PTSD and Use of Outpatient General Medical Services Among Veterans of the Vietnam War

Studies have shown higher rates of service use for general medical conditions among veterans with posttraumatic stress disorder (PTSD). Analyses of data from the National Vietnam Veterans Readjustment Study (NVVRS) found that Vietnam veterans with PTSD were more likely than those without PTSD to use general medical services, particularly services provided by the Department of Veterans Affairs (VA) (1–3). More recent studies of health service utilization among Operation Iraqi Freedom/Operation Enduring Freedom veterans in which VA administrative data were analyzed demonstrated similar patterns of greater general medical service use among veterans with PTSD compared with those with other psychiatric diagnoses and those without a psychiatric diagnosis (4).

Higher utilization of general medical care by individuals with PTSD may be driven by greater needs related to physical health. It is well established that veterans with PTSD suffer from a higher burden of general medical disease compared with veterans without PTSD (5,6). PTSD is also associated with somatic symptoms in the general population (7–9) and among veterans (5,10), potentially motivating higher use of general medical services in this population (11).

However, higher utilization of general medical services by veterans with PTSD could also reflect improved integration of mental health services into ambulatory care settings, such as primary care (12,13). Although recent work suggests that many veterans who screen positive for PTSD have accessed specialty mental health care (14,15), older veterans with a positive PTSD screen (16) and with PTSD (15) have been found to be less likely than younger veterans to receive specialty mental health care. Stigma as a barrier to using specialty mental health care is an ongoing concern in some (14,17,18) but not all (15) studies and may be one reason why older adults are more likely to access behavioral health services in integrated health care settings (19).

Understanding the relationship between PTSD and use of general medical services among aging veterans is necessary to assess VA service needs across the life span. Some evidence suggests that PTSD manifests itself differently among older persons and may be exacerbated by aging (20), resulting in higher rates of disability (21). Higher rates of exposure to trauma have also been found to be associated with increasing general medical concerns as veterans age (22). Despite these associations, few studies have specifically examined the relationship between patterns of current use of outpatient general medical services and veterans’ PTSD symptomatology over time.

The data for this analysis were from the National Vietnam Veteran Longitudinal Study (NVVLS), a 25-year follow-up of the NVVRS cohort. The primary goal was to test whether recent outpatient service utilization for general medical concerns by Vietnam veterans varies according to the level of PTSD symptomatology over time. To further explore patterns of help seeking in outpatient ambulatory care, separate analyses examined whether PTSD symptomatology was associated with veterans’ reports of discussing behavioral health issues as part of a general medical visit.

Methods

The NVVLS was a follow-up study of veterans who participated in the NVVRS (1). The analysis assessed health service use among the NVVLS participants. All research protocols were reviewed and approved by the Abt Associates Institutional Review Board. The sampling, weighting, and other methods for the follow-up have been described in detail elsewhere (23).

Briefly, for the NVVRS, a stratified national probability sample of 2,348 Vietnam veterans was drawn from military records, including 1,632 who were deployed to the Vietnam theater of operations (83% NVVRS response rate) and 716 Vietnam era veterans who served in the military during the years of the Vietnam War but were not in the war (77% NVVRS response rate). The NVVLS information analyzed here came from a self-report mailed questionnaire (phase 1) and a computer-assisted telephone survey (phase 2). Among 1,839 living respondents (1,276 theater and 563 era), 1,450 (79%) participated in at least one phase of the NVVLS, 1,238 (67%) completed both phase 1 and 2, 171 (9%) completed phase 1 only, and 41 (2%) completed phase 2 only. The sample for this analysis included 848 theater veterans and 361 era veterans who had sufficient data from NVVRS and NVVLS to characterize their PTSD symptomatology over time. We focused primarily on theater veterans for the analyses reported here because our interest was the relationship between war zone PTSD and health service utilization. Because data were weighted to account for sample design and nonresponse, the theater veterans represent the approximately 2.5 million Vietnam theater veterans living at the time of the NVVLS.

Measures

To characterize the course of PTSD symptomatology from time 1 (1987) to time 2 (2012), we used the Mississippi Scale for Combat-Related PTSD (M-PTSD) (24), which was the only PTSD assessment administered at both time points. The M-PTSD is a validated 35-item self-report measure (1,24,25) that assesses combat-related PTSD in veteran populations with a 5-point Likert response scale. An empirically determined cut point of 94 for probable PTSD diagnosis was used to maintain consistency with the NVVRS (26). For male veterans, we defined four categories of PTSD symptomatology between the NVVRS and the NVVLS as follows: stable high, M-PTSD scores of ≥94 at both time points; increasing, scores of <94 in NVVRS but ≥94 in NVVLS; decreasing, scores of ≥94 in NVVRS but <94 in NVVLS; and stable low, scores of <94 at both time points. Because of smaller cell sizes among women veterans, we defined two categories for course of PTSD by combining the stable-high and increasing groups in one category and the stable-low and decreasing groups in another.

