Background
Currently, women are one the fastest growing demographic groups in the military, and the proportion of female military service members and veterans is at its highest level ever in the United States (US) and other industrialized countries [
1,
2]. Although women currently comprise only 17% of US active-duty forces, and about 10.5% of current veterans, this percentage is expected to grow. Growth is even greater in the National Guard/Reserve component of the US military. Given current trends, by 2042, women will comprise over 16% of the total US veteran population [
1]. These trends are even more pronounced in other advanced industrial countries [
3,
4]. Thus, it is critical that we conduct research on factors affecting the well-being of women serving in the armed forces. The goal of this study is to assess both military deployment factors and post-deployment experiences that may contribute to lifetime psychological disorders, especially posttraumatic stress disorder (PTSD) in female veterans, relative to their male counterparts, to optimize future training and treatment planning.
Based on previous research, we hypothesize that female veterans receiving healthcare will have higher rates of PTSD and other mental health problems, compared to male veterans. This hypothesis is tentative, however, as research on sex differences on veterans’ mental health is inconsistent. Some research reports similar combat experiences and stress exposures for men and women among active duty military personnel, as well as commensurate rates of mental health problems [
5,
6]. Conversely, other research shows that women’s military experiences and their responses are often different from men’s, placing them at higher risk for psychological problems [
3,
7‐
9]. For example, women experience significantly more sexual harassment and sexual assault prior to and during military service [
3,
10]. In addition, some research finds that male veterans are at greater psychological risk for mental health problems [
11]. Possible reasons for these sex differences include exposure to different types of trauma, genetics, emotional learning, gender socialization, and memory processing [
2]. Our analysis examines if (1) trauma experiences (both military and non-military) and psychological well-being differ between male and female veterans, (2) if we can explain the differences in well-being using multivariate statistical analysis controlling for confounding and other risk factors, and 3) examine sex differences in treatment seeking.
Many of the inconsistent results related to sex differences in US veterans are also seen in studies from other industrialized countries. In their study of Canadian veterans, Brunet et al. report that females were less likely to experience combat related traumas, but more likely to suffer from sexual assaults compared to male Canadian veterans [
3]. Like some US studies [
7], Canadian female veterans were also more likely to meet criteria for PTSD than males. Woodhead et al., in contrast, report few mental health differences between male and female UK veterans, although these results may be due to the relatively fewer women in the study sample [
4].
Much of the previous research on deployment and veteran well-being analyze data from veterans seeking services from the US Department of Veterans Affairs (VA) [
11‐
13], or other government-funded healthcare systems [
4]. Our sample, in contrast, comes from a community population of veterans receiving healthcare from a large non-VA system. Many veterans in the US do not use VA healthcare services and recent policy changes will likely increase the number of veterans seeking care from other providers in the future [
14‐
16]. It should be noted, however, that many participants in our study also receive healthcare from the VA. Thus, this study provides insight into a population of veterans that may overlap with VA-based samples but is different from those used in previous studies. Seeking treatment outside of traditional military healthcare systems may also inform policy planning in other countries to the extent that their military institutions are undergoing change and veterans are seeking care in the civilian healthcare system.
Finally, in recent years, the growth of women’s veterans service organizations (VSOs) in the US has been extensive. To date, over 150 active women’s VSOs and related auxiliary groups have been identified (
https://womenvetsusa.org/about.php). These VSOs report connecting women with active duty, Reserves, and National Guard military service members and their families, and with caregivers, advocates, and with local, state, and federal resources. In the discussion section below, we briefly discuss the potential impact of these VSOs on mental health outcomes for woman. While we do not directly address the impact of these activities have on women veterans, we would expect that they would likely reduce sex differences in well-being and treatment seeking over time.
