The Health of Former Prisoners of War: Results from the Medical Examination Survey of Former POWs of World War II and the Korean Conflict (1992)

The wealth of findings from the preceding chapters are most easily brought together and discussed when organized by disease. This chapter also reviews other pertinent literature and relates it, by organ system, to the disease-specific examination findings noted earlier.

From the earliest mortality follow-up, there has been evidence of excess infectious disease deaths among POWs. Cohen and Cooper (1954) reported mortality rates from tuberculosis that were five times higher than expected, and Nefzger (1970) found a pattern of continued high mortality as a result of tuberculosis persisting in PWP after 20 years. The most recent mortality follow-up by Keehn (1980) showed that although excess tuberculosis mortality in PWP was highest in the years just after repatriation, tuberculosis rates were still roughly twofold higher than expected up to 30 years after repatriation.

Richardson's study of 100 Hong Kong prisoners of the Japanese (1965) found that during 1946–1964, there were 12 deaths from pulmonary tuberculosis whereas only 5 would have been predicted, had the group been dying at rates comparable to those of the Canadian population. In a 1946–1963 mortality follow-up of some 14,000 Australian ex-prisoners of the Japanese, Freed and Stringer (1968) found tuberculosis to be responsible for higher than expected death rates in the later (1951–1963) but not the earlier

(1946–1950) period. A more recent mortality study comparing 908 Australian ex-prisoners with 797 other veterans of the same theater found that mortality differences were pronounced from 5 to 14 years after the war; further analyses, however, did not suggest that these mortality differences could be attributed to particular causes of death (Dent et al., 1989). Moreover, the proportions of subjects whose vital status could not be ascertained differed among POWs (10%) and controls (15%); as a result, the possibility of bias in the study has been raised (Adena, 1989). In a 1973 study comparing samples of New Zealand ex-prisoners (N = 246), ex-servicemen (who went overseas but were not captured, N = 240), and ex-homeservicemen (who did not serve overseas, N = 209), Salmond and colleagues (1977) found a current disablement pension rate for pulmonary tuberculosis of 0% for ex-servicemen, 4.3% for ex-prisoners, and 7.1% for ex-homeservicemen; similar rates for tropical and parasitical disease were 7.1% for exservicemen, 1.1% for ex-prisoners, and 0% for ex-homeservicemen.

Turning to morbidity data, in the 1967 follow-up, Beebe (1975) found significantly higher hospitalization rates among PWP, compared with their controls, for pulmonary tuberculosis, early syphilis, amebiasis, schistosomiasis, and other worm infestation; PWK showed significantly higher rates of pulmonary tuberculosis, amebiasis, dysentery, and other worm infestation. A more recent study of 602 former British prisoners of the Japanese, conducted in Liverpool and published by Gill and Bell (1980), found 88 with strongyloidiasis and 6 with intestinal amebiasis. In a study of 170 Australian former prisoners of war of the Japanese and 172 non-POW veteran controls, Goulston et al. (1985) found 6 current cases of strongyloidiasis, all among POWs (for a rate of 3.5%) and all ascertained by microscopy or culture. However, these investigators reported lifetime prevalence rates of 15% for their Australian POWs and 2% for controls, speculating that their low yield of strongyloidiasis was probably the result of less time spent on direct microscopy. Hill (1988), reporting on former British POWs examined at Princess of Wales Royal Air Force Hospital in Ely, stated that the prevalence rate he found, 16%, was perhaps an underestimate and that 20% might be more realistic. The diagnosis of strongyloidiasis may be aided by the recent development of an ELISA test, evaluated in a sample of American POWs by Pelletier et al. (1988).

In the examination study, lifetime prevalence rates of infectious disease were extremely high, from 95% to 100%, for both POWs and controls, but current rates were roughly 20–30% in both POWs and controls, with the exception of WE, who had a low rate of 7.4% (see Table 5.C.3). There were no noteworthy differences between POWs and controls; therefore, no infectious diseases other than dysentery were singled out for further analysis. However, detailed tabulations were produced (see Appendix C), aggregated to the three-digit ICD [International Classification of Diseases] code

level, in which each mention of an unresolved condition was recorded. Because a particular condition might be mentioned several times in the record of an examination, rates derived from these tabulations are not strictly comparable to the person-based prevalence rates reported elsewhere in this study, nor do they have a straightforward interpretation, given the possibility of multiple mentions of the same condition for a single person.

Nevertheless, these detailed data include only 2 mentions of amebiasis (1 from PWP and 1 from PWK), 10 mentions of dysentery (1 PWP, 2 PWEM, and 7 PWK), and 5 mentions of other helminthiases (all PWP). This latter category included strongyloidiasis, a condition that Gill and Bell (1980) found in nearly 15% of their examinees. Pulmonary tuberculosis was mentioned 16 times (13 PWP and 3 PWK); the most mentioned infectious condition was dermatophytosis: 52 (PWP), 23 (PWEM), 17 (PWE), 118 (PWK), 17 (WP), 0 (WE), and 25 (WK). Dividing these counts by the number of exams gives respective rates of 20.8%, 27.7%, 12.0%, and 28.9% for POWs, and 31.5%, 0%, and 24.3% for controls.

There was no evidence of increased death rates from malignant neoplasms among POWs in either Nefzger's (1970) or Keehn's (1980) follow-ups, and Beebe (1975) found no significantly greater hospitalization rates for any type of cancer. Although a mortality study of British POWs by Gill (1983), based on death certificates sent in by relatives at the request of national and local POW organizations in the north of England, found higher rates of stomach, pancreas, and liver cancer compared with population values, these findings are clearly vulnerable to selection bias. In a recent study of 908 Australian ex-prisoners and 797 controls, there was no evidence of a higher rate of malignancies among former POWs (Dent et al., 1989). In a 1973 study comparing samples of New Zealand ex-prisoners (N = 246), ex-servicemen (who went overseas but were not captured, N = 240), and ex-homeservicemen who did not serve overseas (N = 209), Salmond and colleagues (1977) found 1.4% of ex-servicemen under current medical supervision for neoplasms versus 0% for both ex-prisoners and ex-homeservicemen.

The examination data also support the contention that military captivity is not associated with increased rates of subsequent malignancy. Lifetime prevalence rates of malignancy are around 15–20% for POWs (except for 9% for PWK); current prevalence rates are 2–5% for all POW groups. None of these rates is appreciably different from its comparable control rate. Lifetime rates of benign neoplasm are approximately the same as malignancy rates, whereas current rates—10–12% for all POWs—are somewhat higher than the malignancy rates. Again, however, there are no noteworthy differences between POWs and controls.

Diabetes mellitus occurs in two distinct forms, insulin-dependent diabetes mellitus (IDDM) and non-insulin-dependent diabetes mellitus (NIDDM). Presumably, the military's medical screening process during admission removes applicants with IDDM, leaving NIDDM as the condition of principal interest in this report. Although the prevalence of diabetes increases with age and other factors can place a person at increased risk of NIDDM, the only factor that has been consistently related to an increased prevalence of diabetes mellitus is higher relative body weight (National Research Council, 1989). Because of this link with body weight, diabetes was singled out from endocrine diseases for special consideration. It is unclear, however, what kind of hypothesis should be formed a priori for POWs, because one might argue either of two ways: earlier starvation lowered the risk of diabetes at the time of captivity and lifetime risks should therefore remain lower, or the recovery from starvation, which might have resulted in an excess weight gain, subsequently raised the lifetime risk of diabetes.

Earlier studies of the MFUA cohort showed that POW death rates for diabetes were substantially below comparable U.S. general population figures (Keehn, 1980), but this finding was true for controls as well and may be attributable to the aforementioned medical screening of military service personnel. Beebe (1975) found no excess hospitalization for diabetes.

Data from the current examination show no evidence for either a higher or lower prevalence of diabetes in POWs. Lifetime prevalence rates for POWs are 12–15% compared with 7–11% for controls (except for a WP rate of 18.5%); current rates are nearly identical to lifetime rates. Diabetes prevalence rates were nonetheless slightly higher among POWs who lost more than 35% of body weight during captivity; they showed no uniform gradient with prison camp symptoms. The simple association with weight loss, however, could have been unduly influenced by differences among POW groups; group-specific correlation analyses showed no evidence of noteworthy correlations. There is, in short, no compelling evidence that diabetes is more or less prevalent among POWs than among controls, based on exam data.

The category of other endocrine diseases includes the nutritional deficiency diseases, for which there is ample evidence of elevated prevalence among POWs. The repatriation examinations of 4,618 PWP by the Morgan Board Debarkation Hospitals (Morgan et al., 1946) produced significant evidence of avitaminosis, with beriberi—resulting from a thiamin (vitamin B₁) deficiency—one of the most prevalent prison camp symptoms; a history

suggesting wet beriberi was found in 77% and dry beriberi in 50% of those examined by the Morgan Board. In Beehe's follow-up (1975), significantly higher rates of both beriberi and other avitaminosis among PWP were still found 20 years after repatriation.

The exam data indicated lifetime prevalence rates of other endocrine diseases of between 70–85% among POWs, compared with 50–70% for controls; both PWP and PWK showed appreciably higher rates than their comparable controls. Data from Eberly and Engdahl's (1991) examination study showed lifetime prevalence rates for beriberi of 48% for PWP and 17% for PWK, which are fairly comparable to exam data rates of 61% for PWP and 26% for PWK. There were, however, no noteworthy differences between POWs and controls in current rates of other endocrine diseases, except that PWP had an appreciably lower rate than WP. Thus, appreciably higher rates of other endocrine diseases seen in earlier studies were not seen in the current data. The sequelae of malnutrition and avitaminosis that manifest themselves in other organ systems will be discussed under those systems.

