Military-related experiences and late-life depressive symptomatology among Japanese-American world war II veterans.

Objectives: To examine military service-related variables and late-life depressive symptomatology among older Japanese-American males. Method: This study is a secondary data analysis of a longitudinal, community-based study. A sample of 2669 participants (771 World War II veterans, 1898 civilians) was drawn from the Honolulu-Asia Aging Study. Depressive symptoms were assessed twice across a 9-year period with the Center for Epidemiologic Studies-Depression scale. Covariates included sociodemographic, physical health, health behavior, and psychosocial variables. Combat exposure and symptomatology were examined among a subset of 426 veterans. Cross-sectional and longitudinal designs were analyzed with linear regression. Results: Veterans and civilians did not differ in depression scores. Baseline depression scores significantly predicted follow-up depression scores. For the full sample, lower ratings of quality of life satisfaction, daily activity control and general health were associated with higher depression scores both cross-sectionally and longitudinally. Among veterans, light combat exposure was marginally associated with lower depression scores and longitudinally, previous depression scores and poorer health ratings were significant predictors of depression scores. Conclusion: Results suggest that military service does not affect late-life depressive symptomatology. However, combat exposure may play a marginal role in increased symptoms. Reasons for results include the possibility that other factors are more relevant to late-life depression, symptomatology naturally decreasing over time, or type of combat exposure measurement. Results expand literature by examination of an ethnoracial group not studied often and longitudinal examination of late-life depressive symptoms within a military-related context. Stakeholders should be knowledgeable of the distinct issues presented when serving aging veterans.

Keywords: Japanese-American men; late-life depression; military service; older adults; World War II veterans

Age complicates current understanding of late-life depression. Mirowsky and Ross ([24]) outlined several reasons why depressive symptoms can either increase or decrease with age. For instance, conceptualizing age as a decline suggests that like physiological decline, emotional well-being also might decline with age. However, age alone does not appear to be a significant predictor of late-life depression and age effects often disappear when controlling for other variables (e.g. medical comorbidities, functional impairments; Blazer, [1]; Yang, [41]).

In addition to age, other reasons contribute to limited understanding of late-life depression. One reason is that early-life experiences might affect late-life depression (Davidson et al., [6]). A noted early adult experience is military service, which is generally associated with increased likelihood of mental health difficulties among veterans (Prigerson, Maciejewski, & Rosenheck, [28]). Although the impact of military service on late-life depression is not well-understood, research is increasing (e.g., Boakye et al., [2]; Gould, Rideaux, Spira, & Beaudreau, [12]; Pietrzak & Cook, [26]; Yang & Burr, [40]). Studies suggested that service era was a stronger predictor of depressive symptoms among older veterans over military service alone (Boakye et al., [2]; Gould et al., [12]); likely due to social and political circumstances surrounding different wars (e.g., Vietnam vs. WW II). Additionally, Yang and Burr ([40]) found that combat exposure became non-significant after controlling for difficulties with health and activities of daily living.

A second reason for limited understanding of late-life depression is research design. Many studies use a cross-sectional approach, which might introduce a cohort effect, obscuring a true age effect (Wu, Schimmele, & Chappell, [39]). Thus, use of a longitudinal approach can address this shortcoming and isolate possible age effects (Wu et al., [39]). Because of the associations between other factors (e.g., medical illnesses) and current depression in late life, tracking these factors may help determine how they, along with age and military service, might affect depressive symptoms differently over time. Studies have examined depressive symptoms longitudinally (e.g., Wu et al., [39]; Yang, [41]) but did not include military-related variables.

Examination of veterans can address these issues and late-life depression is a relevant and timely topic within the field. With 19 million veterans (U.S. Census Bureau, [34]) and counting, the U.S. Department of Veterans Affairs (VA) must continuously identify and plan for veterans' needs to provide appropriate and effective care. Alongside the anticipated growth, the group composition is changing. Notably, in 2016 the median age of male veterans was 65, meaning half of veterans are aged over 65 (U.S. Department of Veterans Affairs, National Center for Veterans Analysis and Statistics, [36]). As such, the VA must meet the needs of the aging veteran with respect to new issues or long-term difficulties stemming from military service years or decades earlier.

