Sexual violence is associated with a host of gynecologic problems,1,2 and increased risk of sexually transmitted diseases3 is a consequence of particular concern. Consistent receipt of pelvic examinations for routine surveillance and preventive care is important to the health and well-being of this population. However, despite their increased need for regular screening, women with a history of sexual violence may in fact be at risk for less regular care.4
One reason that women with prior exposure to sexual violence may receive less consistent surveillance is that core components of the pelvic examination, including insertion of the speculum, may function as powerful cues that can elicit a trauma response. This response, which may include a temporary increase in traumatic memories, intrusive thoughts, and even flashbacks of a prior assault,5 is often accompanied by intense negative affect, particularly among women with posttraumatic stress disorder (PTSD).6–8 As such, one might expect that women with a history of sexual violence, particularly those with PTSD, could experience significant distress during pelvic examinations.5,9,10 Such distress could, in turn, impede patients' ability to relax the pelvic muscles, leading to difficulty with speculum insertion and thus heightened examination-related pain.11 The experience of distress and pain during the pelvic examination may motivate women with sexual violence, particularly those with PTSD, to avoid future pelvic examinations.
Despite these theory-based considerations, the possibility that sexual violence may lead to increased distress and pain during pelvic examinations has received little attention in the research literature. To address this gap, we examined the following hypotheses: 1) distress and pain associated with the pelvic examination are more severe among female veterans with a lifetime history of sexual violence than among those without, and 2) examination-related distress and pain is highest among women with sexual violence–related PTSD.
MATERIALS AND METHODS
The present work is a cross-sectional cohort study in which participants using the Women's Health Center of a Veterans Health Administration Medical Center were recruited between October 2005 and October 2006. During this time, female veterans between the ages of 18 and 65 years who were scheduled for a women's health visit (to include a pelvic examination) were considered for participation in this cross-sectional cohort study. Initial assessment for eligibility involved a medical record review. This yielded information about age, racial/ethnic background, physical (including gynecologic history, eg, prior hysterectomy) and mental health status. All participants were seen by the same female nurse practitioner for their women's health visit. During each visit, the nurse practitioner also conducted a brief, in-person screening to assess the woman's eligibility to participate. Figure 1 presents an overview of our recruitment and retention process.
Upon conclusion of the visit (Time 1), potentially eligible participants were further screened by the first author, a clinical psychologist with expertise in the diagnosis and treatment of PTSD in women. Eligible women who wished to participate provided written consent and were enrolled in the study which was approved by the Stanford University Institutional Review Board. Participants then completed a series of self-administered questionnaires that included retrospective reports of distress and pain associated with specific components of that day's pelvic examination. Approximately two weeks after their examination (Time 2), the first author contacted the participant by telephone. During this telephone call she (first author) administered the Clinician Administered PTSD scale,12 a semistructured clinical interview that allowed for the formal evaluation of PTSD.
Despite the use of two time points, we still consider our design to be cross-sectional for the following reason. Our PTSD interview assessed 1) the nature and timing of the traumatic event and 2) the onset and duration of trauma symptoms. As such, we were able to determine that all instances of sexual violence and the onset of PTSD, when present, occurred before each woman's pelvic examination. Having two assessment points may have eased initial participant burden, but it increased risk of attrition. However, significant efforts, including multiple call attempts, were made to ensure maximal retention of participants.
Female veterans were eligible to participate if 1) they were between the ages of 18–65 years and 2) were scheduled for a women's health visit that was to include a pelvic examination for routine surveillance and preventive care. Participants who met these initial criteria were ineligible if they 1) were psychiatrically or medically unstable; 2) had a prior hysterectomy; 3) were currently pregnant; 4) did not have capacity for autonomous consent (eg, any diagnoses of cognitive impairment); 5) had previously served as subjects in the pilot phase of this project; or 6) had a visit that did not fall within a predetermined selection interval.