Although the repatriation examinations of PWP (Morgan et al., 1946) reported a high rate of anemia (52% of the first 1,500 prisoners surveyed), by the time of Beebe's 1967 study, hospitalization rates of POWs and controls for diseases of the blood and blood-forming organs showed no significant differences (Beebe, 1975). It was therefore somewhat surprising to find appreciably lower lifetime prevalences of blood diseases among PWP and PWK—14% and 12%, respectively—compared with WP and WK—29.6% and 22.3%, respectively (see Table 4.2). Current prevalence rates of blood diseases showed similar and, again, noteworthy differences of 12.4% and 11.5% versus 25.9% and 21.4%, respectively, for PWP and PWK versus WP and WK.

Further analyses of these differences in blood disease prevalence showed that most of this deficit was attributable to lower rates of anemia (ICD codes 280–285). When anemia alone was considered (see Chapter 6), again, there were noteworthy differences in prevalence between PWP and PWK and their respective control groups: 6.8% and 5.2% for PWP and PWK, versus 20.4% and 18.5% for WP and WK. The possibility exists, of course, that it is the control rates that are elevated; however, the corresponding WE rate, 11.1%, although lower than either the WP or WK rate, was nevertheless still higher than the PWP and PWK rates, suggesting that it is the POW rates that are low and not the control rates that are high.

Yet further analyses of the data on the prevalence of anemia shed no additional light on this unanticipated finding; no noteworthy associations

were found between anemia prevalence and weight loss or anemia prevalence and symptom score in any of the univariate or multivariate analyses. Even so, the correlation analysis showed small, negative correlations between anemia prevalence and both weight loss and prison camp symptoms, indicating that the greater the weight loss and number of prison camp symptoms reported, the lower the current prevalence of anemia. Thus, although the finding that anemia is less prevalent among former POWs than among comparable controls appears to be genuine, it is unexplained by simply measured factors such as percent weight loss and the presence of medical symptoms during captivity.

All earlier follow-ups of the MFUA cohort have shown that increased psychiatric illness among POWs was the most striking and persistent aftereffect of military captivity. Cohen and Cooper (1954) found higher death rates for suicide (based on quite small numbers) and for accidental deaths; they also found elevated rates of hospitalization for psychoneurosis in both Pacific and European theater prisoners. Nefzger (1970) found that suicides, although few in number, were about 30% more frequent among Pacific prisoners than among their controls, and Keehn (1980) found later suicide rates to be significantly elevated in PWP but not in the other groups. Such mortality findings, however, are not fully indicative of levels of psychiatric illness, which are more appropriately reflected in morbidity data.

In the 1967 follow-up by Beebe, psychiatric illness was clearly the most prevalent aftereffect of captivity (Beebe, 1975). Compared with controls, PWP, PWE, and PWK groups all showed significantly higher rates of hospitalization for mental, psychoneurotic, and personality disorders. Hospitalization rates for the following specific conditions were significantly higher for POWs than for corresponding controls: schizophrenic disorders (PWP), alcoholic psychoses (PWP), anxiety reaction with somatization (PWP, PWE, and PWK), neurotic-depressive reaction (PWP and PWK), psychoneurosis with somatization (PWP and PWE), other psychoneurotic disorders (PWP), pathological personality (PWP), and alcoholism (PWP). Although all three POW groups showed some elevated hospitalization rates, clearly, the PWP group was most affected.

The 1967 follow-up also included the Cornell Medical Index (CMI), a general health questionnaire with a number of items pertaining to psychological symptoms; when scored in the customary manner, the CMI yields values on several subscales. The scored data showed increased self-reported symptoms for all of the mood and feeling subscales: inadequacy (PWP and PWK), depression (PWP), anxiety (PWP and PWK), sensitivity (PWP and PWK), anger (PWP and PWK), and tension (PWP and PWK), as well as

total emotional score (PWP and PWK). The total CMI score (both somatic and psychological) was found to be significantly related to a number of measures of nutritional stress, which are also used in this report for the analyses underlying Table 6.7.

In the latest completed follow-up of the MFUA cohort, which was centered around the Center for Epidemiologic Studies depression (CES-D) scale, Engdahl and Page (1991) found significant and marked elevations in the prevalence of depressive symptoms. In general, rates of depressive symptoms were three to five times higher for POWs than for the general population, and they were positively associated with both weight loss and prison camp symptoms and negatively associated with years of education, age at capture, and being married. (In other words, the more educated, older [when captured], and married POWs were less likely to have depressive symptoms [Page et al., 1991].) Further analysis of the subcomponents of the CES-D showed that these same demographic factors were generally predictive of lower rates of negative affect and higher rates of positive affect, fewer somatic symptoms, and fewer interpersonal problems (Engdahl et al., 1991).

Other recent studies of former American POWs have uncovered similar findings. The results of Eberly and Engdahl (1991) were discussed in an earlier chapter, but it is worth recalling that they showed striking elevations in diagnoses of posttraumatic stress disorder (PTSD) and generalized anxiety among POWs and some elevation of depressive symptoms, even though the Minneapolis depression data were collected using different, and possibly more stringent, diagnostic criteria than were applied in collecting the MFUA data. In an earlier study of 188 former WW II POWs in the Minneapolis area, Kluznick et al. (1986) found a lifetime prevalence of PTSD of 66% and a lifetime prevalence of generalized anxiety disorder of 53%; the prevalence of chronic, unresolved PTSD in this group was 47%. Speed and associates (1989) found that the strongest predictors of PTSD in a smaller group of 62 Minneapolis area POWs were proportion of weight lost and experience of torture during captivity; family history of mental illness and preexisting psychopathology were at best only weakly correlated with persistent PTSD symptoms.

Oboler's report (1987) of 190 consecutive protocol examinations at the Denver VA Medical Center stated that 82% of Pacific prisoners (in all, numbering 50) were found to have psychiatric impairment; 60% suffered from anxiety disorder, 28% from PTSD, and an additional 18% from depressive disorder. Among European prisoners, the corresponding percentages were 60%, 33cA, 11%, and 21%; for Korean prisoners (who numbered 15), 73% had psychiatric impairment, 60% had anxiety disorders, and 47% had PTSD. One PWK was diagnosed with dysthymic disorder.

Sutker and Allain (1991) of the New Orleans VA Medical Center have recently reported on Minnesota Multiphasic Personality Inventory (MMPI)

profiles of 168 former POWs and 67 controls from WW II and the Korean conflict. Their results confirm previous reports that former POWs have higher scores on the hypochondriasis, depression, and hysteria scales of the MMPI. Sutker and colleagues (1991) reported on a battery of tests administered to 22 Korean prisoners and an equal number of controls drawn from the catchment area of the New Orleans VA Medical Center. They found dramatic differences between the two groups on the MMPI subscales for depression and schizophrenia and a higher Mississippi PTSD score among Korean prisoners. Scores on the Beck depression scale and on state and trait anxiety measures were higher, but not significantly, for POWs. Another study based on a sample of 20 Korean conflict prisoners (Sutker et al., 1990) found PTSD in 18 cases (90%); mood disorders were cited in 75% of cases, anxiety disorder other than PTSD in 45%, and alcohol abuse in 20%.

Goldstein and colleagues (1987) studied 41 Pacific prisoners in the Pittsburgh area and found that half met the full diagnostic criteria for PTSD, with 97% reporting some sleep disturbance. The MMPI profile of the group showed elevations of the hypochondriasis, depression, and hysteria scales, suggesting the presence of a pronounced anxiety state with depressive features. Their findings were similar to those of Sutker et al. (1991) noted above and of Klonoff et al. (1976) noted below. A subsequent study of 10 PWP, all of whom reported sleep disturbances and other PTSD-related findings, showed that 6 of the 10 had no stage 4 sleep and had significantly higher mean ventricular brain ratios, which correlated with the number of awakenings (Peters et al., 1990). Without similar data from normal controls, however, these findings could not be specifically related to PTSD.

Zeiss and Dickman (1989) have reported on a statewide sample of 442 Virginia ex-POWs who responded to a questionnaire mailed from the Roanoke VA Regional Office. The three-page questionnaire requested information on PTSD symptoms and included an abbreviated listing of diagnostic criteria; 44% of those mailed a questionnaire responded. Symptoms of a severity consistent with a diagnosis of PTSD occurred in 56% of respondents (55.7% of PWE and 55.1% of PWP), but in contrast to other studies, most predictor variables—duration of internment, age at capture, current age, and duration of current marriage—were not significantly associated with a diagnosis of PTSD. In fact, only rank at capture showed a significant association, and even more surprising, there was no difference in PTSD prevalence between those with a Pacific and those with a European location of captivity. These findings remain anomalous.

Studies of other than American former prisoners have found similar evidence of psychiatric illness. A 1964–1965 study by Richardson of Canadian WW II prisoners of the Japanese who were captured in Hong Kong reported findings from a study group of 100 former prisoners and a control group of 100 of their brothers (Richardson, 1965). Data on the pension

status of all 100 prisoners and controls revealed a higher rate for POWs of compensated neuro-psychiatric conditions—28 POWs were receiving compensation versus 3 controls. Psychiatric examinations were conducted on only a smaller group of 20 prisoners and their brothers (Kral et al., 1967). In this smaller group, there was a significantly higher rate of psychiatric complaints among POWs (12 of 20) than among their brothers (2 of 20), and there were significantly higher rates of tension, anxiety, and depression among the POWs. Klonoff et al. (1976), in their study of Canadian WW II prisoners of the Pacific (N = 34) and European (N = 31) theaters, found an MMPI group profile characterized by elevated hysteria and depression scales.

More recently, Tennant et al. (1986a) studied 170 Australian prisoners of the Japanese and 172 veteran controls. They found elevated rates of clinically diagnosed anxiety and depressive disorders among the POWs—although the finding of a difference in the rates of anxiety in the two groups was not statistically significant—but no difference in alcoholism rates. Their questionnaire study similarly showed excess rates of depressive symptoms among POWs (Tennant et al., 1986b). Further analysis of the data on depressive symptoms showed that self-reported nervous illness during the war and depressive illness since the war were independently and significantly associated in both groups with higher rates of depression, as was unemployment among POWs and not being married among controls. Among controls, education had a significant inverse relationship to depression (i.e., higher education and lower depression); the same type of relationship was seen for socioeconomic status among POWs (Dent et al., 1987).

