An Evidence Map of the Women Veterans’ Health Research Literature (2008–2015)

We searched MEDLINE (Ovid), the Cumulative Index to Nursing and Allied Health Literature (CINAHL), and the VA Health Services Research and Development database for articles published between January 2008 and December 2015. The previous review period ended in September 2008, allowing a short overlap period to capture pending or unindexed publications. The search included the Medical Subject Headings (MeSH) terms Women; Women’s Health; Women’s Health Services; Veterans; Veterans Health; and Hospitals, Veterans.

We excluded studies that were not relevant to health/healthcare, did not include female US Veterans, or only included active duty military. Studies with fewer than 100 participants were excluded if less than 10% of participants were women, and studies with 100 to 1000 participants were excluded if less than 5% were women. Studies with more than 1000 participants were eligible if they included any women. For studies with a female or Veteran proportion < 75% of the total study population, we excluded studies that did not stratify results by sex or Veteran status, respectively. We also excluded case reports, letters, meeting abstracts, dissertations, editorials, reviews, conceptual frameworks, and protocols.

Abstracts were independently reviewed by a trained investigator (ED and NG) or research associate (EK, TV, and RM). A random selection of 18% (404 abstracts) were dual-reviewed; for these, agreement on inclusion was 87% (κ = 0.747), which is considered by convention to be substantial or moderate agreement.11 The full texts of eligible studies were then independently reviewed for inclusion by an investigator or research associate. A second reviewer independently reviewed a 10% random sample of full-text articles, as well as any additional articles that the original reviewer requested. If the two reviewers disagreed, a group arbitration system was used.

For each included study, 15 study characteristics were extracted onto evidence tables by one investigator or research associate. We selected and defined these study characteristics after discussion with key stakeholders in the clinical and research offices of women’s health at the VA and an expert panel composed of VA women’s healthcare providers and researchers, and then refined the categories within each study characteristic through multiple small subsample extractions. The characteristics extracted were healthcare topic, study design, sample size and proportion female, reporting of age and race, focus on special populations (e.g., lesbian, gay, bisexual, and transgender (LGBT), racial and ethnic minorities, or homeless Veterans), follow-up/duration, research setting, use of electronic health record, period of service, Veteran engagement (i.e., participation of patients in the design or conduct of the study), population studied, type of outcomes reported, publication year, and funding source. Each included study was designated one of 39 healthcare topics based on the primary focus of the article, and we grouped these under four subheadings (Table 1). Articles that reported on physical or mental health topics but primarily addressed issues of prevention and screening, healthcare organization and delivery, access and utilization, homelessness, or post-deployment health were placed in the latter groupings. We then performed an iterative, cross-tabular review of the abstracted data. A randomly selected 10% sample of studies were individually dual-reviewed across each column of study characteristics. Inter-rater discrepancies were noted within particular columns that relied on subjective interpretation. We addressed discrepancies by either reducing category granularity (e.g., we collapsed all observational studies except prospective cohort studies into a single grouping) or assigning two researchers to dual review all the studies grouped within a particular category, in order to ensure consistent assignment within each column. The principal investigator verified categorizations and addressed inconsistencies while summarizing the findings by study characteristics.

We sorted and compared studies by healthcare topic, study design, sample size and proportion female, publication year, and funding source.

Mental health articles were dominated by conditions often associated with military service, primarily post-traumatic stress disorder (PTSD) (71/208, 34%), military sexual trauma (MST) (37/208, 18%), and substance abuse (20/208, 10%). Four observational studies primarily addressed depression (3) and anxiety (1), and eight others addressed depression comorbid with other mental or physical health conditions. Four articles described reproductive mental health issues (e.g., postpartum depression). Twelve articles presented the primary findings (4 studies) or secondary analyses (8 studies) of randomized trials related to PTSD, MST, or multiple mental health diagnoses.

No specific clinical condition dominated the physical health articles, and few articles were found regarding common chronic conditions such as obesity (9), chronic pain (7), diabetes (3), and hypertension (0). The four most common topics were prevention and screening, reproductive health, long-term care and aging, and cardiovascular disease; together, they made up half (66/133, 50%) of the physical health articles. Though most studies were observational, there was one randomized controlled trial (RCT) on mammography screening promotion among women Veterans published in 2008,12 three subsequent secondary analyses of that study, and one small, single-site 6-month RCT of aerobic exercise for mild cognitive impairment.13

Thirty-one studies evaluated healthcare organization and delivery, 45% of which were published in 2015. These studies described the challenges, methods, and outcomes related to healthcare delivery for female Veterans. Half focused on comprehensive primary care for female Veterans, including a single VA-funded RCT of VA healthcare providers that tested the effects of a 30-min computerized educational program on gender awareness.14 We identified nine studies related to mental healthcare delivery for female Veterans, and three studies each related to emergency care delivery and virtual or telehealthcare delivery methods.

We identified 57 articles related to access and utilization (24), rural healthcare (3), homelessness (12), or post-deployment health (18). The access and utilization studies assessed barriers to care related to homelessness, mental healthcare, financial concerns, and factors that explain delayed care and attrition, and described VA and non-VA healthcare utilization. Over a third of these specifically addressed Veterans of Iraq and Afghanistan conflicts (9/24, 38%). Nearly half of the 18 studies related to post-deployment health (44%) were published in 2015. One large VA-funded RCT studied the impact of online expressive writing on readjustment difficulties among Veterans of Iraq and Afghanistan conflicts.15

In 2011, the VA WHRN set forth an ambitious research agenda with six key topic areas: (1) mental health, (2) primary care and prevention (including primary care delivery), (3) reproductive health, (4) complex chronic conditions/aging and long-term care, (5) access to care and rural health, and (6) post-deployment health.8 We found evidence that four of these areas advanced considerably in subsequent years, as did the subsection of primary care related to healthcare delivery. Complex chronic conditions/aging and long-term care, and the remainder of primary care and prevention, did not show substantial growth and are addressed separately below.

