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 Medical Follow-up Agency (MFUA) of the Institute of Medicine, National Academy of Sciences, conducted a medical examination survey of former prisoners of war (POWs) of World War II (WW II) and the Korean conflict. This survey, which is part of a longitudinal follow-up study begun shortly after WW II, was the first to be based on medical examinations as well as on questionnaires and records for POWs and controls. The survey focuses solely on morbidity; because no deliberate efforts were made to collect complete, cause-specific mortality data, only brief, anecdotal mortality information is presented.

The goal of the research was to gather and analyze medical examination information from former POWs and comparable controls. The study design linked the MFUA's ongoing POW research and the Department of Veterans Affairs' (VA) POW protocol program to obtain information most efficiently. In brief, subjects in the MFUA study cohort were invited to a nearby VA medical center to undergo the VA protocol exam, a comprehensive physical and psychiatric examination conducted by VA medical personnel. In addition, a face-to-face psychiatric interview and a battery of psychological tests were given—to investigate the trends found in data from MFUA's 1984–1985 mail questionnaire survey on depressive symptoms.

Because the current follow-up is grounded in earlier work, a brief summary of previous study samples is warranted; further details of the sampling plan can be found in Chapter 1 and earlier reports. For the original study samples, MFUA used the Army's official roster of all known WW II prisoners of war to select independent samples of white male Army servicemen who had been captured in the Pacific theater and the European theater (later samples of Korean conflict POWs included other races; see below). Comparable control groups were drawn from Army payroll rosters; efforts were made to balance the proportion of air and ground personnel to correspond to the composition of the POW group. Later, these original samples were augmented with several others: a group of prisoners from the European theater who were sampled from POW admissions to Army hospitals for malnutrition immediately following repatriation, a group of Korean conflict prisoners of war, and a group of comparable nonprisoner controls, selected from a file of all known U.S. Army casualties wounded in action and returned to duty in Korea. The Korean conflict POW and control groups were not limited to whites.

The current survey continues the follow-up of the same sample of POWs and controls originally assembled some 40 years ago. A total of seven study groups were investigated; their abbreviated titles and number of eligible subjects are shown below:

PWP (Prisoners of War, Pacific theater, WW II), 670 subjects;

WP (War veterans, Pacific theater, WW II), 737 subjects;

PWE (Prisoners of War, European theater, WW II), 382 subjects;

PWEM (Prisoners of War, European theater, WW II, Malnourished), 258 subjects;

WE (War veterans, European theater, WW II), 383 subjects;

PWK (Prisoners of War, Korean conflict), 851 subjects; and

WK (War veterans, Korean conflict), 861 subjects.

The current study was charged to address the following questions.

1. Will rates of psychiatric illness, as ascertained by interview and psychological evaluation, be higher among former WW II prisoners of war of the Pacific theater (PWP) than among their nonprisoner controls? Will this also hold true for WW II prisoners of war of the European theater (PWE) and prisoners of war of the Korean conflict (PWK) when compared with their respective controls? Will rates of psychiatric illness be higher among PWP and PWK than among PWE, as observed in earlier studies?

2. What differences, if any, will there be between psychiatric illness assessed by interviewer and illness assessed by using the questionnaire? If there are differences, how will they influence the interpretation of results from (a)?

3. How have differences in illness levels changed over time? In particular, have the earlier differentials between the PWP and PWK groups, on the one hand, and the PWP and PWE groups, on the other, decreased with time?

4. How do the physical examination findings compare with the self-reported diagnoses, symptoms, and complaints from the 1984 questionnaire? Which physical findings are underreported or overreported, and how do nonmedical factors influence this reporting?

5. Is there a pattern of abnormal physical findings in the subset of PWE veterans who were seriously malnourished at repatriation?

Since March 1983, the Veterans Health Services and Research Administration (then the Department of Medicine and Surgery) of the VA has been conducting medical evaluations under a "POW protocol" program. The VA protocol includes a medical evaluation of standardized format, along with a standard questionnaire for collecting the POW's medical history. (This is a fairly detailed compilation of information on the POW's captivity, repatriation, and postwar adjustment.) The MFUA study was designed to use the VA's existing protocol program as the primary vehicle for data collection by simply inviting members of MFUA's longitudinal cohort to undergo a protocol examination at a nearby VA hospital. Thus, the VA's ongoing program was to do double duty—first, as an outreach program to all former POWs and second, as a data collection mechanism for a research study.