Results
The basic characteristics of the sample (Table
1) show that 5% (n = 85) were female. Other features of the sample show numerous large differences between male and female veterans. For example, about 95% of the females were less than 65 years old, while only about 40% of the male veterans were in that age range. About half the women were married, but almost 80% of the men were, and female veterans were also more likely to be college graduates, compared to their male counterparts (47% vs. 24%). Female veterans were more likely to see service during the Afghanistan/Iraq War (54% vs. 21%), be deployed in National Guard/Reserve units (66% vs. 37%) and were much less likely to report high combat exposure (2% vs. 25%), compared to men. The sample shows smaller, but statistically significant differences for race, multiple tours, and unit support, with women being more racially diverse, fewer reporting multiple tours, and higher percentages scoring low on unit support/morale. Data also revealed psychological and health differences between female and male veterans (Table
1). Women were more likely to score low on psychological resilience, less likely to report service-related concussions, and more likely to report poor health. Interestingly, there were no sex differences for childhood abuse/neglect, reported stressful events in the past year or lifetime traumas, social support, alcohol misuse, or self-esteem.
Female veterans in our study were more likely to meet criteria for lifetime and past year PTSD, meet criteria for lifetime and past year depression, and lifetime suicidal thoughts, but not recent ones. Health service results show no sex differences in the current use of VA services, but women were less likely to report a current VA disability, and more likely to have used any (VA or non-VA) psychological services and psychotropic medications in the past year.
Multivariate logistic regression results (Table
2) revealed that women were more likely to meet study criteria for lifetime PTSD (OR = 5.28), depression (OR = 3.09), suicidal thoughts (OR = 2.59), and more likely to report lifetime use of psychological services (OR = 1.72), after adjusting for other demographic factors, stressful events, alcohol misuse, psychological resources, and social support. Other statistically significant factors predictive of lifetime PTSD were childhood adversities (OR = 1.67), past year stressors (OR = 3.30), lifetime trauma (OR = 2.36), combat exposure (OR = 3.07), concussion history (OR = 2.31), positive AUDIT-C results (OR = 1.59, low self-esteem (OR = 3.03), low psychological resilience (OR = 2.24), and low social support (OR = 1.69).
In addition to sex, age was statistically related to lifetime depression (OR = 0.51), as were childhood adversities (OR = 2.13), stressful events (OR = 2.27), lifetime trauma (OR = 1.64), combat exposure (OR = 1.86), concussion history (OR = 1.50), low self-esteem (OR = 3.40), low resilience (OR = 2.32), and low social support (OR = 1.62). The model for lifetime suicidal thoughts showed that childhood adversities (OR = 2.25), low self-esteem (OR = 4.25), and low resilience (OR = 1.95), along with sex, were statistically related to this psychological problem. Finally, besides statistically significant differences by sex, the model for lifetime psychological service use showed that child abuse (OR = 1.87), high stress in the past year (OR = 2.06), high lifetime trauma (OR = 1.54), high combat exposure (OR = 1.69), service related concussion (OR = 2.05), low self-esteem (OR = 2.37), and low resilience (OR = 2.02) were associated with a higher likelihood of service use, while being older than 65 (OR = 0.53) or married (OR = 0.66) was related to a lower likelihood.
As an additional check on sex differences in well-being among our sample of veterans, we replaced lifetime outcomes with current measures of each variable. That is, we replaced lifetime PTSD, depression, suicidal thoughts, and use of psychological services, with PTSD past year, current depression, recent suicidal thoughts, and use of services in the past year. As shown in Table
1, statistically significant bivariate sex differences were found for PTSD past year, current depression, and psychological service use in the past year, with female veterans more likely to have these mental health problems and to report the use of psychological services. When we controlled for the same factors in multivariate models shown in Table
2, sex differences remained for PTSD (OR = 2.53, p = 0.036) and for use of psychological services (OR = 2.43, p = 002), but not for current depression (OR = 1.65, p = 0.202) or suicidal ideation (OR = 1.09, p = 0.856). We also estimated multivariate models with interaction terms for gender and the stress variables (i.e., child abuse, stress past year, lifetime trauma, and combat exposure) to see if women responded differently to these events relative to men, as some have suggested [
2]. None of these interaction terms were statistically significant. (Results are available from the corresponding author.)