Further, the number of minority veterans is growing. In 2014, minorities comprised 22% of the veteran population. By 2040, minority groups are projected to make up approximately 35% of the living veteran population (U.S. Department of Veterans Affairs, [35]). Asian-American veterans are the largest group of veterans over the age of 75. VA benefit or service use by minority veterans has also increased from 2005, a trend expected to continue as the number of minority veterans grows.

A key need addressed by the VA is mental health care (Preston, [27]). A recent analysis of VA health care utilization by veterans of recent conflicts indicated that depressive disorders are among the top three most prevalent mental health diagnoses (Epidemiology Program, [9]). This is consistent with previous research documenting the prevalence and negative outcomes associated with depression among World War II, Korean Conflict, and Vietnam-era veterans in mid-life (Eisen et al., [8]; Hyer, Stanger, & Boudewyns, [15]; Marmar et al., [23]). Insight into long-term or late-onset issues or changes in depressive symptomatology, related to military service may inform the VA on the type and amount of services needed among older veterans (Pietrzak & Cook, [26]).

The current study seeks to expand previous research by using data from a longitudinal, health-related study to examine depressive symptomatology among a community-based group of older males of Japanese descent from Hawaii. To our knowledge, this is the first study to examine whether a community sample of older veterans and civilians differ in depressive symptomatology across time. This study is unique in that it will address the aging veteran population as well as investigating a minority group of veterans not often studied.

We investigate the following four questions:

• How does WW II military service affect late-life depressive symptomatology among veterans and civilians?
• How does WW II combat exposure affect late-life depressive symptomatology among veterans?
• How does WW II military service affect longitudinal changes in depressive symptomatology among veterans and civilians?
• How does WW II combat exposure affect longitudinal changes in depressive symptomatology among veterans?

The current study employs data from the Honolulu Heart Program (HHP) and the Honolulu-Asia Aging Study (HAAS). This report is a secondary analysis of longitudinal, epidemiologic HHP data (1965–1975, 3 examinations) that assessed heart disease among men of Japanese descent living in Hawaii (N = 8006; White et al., [38]). In 1991, the study continued as the HAAS with approximately 80% of the original 1965 sample. The HAAS examined dementia and associated variables (Gelber, Launer, & White, [11]) and continued through 2012 with examinations every 2–3 years. The first HAAS exam in 1991 (hereafter referred to as Exam 4 and subsequent exams termed sequentially) served as baseline for the current study. Participants were included in the current study if they were born between 1910 and 1919, had non-missing veteran status, completed the depression assessment at Exam 4, and were cognitively intact via an Exam 4 Cognitive Abilities Screen Instrument (CASI) score > = 60.0. To be included in the longitudinal analysis, participants had to meet the aforementioned criteria as well as completion of depression assessment at Exam 7 and scored at least 60.0 on the CASI administered at Exam 7.

The fourth exam (1991–1994) served as baseline for the current study. A total of 2669 participants were included in the analysis at Exam 4 (Table 1). Average age at Exam 4 for all participants was 76.4 years (SD = 2.7) and number of children was 2.6 (SD = 1.5). Most participants were married (84.9%). A subset of 426 veterans was used for the combat exposure analysis at Exam 4.

Table 1. Baseline characteristics at Exam 4 and CES-D scores at Exams 4 and 7.

1 N = 426.

A total of 1231 participants completed both Exam 4 and 7 (1999–2000) and were included in the CES-D longitudinal analysis. Average age at Exam 7 was 83.7 (SD = 2.6). A total of 1438 participants did not meet inclusion criteria for the longitudinal analysis. Attrition analyses identified that 28% of the baseline sample died between the end of Exam 4 and the end of Exam 7, 18% did not complete Exam 7, and 7% had incomplete information. Finally, a subset of 331 veterans was analyzed for combat exposure and depressive symptom change.

The main outcome was level of depressive symptomatology measured by the 11-item version of the Center for Epidemiologic Studies Depression (CES-D) scale. Originally 20 items, the CES-D is a self-report questionnaire used to measure depressive symptomatology in the general population (Radloff, [29]). Items are scored on a 0–3 scale, which are added for a total score, with higher scores indicating presence of more depressive symptoms. The use of a shortened form is comparable to the full-scale version (Covinsky et al., [5]; Kohout, Berkman, Evans, & Cornoni-Huntley, [19]).