All respondents were asked whether they received any outpatient treatment and the number of visits in the past six months for a general medical problem from a doctor or other medical person in an office, clinic, or emergency room. Respondents were then asked to identify the places they received care from options that included “at a VA clinic.” Three dichotomous service use measures were created to characterize levels and types of recent outpatient visits for general medical conditions reported in the past six months: outpatient general medical service use (VA and non-VA combined), VA outpatient general medical service use, and three or more outpatient visits (VA and non-VA combined) for general medical concerns (to assess intensity of use).

To indirectly assess screening and treatment for behavioral disorders in ambulatory care, we constructed a dichotomous measure of veterans’ discussion of behavioral health concerns in an outpatient visit. All who reported an outpatient visit for general medical reasons were asked, “During (this visit/any of these visits), did you and the health professional you saw talk about any problems you had with your emotions or nerves?” The same question was then asked substituting “alcohol and drugs” for “your emotions or nerves.” All who endorsed at least one item were coded positive for discussion of behavioral health concerns. In addition to age, sex, and race-ethnicity (for males), several other covariates shown to be associated with higher use of outpatient general medical services were included in the models. Current moderately severe depressive symptoms were assessed by using the eight-item Patient Health Questionnaire with a cutoff score of ≥15 (27), and current high risk of alcohol problems was measured by the Alcohol Use Disorders Identification Test with a cutoff score of ≥16 (28). A measure of the number of chronic health conditions in the past 12 months consisted of a self-report measure of chronic health conditions from the National Health Interview Survey, with separate lists for males (27 conditions) and females (30 conditions) (29).

Statistical Analyses

To assess bivariate associations between PTSD symptomatology and veteran characteristics, we used the Rao-Scott chi square test for categorical characteristics and analysis of variance for continuous measures. We tested unadjusted associations between the course of PTSD symptomatology and the four dichotomous service use outcomes by using the Rao-Scott chi square test, separately for male and female veterans. For the service use measures, and separately for men and women, we estimated two logistic regression models. In the first model, we entered course of PTSD symptomatology, controlled only for age and race-ethnicity. In the second model, we entered high risk of alcohol problems and moderately severe depressive symptomatology to determine whether relationships between PTSD symptomatology and outpatient general medical service use were partially explained by these conditions. We also entered chronic general medical conditions to assess whether PTSD maintained an association with use of general medical services even in an aging population in which general medical conditions are more prevalent. All analyses were weighted so that estimates could be generalized to the population of living Vietnam theater veterans. Analyses accounted for the stratified sample design via Taylor series estimation. Analyses were conducted using the SURVEY procedures in SAS 9.4.

Results

Table 1 presents data on demographic characteristics, symptoms of behavioral disorders, and number of chronic general medical conditions by PTSD course categories. Across both time points (NVVRS and NVVLS), three-quarters (75.3%) of veterans reported low PTSD symptoms, and 6.4% reported high PTSD symptoms. However, 13.7% reported increasing PTSD symptoms and 4.6% reported decreasing symptoms. These rates are elevated in comparison to rates among the Vietnam era veterans, in which 88.7% reported low PTSD symptoms across both time points, with 2.1% in the stable-high and 2.8% in the increasing categories (data not shown).

Those with stable-high and increasing PTSD symptoms were on the whole younger than those with decreasing or stable-low symptoms (p<.001), and a higher proportion of women were in the stable-low category (p=.004). Blacks and Hispanics were overrepresented in the stable-high and increasing categories, compared with whites (p=.001). Specifically, although blacks accounted for 9.5% of all theater veterans, they represented 18.1% of the stable-high group and 17.7% of the group with increasing symptoms; in turn, Hispanics accounted for 5.4% of all theater veterans but 11.8% of the stable-high and 11.5% of the increasing groups.

Risk of alcohol problems was most prevalent among those with stable-high PTSD symptoms (14.0%) and least prevalent among those with decreasing (1.4%) or stable-low symptoms (3.0%) (p=.006). Across both time points, depressive symptoms were also elevated among those with stable-high PTSD symptoms (39.7%) and those with increasing symptoms (30.4%) (p<.001). The mean number of general medical conditions for men was highest for those with stable-high PTSD symptoms (4.9 conditions), followed closely by those with increasing symptoms (4.1 conditions) (p<.001). For women, the mean number of health conditions in the stable-high category was 6.1, markedly higher than in the other categories (p<.001).

Overall, 57% of theater veterans (N=539) reported recent outpatient visits for physical health, and 21% (N=207) reported visits within the VA. Table 2 presents the bivariate analyses of the associations between the four outcomes and course of PTSD symptoms. Among men, those with stable-high and increasing PTSD symptoms were more likely to report recent (in past six months) VA outpatient visits, three or more outpatient visits in the past six months, and speaking with a provider about behavioral health, compared with those with decreasing or stable-low symptoms (all p≤.005). Among women, 69.4% with stable-high and 38.0% with increasing PTSD symptoms reported VA outpatient service use for general medical concerns, compared with only 10.3% of those with stable-low symptoms (p<.001).

