Epidemiology in Military and Veteran Populations: Proceedings of the Second Biennial Conference, March 7, 1990 (1991)

Brian E. Engdahl, William F. Page ^*

Studies of former prisoners of war (POWs) provide valuable insights into post-traumatic adaptation because they gather information from a large population that survived the traumatic experiences of military captivity. Previous studies of POWs have shown elevated rates of psychiatric symptoms and disorders. This report presents evidence from a longitudinal study of three large, representative national samples of former POWs. The study finds that depressive symptomatology, as measured by the Center for Epidemiologic Studies depression scale, is elevated in World War II POWs from the Pacific and European theaters and in Korean conflict POWs. Decades later, depressive symptomatology is found to be strongly associated with prior treatment in captivity. Differences in depressive symptomatology among the three POW groups can be attributed to captivity-related factors and to buffering factors such as age at capture and education.

Studies of former prisoners of war, a large population that survived the traumatic experiences of military captivity, are important in their own right, have relevance to survivors of other captivity maltreatment (Engdahl and Eberly, 1990), and provide insights into the phenomenology of general

post-traumatic adaptation. Although a few studies find little long-term negative effect and even psychological growth among POWs (Ursano, 1981; Sledge, Boydstun, and Rabe, 1980), most studies report elevated incidence of psychiatric symptoms and disorders. Cohen and Cooper (1954) found a four- to five-fold excess of hospitalizations for psychoneurosis but not psychosis among World War II Pacific theater POWs and European theater POWs compared with their controls. Beebe (1975) extended this follow-up study and found significantly more hospitalizations of POWs for a variety of psychiatric illnesses, including schizophrenic disorders, anxiety reactions without somatization, alcoholism, “nervousness and debility,” and other psychoneurotic reactions. Psychoneuroses (particularly anxiety reactions and somatization) and psychoses (e.g., schizophrenia) were especially frequent among Pacific World War II and Korean conflict POWs when compared with their non-POW controls. Kluznik et al. (1986) retrospectively diagnosed psychiatric disorders among 188 World War II and Korean conflict POWs. Within one year of their release, 67% fulfilled DSM-III criteria for post-traumatic stress disorder, and more than half of those continued to have symptoms over 40 years later. Generalized anxiety disorders and depressive disorders also were frequent.

Psychiatric symptoms, particularly depressive symptoms, were elevated among Australian World War II POWs in a recent series of studies (Tennant et al., 1986a,b; Dent et al., 1987). Examination of 170 POWs and comparable controls revealed that POWs were significantly more depressed than non-POW controls some 40 years after repatriation. No differences were found in state anxiety, trait anxiety, neuroticism, psychoticism, and hostility. To explain the persistence of depressive symptoms for 40 years after release from captivity, the investigators linked these findings, suggesting that over the long follow-up period anxiety might diminish but depression might increase as a reaction to chronic post-traumatic impairment. Dent et al. (1987) reported separate regression analyses for the Australian POWs and controls showing the following variables to be predictive of present-day depressive symptoms: experiencing a nervous illness during World War II or a depressive illness after World War II, having a lower level of education or socioeconomic status, and being unmarried, unemployed, or retired.

The present study collected data on the prevalence, severity, and correlates of depressive symptoms in a large national representative sample of American POWs. Other recent reports of POWs' adaptation have been based on smaller samples drawn from a single region of the country (such as Goldstein et al., 1987, at the Pittsburgh Veterans Administration Medical Center (VAMC); and Speed et al., 1989, at the Minneapolis VAMC). The size of the samples in the present study enables more detailed analyses within the three POW eras (POWs of the Pacific and

European theaters, plus POWs of the Korean conflict) as well as comparisons among these eras. Also, the representative nature of the national samples strengthens generalizations made from them to the population of all POWs. We present descriptive data first, followed by data on the correlations between individual predictor variables and depressive symptoms. Finally, we use multiple regression to examine the combined effects of predictors on depressive symptoms.

