning and military service in Vietnam (Centers for Disease Control, 1988a). This study did not directly assess the association between Agent Orange exposure and cognitive functioning. In fact, pilot testing suggested that the two cohorts did not differ signicantly on serum dioxin levels (Centers for Disease Control, 1988b). This study found small but statistically signicant differences between deployed and non-deployed veterans on tests of visual memory, visuospatial abilities, and abstract thinking. In the same study, 41 of 2490 of Vietnam

D.H. Barrett et al. / NeuroToxicology 22 (2001) 491502 Table 5 Adjusteda Wechsler memory scale results by dioxin category Comparison Ranch Hand Background Logical memory Immediate recall Mean Differenceb 95% CId P-value Delayed recall Mean Difference 95% CI P-value Visual reproduction Immediate recall Mean Difference 95% CI P-value Delayed recall Mean Difference 95% CI P-value Associate learning Mean Difference 95% CI P-value Low High

499

7.45 0

7.28 0.17 0.53, 0.18 0.33 4.97 0.03 0.31, 0.36 0.88

7.23 0.23 0.63, 0.17 0.27 4.63 0.32 0.69, 0.06 0.10

6.95 0.5c 0.91, 0.09 0.02 4.53 0.42 0.8, 0.03 0.04

4.95 0

9.62 0

9.74 0.11 0.18, 0.41 0.45 8.51 0.17 0.17, 0.52 0.32 16.08 0.03 0.39, 0.33 0.86

9.73 0.1 0.23, 0.43 0.54 8.37 0.04 0.35, 0.43 0.84 15.95 0.16 0.57, 0.25 0.44

9.59 0.03 0.37, 0.31 0.86 8.26 0.07 0.47, 0.3 0.73 16.08 0.02 0.44, 0.39 0.91

8.33 0

16.11 0

a Adjusted for military occupation, age, race, drink-years, marital status, combat exposure, organic psychotic conditions, other psychoses, neurotic personality and other non-psychotic disorders, substance abuse, and use of psychotropic medications. b Confidence interval. c Difference of means. d Values in bold indicate statistical significance.

veterans (1.6%) and 21 of 1972 non-Vietnam veterans (1.1%) reported problems with concentration or memory in a telephone interview (Centers for Disease Control, 1989). Contrary to the above mentioned studies and the ndings of our study, others have found no support for cognitive decits. Korgeski and Leon (1983) administered tests of immediate and delayed verbal and spatial memory, problem-solving ability, and learning ability to 100 Vietnam veterans whose herbicide exposure was estimated both by self-report and by using military records giving the location of each veteran's duty station in Vietnam. By comparing the military record information with computer records of herbicide spraying missions, researchers calculated a numerical estimate of the likelihood of herbicide exposure. No differences in neuropsychological functioning were

found regardless of whether the veterans were grouped according to self-ratings or records-based ratings of herbicide exposure. Studies of occupational and environmental exposures to high levels of dioxins have provided inconsistent results. Klawans (1987) assessed cognitive functioning in 45 railroad workers exposed to dioxin during cleanup of a chemical spill of polychlorinated phenols. Two years after exposure, 49% of the exposed workers were found to exhibit decreased attention and concentration. Stehr-Green et al. (1987) assessed cognitive functioning among residents of eastern Missouri who lived in areas sprayed with dioxin-contaminated sludge waste for dust-control purposes. In contrast to the ndings of Klawans, the only difference in cognitive functioning found between 154 residents who had lived in the contaminated area for 6 months or longer

500

D.H. Barrett et al. / NeuroToxicology 22 (2001) 491502

Table 6 Adjusteda memory function results among enlisted US Air Force veterans who reported Agent Orange skin exposure and enlisted comparison veterans Wechsler memory scale Logical memory Immediate recall Mean Differenceb 95% CId P-value Delayed recall Mean Difference 95% CI P-value Enlisted comparison Enlisted Ranch Hand (N 659) who reported skin exposure (N 460)

6.66

6.15 0.51c 0.87, 0.16 <0.001 3.65 0.40 0.73, 0.08 0.02 6.24 0.18 0.36, 0.00 0.05

4.05

HalsteadReitan tactual memory Mean 6.42 Difference 95% CI P-value

a Adjusted for age, military occupation, race, drink years, marital status, substance abuse, combat exposure, organic psychotic conditions, other psychoses, neurotic personality, other non-psychotic disorders, and use of psychotropic medications. b Mean difference. c Values in bold indicate statistical significance. d Confidence interval.

and 155 unexposed people who lived in a comparable area without dioxin contamination was on the vocabulary subtest of the WAIS, a measure of baseline verbal intelligence (the direction of the difference was not reported). A review by the Institute of Medicine (IOM), National Academy of Sciences, of the strength of evidence for human health effects among veterans and workers exposed to herbicides used in Vietnam concluded that there was inadequate or insufcient evidence to determine whether an association existed between cognitive and neuropsychiatric effects and exposure to herbicides used in Vietnam (Institute of Medicine, 1994; Goetz et al., 1994). The IOM committee also concluded that studies in laboratory animals do not support the biological plausibility of a relationship between herbicide exposure and cognitive or neuropsychiatric disorders. Our study suggests that dioxin exposure is not associated with global decrements in cognitive abilities. Overall, we found few signicant differences between Ranch Hand and comparison veterans in our assessment of a wide range of cognitive skills.

