INTRODUCTION

Recognition and reporting of sexual assault and/or sexual harassment during military service (military sexual trauma, or MST) is increasingly common.1,2,3 MST has detrimental effects on psychological and physical health,4 contributing to posttraumatic stress disorder, depression, chronic pain, and cardiovascular disease.10,12,13 EHR screening leads to documentation of MST in the medical record, which participants may not desire and/or believe is relevant to their health.14,15 Interpersonal dynamics with the screening clinical staff and experiences within the VA may contribute to patients feeling uncomfortable disclosing MST.12 Systematic factors in EHR screening for MST can also act as barriers to disclosure. Although longer-term patient-provider relationship may facilitate disclosure, screening for MST is often completed once, in-person, and not repeated after an initial negative response.16,17

Undisclosed sexual violence is also associated with worse health outcomes compared to disclosed violence.12,18 Identifying women who experienced MST but are not captured by current EHR screening may therefore improve health outcomes and equitable referral to treatment services. We aim to advance current understanding of the factors associated with MST non-disclosure during EHR screening. Few comparisons of EHR screening and survey results have been conducted within the same sample of women enabling evaluation of the contributions of individual, interpersonal, and systematic factors. Prior studies have also been done in relatively small populations of women Veterans.7,11,19 By studying a multi-state sample of women, we can further define barriers and disparities associated with MST disclosure during EHR screening.

METHODS

Design and Study Population

We used baseline data from a cluster-randomized controlled trial among women Veteran VA primary care users. Study methods have been previously reported.20 We recruited women with 3 + primary care visits in the prior year from 12 VA medical centers in nine states from January to March, 2015. Overall, 1395 eligible women (response rate 45%) completed a computer-assisted telephone interview (CATI) survey, with 93% (n = 1287) consenting to linkage of surveys to their VA EHR data. The VA Greater Los Angeles Healthcare System and RAND Santa Monica Institutional Review Boards approved the study.

Military Sexual Trauma

In EHR screening, MST experience is captured using standardized questions: “While you were in military service: 1) did you ever receive uninvited or unwanted sexual attention such as touching, cornering, pressure for sexual favors or inappropriate verbal remarks, etc., and 2) did anyone ever use force or threat of force to have sexual contact with you against your will.”4,16 “Yes” to either question is categorized as MST positive; EHR data do not distinguish between sexual harassment and assault. Providers are prompted by the EHR to repeat MST screening annually for patients who respond, “Decline to Answer” until they answer “Yes” or “No.” Although there is not a prompt, providers can screen again after a “Yes” or “No.” Any “Yes” in the EHR was considered a positive EHR screen in our study. In the study survey, MST was assessed with the same standardized questions, and subcategorized as sexual harassment if “Yes” to question 1 and sexual assault if “Yes” to question 2, consistent with previous studies.16 We paired participants’ responses in the EHR to their responses on the survey. From the EHR, we also extracted the number of times MST was screened from 1998 to 2015 (the year of the study survey) and obtained the time between completion of the most recent EHR screen and the survey.

Patient Demographics and Mental Health

Participants self-reported age, race, ethnicity, marital status, education, and sexual orientation in the study survey (Appendix Table 1). They completed the Patient Health Questionnaire 2 (PHQ-2) to measure depression, the Generalized Anxiety Disorder 2 scale (GAD-2) for anxiety, a two-item checklist for probable PTSD,21,22,23 and the Alcohol Use Disorders Identification Test-Concise (AUDIT-C) for disordered alcohol use (score ≥ 3).24 The survey captured smoking status and childhood sexual violence.

VA Experiences/Perceptions

We asked women if they had experienced stranger harassment or felt unsafe in the VA. We measured anxiety or distress during a gynecologic exam (scale 0–10). We asked if primary care physicians (PCP) had inquired about MST: “As best you can recall, has your VA provider ever asked if you have experienced any sexual harassment or unwanted sexual contact while in the military?” After screening for lifetime sexual trauma (during childhood, the military and after the military), we assessed if patients had talked to their providers about these experiences: “Have you ever talked to your VA provider about these experiences?” We asked if patients felt comfortable talking to their PCP about emotional issues, and whether they had a PCP gender preference. We obtained measures of delayed/missed medical care or gynecological exams and asked if the delay was due to concern about interacting with other Veterans. Participants reported their insurance coverage. For healthcare utilization, we extracted the number of visits participants had made in a VA emergency room (ER) in the last year from the EHR.