The Medical Follow-up Agency of the Institute of Medicine, National Academy of Sciences, has studied the health of former POWs since the early 1950s. Cohen and Cooper (1955) assembled representative rosters of former World War II POWs and non-POW controls and characterized their mortality, morbidity, and disability after liberation. Nefzger (1970) added Korean conflict prisoners and controls and studied their mortality. Beebe (1975) conducted a 20-year morbidity follow-up, collecting data from military and Veterans Administration (now Department of Veterans Affairs) records and from questionnaires. Keehn (1980) continued the mortality follow-up through 1976. We report data collected from a questionnaire follow-up of these earlier-defined cohorts.

Cohen and Cooper (1955) used the Army's official roster of all known World War II POWs to select random, independent samples of white Army servicemen who were captured in the Pacific and European theaters. Nefzger (1970) subsequently doubled the number of Pacific prisoners and controls in the study cohort and added a group of Korean conflict POWs and nonprisoner controls.

The follow-up reported here began with a review of VA mortality records, which ascertained that 1,319 men from the earlier studies were alive as of mid-1984 and thus eligible for the questionnaire study. Addresses were obtained from the Internal Revenue Service (under an arrangement with the National Institute for Occupational Safety and Health) and from two commercial tracing firms. IRS provided addresses for roughly 90% of the total sample, and tracing firms provided addresses for an additional 3% of subjects. Next, up to three mailings per person were made to each address, with a mailgram preceding the third mailing to alert the addressee to the forthcoming questionnaire. The mailings contained a cover letter from the study director at the time, Robert Keehn, and the questionnaire. Also included in the third mailing was a letter from the National Commander of the American Ex-POWs encouraging participation in the study. The actual mailing of questionnaires began late in 1984, and replies were accepted through December 1985. The

questionnaire included the 20-item Center for Epidemiologic Studies-Depression scale (CES-D; Radloff, 1977), the Lie and Hysteria scales of the Minnesota Multiphasic Personality Inventory (MMPI; Hathaway and McKinley, 1951), questions on smoking and drinking, history of hospitalizations since 1965, medical conditions under treatment, and medical conditions not being treated.

The CES-D, a standardized self-administered rating instrument (see Table 1 ), was chosen as the measure of depressive symptoms, which are known to be elevated among POWs (Tennant et al., 1986a). The CES-D is well suited to estimate the severity and prevalence of depressive symptoms among the general population of POWs, as it is widely used in epidemiologic studies of the prevalence and correlates of depression in non-clinical populations (e.g., Murrell et al., 1983; Weissman et al., 1977). The use of the CES-D allowed a large number of subjects to be surveyed and permitted comparisons with previous CES-D-based general and special population surveys. Although CES-D scores correlate modestly with clinical diagnoses of depression (Myers and Weissman, 1980; Boyd et al., 1982), the CES-D does not yield a diagnosis. Its primary utility is in the estimation of symptom prevalence and in clinical or research efforts as a first-stage screening test. Breslau (1985) and Roberts, Vernon, and Rhoades (1989) conclude that the CES-D detects generalized anxiety about as well as it detects major depression. Roberts et al. (1989) hypothesize that its content indicates a single dimension that they label “demoralization.” Alternatively, Golding and Aneshensel (1989), in reviewing the CES-D's factor structure, conclude that four factors best represent its content: Negative Affect, Positive Affect, Somatic Symptoms, and Interpersonal Problems. We view the CES-D as a measure of general psychological impairment, primarily indexing depression. Each of the 20 CES-D items was assigned a score of 0 to 3, and all the responses were summed. Scores may range from 0 to 60; the standard indication of significant depressive symptoms is a score of 16 or above. Missing item responses were assigned a weight of 0, and 5 or more missing responses on a single questionnaire resulted in the assignment of a missing total score for that person.