In contrast, a 1973 study by Salmond and colleagues (1977), comparing samples of New Zealand ex-prisoners (N = 246), ex-servicemen who went overseas but were not captured (N = 240), and ex-homeservicemen who did not serve overseas (N = 209), found low and quite comparable rates of nervous disorders currently under medical supervision among the three groups—7.1%, 7.9%, and 5.9%, respectively. Rates of disability, however, showed a different pattern, with 19.4% of ex-prisoners being disabled for other nervous conditions versus 9.5% of ex-servicemen and 0% of ex-homeservicemen.

Gill and Bell's (1981) study of 602 British former Far East POWs who were examined in Liverpool found 209 (34.8%) with significant psychiatric illness; in only 7 cases was this unrelated to the POW experience. Of the 209, 90 had depression, 57 had anxiety neurosis, and 62 had both. In 1981, Patrick and Heaf published a review of this and other British work, as well as worldwide findings. The review also included major results from studies of other, non-POW captives, such as the work by Eitinger (1964) and Strom (1968) on Norwegian concentration camp survivors and Thygesen et al. (1970) on Danish concentration camp survivors.

The preceding brief review of psychiatric findings among POWs merely indicates the depth of available material in the specific area of POW

studies; for example, it omits recent research on Vietnam-era POWs and more general review material on the POW experience (e.g., Ursano and Rundell, 1990). There are additional, more tangential areas that could be profitably reviewed as well if space and time permitted. Certainly, the more general study of combat stress and resulting psychiatric disability is especially pertinent to POW studies; indeed, there are clear parallels between the psychological aftereffects of all types of trauma. This point of view is manifested in current arguments about the diagnostic construct of PTSD. This disorder appears for the first time as such in the third edition (1980) of the American Psychiatric Association's Diagnostic and Statistical Manual of mental disorders (known colloquially as DSM-III) but is said to have been known for hundreds of years, although under different names (Trimble, 1985). Clearly, the study of the psychological effects of military captivity could be set in a wider context, but that is not a part of the current report.

The results of the medical examination survey complement those of previous follow-ups and present new material on PTSD, a particular focus of the exam follow-up because it was an expected comorbid or underlying diagnosis associated with the kinds of depressive symptoms seen earlier in the 1984–1985 follow-up. Overall, the most striking psychiatric sequelae seen in the exam are depressive symptoms, PTSD, and generalized anxiety disorder, paralleling the findings of most other studies of former POWs. The exam data provide additional information, however, on the persistence of psychiatric illnesses—for WW II POWs, now as long as 45 years after repatriation. These exam data show that psychiatric illnesses not only persist but that their rates have diminished little over the years. Lifetime prevalence of depressive disorder, for example, was 52%, 36%, 53%, and 56% among PWP, PWE, PWEM, and PWK, respectively; the corresponding rates for current (i.e., unresolved) conditions are 47%, 32%, 51%, and 51%. The data on PTSD are similar, with lifetime rates of 41%, 23%, 28%, and 39% compared with current rates of 40%, 20%, 28%, and 37%. The data for generalized anxiety disorder constitute a third example with lifetime rates of 44%, 47%, 63%, and 59% compared with current rates of 39%, 39%, 55%, and 54%, respectively. In contrast, the structured clinical interview for DSM-III-R (SCID) data show lifetime prevalence levels of PTSD that are about equal to exam levels—but current PTSD rates that are half that size (except for PWK), a finding more in line with those of other studies. Thus, according to physician diagnosis, the rates of psychiatric illness among POWs have decreased only slightly with the passage of time, no matter what their baseline levels, although this observation is not supported by the SCID data. In any case, it is clear that both PWP and PWK rates of psychiatric illness are appreciably higher than rates for PWE. It should be noted that the diagnoses of PTSD, depressive symptoms, and

generalized anxiety—in particular, the first two—overlap substantially and that much additional work would be needed to disentangle them. For example, there is a good deal of overlap in the symptoms—and therefore the diagnoses—of PTSD and depressive symptoms. Moreover, it is conceivable that the POW experience causes depression, which in turn causes PTSD symptoms, or vice versa. Such issues are not addressed in this report.

Before leaving the subject of lifetime prevalence of PTSD, depression, and generalized anxiety disorder, it is worthwhile to place the exam results against a more general backdrop to see the extraordinary group that these former POWs constitute. Various disclaimers have been made—based on the low response rates and the very real potential for serious nonresponse bias—about the ability to draw conclusions from the exam data collected. For the three psychiatric conditions listed above, however, the disclaimers are moot. This assertion will be argued using data from the Epidemiologic Catchment Area (ECA) study, sponsored by the National Institute of Mental Health, which surveyed almost 20,000 Americans at five sites across the country. The study yielded some of the most comprehensive data ever collected on psychiatric illness in the U.S. general population.

Consider first the lifetime prevalence of PTSD. Based on ECA data, Helzer et al. (1987) estimated the prevalence of PTSD among men in the general U.S. population at 0.5%. In contrast, the SCID, for example, estimates the prevalence of PTSD in PWP at 33% and in PWK at 41%—rates some 60 to 80 times higher than those found in the national sample. If one were to recalculate the SCID prevalence rates to include all eligible subjects—respondents and nonrespondents alike—and even if one were to count every nonrespondent as never having had PTSD, the revised SCID prevalence would be roughly cut in half among PWP (who had a 50% response rate) and reduced two-and-a-half-fold among PWP (who had roughly a 40% response rate). These revised, ''worst-case'' estimates are still around 16% for PWP and PWK—and thus 30-fold higher than the national estimate. Although statistical inferences based solely on the sample data are inappropriate, the kind of worst-case analysis presented here shows that despite the sizable problems caused by nonresponse, it is still possible to conclude that PWP and PWK have a significantly higher prevalence of PTSD than the general population.

One can perform similar kinds of calculations for depressive disorders and generalized anxiety disorder. Among men aged 65 and older (comparable to the WW II POW groups), the ECA's estimate of lifetime prevalence of affective disorders (which includes bipolar I and bipolar II disorders, dysthymia, and major depression) is only 1.6%; for men aged 45–64 (closer in age to the Korean conflict group), it is 3.6% (Robins and Regier, 1991). Again, the comparisons to the MFUA estimates are dramatic: physician estimates of lifetime depressive disorders are more than 30 times higher for

PWP and 16 times higher for PWK. Revising the MFUA rates downward in a worst-case scenario still leaves significantly higher prevalence estimates among the MFUA group. Finally, the ECA lifetime prevalence figures for generalized anxiety disorders range from 2.5% to 4.5% for those 65 years of age and older and from 6% to 7% in the age group 45–64. Physician-based prevalence estimates for generalized anxiety disorder were 44% for PWP and 59% for PWK: again, these are quite sizable discrepancies that stand even if worst-case estimates are calculated. Clearly, PTSD, depressive disorder, and generalized anxiety disorder are significantly more prevalent among PWP and PWK than in the general population.

The examination data also provide additional information on the long-term effects of combat, absent captivity, in the follow-up of the WK control group, which consists of men who were lightly wounded and returned to action. The association of combat trauma with psychiatric illness is well established; thus, it is not surprising that the WK group shows almost uniformly higher rates of psychiatric illness than other control groups. Here, the prevalence of psychiatric illness, although somewhat lower in magnitude than in POWs, appears to be as persistent: this pattern is seen for depressive symptoms (28% lifetime versus 22% current), PTSD (12% lifetime versus 11% current), and generalized anxiety (25% lifetime versus 22% current). Again, however, the SCID data tell a different story for PTSD, showing current prevalence rates that are about half the size of lifetime rates. The kind of worst-case analyses made above are not as enlightening in the WK groups, which had only a 15% response rate. Presuming that nonrespondents are free of psychopathology results in worst-case estimates that are reduced six-and-a-half-fold; estimates of this size are too near those of the general population to conclude that the WK group suffers significant psychiatric ill health. Thus, the study results for WK suggest excess psychological problems but do not constitute definitive evidence.

The Structured Clinical Interview data show lifetime rates that are approximately the same as those from the exam but lower current rates; the data from the Mississippi scale are generally a little higher. Again, there are some noteworthy differences between POWs and controls, but these appear to be war-era specific; the current levels of PTSD according to the Mississippi scale are roughly in line with those found either in the SCID or in the exam. The development and use of this instrument among Vietnam veterans may mean that the current standardized scoring and cut-points should be reevaluated for use in WW II and Korean conflict populations. Data from the Hopkins Symptom Check List (SCL-90) suggest that, in addition to the specific conditions noted above, psychiatric symptoms of all sorts are elevated among former POWs, especially among PWK, PWEM, and PWP.

The additional analyses undertaken in Chapter 6 contribute new information to the subject of psychiatric sequelae in former POWs. Both depressive symptoms and PTSD showed noteworthy simple associations with increased weight loss, a finding confirmed by Eberly and Engdahl in their Minneapolis POWs; depressive symptoms, schizophrenia, PTSD, and generalized anxiety all showed noteworthy simple associations with number of prison camp symptoms. In the logistic regression analysis, however, which controlled simultaneously for weight loss, prison camp symptoms, and group differences, depressive disorder showed a noteworthy association only with other symptoms (i.e., not with edema or visual symptoms); PTSD was associated with visual symptoms and with the PWP and PWK groups, generalized anxiety with other symptoms and PWK, and schizophrenia with PWK status. It is worth noting that because weight loss is also a symptom of depression, for example, controlling for it in multivariate analyses may result in a biased estimate of association.