Such knowledgeable bodies as the VA's Advisory Committee on Former POWs have remarked on the value of the VA protocol exam program and the potential value of the data it collects, recommending that the information contained in the program's medical history and examination forms be made available for research use. Caution is required, however, in using these data. Although more than 30,000 examinations have been performed with the VA protocol, as a body they are limited in what they can contribute to our knowledge of POW health problems because they were gathered from a potentially biased sample about which very little is known—that is, the men who presented themselves for examination, a self-selected sample. In contrast, the members of the MFUA cohort were statistically sampled to be representative of WW II and Korean conflict POWs (and comparable veteran controls). If a large proportion of the MFUA cohort could be examined, they would presumably constitute a representative sample, the

findings from which could be used to develop sound inferences and generalizations regarding the whole population of former POWs.

Mention should be made here of what might best be called the "survivor bias." Obviously, only men who survived to respond to the examination invitation could provide data for the study; presumably, less healthy men would have died earlier, and morbidity findings might thus be under-represented. Explicit, cause-specific mortality information on this group is lacking; however, at the time of the last completed mortality follow-up in 1975, overall death rates for former POWs did not differ significantly from those of the U.S. male population of comparable age and color (i.e., white or nonwhite). Moreover, excess cause-specific mortality could be attributed to three causes: trauma, tuberculosis, and cirrhosis of the liver. In general, most of the differences between POW and general population mortality have lessened over time. Thus, there is no striking evidence from earlier mortality follow-ups that POWs die sooner than men in the general population.

Despite the careful construction of the MFUA cohort, it must be noted that its members were in a sense self-selected in that they chose to respond or not to respond to the invitation for examination. Unfortunately, response rates in the current study for both POWs and controls were very low: 40–50% among POWs and 10–14% among controls (see the findings section of this summary and also Chapter 3). However carefully the original sample was assembled, there can be no confidence with response rates as low as this that the group of respondents accurately reflects the composition of the group of all former POWs. Yet neither are such low response rates proof, in themselves, that the group of respondents is nonrepresentative. All that may rightly be said is that the statistical principles governing what one may infer from a sample—and the amount of confidence with which such inferences can be made–do not apply. A case for the representativeness of a sample based on such low response rates simply cannot be made.

Thus, the applicability of these results to the group of all POWs is unknown. There is some indication from the data comparing respondents and nonrespondents that large differences do not exist between actual respondents and all eligible subjects; nevertheless, the weight of these limited data cannot overcome the overwhelming importance of the low response rates. Consequently, the results of the study are presented as descriptive data, for two reasons.

First, even if the examination data from POWs are not strictly representative, they constitute an important case series—the largest national collec-

tion of POW examinations ever gathered together and analyzed. A number of findings in the series are worthy of note. Some confirm earlier research, and some are new; the confirmatory findings, at least, lend some weight to the validation of earlier results. New results, although they must be viewed cautiously, serve a valid scientific purpose in generating new hypotheses for more definitive studies.

The second reason to publish these data is not a scientific one. Although the report does not deal with nonscientific issues explicitly, it is recognized that the examination data will have uses beyond the scientific ones—for example, in providing material for discussion of military service-connected disabilities among former POWs. Despite the fact that sound inferences about the group of all former POWs cannot be drawn from the exam data in this report, policymakers who must deal with such issues should be able to review this descriptive information. The men who participated in these examinations also deserve an account of the process. Not to report the findings of the examination would surely raise more questions than the report, with its careful documentation of the study's limitations, would raise. The results of this examination study are thus discussed below, and we urge a maximum of reasonable caution in their interpretation and use.

The following provides general information about the study methodology (see Chapter 2 for further details). The use of VA protocol data made it possible to perform the current study, but it also introduced several complications. Because the VA protocol examination program began before the MFUA study, a sizable proportion of the POWs in the study had been examined previously by the VA. Knowing this, study designers matched the MFUA cohort against the VA's file of completed examinations, and lists of MFUA study POWs who had already been examined were produced and sent to VA medical centers. Copies of most of these examinations have found their way to MFUA; however, the situation became more complicated when the VA asked MFUA to invite these subjects to be reexamined in order to collect additional data.