The MMPI Lie scale, which detects tendencies to place oneself in an unusually favorable light, was included to check for response bias. Concern that POWs might misreport their psychological symptoms led to inclusion of the MMPI Hysteria scale which focuses on symptoms and includes both over-reporting (Admit) and under-reporting (Deny) subscales (Little and Fisher, 1958).

For most subjects in the three POW groups studied here, earlier data collected by Beebe were available from medical and personnel records as well as from the 1965 questionnaire on medical problems suffered during captivity. Table 2 outlines these captivity medical problems and symptoms collected via self-report from earlier surveys. A total captivity symptom index score was calculated by adding the number of “yes” responses for each condition on the symptom list. Percent of body weight lost was calculated as the difference between self-reported weight at induction and self-reported lowest weight during captivity, divided by self-reported weight at induction.

The total number of responses received from the POWs was 989. After excluding deceased subjects discovered by the survey, the final response rates were 74.5% for the Pacific, 75.3% for the European, and 68.8% for the Korean POWs.

Pacific and Korean POWs reported more captivity symptoms and greater weight loss than European POWs. We note that these self-reported data were collected some 20 years ago, and the accuracy of recall should be greater than that for data collected in the late 1980s. These data probably also are less susceptible to any recall biases introduced by age-associated declines in health. When exposed to comparable trauma, individuals who currently are more ill tend to recall (and report) greater severity of stress than those who are less ill.

Table 3 reveals notable demographic differences among the three POW groups. Relative to the Korean group, both World War II groups were older and included a higher proportion of high school graduates and college-educated men; further, they only include whites. There also was a lower proportion of infantry in both World War II groups with a correspondingly higher proportion of Army Air Corps personnel and a higher proportion of officers, warrant officers, and sergeants in the Pacific POW group.

In addition to these differences in age, rank, and education, there were differences between the two World War II groups. On average the Pacific POWs were slightly older than European POWs at capture, had higher rank and education, and had a lower percentage of draftees. These differences primarily reflect underlying differences in military theaters and eras. As detailed in another report (Page, 1988), differences between

questionnaire respondents and nonrespondents were small but uniform, and limited primarily to rank and education; nonrespondents tended to have lower rank and less education.

Using standard conversion formulas for the MMPI, we converted raw scores to standardized T scores with means of 50 and standard deviations of 10. The MMPI Lie scale and the Deny subscale detected very little response bias, with mean T scores close to the expected value of 50. For both scales across the three groups, mean scores ranged from 48.1 to 52.6. For the three groups, Hysteria scale scores and its Admit subscale scores were elevated, ranging from 66.4 to 74.7. The European POW group means were lower than the Pacific and Korean groups, as discussed in more detail below.

Community studies find the proportion of older males with high CES-D scores to range from 3% to 18%, with an average of roughly 10% (Comstock and Helsing, 1976; Frerichs, Aneshensel, and Clark, 1981; Murrell, Himmelfarb, and Wright, 1983; Eaton and Kessler, 1981). The proportions of CES-D scores at or above the cut-off point of 16 ranged from 54% in the Korean group and 50% for the Pacific group to 37% for the European group. The Pacific POW proportion above the cut-off was greater than the European proportion (chi-square = 12.0, df = 1, p = 00068), and the Korean proportion was greater than the European proportion (chi-square = 20.5, df = 1, p = 00002). The ranking of mean CES-D scores was identical to the ranking of proportions above the cutoff, and the means for the Pacific group (17.9) and the Korean group (18.8) were both above the cut-off. The European group mean CES-D score (13.4), which is significantly lower than those of the Pacific and Korean groups, is low only in comparison to them. Compared with figures for the general population (where normal groups average 5 to 8; National Center for Health Statistics, 1980), the European group mean CES-D rate is markedly high. Psychiatric populations have shown mean scores ranging from 37 for male acute depressives to 19 for male recovered depressives (Weissman et al., 1977).