The CES-D was administered at Exams 4 and 7 (1999–2000). Participants with at least 9 items completed at each exam were included in the analysis and missing item scores were derived by using that individual's mean item score to replace the missing values. At Exam 4, mean scores (SD) did not differ between veterans 3.7 (3.7) and civilians 3.6 (3.5). Ranges for each group were 0–21 and 0–26 for veterans and civilians, respectively. Hereafter, we will refer to elevated depressive symptomatology as depression for brevity, but we emphasize the presence of elevated depressive symptomatology, not a Depressive Episode or Disorder.

Variable information was garnered through self-report with standard questions at Exam 4 unless otherwise noted.

WW II military service status was derived from information gathered from two exams. At Exam 6 (1997–1999), participants with military service were administered the Combat Exposure Scale (CES), a 7-item scale assessing combat exposure (Keane et al., [17]). Total score (range 0–41) is divided into 5 categories, indicating severity of combat exposure. For current analyses, categories were collapsed into 3 categories: no combat exposure, light exposure (1–16) and moderate to severe exposure (17+).

Sociodemographic variables included age at Exam 4, highest education level (less than high school, high school, and more than high school) and yearly income (<$15k, $15–20k, $20–29k, $30 + k). Based on previous research (Zivin et al., [42]), we included the following health-related variables: BMI, difficulty with instrumental activities of daily living ([IADLs]; including finances, telephone use, shopping and preparing meals), divided into two categories (0, 1+) and presence of chronic illnesses (diabetes, hypertension, chronic heart disease, emphysema, cancer), divided into three categories (0, 1, 2+). Health Behaviors/Lifestyle factors included regular exercise and any previous or current use of cigarettes and alcohol. These variables were dichotomized into yes/no responses. Psychosocial factors included control over daily activities (a lot vs. some/very little/none), current general health rating (excellent/good vs. fair/poor), and quality of life (very satisfied vs. satisfied/reasonably satisfied/dissatisfied). The categorical variable responses were collapsed as described to ensure adequate numbers for statistical analyses and for ease of interpretation.

Linear regression was used to analyze the effects of predictor variables on continuous depression scores. Results were replicated using a zero-inflated Poisson model and results were comparable; because readers are more likely familiar with linear regression, those findings are reported. Due to the limited amount of missing data, a pair wise deletion method was used to handle missing control variable data. Ten or fewer cases had missing data on each variable except for income, which was missing 178 cases. Bootstrapping with 1000 samples was used to generate 95% CI and adjusted R-square results are reported. Analyses were completed using SPSS (Version 25).

The overall model was statistically significant (F[17, 2470] = 14.89, p ≤ 0.001) and accounted for 9.4% of the variance in CES-D scores. Military service was not significantly associated with depression scores at Exam 4 (Table 2). However, several variables were significant. Variables associated with higher depression scores included a high school education (β = 0.35; 95% CI [0.03–0.66]) and at least one IADL difficulty (β = 0.74; 95% CI [0.33–1.16]). Further, participants who reported poorer psychosocial functioning had higher depression scores, which included lower satisfaction with quality of life (β = 1.04; 95% CI [0.73–1.36]), less control over daily activities (β = 0.94; 95% CI [0.63–1.24]) and poorer general health rating (β = 0.93; 95% [0.62–1.24]). On the other hand, those who regularly exercised (β = −0.40; 95% CI [−0.70 to −0.09]), those with higher BMI (β = −0.07; 95% CI [−0.12 to −0.03]) and those who earned $20,000 to $29,999 (β = −0.63; 95% CI [−0.92 to −0.09]) and over $30,000 (β = −0.62; 95% CI [−1.04 to −0.23]) had lower depression scores.

Table 2. Full model regression results for depression scores at Exam 4 for military service and combat exposure.

• 2 Note.†≤0.10; *≤0.05; **≤0.01; ***≤.001.
• 3 N = 426.

The overall model analyzing combat exposure and depression among veterans at Exam 4 was statistically significant (F[18, 402] = 1.87, p = 0.017) and accounted for 3.8% of the variance in CES-D scores. Level of combat exposure was not significantly related to depression scores; though the effect of light combat exposure compared to no combat exposure approached statistical significance (see Table 2). The only variable predicting higher levels of depression was poorer life satisfaction ratings (β = 0.74; 95% [0.03–1.44]). No protective factors were identified in this subsample.