Those subjects who had not been examined previously were handled somewhat differently. First, addresses were sought for POWs in each of the three groups in the study (WW II Pacific, WW II European, and Korean conflict) and their corresponding controls. (These addresses came from the VA itself, from the Internal Revenue Service [under a special arrangement with the National Institute for Occupational Safety and Health], and from a commercial tracing firm.) Letters of invitation were mailed, as many as three per person for each of two potential addresses, urging participation in

the VA medical evaluation protocol and requesting the veteran's permission to forward his name to the closest VA medical center for scheduling the examination. The VA medical centers then performed the examinations and forwarded copies of the results to MFUA. While data collection was under way in the field, MFUA staff made a number of site visits to VA medical facilities to observe the examination process.

The examination of members of the MFUA study cohort was identical to the usual VA protocol examination except that both POWs and controls underwent additional psychological testing and a more intensive psychiatric interview. The psychological testing consisted of four standardized psychological questionnaires (see Chapter 2); the usual VA psychiatric consultation, typically an unstructured part of the protocol exam, was augmented by including the posttraumatic stress disorder (PTSD) portion of the Structured Clinical Interview for DSM-III-R, Non-patient Version (SCID). These psychological tests and structured interviews provided a common framework for measuring and reporting psychological problems, which facilitated the comparison of MFUA study results with those of other research.

Toward the end of the study, it became clear that for administrative reasons, a number of examinations were being conducted without the required psychological testing. A decision was made to send blank psychological questionnaires to all potential subjects who had not completed them, regardless of examination status. Funding for this additional data collection (an initial mailing of some 2,000 questionnaires) was obtained from the National Institute of Mental Health (NIMH); the additional psychological data were added to the study's master file. A second questionnaire mailing occurred too late for its data to be included in this report.

Copies of the medical examinations sent by the VA medical centers were abstracted, coded, and computerized, as were the SCID and psychological questionnaires. Unfortunately, the overall rate of completed and coded examinations was disappointingly low—as noted earlier, around 40–50% in POWs and 10–14% in controls. Some of the large apparent difference between POW and control response rates is attributable to the fact that a number of POWs came in for examinations before the formal research program began. The rate of response to the supplemental psychological questionnaire mailing sponsored by NIMH was also low: 25–30% for POWs and only 10–25% for controls. (In general, response rates among controls have been lower than those for POWs in previous morbidity follow-ups.)

The low response rates raise justifiable concerns about potential nonresponse bias, and Chapter 3 focuses on differences in demographic data for subjects who completed the entire exam and whose exams were received,

abstracted, and analyzed by MFUA (''completed exams''), and all eligible subjects. Among POWs, there were no statistically significant differences between the two groups (respondents and nonrespondents) for year of birth, race, component (inductee or not), or marital status; PWP and PWK respondents, however, apparently had more education than nonrespondents, and WW II POW respondents were more apt to have served in the Army Air Corps. Data on VA hospitalization rates are displayed for three groups: subjects with completed exams, all subjects who responded to the invitation to participate (including those with completed exams), and nonrespondents. The data cover 1969–1985, predating the examination survey, and so provide evidence about differences among the three groups. There were no significant overall differences between respondents' and nonrespondents' VA hospitalization rates in any of the POW or control groups. Consequently, although the low response rates prevent any consideration of the exam data as representative, the demographic data and VA hospitalization data at least suggest that there are no overwhelming differences between respondents and nonrespondents in background characteristics or in morbidity as measured by 26 years of VA hospitalization data.

Chapter 4 of the report presents data from the 1,067 coded examinations elicited by the study. These exams provide diagnostic data on more than 65,000 coded medical conditions and are used to calculate "lifetime" prevalence rates, the probability that a given person has ever had (or still has) some specified disease, up through the time of examination. POWs were shown to have high lifetime prevalences of a number of medical conditions, but the most striking findings pertained to psychiatric illness.

Lifetime prevalence rates for selected conditions were compared with similar rates from Eberly and Engdahl's 1991 study of POW examinations at the Minneapolis VA Medical Center. The MFUA data showed elevated rates of prevalence among POWs for medical conditions related to prison camp treatment—for example, dysentery, malaria, and frozen feet—as well as for psychiatric illness. Compared with the Minneapolis POWs, the MFUA POWs had higher lifetime prevalence rates of several medical conditions, a higher rate of depressive disorder, a lower rate of PTSD, and a roughly equal rate of generalized anxiety disorder.