To understand further this basic finding of elevated depressive symptoms among POWs, we conducted additional analyses of the depressive symptoms ' relationships to captivity intensity and individual factors. This approach assumes both the existence of a dose-response

relationship between severity of trauma and the post-traumatic depressive symptoms, plus the presence of individual factors that may moderate the dose-response relationship.

In Table 4 , the proportions of CES-D scores at or above the cut-off point are shown for the various demographic characteristics described earlier. Also shown are the results of chi-square tests comparing the proportions of men with elevated CES-D scores across the various levels of demographic characteristics, by POW group. The differences in the proportion of elevated CES-D scores among categories for race, age, and type of service (e.g., infantry) were somewhat smaller than the corresponding differences for rank/component of service, marital status at entry into service, years of education, and rank at separation. To illustrate, the marked decrease in the proportion of high CES-D scores as the number of years of education increases is evidence of a strong association. Similarly, rank at separation has a strong, graded association with the proportion of high CES-D scores. The consistent CES-D differences by education and rank suggest that a more general (though unmeasured) underlying factor, such as socioeconomic status, may moderate the prevalence and severity of depressive symptoms.

Self-reported number of drinks per day showed an association with CES-D score, in which both non-drinkers and heavy drinkers reported a higher prevalence of depressive symptoms than moderate drinkers. The non-linear nature of this relationship (a U-shaped distribution) argues against the inclusion of self-reported alcohol consumption in the current linear regression models; future analyses will focus on this variable.

Other crosstabulation analyses (not shown) revealed roughly the same proportion of men with high CES-D scores across the three POW groups when comparing men who reported the same range of weight loss or captivity symptom score. This suggests that more severe treatment, as reflected both by greater weight loss and greater number of captivity symptoms, is linked to a higher level of subsequent depressive symptoms, and that differences in severity of treatment at least partially will explain the differences in those symptoms across these POW groups. This observation, like the earlier observations, may be subject to confounding effects--therefore the need for the following multivariate analyses.

We estimated the joint effect of multiple potential causal factors on CES-D scores through multiple regression analyses, treating the CES-D

total score as a continuous dependent variable. The rationale for variable selection was based on previous research, particularly the studies of Australian POWs. We examined the roles of several demographic and captivity-related factors: age at time of capture, marital status at entry into active duty, years of education, weight loss in captivity, captivity symptom score, and, for Korean POWs only, Army General Classification Test (AGCT) score, rank at capture, and race. We treated all but marital status, rank at capture, and race as continuous measures, with rank at capture assigned three ordinal categories (highest to lowest): commissioned officers, sergeants and corporals, and privates. We divided marital status into two categories: married and unmarried (single, separated, or divorced); and coded race into two categories: white and nonwhite.

Rank at capture and AGCT score data had too many missing observations for the Pacific and European POWs, and race was white only in these two groups. Thus rank at capture, AGCT score, and race were included only in a preliminary regression for the Korean POWs which showed that none of these factors was significant; therefore they were omitted from further analyses. Similarly, weeks of captivity had too many missing observations to be included in any of the regressions. We note, however, that weeks of captivity varied little within a POW group and thus would have explained little of the within-group variance in CES-D scores. Notably high among variable intercorrelations for the combined samples were rank at capture with years of education (r = 436) and weight loss with captivity symptom score (r = 457).

The final models for all three groups are statistically significant, suggesting that both captivity severity factors and demographic factors are directly predictive of later depressive symptoms. As Table 5 shows, the strongest predictors of subsequent depressive symptoms are the captivity symptom score and the number of years of education; age at capture is a significant predictor in both Pacific and Korean groups, and percent of body weight lost is a significant predictor only for Pacific POWs. As might be expected, subsequent depressive symptoms are positively associated with more captivity symptoms and greater weight loss, and inversely related to years of education, age at capture, and marital status. That is to say, POWs who had more education or were married at entry into active duty, or were older at capture, had less subsequent depression.