In Table
3, multivariate logistic regression results were presented for the 1:1 matching, 1:3 and 1:5 propensity score matching using “nearest neighbor” statistical methods, as discussed above, to evaluate the odds ratio of sex differences in predicting lifetime PTSD, lifetime depression, and lifetime suicidal thoughts. As can be seen, these are all statistically significant, except for 1:1 matching for lifetime use of psychological services. The differences in the number of cases used in the matching are likely a reason for the divergent statistical findings reported in the table. Nevertheless, these results add strength to our conclusions about the differences between male and female veterans for the study outcomes. That is, matching male and female veterans using propensity scores showed that female veterans were five times more likely to meet criteria for PTSD, and two and a half times at greater risk for major depression and suicidal ideation, and more likely users of psychological services compared to male veterans. As a further test of our model, we conducted a sensitivity analysis which showed that the values of Gamma were close to 1 for lifetime PTSD, depression, and suicidal ideation, which suggests that these models may have unmeasured confounders [
37]. (Results available from the corresponding author [JAB].) We discuss these limitations in the study conclusion.
Discussion
Using data collected from a sample of post-deployment veterans receiving their care from a non-VA healthcare system, we assess sex differences in health outcomes. Regarding our three research questions, we find significant sex differences in military and non-military experiences, with female veterans less likely to experience high combat and in-service concussion, but more likely to perceive low unit support. Women veterans were also more likely to meet criteria for lifetime and past year PTSD, lifetime depression, and lifetime suicide ideation. They are more likely to report using mental health services in their lifetime and in the past year. These sex differences in health outcomes persisted even after controlling for other variables. Our study is consistent with earlier research which finds female veterans at greater risk for lifetime PTSD and other mental health problems, and higher users of psychological services [
2,
5,
6,
9,
11,
18]. We further checked our findings by presenting propensity score and sensitivity analyses. All are consistent in supporting our conclusions. Our findings are especially important in that women are at greater risk for lifetime suicidal ideation and for the use of psychological services, both lifetime and in the past year. Finally, in line with other research, women in our study were different from male veterans in that they were younger, more educated, less likely to be married, and more likely to be deployed as part of the National Guard/Reserve, consistent with other studies of US veterans [
6,
7,
11], as well as studies of veterans who served in other countries [
4].
Past research by Lehavot et al. analyzed data from the 2012–2013 National Epidemiologic Survey on Alcohol and Related Conditions-III and, like our findings, report that women veterans have the highest rates of lifetime and past year PTSD and treatment utilization, compared to male veterans [
2], and that controlling for different types of trauma (e.g., early childhood abuse, interpersonal violence, and recent stressful events) reduced sex differences, but did not eliminate them [
39]. Using VA administrative data, Haskell and her colleagues find that female veterans have greater mental health problems, such as depression and adjustment disorder, and service use, compared to men, but, in contrast to our study, male veterans had higher rates of PTSD, after adjusting for demographic differences [
11]. To our knowledge, though, ours is one of the first studies to report findings for female veterans seen in non-VA hospitals [
14]. Clinicians in both VA and non-VA facilities need to be aware of demographic, pre- and post-deployment experiences, and the needs of female veterans. Moreover, in their study of women who use the VA for mental health care, Kimerling et al. report that only half of the female veterans found that the VA met their needs [
40]. Given the increasing concern among the Department of Veterans Affairs and health policy planners about the rise in suicide rates among veterans [
7], our findings strongly argue for more research on unmet needs among female veterans using non-VA facility data to ensure that this population does not suffer from the lack of appropriate care.
There are two points about our results that should be noted, one related to the propensity score and sensitivity analyses and one on the sex differences observed in the lifetime and past year psychological service use. The logistic regression models for our four outcomes included demographic, military experiences, and non-military experiences variables. Nevertheless, all four of the models continued to show sex differences. The propensity score analysis confirmed these results, but the sensitivity analysis indicated the possibility of omitted factors in the models for PTSD, depression, suicidal thoughts, and use of psychological services. Research shows that female veterans not only have different backgrounds (e.g., are more educated and less likely to be married), but also have different life experiences (e.g., more likely to be sexually harassed) than male veterans [
2,
5,
6,
9,
41]. The result of failing to capture these differences in our models is that we continue to see statistically significant differences for sex. A model that adequately assess these experiences should show no or few statistically significant differences for sex. Future research should carefully consider how pre-military experiences, self-perceptions, and interactions with others affects men and women differently, especially within a military context.