Statistical Analysis and MST Categorization

Our analysis sample included 1287 participants who had responded to both EHR and survey MST questions. We categorized Veterans into “no MST” (no survey or EHR MST), “MST captured by EHR and survey” (both survey and EHR MST), “MST not captured by EHR” (survey MST but no EHR MST), and “MST not captured by survey” (EHR MST but no survey MST). We compared characteristics among the four categories using Pearson’s chi-squared test for categorical variables and adjusted Wald test for continuous variables. The data were weighted for percentages, accounting for survey non-response and design. Fisher’s test could not be extended to these complex survey data given small cell sizes.25 Using the same methods, we compared characteristics between women with “MST not captured by EHR” and women with “MST captured by EHR and survey” (n = 783). We used stepped multivariable logistic regression to examine associations between patient characteristics and “MST not captured by EHR” (vs. “MST captured by EHR and survey”). The models were first adjusted for selected sociodemographic characteristics, then MST screening factors (e.g., MST subtype, number of EHR screens, time between EHR screen and survey), and finally VA experiences/perceptions. We used STATA version 15 (College Station, TX: StataCorp LLC) for all analyses.

RESULTS

Descriptive Statistics

Most participants (54%) were age 45–64 years (mean age = 50, SD = 15). The majority (62%) were non-Hispanic white, 22.8% non-Hispanic Black, and 5.2% Hispanic/Latina. Over half (61%) were not married and 39% had a college degree. Approximately 14% identified as lesbian, gay, bisexual, or transgender (LGBT), and 38% had other sources of insurance in addition to the VA coverage. Among all participants, 38% had “no MST,” 34% had “MST captured by EHR and survey,” 26% had “MST not captured by EHR,” and 1.3% had “MST not captured by survey” (Table 1). Approximately 35% had a positive MST EHR response, whereas 61% had a positive MST survey response. In the MST survey responses, 32% of women reported histories of military sexual harassment and sexual assault, 27% reported sexual harassment without sexual assault, and 1.3% reported sexual assault without sexual harassment.

Table 1 Sample Characteristics by MST Responses in the VA Electronic Health Record (EHR) Compared to the Study Survey in Full Source Sample of all WV-HUES Respondents, N (Weighted Percentage)
Full size table

We compared women who had “MST not captured by EHR” (survey MST but no EHR MST) to those who had “MST captured by EHR and survey” (both survey and EHR MST) (n = 783) (Table 2). Of the women who had “MST not captured by EHR,” 68% reported sexual harassment without sexual assault and 31% reported sexual harassment and sexual assault. In contrast, among women with “MST captured by EHR and survey,” 28% reported harassment without assault and 70% reported harassment and assault. Latina and Black women more frequently reported only sexual harassment (36% and 31%, respectively) compared to white women (24%). Among the women with “MST not captured by EHR,” 15% were screened more than once compared to 44% of women with “MST captured by EHR and survey” (p < 0.001). A lower percentage of Latina participants (16%) were screened more than once compared to white (30%) or Black women (34%).

Table 2 Select Sociodemographic Characteristics Comparing “MST Not Captured by EHR” to “MST Captured by EHR and Survey” in the Primary Analytic Sample (Women with Survey and/or EHR-Documented MST; N (Weighted Percentage))
Full size table

With respect to VA experiences/perceptions, MST was more often captured by EHR screening in women who indicated that they had experienced VA stranger harassment (39% captured vs. 26% not captured; p < 0.001), felt unsafe at the VA (24% captured vs. 15% not captured; p = 0.001), reported talking to their PCP about lifetime sexual trauma (28% captured vs. 7% not captured; p < 0.001) and who preferred a female provider (57% captured vs. 43% not captured; p = 0.001). Women with captured MST also had higher ER utilization (mean 0.79 (SD = 0.64–0.94) vs. 1.3 (SD = 0.96–1.6)) and reported more delayed medical care due to concern about interacting with other Veterans (24% vs. 13%; p = 0.04) compared to women with MST not captured by the EHR.