A regression equation based on the combined Pacific and Korean samples is presented in the last column of Table 5, including a dummy variable for POW status (in effect, Pacific POW status = 1 and Korean POW status = 2). The combined regression shows that both captivity factors and all demographic factors but one, marital status, are significant predictors of subsequent depressive symptoms. The strongest captivity-related predictive factor is captivity symptom score, and the strongest

demographic predictive factors are number of years of education and age at capture, each with roughly equal importance. It is interesting that the dummy variable for POW status has a small and insignificant predictive weight. All other things being equal, predicted Pacific and Korean POWs' depressive symptoms are essentially the same; the captivity and demographic factors shown here account statistically for the differences in subsequent depressive symptoms.

The European group was not combined with the others for several reasons. Because European POWs' captivity treatment was less harsh than that of Pacific or Korean POWs, it may be related differently to subsequent depressive symptoms. Also, because all three nonsignificant terms in the European group regression had opposite signs from those in the other groups, the regression estimates for the European group may be unstable, perhaps in part because this group is the smallest of the three.

Although weight loss is viewed as a strong indicator of harsh treatment (Beebe, 1975) and as a robust predictor of psychiatric sequelae (Speed et al., 1989), it is a significant predictive factor only for the Pacific group. In the European group it is possible that the lack of variability in weight loss accounted for its lack of statistical significance, but in the Korean group this explanation is not applicable. Overall, the captivity symptom score appears to be the better predictor of depressive symptoms.

The data in this report differ from those in other studies in two important ways. First, they are drawn from the largest and most representative longitudinal samples of American POWs available. The size of the samples allows the kind of multivariate analyses presented here, and their representativeness ensures that generalizations to the population of all POWs will be sound. The second feature of the data, the inclusion of independent, nationwide samples from three different war theaters, allows separate group analyses and intergroup comparisons and contrasts.

The results from the MMPI Lie and Deny scales reflect no tendency to deny symptoms in these samples, but the elevated Hysteria and Admit scales raise the possibility of over-reporting of current symptoms. Keeping in mind that the Admit items are a subset of the Hysteria items, it is logical that Little and Fisher (1958) noted their high intercorrelation and recommend viewing them as interchangeable. In the present sample their correlation was 77. Also, systematic increases in Hysteria scores over time have been reported in normal samples (Leon et al. 1979). These increases

point to the presence of physical complaints and bodily concerns that might be quite realistic among older men, lowered energy levels, and depressive symptoms. Hysteria scores frequently are elevated in valid POW MMPI profiles (Klonoff et al., 1976; Goldstein et al., 1987). Although item-level analyses are required to resolve this question, we believe the elevated Hysteria and Admit scores reflect normal somatic symptoms and concerns plus depressive symptoms, not a response bias of symptom over-reporting.

Because internal consistency is an important component of validity, we again note that the high rates of depressive symptoms were associated with treatment in captivity, and the predicted effects of the majority of the buffering demographic variables were in the expected directions. When similar factors operate in similar ways in differing groups, we have evidence for the validity of results. In addition, unpublished comparisons with a control group of non-POW Korean combat veterans showed that they have significantly lower rates of depressive symptoms than Korean POWs. Finally, we note that Beebe's earlier regression analysis found a statistically significant relation between captivity factors and total score on the Cornell Medical Index (Beebe, 1975), producing results similar to the regression results presented here.

External consistency with other studies provides additional evidence of validity. Most important is the consistency of our findings with independent evidence from the studies of World War II Pacific theater Australian POWs (Tennant et al., 1986a,b) mentioned earlier. These independent findings not only parallel the results in this report but, significantly, are based on data collected using a different psychological instrument, the Zung depression scale (1965). Because the Australian studies measured different variables by different means, their significant regression analyses variables could not be duplicated by those found in this study. Nevertheless, there are clear parallels between the two sets of significant variables (education in both models and the related measure of socioeconomic status in the Australian model) and more tenuous parallels (nervous illness during World War II in the Australian model and captivity symptoms in the American). In all, the kinds of factors associated with the depressive symptoms found among American POWs some 40 years after capture are quite similar to those found among Australian POWs.