Interpreting these different associations is not at all straightforward. Because all of the prison camp symptom measures also serve as more general measures of harshness of captivity, they may very well be confounded not only with some outcomes (e.g., weight loss and depression) but also with other measures of general and psychological stress, such as beatings and torture. Thus, although the presence of visual symptoms (associated with PTSD) indicates a deficiency of vitamin A, it would be inappropriate, for example, to speculate on a nutritional link between vitamin A deficiency and subsequent PTSD without some biological basis. Similarly, little can be said about the association of other prison camp symptoms with depressive disorder and generalized anxiety. The noteworthy group differences derived from the use of the logistic model, however, are more easily interpreted: both PWP and PWK have higher rates of PTSD than PWE, even after accounting for weight loss and prison camp symptom measures; and generalized anxiety is especially high among PWK, even after accounting for all other factors.

The data on schizophrenia deserve separate discussion. Lifetime prevalence rates for schizophrenia were around 1%, the rate Eberly and Engdahl (1991) quote as a general population estimate, for both PWP and PWE; PWEM and PWK lifetime rates were 2.3% and 3.4%, respectively. Current and lifetime rates differ little. Current schizophrenia in the combined POW group had a noteworthy association with the number of prison camp symptoms (Table 6.4); in addition, there was a noteworthy negative correlation of weight loss and schizophrenia in PWK (Table 6.5), indicating a higher rate of schizophrenia in those who had less weight loss. The logistic model, however, found an appreciable association only with PWK status (a 4.4-fold higher risk), indicating a clear link with PWK status but no noteworthy associations with weight loss or prison camp symptoms. Although the data

on schizophrenia are quite sparse, it still appears that PWK have a higher prevalence rate and that for PWK alone, less weight loss is associated with a higher prevalence of schizophrenia. This finding argues against nutritional deficiency as a cause of later illness.

In summary, the data show that a number of specific psychiatric conditions occur more frequently among former POWs—in particular, depression, PTSD, and generalized anxiety disorder. The current exam survey data show that POW rates for these conditions are still higher than control rates some four-and-a-half decades after repatriation and that these excesses are found in data collected by examination, structured interview, and questionnaire. The finding of elevated rates of PTSD, a relatively new diagnostic construct, occurs for the first time in this cohort.

In addition to the simple prevalence rate comparisons of POWs and controls, more sophisticated analyses have linked these increased rates with earlier treatment during captivity. Unfortunately, the markers of captivity stress, weight loss and prison camp symptoms, are not only measures of malnutrition but also of general maltreatment. As a result, their noteworthy associations with psychiatric illness do not tell us a great deal about the specific factors underlying subsequent morbidity.

Among PWP, the observations of prison camp physicians linked symptoms of peripheral nerve damage to nutritional deficiencies during captivity. Among American prisoners taken at Bataan and Corregidor, Hibbs (1946) reported that everyone in the camp had one form of beriberi or another at some time. He also noted that more than 75% of men in the camp had the predominately sensory disturbances or painful feet characteristic of "dry" beriberi, although less than 2% of the men developed motor paralysis; at the height of the disease, about 40% had 1⁺ to 2⁺ exaggeration of deep tendon reflexes. A recent analysis of clandestine medical records from two small camps in Japan—those of Maj. William Stewart of the Niigata POW Camp and of Capt. LaMoyne Bleich of the Oeyama POW camp—document the cumulative incidence of various categories of disease to August 1945 (Roland and Shannon, 1991). In Niigata there were 166 neurological disorders among 448 prisoners, a rate of 37%; in Oeyama there were 87 such disorders among 177 prisoners, a rate of 49%.

A report of Canadian prisoners of the Japanese stated that 84% of the repatriates gave a history of neurologic damage during internment; neurologic damage was still evident in 51% of them after they returned to Canada (Crawford and Reid, 1947). The repatriation examinations of some 4,618 former American POWs of WW II reported by Morgan et al. (1946) docu-

mented minimal polyneuritis in 568 cases (184 with diminished tendon reflexes and 409 with impairment of sensation) and pronounced polyneuritis in another 29 subjects (16 with diminished tendon reflexes and 22 with impairment of sensation).

Kral and colleagues (1967) examined a group of 20 Canadian former prisoners of the Japanese and 20 of their brothers then living in the Montreal area. They found a significantly higher prevalence of neurological complaints among POWs compared with controls (15 versus 4); their neurological findings included increased rates of superficial hypesthesia, particularly in the lower extremities (12 versus 3), diminished vibration and position sense (8 versus 0), ataxia of the spinal or motor type (9 versus 1), and cranial nerve involvement—sluggish pupillary response and weakness of the facial nerve (6 versus 0). In a 1973 study comparing samples of New Zealand ex-prisoners (N = 246), ex-servicemen who went overseas but were not captured (N = 240), and ex-homeservicemen who did not serve overseas (N = 209), Salmond and colleagues (1977) found a higher rate (5.3%) of central nervous system disorders under current medical supervision among ex-POWs than among ex-servicemen (1.4%).

At Queen Mary's Hospital, Roehampton, some 4,684 British Far East ex-POWs were examined (Gibberd and Simmonds, 1980), and 679 were found to have a neurological condition. Of these, 593 had either optical atrophy or peripheral neuropathy and no other neurological disease. Gill and Bell (1980) reported that 34 of 602 British ex-POWs examined in Liverpool between 1968 and 1978 had evidence of nutritional neuropathy; their subsequent report (Gill and Bell, 1982) of 898 ex-POWs revealed that 5.5% displayed persistent symptomatic neurological disease. A report by Cruickshank (1961), who spent three-and-a-half years in the Changi Military POW Camp on Singapore Island as a medical specialist, provides observations on diet during captivity and its relation to various deficiency syndromes, most of them affecting the nervous system. He noted that neurological features were present in 229 (57%) of the 400 cases suffering from beriberi; among the first 171 such cases, 22% had motor symptoms only, 31% had sensory symptoms only, and 47% had both kinds of symptoms.

Findings of increased neurological problems were still being seen in Beebe's 1967 follow-up of the MFUA study groups. PWP had significantly higher rates of hospital discharges in the broad category of "nervous system. other than central, and sense organs" (Beebe, 1975), and when attention was restricted to 3-digit ICD codes, a significantly higher rate of PWP hospitalizations appeared in the category "other diseases of the peripheral nerves." PWP also had a significantly higher rate of military service-connected compensation for peripheral neuritis.

In more recent studies of American POWs, Eberly and Engdahl (1991) do not mention peripheral neuropathy, probably owing to the lack of suit-

able data for control rates, but Oboler (1987) found on current examination that 48% (24 of 50) of Pacific WW II prisoners had objective evidence of a persistent peripheral sensory neuropathy. The majority reported only mild symptoms of numbness and tingling, but there were some cases with more severe symptoms, such as "burning" or "electric" sensations that were aggravated by movement and pressure. Among European prisoners, the prevalence of mild peripheral sensory abnormalities was only 5% (6 of 121); among Korean prisoners the prevalence was 7% (1 of 15).

Hong (1986), of the Livermore, California, VA Medical Center, reported the results of neurological examinations and nerve conduction studies on 52 (32 Pacific and 20 European) WW II POWs. Based on the neurological exam, 25 of the PWP and 9 of the PWE were clinically judged to have persistent peripheral neuropathies. All those with a clinical diagnosis had neurological signs of reduced sensation, and about 60% had symmetrically reduced deep tendon reflexes. Electrodiagnostic findings were consistent with distal polyneuropathy of the axonal degeneration type in 28 (88%) of the 32 PWP with a history of beriberi neuritis; in the PWE, however, only 8 (40%) of 20 had electrodiagnostic abnormalities suggesting chronic neuropathies.

In the current MFUA examination survey, lifetime prevalence of disease of the nervous system was appreciably higher in PWP and PWK, compared with their respective controls, as was the current (unresolved) prevalence; indeed, lifetime and current rates changed little—36% compared with 31% for PWP and 25% compared with 21% for PWK. In the analysis of question E in Chapter 5, PWEM were shown to have higher rates of lifetime peripheral nerve disease, which were mostly due to higher rates of mononeuritis of the upper limb and mononeuritis multiplex, mononeuritis of the lower limb, and hereditary and idiopathic peripheral neuropathy.

The analyses of Chapter 6 thus focused specifically on peripheral neuropathy and showed higher rates of current peripheral nerve disease among PWP and PWK. Peripheral neuropathy had a noteworthy association, among all POWs, with both weight loss and prison camp symptoms; in PWP it was appreciably correlated with prison camp symptoms and in PWE, with both weight loss and prison camp symptoms. Logistic regression analysis showed that when prison camp symptoms were redefined into three groups—edema, visual symptoms, and other symptoms—peripheral nerve disease had an appreciable association only with edema: those POWs who reported swelling in the ankles, feet, or legs while in the prison camp had 2.4 times the risk of developing subsequent peripheral nerve disease.

Although all three measures of prison camp symptoms in some part reflect overall harshness of treatment, symptoms of edema have a more specific interpretation. A grossly deficient diet produces protein deficiency and edema, which may be intensified by beriberi heart disease. In addition, the distribution of edema provides important information about its cause:

protein deficiency produces a more generalized edema, whereas edema associated with heart problems tends to be more extensive in the legs (Wilson et al., 1991, p. 232). Thus, self-reported edema in the feet, ankles, and legs is presumably related to beriberi heart disease ("wet" beriberi) in prison camp, which is caused by thiamin deficiency, and that same B₁ deficiency is also the cause of acute peripheral nerve disease ("dry" beriberi). The acute relationship between edema and peripheral nerve disease in prison camp is well established. An appreciable association between edema and chronic peripheral nerve disease, now observed some four-and-a-half decades later, has also been established. However, the pathophysiological processes that might link the two, if there are any, have not yet been established.

Finally, optic nerve disease has been associated with malnutrition in POWs. Bell and O'Neill examined 560 members of the Winnipeg Grenadiers who had been held captive by the Japanese for nearly four years and reported a 20% rate of partial optic atrophy (Bell and O'Neill, 1947). This condition was rarely found, however, during the current exam. As mentioned earlier, detailed tabulations, aggregated to the 3-digit ICD code level, recorded each mention during the exam of an unresolved condition. Because a particular condition might be recorded several times in an examination, the rates derived from the detailed tabulations are not strictly comparable to the person-based prevalence rates reported elsewhere. These detailed diagnostic data (see Appendix C) included only 26 mentions of current disorders of the optic nerve (ICD code 377): 9 for PWP, 1 for PWE, and 16 for PWK. Optic nerve disorder was not mentioned in any of the control examinations.