Women were deemed ineligible to participate due to psychiatric instability if one or more of the following criteria were met: 1) receipt of inpatient psychiatric care, treatment for alcohol/drug abuse, or presentation of suicidal or homicidal ideation within the past 90 days; 2) prior diagnosis of a psychotic spectrum disorder; or 3) history of conservatorship. Medical instability was defined as 1) acute medical illness or injury requiring immediate intervention or 2) presence of a terminal illness. Psychiatric and medical instability were determined by chart review and a brief assessment conducted by the nurse practitioner at the beginning of each woman's visit.
Regarding the last exclusion criterion, selection of participants by appointment interval, we determined through pilot testing that it was not feasible for a single interviewer to assess all eligible women attending the clinic on a particular day. The time needed to administer the consent form and the study instruments for one patient sometimes exceeded the time that it took for the following patient to complete her clinic visit and depart. To avoid selection bias, ie, the risk that women with briefer visits would have less opportunity to participate, we devised a systematic strategy that allowed for the unbiased selection of a subset of participants attending clinic on any particular day. First, all eligible women were approached sequentially in their order of departure from the examination room until we obtained our first study enrollee. Next, the second participant was selected only if she met all eligibility requirements and her clinic visit concluded at least 30 minutes after the first participant had completed data collection. All subsequent participants were selected using this strategy.
Of the 94 eligible women, 68 (73%) agreed to participate at Time 1—immediately after the pelvic examination. Fifty completed the Time 2 assessment, which yielded an attrition rate of 26%. However, upon inspection of the data we discovered that only one participant had PTSD but no exposure to sexual violence. To allow subgroup analyses (described later), data from this participant was excluded from all analyses, leaving 67 women at Time 1 and 49 at Time 2.
In an effort to determine if prior exposure to sexual violence or PTSD status affected participation, we used data from a chart review. Veterans Health Administration patients are screened for exposure to military sexual trauma or sexual assault or harassment experienced during their military tour of duty. This variable, along with any chart documentation of prior diagnosis of PTSD, was used to compare women who participated and those who declined. No difference in rate of military sexual trauma was found between women who participated compared with those who declined: 36% compared with 33%, respectively, P=.92. Similarly, no significant differences emerged between groups regarding PTSD status: 26% compared with 21% respectively, P=.61.
We compared exposure to sexual violence and PTSD among the 49 women (72%) who completed the Time 2 assessment (and were retained in the study) and those who dropped out. Using their responses to the Sexual Experiences Survey,13 described below, we found that 67% of dropouts and 82% of completers had a history of sexual violence, a nonsignificant difference P=.09. Similarly, 23% of dropouts and 32% of completers had a diagnosis of PTSD documented in their medical records, P=.86.
Upon completion of the study, we compared data from the medical record review with the results of our own evaluation of patients' history of exposure to sexual violence and current PTSD status. Concurrence between military sexual trauma (chart review) and military sexual assault reported on the Sexual Experiences Survey was 74%; concurrence between medical record evidence for PTSD and the results of our assessment was 64%.
Demographic data, including patient age, race, and ethnicity were extracted from chart review and confirmed with patients' self-reports during the Time 2 interview. Information on race and ethnicity was included to ensure that the sample was representative of the typical Veterans Health Administration patient population.
Sexual violence was measured using the Short Form (Version V) of the Sexual Experiences Survey.13 This six-item scale assesses lifetime exposure to sexual assault (ie, coercive or otherwise nonconsensual sexual intercourse that is either attempted or completed against the woman's will). We developed a binary (Yes/No) variable, any sexual violence, where a “yes” indicates one or more lifetime instances of attempted/completed sexual assault. One participant initially did not endorse any instances of sexual violence on this scale. However, during the subsequent clinical interview she disclosed that she was raped in childhood, and thus we included her in our sexual violence group.
The Clinician Administered PTSD Scale12 is a semistructured clinical interview designed to establish a diagnosis of PTSD; it has excellent psychometric properties.12 Patients were considered to have PTSD if their symptom ratings were consistent, by conventional scoring standards, with a current (ie, past month) diagnosis of the disorder.