As mentioned earlier, there are several possible interpretations of elevated CES-D scores, any one of which may hold for a particular individual. Elevated CES-D scores in non-clinical samples may be part of a larger constellation of problems, particularly those associated with

medical illnesses (Murrell, Himmelfarb, and Wright, 1983) or lack of social support or economic resources (O'Harah, Kohout, and Wallace, 1985). Higher scores also may reflect “demoralization” (Roberts et al., 1989). CES-D symptoms may appear not only as a manifestation of depressive illness as such, (i.e., dysthymia or major depression), but as a manifestation of another psychiatric disorder. Two in particular are known to occur with elevated frequencies among POWs: posttraumatic stress disorder (PTSD) and generalized anxiety disorder (Kluznik et al., 1986). Depressive symptoms are an associated feature of PTSD, and major depression actually shares three diagnostic criteria with PTSD--loss of interest in activities, sleep disturbance, and impaired concentration. The ongoing study of these cohorts includes measurement of both PTSD and depression by questionnaire and by direct examination. This should help obtain a clearer understanding of POWs' adjustment.

In comparison with CES-D general population studies, the depressive symptom rates in this study are high. These groups of former POWs differ significantly from the general population and, in fact, they most resemble a clinical population of recovering depressives.

The statistical evidence from the regression analyses supports two conclusions: (1) the treatment of POWs during military captivity, at least as measured by self-reported medical symptoms (in all three groups) and weight loss (in Pacific POWs), is statistically linked with subsequent depressive symptoms; and (2) differences in depressive symptoms can be attributed to differences in these captivity-related factors, even when moderating factors such as age and education are considered. The events of military captivity suffered decades ago are predictive of current, chronic post-traumatic depressive symptomatology.

Beebe GW (1975) Follow-up studies of World War II and Korean War prisoners, II: Morbidity, disability, and maladjustments Am J Epidemiol 101 : 400-422

Boyd JH Weissman MM Thompson WD et al (1982) Screening for depression in a community sample Arch Gen Psychiatry 39 : 1195-1200

Breslau N (1985) Depressive symptoms, major depression, and generalized anxiety: A comparison of self-reports of CES-D and results from diagnostic interviews Psychiatry Res 15 : 219-229

Cohen BM Cooper MZ (1954) A follow-up study of World War II prisoners of war Veterans Administration Medical Monograph Washington, DC : US Government Printing Office

Comstock GW Helsing KJ (1976) Symptoms of depression in two communities Psychol Med 6 : 551-563

Dent OF Tennant CC Goulston KJ (1987) Precursors of depression in World War II veterans forty years after the war J Nerv Ment Dis 175 : 486-490

Eaton WM Kessler LG (1981) Rates of symptoms of depression in a national sample Am J Epidemiol 114 : 528-538

Engdahl BE Eberly RE (1990) The effects of maltreatment In P Suedfeld (Ed). Psychology and Torture . Washington, DC : Hemisphere Press

Frerichs RR Aneshensel CS Clark VA (1981) Prevalence of depression in Los Angeles county Am J Epidemiol 113 : 691-699

Golding JM Aneshensel CS (1989) Factor structure of the Center for Epidemiologic Studies Depression Scale among Mexican-Americans and non-Hispanic whites Psychol Assess 1 : 163-168

Goldstein G van Kammen W Shelly C et al (1987) Survivors of imprisonment in the Pacific theater during World War II Am J Psychiatry 144 : 1210-1213

Hathaway SR McKinley JC (1951) The Minnesota Multiphasic Personality Inventory Manual-Revised . New York : The Psychological Corporation

Keehn RJ (1980) Follow-up studies of World War II and Korean Conflict prisoners: III. Mortality to 1 January 1976 Am J Epidemiol 111 : 194-211