During the first six years after liberation, Cohen and Cooper (1954) found higher hospitalization rates among PWP than among controls for deafness (3% versus 0.2% in WP), other ear conditions (3% versus 1% in WP), and eye conditions that were not due to nutritional deficiency (3% versus 0.6%). In similar data for PWE there were no differences worth noting. Beebe's 1967 follow-up found significantly higher hospitalization rates among PWP for other diseases of the eye and for otitis media without mastoiditis; among PWK, higher rates were seen for other deafness.

In the current exam survey, lifetime prevalence rates of sense organ disease were quite high among both POWs and controls, generally from 80–90%, with no noteworthy differences seen between the two groups. The prevalence rates for current (unresolved) conditions were nearly identical to the lifetime figures. Given the lack of specific medical hypotheses, no further formal analyses were undertaken.

In the first follow-up of the MFUA cohort, Cohen and Cooper (1954) reported that there were 58 PWP deaths in the first six years after liberation, compared with an expected 25.3 deaths (based on U.S. white male mortality rates). Eight of these deaths had cardiovascular disease as an underlying cause; 4.1 such deaths would have been the expected number. However, by the time of Nefzger's follow-up, through 1965, cardiovascular death rates were below comparable U.S. general population rates for every POW and control group save PWK. Nefzger (1970) further noted that mortality ratios (calculated by dividing observed by expected deaths) for arteriosclerotic heart disease were slightly larger than those for all vascular disease in the WW II groups, whereas the reverse held true for PWK. This finding suggested that the rate of arteriosclerotic heart disease deaths might have been reduced in the first few years after liberation. Closer inspection of mortality ratios in 4-year intervals, however, offered no support for this hypothesis. All study groups were still too young for cerebrovascular death to have been a frequent cause of death.

Keehn (1980), like Nefzger, found that mortality from cardiovascular disease was below comparable U.S. rates for all six groups of POWs and controls. No relationship between former POW status and level of mortality was apparent, even when an additional 155 deaths mentioning cardiovascular conditions (but not as the underlying cause) were included. Hypertension was mentioned with similar frequency on the death certificates of former POWs and controls.

In Beebe's 1967 morbidity follow-up, hospitalizations for circulatory system disease were significantly higher among PWP compared with controls and, in the early follow-up years, among PWK (Beebe, 1975). Investigations of detailed three-digit ICD codes showed that PWP had rates of arteriosclerotic heart disease and benign essential hypertension that were significantly higher than those of controls.

Gill's (1983) report of mortality and autopsy findings among former British Far East POWs showed a proportion of deaths from ischemic heart disease that was roughly equal to the rate seen in the general population of England and Wales and lower than the rate seen in Scotland. The proportion of deaths that mentioned rheumatic heart disease as the underlying cause was larger among POWs than for these comparison groups, but this amounted to only 6 cases; the proportion of deaths attributable to cerebrovascular disease was roughly the same in ex-POWs as in the general population. Autopsy findings were available for one case with cardiac

beriberi, but a further case, although consistent with a diagnosis of chronic cardiac beriberi, could not be confirmed because of the presence of longstanding rheumatic mitral valve disease. Gill states that the difficulty with the last case is fairly common, noting that "the histological changes induced by cardiac beriberi are fairly stereotyped, but are not pathognomic." It should be kept in mind, however, that this study was based on death certificates sent in by relatives at the request of national and local POW organizations in the north of England. The study findings thus are clearly vulnerable to selection bias.

In Richardson's study of 100 Canadian prisoners of the Japanese, which used 100 of their brothers as controls, there was no evidence of increased hypertension; the proportion of POWs with diagnosed hypertension was 15% versus 14% for controls. Atherosclerotic heart disease was present in 3 POWs and 4 controls, and nondiagnostic electrocardiogram data had almost identical rates (12% and 13%). There was, however, a notable difference between former POWs and their brothers in those reporting symptoms of dyspnea on effort without chest pain (28% versus 12%, respectively) and in the reported history of feet sensitive to cold (45% versus 24%, respectively). Thus, Richardson concludes that "the data obtained by clinical examination in this survey do not suggest a significant difference between ex-prisoners and their brothers in the prevalence or severity of cardiovascular disease," adding, however, that "it is possible that significant differences would be found in a larger series" (Richardson, 1965, p. 50).

Richardson found a different story in the Canadian POW mortality data. Using Canadian vital statistics data for the years 1946–1964, he calculated the number of expected deaths that would have been observed in the group had they died at the published Canadian death rates. During this period there were a few more deaths from all causes than expected (135 versus 119.08), but this increase was not statistically significant. In contrast, deaths from atherosclerotic heart disease during the period numbered 47—compared with only 29.94 expected—which was a highly significant excess (p = .01). Richardson notes that the abnormally high atherosclerotic heart disease rate and approximately normal overall death rate are difficult to explain, yet judges that for whatever cause, "it seems necessary to conclude that service in the Far East has probably played some part in the unfavourable mortality experience from atherosclerotic heart disease."

In contrast, a 1973 study by Salmond and colleagues (1977) that compared samples of New Zealand ex-prisoners, ex-servicemen who went overseas, and ex-homeservicemen who did not serve overseas found current disablement pension rates for circulatory system disorders of 4.3%, 0%, and 7.1%, respectively. Gill and Bell's (1981) study of British POWs examined at Liverpool reported that 17% of their series of 602 examinees had ischemic heart disease diagnosed either clinically or by electrocardiography (ECG),

and that the overall ECG abnormality rate was 13%. These rates were not thought to be excessive.

Freed and Stringer's (1968) study of mortality among more than 14,000 Australian POWs during 1946–1963 showed an overall significant deficiency of deaths from arteriosclerotic and degenerative heart disease, both during the earlier period (1946–1950) of follow-up—23 observed versus 40.7 expected—and later (1951–1963)—293 observed versus 370.3 expected. Both of these differences were statistically significant. The authors also observed that this deficit was most marked in the older age groups. A recent mortality study of 908 Australian ex-POWs and 797 other veterans found a higher proportion of deaths from ischemic heart disease among POWs but a lower proportion of ischemic heart disease, other heart disease, and cerebrovascular disease. None of these differences, however, was exceptional, and overall, the cause of death was not associated with POW status. This study relied on other than official records; because the proportions of subjects whose vital status could not be ascertained differed among POWs (10%) and controls (15%), the possibility of bias has been raised by other researchers (Adena, 1989).

Recent morbidity data on cardiovascular disease in former POWs come from the Eberly and Engdahl and the Oboler studies. As Chapter 4 noted, Eberly and Engdahl (1991) found lower than expected lifetime prevalence rates of hypertension, cerebrovascular accident, myocardial infarction, and intermittent claudication among Minneapolis POWs, compared with general population rates. However, the rates in the Minneapolis group for conditions other than hypertension were also lower than examination rates in the MFUA study for both POWs and controls, raising the possibility that the Minneapolis POW group was healthier than the MFUA's national sample. Recent data collected by Oboler (1987) at the Denver VA Medical Center report the occurrence of hypertension in 34% of PWP, 33% of PWE, and 27% of PWK; these figures are a little lower than those of Eberly and Engdahl but still roughly comparable to the MFUA rates. Oboler reports no findings consistent with chronic beriberi heart disease and states that "no definite connection could be drawn between these veterans' POW confinement and their current cardiovascular disease."

The examination data showed no appreciably higher lifetime prevalence rates among POWs for either heart disease, cerebrovascular disease, or hypertension and other circulatory diseases (Table 4.2), nor were there any differences in current prevalence rates (Table 5.C.3). In Chapter 6, however, the analysis focused on more narrowly defined medical conditions. Although there were no noteworthy differences between POWs and controls in the prevalence of hypertension, cerebrovascular disease, or myocardial infarction, PWEM showed an appreciably higher rate of ischemic heart disease than did WE—28% compared with 7%. Further analyses showed no

noteworthy associations with either weight loss or prison camp symptoms, but the Cochran-Mantel-Haenszel analysis found an appreciable association of ischemic heart disease with prison camp symptoms, after controlling for POW group differences.

Logistic model analyses showed the prevalence of ischemic heart disease to be 1.7 times higher among POWs who reported edema in prison camp than in POWs who did not report it—a new finding; they also demonstrated that ischemic heart disease was, in addition, only about half as prevalent among PWK, probably owing to their younger age. The prevalence of cerebrovascular disease was possibly associated with the number of visual symptoms (persistent difficulty seeing in the dark, etc.), each additional symptom increasing the risk of cerebrovascular disease 1.7 times—again, a new finding among POWs.

Although the appreciable association between increased prevalence of ischemic heart disease later in life among POWs and self-reported symptoms of edema in prison camp is new, the acute cardiac effects of nutritional deprivation are well known. Unlike the situation for psychiatric conditions, the reporting of edema in prison camp is not only a general measure of stress but also indicates a specific nutritional deficiency, beriberi. A grossly deficient diet produces protein deficiency and edema, which may be intensified by beriberi heart disease. As noted earlier, the distribution of edema provides important information about its cause, with protein deficiency producing a more generalized edema and heart disease producing an edema that tends to be more extensive in the legs (Wilson et al., 1991, p. 232). Thus, the self-reported edema in the feet, ankles, and legs is presumably related to beriberi heart disease (''wet'' beriberi) in prison camp, which is caused by thiamin deficiency. There have been difficulties in linking beriberi with subsequent heart disease in clinical studies because typical medical findings are consistent with beriberi heart disease but are not pathognomic (recall Gill above). The examination data from the current study now provide epidemiologic evidence to suggest that earlier treatment in prison camp is linked to increased ischemic heart disease four-and-a-half decades later. However, one must interpret this association cautiously because it has no established pathophysiologic basis.