The next analysis examined CES-D at Exam 7, controlling for scores at Exam 4. Overall, most participants reported low depression scores at Exam 7. The overall model for predicting depression scores at Exam 7 among veterans and civilians was statistically significant (F[18, 1225] = 6.61, p <.001) and accounted for 7.1% of the variance in CES-D scores. Three variables predicted higher depression scores at Exam 7. The strongest predictor was CES-D scores at Exam 4 (β = 0.25; 95% CI [0.18–0.31]). Also, those with at least some college reported higher depression scores (β = 0.60; 95% CI [0.01–1.20]). Finally, those who reported poorer general health had higher depression scores (β = 1.11; 95% CI [0.62–1.60]). No protective factors were identified.

This analysis examined changes in CES-D scores among veterans with and without combat exposure (Table 3). The overall model was statistically significant (F[19, 319] = 1.91, p = 0.013) and accounted for 5.1% of the variance in CES-D scores. Similar to the results for Question 2, level of combat exposure was not related to depressive symptom levels. Two variables predicted higher depression scores: higher depression scores at Exam 4 (β = 0.24; 95% CI [0.12–0.37]) and poor general health ratings (β = 0.96; 95% CI [0.04–1.89]). No protective factors were found in this subset.

Table 3. Full model logistic regression results for depression scores at Exam 7 for military service and combat exposure.

• 4 Note.†≤0.10; *≤0.05; **≤0.01; ***≤.001.
• 5 N = 331.

The current study examined questions regarding WW II military service-related variables and depressive symptoms among a Hawaii-based community sample of older males of Japanese descent. Overall, rates of elevated depression were low. This was consistent with other samples of older adult men (Gould et al., [12]). Similar to previous work we found that most participants reported few depressive symptoms across time (Choi & Bohman, [3]).

This study examined the impact of WW II military service on baseline depression symptoms measured in late life (Question 1) and changes in depression scores over time (Question 3). Results indicated that early life military service was not significantly associated with late-life depressive symptoms or changes in depression scores. These findings were similar to recent studies that compared WW II veterans and nonveterans (Boakye et al., [2]; Gould et al., [12]). The current study expands upon those results through inclusion of combat exposure, examination of an ethnoracial group not often studied, inclusion of psychosocial constructs as independent variables, and use of a longitudinal approach.

Several reasons may explain current findings. One, the relationship between depression and military service may wane over time. In the current cohort, participants were asked about depression approximately 50 years after the U.S. became involved in WW II, whereas other studies examined depression soon after deployment return or military separation. Current findings are similar to a study that found veterans of more distant wars (WW II and Korean) were less likely to report current mental distress than civilians of the same period, and veterans of more recent wars (Vietnam) were more likely to report current mental distress compared to civilians (Boakye et al., [2]). However, the sociopolitical characteristics of the war also may have played a role in observed differences; WW II was a different war from others, particularly Vietnam. Life after returning home may have been more supportive for WW II veterans than Vietnam veterans (Boakye et al., [2]). Finally, several sample characteristics could contribute to the lack of differences between military veterans and civilians. The overall level of depressive symptomatology was low among the current sample, which may reflect unique characteristics such as resilience, better life-long health, and psychiatric stability, and who were willing to participate in the original longitudinal research project.

Another reason why military service was non-significant is that among older adults, other variables might be more integral to depression, with health-related issues cited most frequently (e.g. Luppa, Luck, König, Angermeyer, & Riedel-Heller, [22]; Sutin et al., [32]; Vink, Aartsen, & Schoevers, [37]; Wu et al., [39]). Current results suggest that presence of one IADL difficulty was significantly associated with depression at Exam 4. However, IADL difficulty did not significantly predict Exam 7 depression scores. Previous research among older adults found that depression increased leading up to and 1.5 years after onset of disability; after that symptoms did not increase (Fauth, Gerstorf, Ram, & Malmberg, [10]). Thus, the non-significant relationship between depression scores and physical health variables at Exam 7 could be attributed to adaptation of the disability after Exam 4. It is also possible that participants received physical or mental health treatment between the exams, which affected the role of physical health in depression scores at Exam 7.