Women rated their level of emotional distress for various components of their women's health visit on an adapted Subjective Units of Distress Scale.14 The stem for this 7-item scale was: “Thinking about the examination that you have just completed, please use the scale provided below to rate your level of anxiety for each component.” Ratings were anchored on a 6-point Likert-type scale, with “0” being “not at all anxious” and “5” being “very anxious.” (Ratings were subsequently transformed to an 11-point scale for analyses so they could be more directly compared with ratings of pain.)
Three items from this scale tapped distress related to specific components of the pelvic examination: “manual pelvic examination,” “lying in stirrups,” “insertion of the speculum into vagina.” The mean response across these three items became our measure of “distress associated with the pelvic examination.” The remaining items tapped distress related to elements of the visit that were unrelated to the pelvic examination: “disrobing/taking off my clothes,” “getting weighed,” ”having blood pressure taken,” “discussing health history/concerns with my doctor.” The mean response across these three items became our measure of “other distress.”
Immediately upon completion of the visit, patients also rated their perceptions of pain experienced during speculum insertion into the vagina. Using an established technology for assessing pain,15 ratings were anchored on an 11-point Likert-type scale with “0” being “no pain” and “10” being “severe pain.” Hilden and colleagues10 used a similar scale and rating technique to assess pain associated with pelvic examinations.
Before beginning our study, we piloted our research protocol on 20 women veterans who met study eligibility criteria. This allowed us to 1) assess the readability, comprehensibility, and face validity of our instruments; 2) monitor participant reactions; and 3) incorporate their feedback into the development of our final set of instruments. Moreover, the pilot study enabled us to optimize the logistics (ie, screening, selection, administration of measures and follow-up telephone calls) of our research protocol.
Our analyses were influenced by the distributions on the core predictor variables of sexual violence and PTSD status within our sample. As noted earlier, preliminary descriptive analyses revealed that, in all cases but one, women with PTSD also had a prior history of sexual violence. Data from this participant was excluded from all analyses. As such, traditional interaction models that could have assessed PTSD as a potential moderator of the relationship between sexual violence and distress and/or pain associated with the pelvic examination or pain were not feasible.
Given factors such as our sample and use of measures featuring ordinal scales, we applied nonparametric statistical techniques to minimize statistical inference problems. Preliminary analyses compared women with and without sexual violence on demographic variables, including race, ethnicity, and age using χ2 and nonparametric rank sum tests. To evaluate our first hypothesis, we performed the Mann–Whitney U test on distress and pain associated with the pelvic examination between those with and without sexual violence (N=67). To evaluate our second hypothesis we compared differences in distress and pain across three groups of women (those with no sexual violence, those with sexual violence but no PTSD, and those with sexual violence and PTSD) using the Kruskal-Wallis test. These latter analyses were restricted to the subset of patients for whom PTSD status was known (n=49).
Graphical representations of the data are presented in box plots, which depict the distribution of data and compare these distributions across groups. The “boxed area” of a plot represents the location of the 50% of the data that falls between the 25th and 75th quartile, also known as the interquartile range. The median, or 50th percentile, is depicted with a heavy line, whereas the mean is indicated by a triangle. The “whiskers,” or dashed lines leading up from the box top (75th quartile) and down from the bottom of the box (25th quartile), illustrate the complete range of responses, and empty circles on the plots represent extreme outliers.
For all analyses, α was set at P<.05. Analyses were conducted in SPSS 11.5 for Windows (SPSS Inc., Chicago, IL); R 2.4.1 was used to generate the graphics. Our outcomes of distress and pain associated with the pelvic examination were correlated (P<.001), but the overlapping variance (36%) was sufficiently low that we analyzed them as separate endpoints.
Rate of sexual violence was 77% (95% confidence interval 0.67–0.87) in our population. Demographic characteristics of our sample, stratified by sexual violence status, are presented in Table 1. There were no statistically significant differences in age, race, or ethnicity between women with and without prior exposure to sexual violence.