However, there were a few signicant differences that should be further explored. Our assessment of the Halstead impairment index, a crude method for categorizing brain damage, indicated that Ranch Hand veterans with low dioxin exposure were more likely than comparison veterans to be rated as severely impaired. However, the number of veterans categorized as severely impaired was small (N 64) and the average impairment index score for Ranch Hand veterans was not signicantly increased, regardless of their exposure category. We found consistent and signicant decrements in Ranch Hand veterans with the highest dioxin levels relative to comparison veterans on tasks involving verbal memory functioning. We found the same decrements in Ranch Hand enlisted ground personnel with reported skin exposure to herbicides relative to enlisted comparison veterans. Although the differences were statistically signicant, they were relatively small, and additional research is needed to determine the robustness of these ndings and to assess their clinical signicance. This research should attempt to obtain a better understanding of the biological basis for a relation between dioxin exposure and specic memory functions. Finally, we found signicant differences with regard to measures of motor functioning. In contrast to the comparisons, Ranch Hand veterans with low dioxin levels showed slower motor speed (measured by the nger-tapping test on both the dominant and the nondominant hand) and those with background dioxin levels showed decreased motor strength (measured by the grip strength test with the non-dominant hand only). These differences were small, of uncertain clinical signicance, and difcult to interpret because a clear dose response was not demonstrated and because Ranch Hand veterans with the highest dioxin levels did not differ from the comparisons on these measures of motor functioning. In its review of the literature, the IOM concluded that there was inadequate or insufcient evidence to determine whether an association existed between dioxin and motor and coordination problems (Institute of Medicine, 1994). However, the IOM committee recommended continued follow-up of motor and coordination outcomes among exposed persons. Although our ndings regarding motor functioning were most likely due to chance, we recommend that motor functioning be measured in future studies of the Ranch Hand cohort to further examine this issue. Alternative analyses contrasting veterans in the second, third, fourth, and fth dioxin quintiles with those in the rst quintile of the combined cohort produced

D.H. Barrett et al. / NeuroToxicology 22 (2001) 491502

501

results consistent with those of the dioxin category contrasts. Veterans in the fth quintile of the dioxin distribution scored signicantly worse than veterans in the rst dioxin quintile on tests of verbal memory (immediate and delayed recall scores of the WMS logical memory task). Our ability to detect associations was limited by the xed size of the Ranch Hand cohort. Since all Ranch Hand veterans have been identied and invited to participate in the study, their number could not be increased. The rarity of some of the abnormalities we studied led to imprecise measures of association indicated by wide condence intervals; small numbers prevented us from making strong inferences regarding the most heavily exposed veterans. Confounding was another concern. Although, we adjusted for all known confounders, there is the possibility that there were others that we did not take into account. Also, of concern was the number of statistical tests we performed. We found nine statistically signicant differences (P 0:05) out of the 120 tests we performed. Although, we cannot rule out that these signicant ndings reect chance differences resulting from the many analyses we conducted, the apparent clustering of certain memory decits in Ranch Hand veterans with the highest dioxin levels suggests that these ndings may not be due to chance. This study was limited by uncertainties regarding dioxin exposure. The serum dioxin measurements were accurate (Michalek et al., 1996b) and correlate with reported skin exposure to herbicide in Vietnam (Michalek et al., 1995), but were made up to 30 years after exposure. Our initial dose calculation was based on a rst-order decay law with an assumed constant halflife. The accuracy of our extrapolated initial dose estimate is unknown. Further, we regard the background category as a mixture of exposed and unexposed veterans whose true status could not be determined with available data. The time, since exposure varied between 15 and 26 years, or 2 or 3 dioxin half-lives. The elimination of dioxin in the intervening years and lack of alternative evidence of exposure left the exposure status of Ranch Hand veterans with background levels (below 10 ppt) unresolvable. The ongoing nature of the Air Force Health Study provides an excellent opportunity to further examine the association between dioxin exposure and memory function. Future studies of this cohort should include a more rened assessment of memory abilities. These additional studies should more fully examine possible confounding associated with psychiatric conditions that interfere with memory functioning and are known