Logistic regression models examining the three service use outcomes for male theater veterans are presented in Table 3. Course of PTSD symptoms was associated with VA outpatient visits (model 1); the likelihood of VA outpatient visits was more than three times as high for those in the stable-high and increasing PTSD categories compared with those in the stable-low category. However, when the number of health conditions was entered in the model (model 2) the odds remained significantly elevated only for those in the increasing PTSD category (odds ratio [OR]=2.27).

The likelihood of three or more outpatient visits (VA or non-VA) for men was also related to course of PTSD symptomatology, although the pattern was different. In the fully adjusted model, those with decreasing PTSD symptoms over time had lower odds than those in the stable-low category of reporting three or more outpatient visits (OR=.19). Pairwise comparisons (not shown) indicated that those with decreasing PTSD symptoms were less likely than those with increasing symptoms to report three or more outpatient visits (OR=.18, 95% CI =.06–.60, p=.005).

For female veterans, stable-high or increasing PTSD symptoms had an independent association with use of VA outpatient services, even in the fully adjusted model (OR=5.46) (Table 4). Stable-high or increasing PTSD symptoms were associated in the unadjusted model for women with significantly greater odds of having three or more outpatient visits; however, when depressive symptoms and number of health conditions were added to the model (model 2), the difference was no longer significant.

The final models (Table 5) showed that among male theater veterans who had an outpatient visit in the past six months, those with stable-high or increasing PTSD symptoms were much more likely than those in the stable-low category to discuss behavioral health issues with their general health care physician. Black males were also more likely than whites to discuss these issues. Men with more health conditions were more likely than men with fewer conditions to discuss behavioral health issues. For women, PTSD symptoms were not associated with a greater likelihood of such discussions.

Discussion

These analyses found a relationship between PTSD symptomatology over time and recent VA outpatient health service use among male and female Vietnam theater veterans. Although only a minority of Vietnam veterans sought care within the VA, higher proportions with stable-high or increasing PTSD symptoms reported use of ambulatory care services. These findings are consistent with recent studies that show increasing use of any Veterans Health Administration services by Vietnam veterans with PTSD (30).

The relationship between PTSD and outpatient VA service use may be a function of how PTSD manifests clinically with chronic general medical concerns among aging veterans (31). Recent VA policy has increased representation of mental health specialists within VA primary care settings (15,32), which may be even more critical for older veterans. The findings also suggest the importance of continued attention to how the aging process or other stressors (for example, retirement or loss of a spouse) could stimulate PTSD symptomatology (30,33–35), potentially driving increased utilization. Given the relationship between combat-related PTSD and general medical conditions for veterans from the Vietnam era and subsequent wars (5,11,22,36), these findings suggest benefits of screening for behavioral and other conditions in outpatient care settings (37). Female veterans with stable-high PTSD symptoms had the highest mean number of health conditions, suggesting that the relationship between aging and PTSD among female Vietnam veterans requires further analysis.

More than half of male veterans with PTSD symptoms brought behavioral health concerns to an outpatient health care visit, even when the analysis controlled for the number of general medical conditions. More analysis is necessary to assess whether discussion of behavioral health concerns among Vietnam veterans in VA ambulatory care settings is a substitute for or a complement to specialty mental health care. Attention to differences between male and female veterans in patterns of help seeking for mental health services in the VA is warranted (38). Black males were more likely than whites to talk with their general health care provider about behavioral health, which was not surprising given some evidence of preferences in this population for receipt of behavioral health care outside the specialty mental health system (39).

One unexpected finding was that male Vietnam veterans with decreasing PTSD symptoms were less likely than those with increasing symptoms and with low-stable symptoms to have three or more outpatient visits in the past six months. Further research is necessary to determine whether attention to PTSD symptoms may have long-term benefits with respect to development of health conditions or their chronicity.

It is important to place the findings from this analysis in a policy context. Veterans who receive Medicare benefits because of disability were found to be more likely to use VA services than veterans who receive Medicare because of age (40), which may account for higher VA service use by veterans with PTSD. Among veterans who sought services in this study, close to 50% with three or more visits in the stable-high and increasing PTSD symptoms groups sought only VA services (results not shown).

A limitation of this study was that VA data restrictions do not allow analysis of information about service-connected disability for NVVLS participants. High rates of use of VA outpatient health care by veterans with PTSD could also be influenced by greater overall access to mental health services in the VA, as a result of enhancements in availability of VA specialty mental health services (33). Our finding that black males were more likely than white males to use VA outpatient services for general medical conditions is consistent with other analyses demonstrating that health service disparities are not evident within the VA system (41). It is important to note that this analysis used self-report measures of service use and PTSD symptoms rather than administrative data. In addition, the behavioral health measures were screeners and not clinical diagnostic assessments. Replication of these findings with other data sources is an important next step.

Conclusions

Vietnam veterans with high and increasing PTSD symptomatology over time were more likely than those with little PTSD symptomatology to present for care in VA outpatient general health services, and behavioral health issues were frequently raised within this context. The relationships between aging, combat-related PTSD, and use of VA health services are important topics for future research with Vietnam veterans and other service members. Clinical and policy attention to address PTSD with other psychiatric and medical comorbidities within the context of outpatient general medical care for older veterans is warranted.