The increased prevalence of cerebrovascular disease among POWs that is associated with increased reporting of visual symptoms appears to be an entirely new finding without strong accompanying clinical evidence. Visual symptoms, like edema, are thought to be fairly specific evidence of nutritional deficiency, in this case vitamin A. Although vitamin A is necessary to maintain epithelial tissue and there has been great interest in the inverse association of vitamin A and cancer, a potential link between earlier vitamin A deficiency and subsequent excess stroke 45 years later must be considered highly speculative.

Peripheral arterial disease (PAD), whose assessment typically includes a history of intermittent claudication, deserves separate mention. Intermittent claudication was seen at such an appreciably higher level in the MFUA sample than in the Eberly and Engdahl (1991) sample that it might have been dismissed as an artifact of the exam process, especially since there were no appreciable differences between lifetime prevalences for MFUA POWs and controls. Instead, the analyses of Chapter 6, discussed below, linked PAD to earlier treatment in prison camp.

There are few systematic prevalence data on PAD in populations, an exception being the Criqui et al. (1985) population-based study of a sample of whites averaging 66 years of age. The study showed that large-vessel PAD was present in 12% of that group, which was within the range of the examination findings. In general, however, PAD is characteristically a disease of old age, and its risk factors are similar to those for cardiovascular heart disease and stroke—cigarette smoking, diabetes, systolic blood pressure, and serum cholesterol (Fowkes et al., 1992). Unfortunately, data on PAD risk factors such as smoking and cholesterol were not routinely collected in the examination.

In the exam data, the current prevalence of intermittent claudication and of arterial vascular disease had separate, noteworthy associations with percent weight loss and number of prison camp symptoms. The logistic regression analysis, however, showed that for both conditions the only noteworthy association was with weight loss, with each additional 10% of pre-captivity weight loss increasing the prevalence of disease by 30–40%. The lack of association with symptoms indicating a specific deficiency, such as in the case of ischemic heart disease and cerebrovascular disease, combined with little clinical information on nutritional risk factors, makes further explication of this finding impossible.

For the six years following liberation, Cohen and Cooper (1954) reported higher hospitalization rates for PWP compared with controls for respiratory diseases other than pneumonia—17% for PWP versus 5% in controls; Beebe (1975) also reported higher rates for PWP and for PWE in the broad rubric of respiratory conditions. However, there were no significant differences between POWs and controls in any of the detailed three-digit diagnostic data.

Gill and Bell (1981) note that no other surveys have sought data on the prevalence of respiratory disease in former Far East POWs and cite a rate of chronic bronchitis of 22% (135 of 602 former POWs examined in Liverpool); this rate is thought to be about double that of the comparable general population. They investigated respiratory disease further by reviewing chest

x rays and comparing them with a randomly chosen group of controls, matched on age, sex, and social class. These efforts produced radiologic evidence of significantly more thickened pleura, old rib fractures, emphysema, and diffuse fibrosis among the POWs. Unfortunately, Gill and Bell had no data on smoking, but they noted that recurrent attacks of bronchitis were common in captivity. They also noted that work in dusty environments was common among POWs, which could have relevance for their findings of excess diffuse pulmonary fibrosis. In Gill's study of mortality and autopsy findings (Gill, 1983), however, the proportion of deaths from chronic bronchitis was nearly identical in former POWs and in the general population of Scotland; it was smaller in POWs than in the general population of England and Wales. Dent et al. (1989) likewise show a smaller proportion of deaths resulting from respiratory disease in POWs than in comparable controls.

Compared with controls, the examination data show an appreciably higher lifetime prevalence of acute respiratory disease among PWK only and no noteworthy differences for chronic disease. For current (unresolved) respiratory disease, there were no noteworthy differences between POWs and controls for either acute or chronic conditions. In Chapter 6, asthma was specifically selected for further study, based on clinical judgment, but, again, there were no appreciable differences in prevalence between POWs and controls. Analyses of the relationship of asthma and prison camp treatment, however, revealed a noteworthy association of asthma with prison camp symptoms, and in PWP the current prevalence of asthma was appreciably correlated with both weight loss and prison camp symptoms.

In the logistic regression analysis, an increased prevalence of asthma was associated with increased visual symptoms, and each additional reported symptom increased the prevalence by roughly 29%. Other factors being equal, PWK had about half the prevalence of WW II POWs, a finding that is probably related to age. The noteworthy association of asthma with visual symptoms (similar to that seen for stroke) raises again the intriguing possibility that the observed excess prevalence of asthma might be nutritionally related, because the reporting of visual symptoms is fairly specific evidence of vitamin A deficiency and vitamin A is necessary for the maintenance of epithelial tissue. Again, however, there is little other evidence for such an association, and any hypothesis linking vitamin A deprivation and subsequent asthma (45 years later) must be considered highly speculative.

During the first six years following repatriation, Cohen and Cooper (1954) reported higher rates of hospitalization for gastritis, enteritis, and functional intestinal disorders among PWP (6%) and PWE (5%), compared with controls (1% and 0.2%, respectively). Beebe (1975) found increased

hospitalization rates among PWP for "gastroenteritis and colitis, except ulcerative," and for "other diseases of the intestines and peritoneum." He further stated that VA compensation data suggested an excess of peptic ulcer, both for PWP (with a compensation rate of 4.2% versus 0.6% for WP) and for PWK (4.0% versus 0.7% for WK).

Richardson's (1965) study of 100 former Canadian prisoners and an equal number of their brothers showed an increased history of liver disease in POWs (25 versus 13) and of nonspecific gastrointestinal symptoms with no diagnosis (47 versus 34). A relatively high incidence of irritable bowel syndrome was also noted. Special attention was paid to the diagnosis of peptic ulcer, with the finding that 21 POWs had such a diagnosis (current or during 1946–1964) compared with 8 controls; 11 POWs had been granted a pension for their ulcers compared with 2 controls. The locations of these ulcers were gastric (4 POWs, 2 controls), duodenal (13 POWs, 5 controls), and unspecified (4 POWs, 1 control). Within the past two years, 17 POWs had had symptoms versus only 5 controls. A review of records for all ex-prisoners living in Canada found a prevalence rate of 15.7% for peptic ulcer.

In a 1973 study comparing samples of New Zealand ex-prisoners, ex-servicemen who went overseas, and ex-homeservicemen who did not serve overseas, Salmond and colleagues (1977) found rates of digestive system disorders under current medical supervision of 12.4%, 8.6%, and 7.4%, respectively. Rates for disablement pensions, however, were nearly equal: 7.5%, 7.1%, and 7.1%. Ex-prisoners had a higher rate of current stomach trouble (36.1%) than did ex-servicemen (23.5%) or ex-homeservicemen (14.8%). Eighty percent of those with present symptoms of stomach trouble also had such symptoms at the time of their discharge from the service.

Goulston et al. (1985) studied gastrointestinal morbidity among 170 former Australian prisoners and 172 comparable controls. They reported that duodenal ulcers were significantly more prevalent among former POWs (24.7%) than among controls (10.5%), a finding that was confirmed by the observation that a higher proportion of POWs were taking cimetidine (9.0%) compared with controls (2.3%). They reported no other significant differences between the two groups in gastrointestinal disease.

Gill and Bell (1981) performed single-contrast barium meal examinations of all Far East POWs examined at Liverpool during 1968–1978 who complained of significant dyspepsia. They found evidence of duodenal ulceration in 48 patients, out of a total of 602 POWS examined, for a rate of 8.0%; an additional 6.8% had been successfully treated for duodenal ulcer in the past. These figures were thought to be high, compared with an earlier British estimate.

The examination data showed an appreciably higher lifetime prevalence of digestive disorders among PWK (Table 4.2) than among controls and

appreciably higher current prevalences for both PWP and PWK (Table 5.C.3). Comparison of PWEM and PWE lifetime prevalence data showed that gastroenteritis was appreciably more prevalent among PWEM (43%) than among PWE (28%). The analyses of Chapter 6 showed an appreciably higher prevalence of ulcer in PWK and of gastroenteritis in both PWEM and PWK, compared with controls; among all POWs combined, however, neither condition was appreciably associated with either weight loss or prison camp symptoms, nor were there any noteworthy correlations in the individual POW groups.

The logistic regression analysis, however, showed a noteworthy association of ulcer with both visual symptoms and with PWK status, and gastroenteritis had a noteworthy association with the reporting of other symptoms. Each additional reported visual symptom was estimated to increase the current prevalence of ulcer by 15%; all other things being equal, PWK had about 1.8 times as high a prevalence as the WW II POWs. As noted earlier for stroke and asthma, the reporting of visual symptoms is thought to be fairly specific evidence of vitamin A deficiency (vitamin A is necessary for the maintenance of epithelial tissue). Again, however, there is little other evidence for such an association, and any hypothesis linking vitamin A deprivation directly to subsequent ulcer 45 years later must be considered highly speculative. Likewise, there is no specific explanation for the noteworthy association of other symptoms and gastroenteritis, except that the "other symptom" category included reports of prison camp dysentery.

Cohen and Cooper (1954) reported higher rates of hospitalization in the first six years after repatriation for non-VD genitourinary diseases among PWP (6.5%) compared with WP (1%), but not in PWE (2%) compared with WE (3%). Beebe (1975) observed significantly higher hospitalization rates for genitourinary system disease in PWP than in controls, and, at the three-digit ICD level, significantly higher PWP rates of hospitalization for kidney infections and calculi of the kidney and ureter. Higher rates of genitourinary conditions among POWs have not been cited in the earlier reports of studies on British, Canadian, and Australian POWs, and Salmond et al. (1977) report a lower rate of hospitalization for New Zealand ex-prisoners (6.8%) than for ex-servicemen (8.2%).

The examination data showed an appreciably higher lifetime prevalence of urogenital conditions for PWE compared with WE (Table 4.2), but this is due to a low WE rate rather than a high PWE rate. There were no noteworthy differences in the current prevalence of urogenital conditions (Table 5.C.3). Given the lack of differences and no clear a priori hypotheses, no further analyses were undertaken.