In Chapter 5, the discussion addresses the five basic questions listed earlier. The answer to question A, which asked whether psychiatric illness was more extensive in POWs than in controls, was that the data showed strikingly higher rates for several conditions, although no statistical tests could appropriately be applied. In particular, the rates for PTSD were appreciably higher among POWs when measured independently in each of three different ways: examination, structured clinical interview (SCID), or questionnaire (the Mississippi scale). Rates of depressive disorder were similarly elevated, again, when measured independently in three ways: ex-

amination, Center for Epidemiologic Studies depression (CES-D) questionnaire, or Beck questionnaire. The finding of a higher rate of PTSD was particularly important, given that the 1984–1985 survey had stimulated the hypothesis that PTSD was a possible psychiatric diagnosis underlying an increased rate of depressive symptoms or a comorbid condition. Additional data from the 90-item Hopkins Symptom Check List (SCL-90), however, suggested that psychiatric illness was not necessarily limited to PTSD and depressive disorder, and that there are signs of an appreciable increase in general psychopathology among POWs.

Question B asked whether there were important differences among measurement instruments for psychiatric illness. The answer for this case series was that there were appreciable differences among the various ways of measuring PTSD and depressive disorder. For PTSD, the exam and structured clinical interview—which were generally performed by the same person—agreed much more closely with each other than with the questionnaire; for depressive disorder, one scale (the CES-D) gave higher estimates of depressive disorder while the other, the Beck, gave lower estimates. Every measurement method, however, found appreciable differences between POWs and controls.

Question C asked whether earlier hospitalization rates were comparable to recent rates of self-reported hospitalization. In general, the data showed far fewer differences between POWs and controls now than in 1967. A comparison of lifetime and current prevalence rates from the current exam also showed that morbidity differences appear to have lessened over time.

Question D asked whether examination and earlier self-report questionnaire data provided comparable estimates of the prevalence of illness. Among those who completed the examinations, medical conditions were self-reported by questionnaire much less frequently than they were noted during the examination; typically, self-reported prevalence was only one-half to one-tenth the examination-based value. Some of this disparity appeared to be due to limitations in the questionnaire design (e.g., a fixed number of blanks for responses) as well as to differences between physicians and questionnaire respondents in the use of medical terminology.

Question E addressed differences between the special, albeit small, subset of severely malnourished European prisoners, PWEM, and their European theater counterparts, PWE. The case series of PWEM who were examined displayed a markedly higher lifetime prevalence of dysentery, beriberi, frozen feet, peripheral nerve disease, and gastroenteritis than was found among PWE. The latter two conditions, which were not as clearly related to war camp treatment as the others, were studied in further diagnostic detail. Excess peripheral nerve disease was attributable to excesses in three specific categories: mononeuritis of the upper limb, mononeuritis of the lower limb, and hereditary and idiopathic neuropathy. The higher preva-

lence of gastroenteritis was accounted for almost entirely by noninfectious gastroenteritis and colitis, rather than by irritable bowel syndrome, a disease that has already been presumptively linked to military captivity.

The analyses discussed in Chapter 5 provide evidence of psychological aftereffects of military captivity as long as 45 years after repatriation but less striking results concerning other aftereffects. Chapter 6 presents a different sort of descriptive analyses, undertaken to complement those of Chapter 5, that focused on a specific set of medical conditions taken from published reports of POW examinations at two VA medical centers, Minneapolis and Denver. Preliminary analyses of these conditions showed higher current prevalence rates among POWs for ischemic heart disease (PWEM only), peripheral nerve disease (PWP and PWK), ulcer (PWK only), gastroenteritis (PWEM and PWK), depressive disorder (PWP, PWE, PWEM, and PWK), posttraumatic stress disorder (PWP, PWEM, and PWK), and generalized anxiety disorder (PWP, PWE, PWEM, and PWK); PWP rates for malaria were appreciably lower than control rates. These conditions, with the exception of malaria, were included in the final stage of analysis. The entire list of selected conditions was also subjected to another set of preliminary analyses; these studies focused on determining whether the prevalence of any of the conditions was associated with the specific factors that measured harshness of treatment as a POW, such as percentage of body weight loss. All medical and psychiatric conditions with findings of either a substantially higher prevalence among some POW group or a marked association with some military captivity factor were included in the final stage of analysis.