to be prevalent among Vietnam veterans (such as posttraumatic stress disorder, anxiety disorders, major depression, and substance abuse) (Barrett et al., 1996). Further studies are needed to examine the possible interaction between dioxin exposure in early adulthood and normal age-related changes in the central nervous system. The National Academy of Sciences has recommended a prospective study of people with documented high herbicide exposure after they reach age 60 years (Institute of Medicine, 1994). The Air Force Health Study provides a unique opportunity to examine this question. REFERENCES
Barrett DH, Green ML, Morris R, Giles WH, Croft JB. Cognitive functioning and post-traumatic stress disorder. Am J Psychiat 1996;153:14924. Board of the American Society of Health-System Pharmacists. AHFS drug information. Bethesda, MD: American Hospital Formulary Service, 2000. Centers for Disease Control. Vietnam experience study. Health status of Vietnam veterans: II. Physical health. JAMA 1988a;259:270814. Centers for Disease Control. Serum 2,3,7,8-tetrachlorodibenzo-pdioxin levels in US Army Vietnam-era veterans. JAMA 1988b;260:124954. Centers for Disease Control. Health Status of Vietnam Veterans, vol. IV. Atlanta: US Department of Health and Human Services, Public Health Service, 1989. p. 37. Fiedler N, Gochfeld M. Neurobehavioral correlates of herbicide exposure in Vietnam veterans: a report on the Pointman Projects to the New Jersey Agent Orange Commission, 1992. Goetz CG, Bolla KI, Rogers SM. Neurologic health outcomes and Agent Orange: Institute of Medicine report. Neurology 1994;44:8019. Grubbs WD, Lustik MB, Brockman AS, Henderson SC, Burnett FR, Land RG, Osborne DJ, Rocconi VK, Schrieber ME, Williams DE. Air Force Health Study: An epidemiologic investigation of health effects in Air Force Personnel following exposure to herbicides. NTIS AD A-304-306 through AD A304-316. Springfield: National Technical Information Service, 1996. p. 214. Institute of Medicine. Veterans and Agent Orange. Washington: National Academy Press, 1994. Jarvis PE, Barth JT. HalsteadReitan test battery: An interpretive guide. Odessa: Psychological Assessment Resources Inc., 1984. p. 225. Jastak JF, Jastak SR. The wide range achievement test. Wilmington: Guidance Associates, 1965. Klawans HL. Dystonia and tremor following exposure to 2,3,7,8tetrachlorodibenzo-p-dioxin. Mov Disord 1987;2:25561. Kleinbaum DG, Kupper LL, Morgenstern H. Epidemiologic research. Principles and quantitative Methods. New York: Van Nostrand, 1982. p. 455. Korgeski GP, Leon GR. Correlates of self-reported and objectively determined exposure to Agent Orange. Am J Psychiat 1983;140:14439.

502

D.H. Barrett et al. / NeuroToxicology 22 (2001) 491502 Puckett CD, editor. The educational annotation of ICD-9-CM. Reno: Channel Publishing Ltd., 1995. Reitan RM. Investigation of the validity of Halstead's measure of biological intelligence. Arch Neurol Psychiat 1955;73:2835. Reitan RM. The effects of brain lesions on adaptive abilities in human beings. Seattle: University of Washington, 1959. SAS/STATTM Software. Changes and enhancements through release 6.12. Cary, NC: SAS Institute Inc., 1997. Stehr-Green P, Hoffman R, Webb K, Evans RG, Knutsen A, Schramm W, et al. Health effects of long-term exposure to 2,3,7,8-tetrachlorodibenzo-p-dioxin. Chemosphere 1987; 16:208994. Wechsler D. WAIS-R manual. New York, Psychological Corporation, 1981. Wechsler D. A standardized memory scale for clinical use. J Psychol 1945;19:8795. Wolfe WH, Michalek JE, Miner JC, Rahe A, Silva J, Thomas WF, et al. Health status of Air Force veterans occupationally exposed to herbicides in Vietnam: I. Physical health. JAMA 1990;264:182431.

Levy CJ. Agent Orange exposure and post-traumatic stress disorder. J Nerv Ment Dis 1988;176:2425. Michalek JE, Pirkle JL, Caudill SP, Tripathi RC, Patterson Jr, DG, Needham LL. Pharmacokinetics of TCDD in veterans of Operation Ranch Hand: 10 year follow-up. J Toxicol Environ Epidemiol 1996a;47:20920. Michalek JE, Tripathi RC, Kulkarni PM, Pirkle JL. The reliability of the serum dioxin measurement in veterans of Operation Ranch Hand. J Expo Anal Environ Epidemiol 1996b;6:32738. Michalek JE, Rahe AJ, Kulkarni PM, Tripathi RC. Levels of 2,3,7,8-tetrachlorodibenzo-p-dioxin in 1302 unexposed Air Force Vietnam-era veterans. J Expo Anal Environ Epidemiol 1998;8:5964. Michalek JE, Wolfe WH, Miner JC, Papa TM, Pirkle JL. Indices of TCDD exposure and TCDD body burden in veterans of Operation Ranch Hand. J Expo Anal Environ Epidemiol 1995;5:20923. Peper M, Klett M, Frentzel-Beyme R, Heller W. Neuropsychological effects of chronic exposure to environmental dioxins and furans. Environ Res 1993;60:12435.