Factors Associated with MST Disclosure

In our first model analyzing sociodemographic characteristics associated with “MST not captured by EHR,” Latina and Black women were significantly less likely to disclose MST (Latina: OR = 3.0, 1.6–5.6; Black: OR = 1.6, 1.2–2.2) (Table 3). Age, marital status, and education were not associated with uncaptured EHR MST. LGBT sexual orientation was associated with having MST captured by the EHR and survey (OR = 0.58, 0.36–0.92). When we added MST screening factors (e.g., MST subtype, number of EHR screens, time between EHR screen and survey) to the model, Latina and Black women continued to have significantly higher odds of MST not captured by the EHR. The magnitude of the association decreased for Latina women (OR = 2.0, 1.0–4.2) but not for Black women (OR = 1.7, 1.3–2.2). LGBT sexual orientation was no longer associated with our outcome. Women who reported sexual harassment without sexual assault in the survey (vs. sexual harassment and sexual assault) had fivefold greater odds (OR = 5.1, 3.8–6.9) of having MST not captured by the EHR. Women who were screened for MST more than once in the VA had 73% lower odds of having MST not captured by the EHR (OR = 0.27, 0.17–0.42). The time between EHR screening and survey administration was not significantly associated with uncaptured EHR MST.

Table 3 Associations Between Select VA Perceptions or Experiences and “MST Not Captured by EHR” in Women Veterans. Stepped Regression Model First Adjusted for (1) Sociodemographic Characteristics; Then Adjusted for (2) Sociodemographic Characteristics and MST Screening Factors; and Finally Adjusted for (3) Sociodemographic Characteristics, MST Screening Factors and VA Experiences/Perceptions
Full size table

In our final model, including sociodemographic characteristics, MST screening factors, and VA experiences/perceptions, Black and Latina women continued to have almost twofold greater odds of MST not captured by the EHR (Black: OR = 1.6, 1.2–2.2; Latina: OR = 1.8, 1.1–3.1). Compared to the previous model, the odds of MST not being captured by the EHR remained about 5 times higher for women who reported only military sexual harassment (OR = 4.8, 3.4–6.9) and 71% lower for women who were screened for MST more than once (OR = 0.29, 0.18–0.46). Women who had disclosed any lifetime sexual trauma to their PCP had 73% lower odds of MST not being captured in by the EHR (OR = 0.27, 0.13–0.57).

DISCUSSION

In this population-based study, participants reported MST in a research survey almost twice as often as in EHR screening. Black and Latina women had higher odds of MST not being captured in EHR screening, independent of MST screening factors and VA experiences. Exposure to sexual harassment alone (without sexual assault) was also associated with MST not being captured in the EHR. However, women who were screened more than once and/or had talked to their PCP about lifetime sexual trauma more often had MST captured in the EHR.

Our results align with previous literature showing higher disclosure of MST on research surveys compared to the EHR screen.11,19 The average time between EHR screening and our study survey was ~ 7 years for both participants who were captured and not captured, making it less likely that temporal socio-cultural changes impacted decisions to disclose. However, there have been considerable changes in the public discussion around sexual harassment and assault over the last several years which may influence disclosure.26 Similarly, the captured and uncaptured groups reported being equally comfortable talking to their PCP about emotional experiences.19,27 We were unable to explore other methodological differences between EHR screening and study surveys that could influence disclosure, such as the concern regarding having MST documented in the medical record and in-person direct inquiry compared to survey MST questions. Further, telehealth is becoming increasingly common in the VA, with unknown effects on MST screening.28,29 Rigorous qualitative studies are needed to further understand the nuanced decision-making processes underlying differences in disclosure of MST and how this may be impacted by telehealth. Importantly, several factors associated with having captured MST during EHR screening included women who reported that they had experienced stranger harassment in the VA, felt unsafe at VA, had greater distress during gynecology exams, delayed medical care due to concern for interacting with other Veterans, and preferred female providers. EHR screening may be better at capturing a subset of women who have these additional, often challenging, VA clinical experiences.

