Walsh K, Danielson CK, McCauley JL, Saunders BE, Kilpatrick DG, Resnick HS. National Prevalence of Posttraumatic Stress Disorder Among Sexually Revictimized Adolescent, College, and Adult Household-Residing Women. Arch Gen Psychiatry. 2012;69(9):935-942. doi:10.1001/archgenpsychiatry.2012.132
Author Affiliations: Department of Psychiatry and Behavioral Sciences, Medical University of South Carolina, Charleston.
Context Despite empirical links between sexual revictimization (ie, experiencing 2 or more sexual assaults) and posttraumatic stress disorder (PTSD), to our knowledge, no epidemiological studies document the prevalence of sexual revictimization and PTSD. Establishing estimates is essential to determine the scope, public health impact, and psychiatric sequelae of sexual revictimization.
Objective To estimate the prevalence of sexual revictimization and PTSD among 3 national female samples (adolescent, college, and adult household probability).
Design Surveys were used to collect data from the National Women's Study–Replication (2006; college) as well as household probability samples from the National Survey of Adolescents–Replication (2005) and the National Women's Study–Replication (2006; household probability).
Setting Households and college campuses across the United States.
Participants One thousand seven hundred sixty-three adolescent girls, 2000 college women, and 3001 household-residing adult women.
Main Outcome Measures Behaviorally specific questions assessed unwanted sexual acts occurring over the life span owing to the use of force, threat of force, or incapacitation via drug or alcohol use. Posttraumatic stress disorder was assessed with a module validated against the criterion standard Structured Clinical Interview for DSM-IV.
Results About 53% of victimized adolescents, 50% of victimized college women, and 58.8% of victimized household-residing women reported sexual revictimization. Current PTSD was reported by 20% of revictimized adolescents, 40% of revictimized college women, and 27.2% of revictimized household-residing women. Compared with nonvictims, odds of meeting past 6-month PTSD were 4.3 to 8.2 times higher for revictimized respondents and 2.4 to 3.5 times higher for single victims.
Conclusions Population prevalence estimates suggest that 769 000 adolescent girls, 625 000 college women, and 13.4 million women in US households reported sexual revictimization. Further, 154 000 sexually revictimized adolescents, 250 000 sexually revictimized college women, and 3.6 million sexually revictimized household women met criteria for past 6-month PTSD. Findings highlight the importance of screening for sexual revictimization and PTSD in pediatric, college, and primary care settings.
Sexual victimization (SV) is an endemic societal problem associated with a range of mental health sequelae such as anxiety, depression, posttraumatic stress disorder (PTSD), substance abuse, interpersonal difficulties, and health problems including human immunodeficiency virus.1 Sexual victimization disproportionately affects girls and women, with female victims reporting more than 90% of all sexual assaults.2 Epidemiological studies have documented higher PTSD estimates among women,3,4 leading to some speculation that sex differences in the experience of SV account for higher estimates of PTSD observed among women.5 However, some studies have found higher estimates of PTSD to exist among women even after controlling for sex differences in exposure to different traumatic stressors.6 As recently noted,7 more research on the prevalence of PTSD following specific types of victimization experiences is warranted. One such common victimization experience that has received little attention in the epidemiological literature is sexual revictimization (ie, 2 or more sexual assaults/rapes across the life span).
Studies suggest that 20% to 25% of female children experience sexual abuse.8 Among college women, 15% to 20% report a rape or attempted rape during childhood, adolescence, or adulthood,9 and general population estimates suggest that between 13% and 25% of adult women will experience a sexual assault during their lifetimes.10,11 Robust associations between early sexual abuse and subsequent sexual victimization have been well documented among college and community women.12 Meta-analyses suggest a medium effect size (0.59) for the association between early SV and later SV, with stronger effect sizes (0.64) emerging when more restrictive definitions of victimization (eg, penetration) are used compared with broader definitions (0.38) of victimization (eg, exhibitionism).13 Although reviews suggest that 10% to 20% of adolescents who report child sexual abuse will be revictimized prior to age 19 years, as many as 2 of 3 women with a history of child sexual abuse report sexual revictimization.12 Despite cross-sectional and longitudinal evidence of the link between early SV and later revictimization, to our knowledge, there are no epidemiological studies establishing the prevalence of sexual revictimization. Indeed, using the search term sexual revictimization, 117 journal articles were abstracted in PsycINFO for the years 1981 to 2011 as of January 20, 2012; none describe the lifetime prevalence of sexual revictimization among representative samples of women. Thus, the present study addresses this gap by examining the prevalence of sexual revictimization in 3 different populations using national epidemiologic data from adolescent, college, and household-residing women.
As noted earlier, experiencing a single SV is associated with problems across a number of domains, including substance abuse, interpersonal problems, and psychiatric disorders (eg, PTSD, panic disorder, and depression). Reviews of research with non–nationally representative samples suggest that women who experience revictimization are at even greater risk for such problems, particularly PTSD, in comparison with singly victimized women.12 Posttraumatic stress disorder is costly to the individual, the family, and society at large given its association with increased use of the health care system and both the direct (medical/mental health treatment, pharmacological interventions, and case management) and indirect (eg, loss of wages and productivity) health care costs.14,15 Early screening and treatment may reduce this burden; however, the scope and severity of the problem have not been adequately described in past work. Although the National Women's Study, conducted in 1990, found that nearly one-third of rape victims developed PTSD,16 no representative studies have reported estimates of PTSD among sexually revictimized women specifically. The present study addresses this gap by using data from probability samples to explore the prevalence of PTSD among adolescent, college, and household-residing women reporting sexual revictimization.