High school education was significantly associated with higher depression scores at Exam 4 and having at least some college significantly predicted higher scores at Exam 7. This is in contrast to research that suggests higher education is a protective factor against depression in older age (Djernes, [7]; Sutin et al., [32]). Perhaps this finding is unique to Japanese-American men or is related to Western acculturation and conceptualization of depression, relationships that warrant further study.

At Exam 4, poorer ratings of all three psychosocial variables were significantly related to higher depression scores at Exam 4. This is consistent with previous research that suggested subjective ratings of current well-being and perceptions of one's functioning are relevant to depression (Djernes, [7]; Luppa et al., [22]; Vink et al., [37]). However, at Exam 7, only the general health rating variable at Exam 4 was associated with depression scores, suggesting a longitudinal relationship with depression. However, we note that the directional relationship between these constructs and depression is unknown. Participants could have been depressed prior to Exam 4, which affected health ratings, or poorer feelings about health status contributed to concurrent and future depression.

The strongest predictor of depression scores at Exam 7 was depression scores at Exam 4. This is consistent with previous research that suggests pre-existing mental health difficulties are risk factors of later issues (Cole & Dendukuri, [4]; Djernes, [7]). The current analyses could not determine whether current assessments indicated new onset, recurrence or a chronic issue with depression. While we note that current results revealed relatively low depression scores at both Exams, we encourage future studies to consider these patterns of depression.

Results for associations of WW II combat exposure with late-life depression were mixed. In these veterans-only analyses, compared to veterans with no combat exposure, those with light level of combat exposure were marginally less likely to report depression at Exam 4. But at Exam 7 light combat exposure was no longer significant.

The marginal significant effect of light combat on being less likely to report higher depression scores at Exam 4 is consistent with findings in the current literature (Lee, Vaillant, Torrey, & Elder, [21]; Yang & Burr, [40]). Possible reasons for these findings exist. First, as previously mentioned, sociopolitical circumstances of the war under study may explain decreased likelihood for depression (e.g. Yang & Burr, [40]). Also, as described by Prigerson et al., [28], the long-term effects of combat exposure on the occurrence of depression may be indirect and mediated through other conditions, such as posttraumatic stress disorder (not assessed in this study).

Second, results may be due to different approaches to combat exposure measurement. The current study used the CES, which captures several aspects of combat whereas other studies used one yes/no item (e.g. Pietrzak & Cook, [26]; Yang & Burr, [40]). The CES offers a deeper profile of combat exposure and may have captured the experiences that contribute to depression even after controlling for other variables.

Third, differing results may also be due to possible positive impacts of combat exposure. Those who experienced combat might develop coping skills that transfer to civilian life (Pietrzak & Cook, [26]). For the current sample, perhaps veterans with light levels of combat developed and used coping skills, but for veterans with more severe exposure, engagement in severe combat during service preempted any opportunity to develop effective coping skills. Further, other research suggests the relationship between combat exposure and well-being in later life is mediated by perceived positive aspects of military service (Lee, Aldwin, Choun, & Spiro, [20]). Perhaps veterans in the current study judged their military experience positively, which decreased likelihood of depression at Exam 4. Also, as likely members of the segregated Japanese-American military service groups of WW II, these veterans faced great pressure to be uphold the honor and represent the dignity of their cultural groups, which may inhibit expression of any weaknesses.

Similar to the full sample, depression scores at Exam 4 significantly predicted depression scores at Exam 7 among veterans, suggesting that previous depression scores are a robust predictor of later depression scores. Again, we encourage further research into factors related to onset, recurrence and chronicity as proper treatment and/or prevention is a critical concern for veterans. This is particularly important for veterans who have encountered combat exposure as these veterans often have higher rates of mental health difficulties (Thomas et al., [33]).

Finally, psychosocial variables were significant in both veteran analyses. Veterans who reported poorer satisfaction with quality of life were more likely to report higher depression scores at Exam 4 and poor health ratings at Exam 4 significantly predicted higher depression scores at Exam 7. These results are consistent with research linking perceived health, functional limitations, negative self-perceptions about aging, and negative cognitive appraisals as intervening variables between physical illness and depressive symptoms (Han, [13]; Norton et al., [25]; Schnittker, [31]).