Diseases of the skin resulted in more frequent hospitalizations for PWP than for WP in the six years following liberation—13% versus 3%—but not for PWE compared with WE—5% versus 4% (Cohen and Cooper, 1954), a finding that was later confirmed by Beebe (1975). Among the detailed diagnoses reported by Beebe, only the PWP hospitalization rates for boil and carbuncle were significantly elevated. Gill and Bell's (1980) report of persisting tropical diseases among former British POWs notes a high prevalence of "creeping eruption" skin rash (a result of infection with Strongyloidiasis stercoralis and thus discussed earlier under infectious diseases), of tropical ulcers, and, in one case, of spontaneously occurring keloid scar tissue. There is, however, little or no other mention of skin disease in the published reports cited earlier on Canadian and Australian ex-POWs; moreover, Salmond et al. (1977) report a rate of hospitalization for skin disorders that is lower for ex-prisoners (2.4%) than for ex-servicemen (3.4%).

The examination data showed no appreciably higher lifetime or current prevalence rates of skin conditions among POWs compared with controls (Tables 4.2 and 5.C.3). In the absence of specific hypotheses, no further analyses were undertaken.

The rate of hospitalization for diseases of bones and organs of movement was elevated among PWP (8.5%) compared with WP (2%) and somewhat elevated among PWE (3%) compared with WE (1.5%) (Cohen and Cooper, 1954); only PWP showed a significant excess in Beebe's follow-up (1975). The only detailed condition in this broad rubric with a significantly higher PWP hospitalization rate was osteoarthritis and allied conditions.

Richardson (1965) reports on the results of an x ray and clinical survey of disabilities of the cervical and lumbar spine among 96 Canadian prisoners of the Japanese and 96 of their brothers chosen as controls. The radiologic findings showed no appreciable difference between POWs and controls in osteoporosis, disk space narrowing, or osteophyte formation. There were, however, more reported neck and back troubles among the ex-prisoners, and this finding of more frequent symptomatic problems unaccompanied by radiologic evidence could not be satisfactorily explained. Because the study did not establish a statistically significant difference, Richardson concludes: "On the evidence available it is impossible to determine to what extent service-related factors account for this trend."

In a 1973 study comparing samples of New Zealand ex-prisoners, ex-servicemen who went overseas, and ex-homeservicemen who did not serve overseas, Salmond and colleagues (1977) found rates of bone and muscle

disorders under current medical supervision of 20.1%, 12.1%, and 8.1%, respectively. Rates of current disablement pensions, however, showed a much different pattern: 18.2% for ex-POWs, 11.9% for ex-servicemen, and 42.9% for ex-homeservicemen. Arthritis or back trouble was reported by 55.1% of ex-POWs, 47.2% of ex-servicemen, and 40.8% of ex-homeservicemen. At the time of their discharge from service, 69% of ex-POWs had such symptoms compared with 38% of ex-servicemen and 35% of ex-homeservicemen.

Oboler's (1987) recent report on ex-POWs examined in the Denver VA Medical Center showed a high prevalence of spinal arthritis in both European and Pacific prisoners; in the European group especially, this condition was related to parachuting from or landing with a disabled aircraft. In the European group, 52% (63 of 121 ex-POWs) had radiographic evidence of spinal arthritis, and 24% had evidence of degenerative arthritis that could be related to the effects of POW capture or captivity. Among Pacific prisoners, 68% (34 of 68) had current evidence of spinal osteoarthritis, with 46% having disease related to earlier POW confinement.

In the examination data, the lifetime prevalence of musculoskeletal disorders among POWs was quite high—87% to 94%; it was appreciably higher among PWK than WK (Table 4.2). Current prevalence rates (Table 5.C.3) of musculoskeletal conditions, however, were nearly the same as lifetime rates, and there were no noteworthy differences between POWs and controls. In Chapter 6, attention was narrowed to osteoarthritis and traumatic arthritis, but, again, there were no noteworthy differences in current prevalence rates (Table 6.1).

Osteoarthritis, however, showed a noteworthy association with weight loss in the group of combined POWs: it was present in 35% of POWs who reported a weight loss of 35% or less and in only 28% of POWs reporting a weight loss of more than 35%. This negative association of weight loss and osteoarthritis—more arthritis in those with less weight loss—was the only negative association observed among the conditions in Table 6.3; it was also seen in negative correlations between weight loss and osteoarthritis for both PWP and PWK in Table 6.5. Controlling for POW group, there was still some association of osteoarthritis with weight loss.

The logistic regression analysis confirmed these earlier indications, showing that the relative odds of having osteoarthritis (0.88) are smaller by about 12% for every additional 10% of body weight loss during captivity. As noted previously, the effects of additional weight loss are cumulative, and POWs who lost 40% of body weight, for example, have only 60% of the estimated prevalence of osteoarthritis of POWs who reported a weight loss of less than 10%.

This unusual finding was entirely unanticipated and difficult to explain. The data argue against its being a simple statistical artifact because the

prevalence of traumatic arthritis was higher in the high-weight-loss group, no doubt because weight loss was serving as a proxy for general harshness of treatment. Traumatic arthritis also showed small, but positive, correlations with weight loss and with prison camp symptoms, as expected. Thus, osteoarthritis, a clinical condition distinct from traumatic arthritis, showed a disparate pattern of association. The explanation for this finding remains unclear.

The medical conditions under these headings have been grouped together because they are all somewhat atypical. Congenital conditions should appear infrequently, if at all, because the military's medical entrance exam is meant to screen out disqualifying, preexisting conditions. Both symptoms and ill-defined conditions, as well as V-codes (i.e., factors influencing health status and contact with health services), are categories containing conditions that did not warrant full, detailed diagnostic coding. Medical conditions that are coded to the injury and poisoning category are likewise unusual because although the injury itself is assigned a code in this category (e.g., fracture of vertebral column), its sequelae are specified under another diagnostic rubric (e.g., traumatic arthritis). These conditions do, nevertheless, indicate an increased burden of illness, however ill defined.

Cohen and Cooper (1954) report appreciably higher hospitalization rates for wounds or injuries, including residual effects and their treatment, in both PWP (20%) and PWE (16%), compared with their respective control groups, WP (4%) and WE (5%). Beebe (1975) reported significantly higher hospitalization rates, compared with controls, for symptoms, for observation and examination only (corresponding most closely to the ICD-9-CM category of V-codes), and for "accidents, poisoning and violence" among PWP and PWK; an increased rate of hospitalization for accidents, poisoning, and violence was seen as well among PWE. There were no significant differences between POWs and controls for hospitalizations for congenital conditions.

Detailed diagnoses associated with POW status in Beebe's follow-up included upper gastrointestinal (GI) symptoms (PWP and PWE), abdominal and lower GI symptoms (PWP), symptoms referable to limbs and back (PWK), nervousness and debility (PWP and PWE), observation without medical care (PWP and PWK), and follow-up examination without medical care (PWP and PWK). Other diagnoses with statistically significant elevations among POWs were fracture of the femur (other than neck) (PWK), multiple

open wounds of the face and other sites (PWK), and effects of reduced temperature (PWK). Salmond et al. (1977) report a higher rate of hospitalization for wounds and accidents in New Zealand ex-prisoners (9.6%) than in ex-servicemen (6.8%), but ex-homeservicemen also had a high rate (9.3%). Rates of ill-defined and miscellaneous conditions were 3.4% for ex-servicemen, 2.0% for ex-prisoners, and 4.0% for ex-homeservicemen.

The exam data show low lifetime and current prevalence rates for congenital conditions, with no appreciable differences between POWs and controls—a somewhat reassuring finding. Lifetime and current rates of symptoms and ill-defined conditions are all quite high—97% or more in all but one group—and the only noteworthy difference is between PWK (lifetime rate of 99.3%) and WK (lifetime rate of 94.2%). Lifetime injury and poisoning rates are likewise high, although generally only in the low 90% range, and both PWE and PWEM rates are appreciably higher than the WE rate; here it may be more a reduced WE than increased PWE and PWEM rates that account for this difference. Current rates for injury and poisoning showed no appreciable differences between POWs and controls. Lifetime prevalence rates of V-codes were around 50% in all groups; there were no noteworthy differences. PWP had an appreciably lower prevalence of current V-code conditions than did WP, and there were no other noteworthy differences. The ill-defined nature of these findings and their presumed overlap with other, better defined medical conditions argue against further detailed analyses of these results.

In many instances, the organ-specific findings from this study based on medical examination data are familiar. The appreciably increased prevalence of depressive disorder, PTSD, and generalized anxiety, for example, is not unexpected. Similar findings regarding peripheral nerve disease, ulcer, and gastroenteritis are, likewise, not surprising. Even in these cases, however, there are some intriguing new data on a potential link with nutritional deficiencies, such as between ulcer and earlier visual symptoms (indicating vitamin A deficiency). The noteworthy association between current peripheral nerve disease and earlier edema, itself indicative of a previous vitamin B₁ deficiency, suggests that along with the well-known short-term neurological effects of (dry) beriberi, there may be persistent neurological effects decades after the original nutritional disease has been successfully treated and acute symptoms have abated.

The finding of an increased prevalence of schizophrenia among PWK is new, and an appreciable correlation with weight loss in this group offers further material for speculation. Because schizophrenia has not been linked with psychological trauma, the material basis for the observed association

(e.g., an organic brain syndrome associated with injury or nutritional deprivation) could well be something other than general ill treatment; the association may also be a statistical artifact. Findings of increased asthma and cerebrovascular disease in POWs who reported visual symptoms in prison camp are likewise new, and somewhat unexpected. Again, this is an instance of the identification of aftereffects of military captivity accompanied by evidence of a deficiency of vitamin A in prison camp. The findings concerning both intermittent claudication and arterial vascular disease appear for the first time in this cohort, and their associations with percent weight loss do not contribute much to an explanation. The last new finding, an appreciably lower prevalence of osteoarthritis in POWs who reported greater weight loss, is not only unanticipated but in the opposite direction of all the other findings in Chapter 6. No explanations for it come readily to mind.