The statistical significance for sex in the models for lifetime and past year use of psychological services also suggests the need for an examination of the differences in male and female veteran experiences. It is noted that in the lifetime model, the odds ratio for the logistic regression reached statistical significance, and the propensity score results showed statistically significant sex differences for the 1:3 and 1:5 models. Studies suggest sex differences can influence both the experience of stressful events and how persons respond to such events, including treatment seeking [
40]. Further, more attention should be paid to the different military experiences of men and women veterans as they relate to these factors. In our study, for example, men were more likely to be deployed to Vietnam, while women were more likely deployed to Iraq/Afghanistan. Women were much more likely to be deployed as National Guard/Reserve than men, and more likely to have had low psychological resilience. Women were also significantly younger in our study. The fact that sex differences persist even after including many controls suggest that sex differences play a key role in mental health outcomes and use of psychological services. Greater attention to these social and psychological sex differences, along with other factors, may provide greater insight onto the psychological consequence of military deployments and may help identify gender gaps in services.
In terms of study limitations, our data were cross-sectional, which precludes assessment of causality. Second, we only were able to successfully recruit 85 female veterans. As noted, this may have biased our results and may not represent the larger population of female veterans receiving care. Although our propensity score analysis confirmed the multivariate analyses, it is possible that unmeasured variables, such as a more detailed history of sexual assault, might change our results. It is also possible that only certain types of female veterans seek healthcare in non-VA facilities, and we have not included this confounding factor in our models. Future research should sample veterans receiving care from both VA and non-VA facilities to broaden generalizations for this population. Third, the current study only included previously deployed U.S. veterans seen at a large non-VA multihospital system in Pennsylvania and our results may not generalize to all veterans. Research which includes veterans from both VA and non-VA systems may better clarify gaps in care. Fourth, the findings may not generalize to non-White US veterans because over 90% of our sample was White. Finally, we did not ask about gender identity, sexual assault/harassment in the military, or sexual orientation. Future studies should explore these issues in more detail, as these have been related to poor mental and physical outcomes among veterans [
41]. Many of these limitations are also found in other studies, especially for those of veterans who served in industrialized countries other than the US [
4‐
7,
38‐
40].
In recent years, the growth of Veteran Service Organizations (VSOs) for women has been extensive. While outcome data related to VSOs are limited, preliminary studies are encouraging [
42‐
44]. For example, while these VSOs appear to be well received, several studies suggest that the more the veteran’s VSO engagement, the larger the meetings, the more involvement in activities, the better are the outcomes for women veterans [
42‐
44]. These VSO findings are promising, although further research is required to better assess the VSO’s impact on women’s mental health.
Conclusions
Despite these limitations, our findings support the case for more gender-informed planning, given the projected increase in female veterans in both the US and other industrial countries. These changes might include hiring more female healthcare providers, more explicit training in the health needs of female veterans, and training on greater sensitivity to a female’s unique life experiences, such as sexual harassment [
1,
40]. Further, and consistent with other researchers [
1,
39,
40], we find female veterans have higher use of mental health care services, relative to male veterans, and have found that a higher percent of them accessed psychiatric services in the past year. In the Kimerling et al. study [
40], results suggest that female veterans reported lower use of VA healthcare because there were fewer female doctors and women-only healthcare settings. Our findings, along with other studies on female veterans, need to be used to inform changes in the provision of healthcare services to US veterans at both VA and non-VA facilities [
15]. Studies of non-VA healthcare service delivery to veterans will be important to develop future public–private partnerships, which will be key in addressing gender differences in the healthcare needs of male and female veterans [
40].
Previous presentation
This study was presented at the International Socity for Traumatic Stress Studies, Chicago, IL, November, 2017
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.