Klonoff H Clark G Horgan J et al (1976) The MMPI profile of prisoners of war J Clin Psychol 32 : 623-627

Kluznik JC Speed N Van Valkenburg C et al (1986) Forty-year follow-up of United States prisoners of war Am J Psychiatry 143 : 1443-1446

Leon GR Gillum B Gillum R Gouze M (1979) Personality stability and change over a 30-year period-Middle to old age J Consult Clin Psychol 47 : 517-524

Little KB Fisher J (1958) Two new experimental scales of the MMPI J Consult Psychol 22 : 305-306

Murrell SA Himmelfarb S Wright K (1983) Prevalence of depression and its correlates in older adults Am J Epidemiol 117 : 173-185

Myers JK Weissman MM (1980) Use of a self-report symptom scale to detect the depressive syndrome Am J Psychiatry 197 : 1081-1084

National Center for Health Statistics (1980) Basic data on depressive symptomatology DHEW Pub. No. 80-1666 Hyattsville, MD

Nefzger MD (1970) Follow-up studies of World War II and Korean War Prisoners: I. Study plan and mortality findings Am J Epidemiol 91 : 123-138

O'Hara MW Kohout FJ Wallace RB ( 1985 ) Depression among the rural elderly: A study of prevalence and correlates J Nerv Ment Dis 173 : 582-589

Page WF ( 1988 ) A report of the longitudinal follow-up of former prisoners of war of World War II and the Korean conflict: data on depressive symptoms from the 1984-1985 questionnaire. Final report to the Veterans Administration under contract V101(93)P-1088. Washington, DC

Radloff LS ( 1977 ) The CES-D scale: A self-report depression scale for research in the general population Appl Psychol Meas 1 : 385-401

Roberts RE Vernon SW Rhoades HM ( 1989 ) Effects of language and ethnic status on reliability and validity of the Center for Epidemiologic Studies-Depression Scale with psychiatric patients J Nerv Ment Dis 177 : 581-592

Rundell JR Ursano RJ Holloway HC Silberman EK ( 1989 ) Psychiatric responses to trauma Hosp Comm Psychiat 40 : 68-74

Sledge WH Boydstun JA Rabe AJ ( 1980 ) Self-concept changes related to war captivity Arch Gen Psychiatry 37 : 430-443

Speed N Engdahl BE Schwartz J Eberly RE ( 1989 ) Posttraumatic stress disorder as a consequence of the prisoner of war experience J Nerv Ment Dis 177 : 147-153

Tennant CC Goulston KJ Dent OF ( 1986a ) The psychological effect of being a prisoner of war: Forty years after release Am J Psychiatry 143 : 618-621

Tennant CC Goulston KJ Dent OF ( 1986b ) Clinical psychiatric illness in prisoners of war of the Japanese: Forty years after release Psychol Med 16 : 833-839

Ursano RJ ( 1981 ) The Vietnam era prisoner of war: Precaptivity personality and the development of psychiatric illness Am J Psychiatry 138 : 315-318

Weissman MM Sholomskas D Pottenger M et al ( 1977 ) Assessing depressive symptoms in five psychiatric populations: A validation study Am J Epidemiol 106 : 203-214

Zung WWK ( 1965 ) A self-rating depression scale Arch Gen Psychiatry 12 : 63-70

This paper is based in part on the final report to the Veterans Administration under Contract V101(93)P-1088. Special acknowledgment for help in that earlier report is made to Mr. Robert J. Keehn and Dr. Gilbert W. Beebe; thanks also to Dr. Richard D. Remington, Dr. Gerald L. Klerman and Dr. Raina E. Eberly for their review of it. The staff of the Medical Follow-up Agency also deserve credit for the careful and tireless work in collecting and organizing the data upon which this report is based. Above all, the cooperation of the former prisoners of war themselves made this research possible.