As expected, women with prior sexual violence had higher ratings of distress (median 4.27) than did those with no sexual violence (median 0), P=.03 (Fig. 2). Moreover, inspection of the medians revealed that among women with prior sexual violence, ratings of distress associated with the pelvic examination (median 4.27) were higher than those for “other distress” (distress with elements of the visit unrelated to the pelvic examination) (median 1.37) P<.001). For women with no prior exposure to sexual violence, median scores for distress associated with the pelvic examination and “other distress” were both “0.”
Ratings of pain associated with speculum insertion were generally low in comparison with distress, but the difference between those with (median 2.5) and those without sexual violence (median 0) was significant, P=.04 (Fig. 2).
The analyses for Hypothesis 2 were restricted to the subset of women evaluated for PTSD, n=49. The women were grouped into three categories: 1) those with no sexual violence or PTSD (“no sexual violence”) (n=9); 2) those with sexual violence but no PTSD (“sexual violence only”) (n=25); and 3) those with both (“sexual violence plus PTSD”) (n=15).
As shown in Figure 3, women with sexual violence plus PTSD (median 5.49) had the highest ratings of distress associated with the pelvic examination, those with sexual violence only (median 2.44) the next highest, and those with no sexual violence and no PTSD (median 0) the lowest, P=.015. Women with sexual violence plus PTSD reported significantly higher ratings of examination-related distress compared with women with sexual violence only, P=.02 and to those with no sexual violence, P=.02. Regarding the last comparison in which ratings of distress were compared among women with sexual violence only and those with no sexual violence at all, results showed a high number of ranked ties, suggesting that this P value is of limited interpretability. As an alternative, we also provide the P value for a t test for this comparison, P=.006. Interestingly, no statistically significant differences emerged between women with sexual violence only and those with no sexual violence, P=.32, although the ability to detect a difference between these subgroups was limited by sample size.
Ratings of pain with speculum insertion were relatively low compared with distress ratings, and no differences among the three subgroups emerged P=.49 (Fig. 3).
We found that both sexual violence and PTSD status adversely affected female veterans' reactions to pelvic examinations during routine preventive care. Prior exposure to sexual violence was related to higher ratings of distress associated with the pelvic examination, and PTSD accounted for additional variation in distress above that linked to sexual violence. Higher ratings of pain were also found among women with sexual violence compared with those without. However, PTSD was not linked with increased pain from speculum insertion beyond that accounted for by sexual violence; limited sample size and thus power may have precluded detection of this effect.
Our findings are consistent with the handful of prior studies that have shown a relationship between sexual violence and distress5,9,10 and pain10 associated with the pelvic examination. Our findings also extend the current literature by identifying a high-risk subgroup of women with sexual violence exposure—those with current symptoms of PTSD—who are most at risk for distress during the examination. Thus, although sexual violence alone is related to examination- related distress, clinicians need to be particularly attentive to the potential for distress among sexual trauma survivors with PTSD.
Also, as shown in both Figures 2 and 3, the range of distress and pain responses in women with and without sexual violence was substantial. Some women with no sexual violence history experienced examination-related distress/pain. Conversely, some women with a history of prior sexual violence did not experience examination-related distress/pain. Thus, clinicians also need to be aware of the variability and individuality in responses among women, even among those with a known trauma history. Future research that empirically explores additional factors (beyond trauma history and diagnostic status) that may affect women's examination-related distress is needed.
Several caveats about our study methodology warrant discussion. First, although our design allows for the determination of temporal precedence of core variables, confidence in the causal nature of the relationships is limited. For example, we know that our participants' exposure to sexual violence and diagnoses of PTSD preceded the pelvic examination. However, other variables may have been confounded with these two predictors in our sample. In addition, we were not able to determine whether sexual violence and PTSD interact in accounting for increased distress and pain, because no one in our final sample had PTSD without sexual violence. We also did not identify additional sources of distress during pelvic examinations; thus our ability to determine the effect of sexual violence and PTSD—relative to other factors—on distress is limited. Finally, the size of the subcohort of women with no sexual violence, particularly at Time 2, was small. The limited sample size and statistical power may have affected our ability to detect some differences.