Strengths of the current study include the use of a large community sample of Japanese-American men to examine military service and late-life depressive symptoms longitudinally. The use of a community sample may provide insight to the VA into what community veterans need, how to provide outreach, and how to plan and evaluate programs and encourage veterans to use VA care (Boakye et al., [2]). Current analysis used a longitudinal approach across nine years and examined military-related variables, whereas previous studies used 2 years (e.g., Choi & Bohman, [3]). To our knowledge, this is the first study to examine this topic among older adults. We included many control variables to capture the influences of a variety of factors related to late-life depressive symptomatology.

The study has some limitations. First, no history of mental health in early or middle adulthood, especially prior depression was available. Moreover, becoming a participant in the original HHP study required acceptance of an invitation for life-long participation in a health-related federal research project. Previous reviews identified previous mental health issues as a risk factor among elderly community adults (Cole & Dendukuri, [4]). Knowing if problems are new or ongoing can direct type of intervention or guide prevention efforts. Also, depressive symptoms may be biased due to under-reporting of current symptoms in this population of Japanese-American WW II survivors. This could be affected by discomfort acknowledging symptoms due to self or community stigma as well as remnants of the WW II experiences as probable members of the segregated 442nd regimental combat team or/and the 100th infantry battalion.

Next, we have only minimal information about mental or physical health treatment. Too few people were taking antidepressant medication to be included as a reliable control variable. These individuals were left in the analysis. We did not have information about psychological treatment, which could mediate relationships between variable and outcome (Regan, Kearney, Savva, Cronin, & Kenny, [30]).

Finally, we note that the current sample is exclusively Japanese-American men, which raises questions about the conceptualization and display of depressive symptoms and the validity of using the CES-D among this group. Acculturation also could play a role in depressive symptoms (Harada et al., [14]). Factor analysis of the CES-D among different racial groups revealed differences between groups and varied from the original structure (Kim, DeCoster, Huang, & Chiriboga, [18]). We interpret current results with these recognized concerns.

Current findings highlight potential areas for future research. First, exploration of psychosocial variables is necessary. For instance, poorer general health ratings were consistently associated with higher depression scores. Identifying which components are important to one's general health can go a long way in conducting meaningful research than can translate into practice. Second, identifying characteristics of older adults with remitting depression or resilience (never depressed) could point towards possible areas of intervention or prevention. Third, related to veteran research, pre-war or pre-service functioning would be useful in order to contextualize military service and its relation to late-life depression within one's lifespan. Understanding if late-life symptoms are chronic or late onset can provide valuable information to practitioners and researchers alike. Finally, perception of military service is important to measure. Military service appraisals are being studied in relation to later-life PTSD after early-life military service (e.g., Kang, Aldwin, Choun, & Spiro, [16]); however, research related to late-life depression is limited. Perceptions of service both immediately and years after military separation is an interesting measure to determine if changes in perception occur and how it might play a role in late-life depression.

Caring for the older veteran highlights distinct issues and challenges. However, current analyses suggest that military-related variables had minimal direct effect on late-life depressive symptomatology, although combat exposure demonstrated marginal significance among veterans. Overall, most of the sample reported minimal depressive symptoms. To our knowledge this is the first study to longitudinally examine how military-related variables affect depressive symptomatology among an older Japanese-American community cohort. For the whole sample, psychosocial variables most consistently associated with depression variables both cross-sectionally and longitudinally, which highlight the importance of self-perception and point toward possible areas for the VA and clinicians in general to focus on. Positive social relationships (e.g. available social support, family and friends' proximity), enhancing resilience and targeting negative cognitive distortions have been found to improve subjective life satisfaction and health ratings among older veterans (Han, [13]; Lee et al., [21]; Pietrzak & Cook, [26]; Schnittker, [31]; Yang & Burr, [40]), lowering risk for a late-life depressive episode. The VA serves many veterans, many of whom are over the age of 65. This work highlights the ongoing effects military service may have on veterans' current functioning more than half a century later.

Opinions, interpretations, conclusions and recommendations are those of the authors and are not necessarily endorsed by the Department of Defense. The authors thank Marnie Meyer for reviewing the drafts and providing valuable feedback.

No potential conflict of interest was reported by the authors.