Last, but certainly not least, is the finding of a noteworthy association between ischemic heart disease and earlier reporting of localized edema. Although there has been much interest in heart disease among former POWs, this result was not entirely expected, given the lack of noteworthy differences between POWs and controls and the fact that other studies have found conflicting evidence from both POW morbidity and mortality data. The lack of a clear biological mechanism linking nutritional deprivation and subsequent chronic heart disease requires that one remain somewhat skeptical of this finding of association, especially given the caveats noted earlier in this report regarding the low response rates. Nevertheless, localized edema is a noteworthy risk factor for only two current medical conditions in these POW examinations—peripheral nerve disease and ischemic heart disease—both of which are acutely related to thiamin deficiency, either the "dry" form of beriberi (peripheral nerve disease) or the "wet" form (cardiac problems). The specificity of association between localized edema and the only two medical conditions with well-established acute relationships to thiamin deficiency suggests that the association between earlier nutritional deprivation in prison camp and chronic ischemic heart disease is not an artifact.

Richardson (1965, p. 61) wrote in his earlier report that "the request for progressive increases [in pension] with advancing age would be more impressive if there were evidence of widespread or almost universal deterioration in the health of these veterans with the passing years. There is no evidence that this is the case, although there is evidence that as in any aging population there is an increasing number with serious disability." Excepting psychiatric illness, this report has shown little evidence of widespread ill health among former prisoners of war; even so, the number of specific medical conditions now posited as aftereffects of military captivity continues to grow.

Adena MA. 1989. The health of Australian veterans. Med. J. Austral. 150:356–357.

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.

Bell PG, O'Neill JC. 1947. Optic atrophy in Hong Kong prisoners of war. Can. Med. Assoc. J. 56:475–481.

Cohen BM, Cooper MZ. 1954. A Follow-up Study of World War II Prisoners of War. Veterans Administration Medical Monograph. Government Printing Office, Washington, DC.

Crawford JN, Reid JAG. 1947. Nutritional disease affecting Canadian troops held prisoner of war by the Japanese. Can. J. Res. 25(E):53–85.

Criqui MH, Fronek A, Barrett-Connor E, et al. 1985. The prevalence of peripheral arterial disease in a defined population. Circulation 71:510–515.

Cruickshank EK. 1961. Neuro-psychiatric disorders in prisoners-of-war. In Psychiatrie der Gegenwart, Band III. Springer-Verlag, pp. 807–836.

Dent OF, Tennant CC, Goulston KJ. 1987. Precursors of depression in World War II veterans 40 years after the war. J. Nerv. Menit. Dis. 175:486–490.

Dent OF, Richardson B, Wilson S, et al. 1989. Postwar mortality among Australian World War II prisoners of the Japanese. Med. J. Aust. 150:378–382.

Eberly RE, Engdahl BE. 1991. Prevalence of somatic and psychiatric disorders among former prisoners of war. Hosp. Commun. Psychiatry 42:807–813.

Eitinger L. 1964. Concentration Camp Survivors in Norway and Israel. Oslo University Press, Oslo; Allen and Unvin, London.

Engdahl BE, Page WF. 1991. Psychological effects of captivity. In Page WF (ed.), Epidemiology in Military and Veteran Populations, National Academy Press, Washington, DC, pp. 49–66.

Engdahl BE, Page WF, Miller TW. 1991. Age, education, maltreatment, and social support as predictors of chronic depression in former prisoners of war. Soc. Psychiatry Psychiatr. Epidemiol. 26:63–67.

Fowkes FGR, Housley E, Riemersma RA, et al. 1992. Smoking, lipids, glucose intolerance, and blood pressure as risk factors for peripheral atherosclerosis compared with ischemic heart disease in the Edinburgh artery study. Am. J. Epidemiol. 135:331–340.

Freed G, Stringer PB. 1968. Comparative mortality experience, 1946–1963, among former Australian prisoners of war of the Japanese. Aust. Repat. Dept. Med. Bull. 2:28.

Gibberd FB, Simmonds JP. 1980. Neurological disease in ex-Far-East prisoners of war. Lancet ii:135–137.

Gill G. 1983. Study of mortality and autopsy findings amongst former prisoners of the Japanese. J. R. Army Med. Corps 129:11–13.

Gill GV, Bell DR. 1980. Persisting tropical diseases amongst former prisoners of war of the Japanese. Practitioner 224:801–803.

Gill GV, Bell DR. 1981. The health of former prisoners of war of the Japanese. Practitioner 225:531–538.

Gill GV, Bell DR. 1982. Persisting nutritional neuropathy amongst former war prisoners. J. Neurol. Neurosurg. Psychiatry 45:861–865.

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.

Goulston KJ, Dent OF, Chapuis PH, et al. 1985. Gastrointestinal morbidity among World War II prisoners of war: 40 years on. Med. J. Aust. 143:6–10.

Helzer JE, Robins LN, McEvoy L. 1987. Post-traumatic stress disorder in the general population. N. Engl. J. Med. 317:1630–1634.

Hibbs RE. 1946. Beriberi in Japanese prison camp. Ann. Intern. Med. 25(2):270–282.

Hill JA. 1988. Strongyloidiasis in ex-Far East prisoners of war (letter). Br. Med. J. 296:753.

Hong CZ. 1986. Electrodiagnostic findings of persisting polyneuropathies due to previous nutritional deficiency in former prisoners of war. Electromyogr. Clin. Neurophysiol. 26:351–363.

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

Klonoff H, Clark C, 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, Magraw R. 1986. Forty-year follow-up of United States prisoners of war. Am. J. Psychiatry 143:1443–1445.

Kral VA, Pazder LH, Wigdor BT. 1967. Long-term effects of a prolonged stress experience. Can. Psychiatr. Assoc. J. 12:175–181.

Morgan HJ, Wright IS, van Ravensworth A. 1946. Health of repatriated prisoners of war from the Far East. J. Am. Med. Assoc. 130(15):995–999.

National Research Council. 1989. Diet and Health: Implications for Reducing Chronic Disease Risk. National Academy Press, Washington, DC.

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.

Oboler S. 1987. American prisoners of war—an overview. In Williams T (ed.), Post-Traumatic Stress Disorders: A Handbook for Clinicians. Disabled American Veterans, Cincinnati, Ohio, pp. 131–143.

Page WF, Engdahl BE, Eberly RE. 1991. Prevalence and correlates of depressive symptoms among former prisoners of war. J. Nerv. Ment. Dis. 179:670–677.

Patrick DL, Heaf PJD. 1981. Long-term Effects of War-Related Deprivation on Health: A Report on the Evidence. British Members Council of the World Veterans' Federation, London.

Pelletier LL, Baker CB, Gam AA, et al. 1988. Diagnosis and evaluation of treatment of chronic strongyloidiasis in ex-prisoners of war. J. Infect. Dis. 157:573–576.

Peters J, van Kammen DP, van Kammen WB, Neylan T. 1990. Sleep disturbances and computerized axial tomographic scan findings in former prisoners of war. Compr. Psychiatry 31:535–539.

Richardson HJ. 1965. Report of a study of disabilities and problems of Hong Kong veterans, 1964–1965. Canadian Pensions Commission, Ottawa.

Robins LN, Regier DA (eds.). 1991. Psychiatric Disorders in America: The Epidemiologic Catchment Area Study. Free Press, New York.

Roland CG, Shannon HS. 1991. Patterns of disease among World War II prisoners of the Japanese: hunger, weight loss, and deficiency diseases in two camps. J. Hist. Med. Allied Sci. 46:65–85.

Salmond GC, Geddes MK, Salmond CE. 1977. The Health of Former Servicemen. War Pensions Medical Research Trust Board, Wellington, New Zealand.

Speed N, Engdahl BE, Schwartz J, et al. 1989. Posttraumatic stress disorder as a consequence of the prisoner of war experience. J. Nerv. Ment. Dis. 177:147–153.

Strom A (ed.). 1968. Norwegian Concentration Camp Survivors. Oslo University Press, Oslo.

Sutker PB, Allain AN. 1991. MMPI profiles of veterans of WW II and Korea: comparisons of former POWs and combat survivors. Psychol. Rep. 68(1):279–284.

Sutker PB, Winstead DK, Galina ZH, Allain AN. 1990. Assessment of long-term psychosocial sequelae among POW survivors of the Korean conflict. J. Pers. Assess. 54(1&2): 170–180.

Sutker PB, Winstead DK, Galina ZH, Allain AN. 1991. Cognitive deficits and psychopathology among former prisoners of war and combat veterans of the Korean conflict. Am. J. Psych. 148:67–72.

Tennant C, Goulston K, Dent O. 1986a. Clinical psychiatric illness in prisoners of war of the Japanese: forty years after release. Psychol. Med. 16:833–839.

Tennant C, Goulston K, Dent O. 1986b. The psychological effects of being a prisoner of war: forty years after release. Am. J. Psychiatry 143:618–621.

Thygesen P, Hermann K, Willanger R. 1970. Concentration camp survivors in Denmark: persecution, disease, disability, compensation. Dan. Med. Bull. 17:65–108.

Trimble MR. 1985. Post-traumatic stress disorder: history of a concept. In Figley CR (ed.), Trauma and Its Wake: The Study of Post-Traumatic Stress Disorder, Brunner/Mazel, New York.

Ursano RJ, Rundell JR. 1990. The prisoner of war. Mil. Med. 155:176–180.

Wilson JD, Braunwald E, Isselbacher KJ, et al. 1991. Harrison's Principles of Internal Medicine, 12th ed. McGraw-Hill, Inc., New York.

Zeiss RA, Dickman HR. 1989. PTSD 40 years later: incidence and person-situation correlates in former POWs. J. Clin. Psychol. 45:80–87.