Furthermore, the rate of sexual violence observed in our sample was very high. To an extent, this is consistent with an empirical literature documenting a disproportionately high rate of exposure to sexual violence in women veterans using Veterans Health Administration facilities relative to their civilian peers.16–18 However, our observed rate is higher than reported in some well-known prior studies.16 The rate we report is consistent with findings in a recent study of women veterans using Veterans Health Administration facilities,17 which also used the Sexual Experiences Survey13—a highly sensitive instrument that evaluates lifetime exposure to sexual assault. Thus, we suspect that our observed rate of sexual violence reflects the use of a highly sensitive measure, not sampling bias. Nevertheless, the high rate of sexual violence could raise some concerns about the generalizability of our findings. Sexual violence and PTSD are also widespread in the general population of women,18–19 however, and the clinical presentation of sexual assault-related PTSD would be expected to be the same in all women.6 Thus, although our study needs to be replicated in other populations, it is likely that our findings apply to civilian women as well.
The use of a single provider to conduct the women's health visits (including the pelvic examinations) and a single investigator (first author) for data collection is both a strength and limitation of our work. Although both provide greater consistency of experience across participants, they also limit the generalizability of findings. Finally, for ethical reasons, we excluded 25 potential subjects because they were psychiatrically unstable at the time of the visit. It is likely that severity of psychiatric illness was greater in these women; excluding them may have reduced the magnitude of observed effects. Replication of our findings in larger, more diverse samples (including civilian samples) is needed.
Core aspects of our methodology strengthen the rigor of our study and allow our findings to extend the established literature in several ways. First, unlike prior studies focusing on specific types of sexual violence (ie, childhood exposure or adult exposure only)5,9 we conducted a comprehensive assessment of lifetime exposure to attempted and completed sexual assault. Also, our study simultaneously evaluated lifetime exposure to sexual violence and current PTSD status. This allowed us to determine the contribution of each variable to our outcomes of interest. We also were able to benchmark distress associated with the pelvic examination against distress associated with other parts of the visit. Our careful sequential sampling technique reduced the risk of introducing bias and would be expected to improve the validity of findings. Finally, our methodology, which included extensive pilot testing of the multistep protocol, assessment of distress immediately after the examination, use of validated instruments, in-depth inquiry by a psychologist in a safe setting, and attention to maximizing response rate, would be expected to yield high-quality data.
Taken together, our results suggest several immediate clinical and research implications. First, although sexual violence and PTSD are often linked with distress experienced during the pelvic examination, they are not universally so. Thus, although clinicians should have heightened sensitivity to the possibility that women with prior exposure to sexual violence, especially those with PTSD, may experience distress associated with the pelvic examination, their approach to each particular woman should be individualized. Toward this end, increased efforts directed toward screening for and understanding the effect of sexual violence and PTSD on the lives of individual women seeking preventive gynecologic care is warranted.
Second, clinicians who administer pelvic examinations need to be prepared to anticipate, prevent, and actively manage distress when it does occur during pelvic examinations. Carefully explaining what to expect during the procedure, avoiding rushing the patient, and assuring the patient that the examination can be stopped at any time may help to prevent distress. Tailored training regarding the management of distressed patients, particularly those with PTSD, and consultation with mental health providers might be productive steps toward this goal.
Finally, more research is needed to fully understand the relationship of sexual violence and PTSD and emotional reactions to and pain in the pelvic examination. Efforts also are needed develop, evaluate, and implement therapeutic interventions specifically designed to help women with prior sexual violence, particularly those with PTSD, learn skills to help them better tolerate pelvic examinations. Interdisciplinary collaboration that draws upon the combined expertise of providers of women's health care and mental health would facilitate this aim.