In this stage, the prevalence data for these conditions were reanalyzed using logistic regression to determine the joint effects of weight loss and war camp symptoms, taking into account POW group differences. Prison camp symptoms were further refined into three separate measures—the presence of edema, the number of visual symptoms (such as night blindness) reported, and the number of other symptoms reported—and considered along with percent weight loss. The logistic regression analyses showed that among POWs with completed exams and earlier data on weight loss and prison camp symptoms, edema was significantly associated with a higher prevalence of ischemic heart disease and peripheral nerve disease; visual symptoms were associated with higher prevalences of cerebrovascular disease, ulcers, asthma, and PTSD; and other symptoms were associated with higher prevalences of intermittent claudication, gastroenteritis, depressive disorder, and generalized anxiety. Percent weight loss was markedly associated with a higher prevalence of intermittent claudication and arterial vascular disease and strongly associated with a lower prevalence of osteoarthritis. In most cases the odds ratios, which estimate the size of the increased prevalences, were between 1.0 and 2.0, indicating relatively moderate elevations

in prevalence; however, the estimated effects of visual and other symptoms increased gradually with each additional reported symptom, so that the cumulative effect on the relative odds for POWs with a large number of symptoms would be quite substantial. In addition to these findings, PWP showed a higher prevalence of PTSD, compared with PWE, and among PWK, again compared with PWE, there were higher prevalences of ulcer, PTSD, schizophrenia, and generalized anxiety, as well as a lower prevalence of ischemic heart disease, arterial vascular disease, and asthma; these latter would appear to be age related. In general, the logistic regression analyses linking current medical conditions to earlier prison camp factors suggest more medical conditions related to the POW experience than did the simpler descriptive comparisons between POWs and controls.

After a summary of results in Chapter 7, Chapter 8 presents a detailed discussion of the study findings, comparing them with other research conducted among American and foreign POWs. In many cases, the organspecific findings based on medical examination data were anticipated. The increased prevalence of depressive disorder, PTSD, and generalized anxiety was not unexpected, and similar findings regarding peripheral nerve disease, ulcer, and gastroenteritis likewise were not surprising. Even in these cases, however, some intriguing new data surfaced on a potential link with nutritional deficiencies, such as between ulcer and earlier visual symptoms (indicating vitamin A deficiency). The link between current peripheral nerve disease and earlier edema, indicative of a prison camp vitamin B1 deficiency, was evidence that the well-known short-term neurological effects of (dry) beriberi are associated with persistent neurological symptoms that last for decades, even after the original nutritional disease has been successfully treated and acute symptoms have abated.

The finding of increased prevalence of schizophrenia among PWK in the case series was a new one, and a significant correlation with weight loss in this group offers material for speculation. Possible explanations include the hypothesis that organic brain syndrome arising from injury or malnutrition was incorrectly diagnosed as schizophrenia or that the association was a statistical artifact; settling the matter would require further research. Also new, and somewhat unexpected, were findings of increased asthma and cerebrovascular disease in POWs who reported visual symptoms in prison camp. Again, this involved the identification of an aftereffect of military captivity accompanied by evidence of a deficiency of vitamin A in prison camp. Findings concerning both intermittent claudication and arterial vascular disease appeared for the first time in this cohort, but their associations with percent weight loss do not provide much material to aid explanation. The last new finding, a significantly lower prevalence of osteoarthritis in POWs who reported higher weight loss in captivity, was not only unantici-

pated but was in the opposite direction of all other findings in Chapter 6. No explanations for it come readily to mind.

Last, but certainly not least, was the finding of a statistically significant association between ischemic heart disease and earlier reporting of localized edema. Although this finding is not exactly new, it was also not entirely expected, given that earlier studies have found conflicting evidence in both POW morbidity and mortality data. The most interesting aspect of this finding may be its specificity: localized edema was a significant risk factor for only two current medical conditions—peripheral nerve disease and ischemic heart disease—both of which were 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 raises intriguing questions about the association between earlier nutritional deprivation in prison camp and subsequent chronic ischemic heart disease—even in the absence of a satisfactory explanatory biological mechanism to link the two.

In summary, excepting psychiatric illness, this report shows little evidence of widespread ill health among former prisoners of war compared with their non-POW veteran counterparts. Nevertheless, analyses of the associations between prison camp factors and subsequent disease prevalence have uncovered a number of medical conditions that can be posited as aftereffects of military captivity.