Mental health articles continue to dominate the VA women’s health literature (47% of studies), consistent with the previous review (85/195, 44%).5 PTSD studies remain prominent but now represent only one-third of mental health research, compared with nearly half in the previous review.5 In contrast, sexual trauma and substance abuse have grown considerably as a proportion of mental health research.5 Research related to the delivery of comprehensive primary care for female Veterans shows evidence of coordinated growth, involving varied viewpoints (providers, Veterans, vulnerable subpopulations) and multiple methodologies (observational studies, qualitative studies, and an RCT). Several other topic areas (reproductive health, access to care, rural health, and post-deployment health) with little research at the outset of our study period have grown dramatically in number of publications since being named research priorities, with publication counts rising as much as seven-fold.

The VA WHRN has also emphasized research related to particular subpopulations of female Veterans. Returning Veterans of Iraq and Afghanistan conflicts make up one-third of living female US Veterans.16 Over one-fifth of included articles targeted Veterans of those conflicts, and three-quarters of those have been published since 2012. The majority of studies addressing LGBT Veterans, racial and ethnic minorities, and homeless Veterans have also been published since 2012.

The overall increase in publications in recent years can be at least partially attributed to VA-funded journal supplements in 2011,7 2013,17 and 2015.18 The proportion of female Veterans’ health research that is VA-funded has also grown from 45% (studies from 1978 to 2004)4 to 60% (2004–2008)5 to 69% of studies in this review (2008–2015).

We identified five primary gaps: research on common chronic disease topics, sex-specific results reporting, interventional study design, funding reporting, and Veteran engagement. First, several topics had surprisingly little research relative to their clinical prevalence—specifically, physical health topics in primary care and chronic disease, prevention and screening, and long-term care and aging. For example, we found no studies with a primary focus on hypertension though hypertension affects nearly 40% of middle-aged female Veterans in the VA and over 60% of those over 65.3 Controlling hypertension and other cardiovascular disease risk factors is critical for women, one in four of whom will die of heart disease.19 In addition, mental health topics most often encountered in primary care, including depression, anxiety, and postpartum depression, were largely absent from the literature. Depression is the most common mental health diagnosis among female Veterans at VA,3 including those returning from Iraq and Afghanistan.20 Evidence maps are primarily descriptive and our results do not directly address the causes or consequences of literature gaps. However, the stark disparity between the prevalence and significance of common chronic health conditions and the quantity of published research addressing those topics merits review.

We suggest that the apparent inattention to common chronic health conditions is primarily attributable to (1) the stage of existing evidence for most common conditions in primary care and (2) a lack of sex-specific results reporting for clinical research that includes female Veterans. For conditions such as hypertension and depression, decades of federally funded clinical research has defined best practices for healthcare. As a result, focusing ongoing research on health services delivery may be the most appropriate way to optimize the quality of care for female Veterans with these conditions. A 2008 VA Under Secretary for Health workgroup report on the provision of primary care to female Veterans highlighted the complexity of treating female Veterans with multiple comorbid chronic mental and physical health conditions, and identified fragmentation of care for general and gender-specific health concerns.21 Since 2008, most research related to common chronic conditions among female Veterans has addressed healthcare organization and delivery. For example, though very few articles in our sample primarily addressed depression, we found additional studies evaluating depression comorbid with physical health conditions and exploring integrated mental health and primary care delivery.

At least some research on common chronic conditions is being conducted with female Veterans, but the study results are not consistently reported by sex. We excluded over 350 articles that did not report sex-specific results. Though we did not extract study characteristics for excluded articles, a title search found 24 articles with the words “diabetes” or “depression” in the title (though only three had “hypertension” or “blood pressure,” and none had “anxiety”). The need for sex-specific reporting of scientific research results has been recognized by both the NIH22 and the Institute of Medicine,23 ^, 24 though multiple challenges related to study design, statistical analysis, and results reporting exist.24 VA has long required the inclusion of women in research,8 and encouraging sex-specific results reporting could expand the field of female Veterans health research and allow for future meta-analyses by sex.

Conducting interventional research among female Veterans has been challenging due to the small number of women at any one clinical site.25 We identified only eight published RCTs over the past 8 years. A simple search of excluded studies with the words “randomized trial” in the title revealed at least seven additional RCTs that included female Veterans but did not provide sex-specific results, and four more that included too few women to meet our criteria. Increasing the recruitment of women into existing VA trials and encouraging sex-specific results reporting could augment the field of experimental research related to female Veterans' health.

Reporting the source and role of funding is a quality standard for both experimental26 and observational27 research. Though 78% of studies identified at least one source of funding (a relatively high rate of funding compared to other medical fields),28 ^– 30 20% of included articles did not report a funding source, which likely represents unfunded research.31 Explicitly describing research as unfunded will help stakeholders allocate resources.

Finally, although several studies incorporated Veterans’ perspectives, they all adhered to a traditional model of women as study subjects rather than as research stakeholders or partners. Researchers are increasingly seeking to engage patients and community members in the development of study questions, selection of outcome measures, and interpretation of findings. Incorporating female Veterans’ voices in the production of future research will strengthen the relevance and credibility of that work.