Few studies have examined differences in MST screening results by race and ethnicity. Those that have were done in smaller populations with less ability to detect subgroup effects.11 Our study showed that Black and Latina women had higher odds of not being captured during EHR MST screening. Black and Latina women also more frequently reported sexual harassment alone (without sexual assault), and Latina women were less often re-screened for MST. However, even after controlling for type of MST, frequency of screening, and VA experiences, Black and Latina women were less likely to disclose MST. Disclosure of MST has been associated with availability of a desirable disclosure recipient who has similar characteristics to the survivor.30 Discordant racial and/or ethnic identities between the provider or clinic staff and the patient may contribute to nondisclosure of MST in VA clinical screening. Among women survivors of intimate partner violence (IPV), there are mixed results regarding disclosure rates to healthcare professionals according to concordant race and ethnicity.31 However, trust in providers has been shown to be a strong indicator of disclosure.32 This trust may be disrupted due to experiences of historical, institutional, and/or interpersonal racism demonstrated by racial/ethnic health disparities,33 the history of racial violence in the military34,35 and the larger context of racism and unethical experimentation on people of color in the US medical system.36

Our results reinforce recent research showing that experiences of sexual harassment in the military without sexual assault are less often captured in EHR screening.11 Military sexual harassment includes a variety of adverse experience that have important health implications.37 Previous studies of sexual harassment have indicated that many women do not report it to authority figures because they do not believe that reporting will help,38 that their report will be taken seriously or that the disclosure would benefit them.14,30 Our findings contradict a study of MST which found that women with a history of assault MST were less likely to disclose.10 This contradictory finding could be attributable to using self-report of MST disclosure rather than a direct evaluation of EHR responses.

Women may be more likely to disclose a history of sexual violence if they have positive experiences and interpersonal connections with their healthcare providers.30,32 Cultural or gender similarities, frequent visits, and good rapport with providers have been shown to facilitate discussions about MST and IPV.30,32 In contrast, negative experiences and not feeling safe in the VA can contribute to non-disclosure of MST.12 However, we found that among Veterans with MST, those who experienced harassment in the VA and felt unsafe in the VA were more often captured by EHR screening. This finding may be because a greater proportion of participants captured by EHR screening also reported experiences of miliary sexual harassment and sexual assault (compared to sexual harassment alone), putting the women at higher risk of having negative experiences in the VA. Our multivariable models also indicated that women who reported talking to their PCP about lifetime sexual trauma (not limited to MST) had increased odds of having MST captured by the EHR.

Our study is the first to indicate that women who were screened for MST in the VA more than once were more likely to have their MST experience captured by EHR screening. Our findings are supported by literature showing that many people who have experienced sexual violence do not disclose it during a single clinical encounter, particularly during the first time meeting a provider.39 Over the last several years, the VA has made considerable efforts to improve MST screening and provide training and guidance for healthcare providers, which includes re-screening for MST. Re-screening is not specifically endorsed in VA screening policy and may vary by facility (Directive 1115 on MST).17 More than one screening may allow for more time to build longitudinal relationships with the healthcare team and facilitate disclosure of MST.

Limitations

Our study has several limitations. While within the typical range for academic research,37 our survey response rate was 45%, raising the potential for non-response bias. We therefore weighted our findings to address such potential bias. To minimize recall and social desirability biases, we used CATI with trained interviewers.40 Our “negative” MST category may have included women with MST histories who chose not to disclose these experiences in either EHR screening or the survey, resulting in misclassification. Our study population includes women with 3 + primary care visits in the previous year. Since women Veterans using VA have an average of three primary care visits per year, we consider these routine users. However, this cut point limits our ability to generalize to women who are less frequent VA users or who do not use VA for care. Further, it is important to explore disclosure of MST among men, Veterans with historically marginalized racial and ethnic identities that were not included in this study, as well as transgender and non-binary Veterans.

Clinical Implications

Our work is one of the largest and most diverse studies to evaluate differences in MST disclosure within the same cohort of women. It includes women from multiple geographical regions of differing resources, complexities, and sizes. Our findings are a reminder to healthcare providers that many women with a negative EHR screen have not disclosed their experience of MST. We add to current literature by indicating that MST experiences of Black and Latina women, and women who are screened only once, are less frequently captured by EHR screening. Our findings have potential implications for equitable access to healthcare given the dependence on a positive MST EHR screen to gain free resources for MST-related health conditions, including obtaining service-connection for disabilities related to MST.41,42

We confirm that there is under-reporting of MST in the EHR. However, our findings are not unique to the VA EHR or to women Veterans.43 Sexual trauma is likely under-disclosed in many clinical settings and patient populations, highlighting the importance of universally applied trauma-informed care38 for the VA and the broader healthcare community.