The current study used data from 3 national probability samples of women (adolescent, college, and household-residing women) to better understand the prevalence of sexual revictimization as well as the prevalence of current and lifetime PTSD among revictimized women. For comparison, we also assessed PTSD prevalence among women reporting a single sexual victimization. Based on reviews of studies using samples of convenience,12 it was expected that 10% to 20% of sexually victimized adolescents would report revictimization and 60% to 70% of sexually victimized college and adult household-residing women would report revictimization. Further, given greater severity of PTSD symptoms among revictimized women,1 it was expected that estimates of PTSD among revictimized women would be substantially higher than the 30% prevalence estimates of PTSD observed for singly victimized women. Determining accurate estimates of the prevalence of sexual revictimization, as well as better understanding the role of sexual revictimization in the prediction of PTSD, will inform screening, assessment, and intervention efforts in this domain among mental health providers serving adolescent and adult female populations.
Data for the present study were drawn from 2 separate studies, encompassing 3 separate sampling frames: adolescent participants were part of the National Survey of Adolescents–Replication (NSA-R) and college and household-residing participants were part of the National Women's Study–Replication (NWS-R). All procedures for each of the 3 studies were approved by the institutional review board of the Medical University of South Carolina.
The NSA-R17 is a longitudinal, nationally representative study of adolescents aged 12 to 17 years (n = 3614 at wave 1) designed to assess the prevalence, risk factors, and mental health outcomes of exposure to potentially traumatic events. The study was conducted as a random digit dial telephone survey of households with children between the ages of 12 and 17 years and included an oversample of urban households. The random digit dial method involved use of telephone banks within specified geographic regions using the comprehensive database of telephone hundred banks (defined as each block of 100 potential telephone numbers with an exchange that includes 1 or more residential listings). Once a block had been selected, a 2-digit random number in the range of 00 to 99 was appended to the block to form a 10-digit telephone number. Once household eligibility was determined (ie, the home had at least 1 youth in the desired age range), screening and introductory interviews were conducted with parents to establish rapport. Parents were asked if the (randomly) selected child could also participate in the study and were provided the opportunity to call a toll-free number to confirm the authenticity of the study. When possible, adolescents were interviewed immediately following parent interviews. If adolescents were unavailable, interviewers scheduled appointments and/or called back at different times of the day or days of the week. After obtaining informed consent from a parent and assent from the adolescent, interviews were conducted using computer-assisted telephone interviewing technology by employees of Abt Schulman, Ronca, & Bucuvalas, Inc who were well trained, highly skilled, and experienced in conducting this type of interview. Adolescents were offered an incentive of $10 to complete the survey, which averaged 43 minutes. Of the 6694 parents who were interviewed, 5426 (81.1%) gave permission for a randomly selected adolescent per household to be contacted. A total of 3921 (72.3%) of adolescents with parental permission for contact were located during the field period and informed about the study. The remaining 1505 adolescents with parental permission could not be contacted during the field period. Of the 3921 adolescents who were contacted and informed about the study, 188 refused to participate, 119 started the interview but did not complete it, and 3614 completed the interview. Therefore, the percentage of completed adolescent interviews among households with eligible adolescents (ie, those with parental permission for contact; n = 5426) was 66.6%. The percentage of completed interviews among eligible adolescents who were contacted and informed about the study (n = 3921) was 92.2%, indicating that the vast majority of adolescents who could be contacted completed interviews. Only the 1763 NSA-R female participants at wave 1 (collected in 2005) were included in the present study. To correct for oversampling, data were weighted to bring the sample in line with the adolescent US population based on 2005 census data.17 Mean (SD) age of participants at wave 1 was 14.5 (1.71) years. Regarding racial/ethnic makeup, 69% were white, 13% were African American, 10% were Hispanic, 3% were Native American, and 3% were Asian/Pacific Islander. Demographic characteristics of the female-only sample did not differ significantly from the full sample.
The NWS-R18 is a telephone survey of the prevalence and characteristics of rape that was conducted in 2006. Following informed consent, the 20-minute structured telephone survey was administered by trained female interviewers at Abt Schulman, Ronca, & Bucuvalas, Inc using computer-assisted telephone interviewing technology. College participants were 2000 college women recruited from the American Student List. The American Student List included 6 million students who were attending approximately 1000 US colleges and universities. Abt Schulman, Ronca, & Bucuvalas, Inc purchased a sample containing nearly 17 000 women to generate responses that were similar to the national census representation of college women. Following classification of the sample by region, the list was released to be dialed in proportion to the national census (2000) representation of college women. The sample was classified into 9 regions: New England, Mid Atlantic, East North Central, West North Central, South Atlantic, East South Central, West South Central, Mountain, and Pacific. The sample was then released to be dialed in proportion to the national census representation of college women. This procedure was designed to ensure adequate representation to the US population of college women. There were 253 different colleges included in the sample from 47 different states. Of those contacted (n = 3805), 28.8% (n = 1094) were ineligible for participation. Among those eligible for participation (n = 2711), 8.9% (n = 240) refused to participate and 17.7% (n = 480) did not complete the interview; thus, the completion rate among eligible participants was 73.8%. Mean (SD) age was 20.1 (1.7) years with a range from 18 to 67 years. Approximately 75% (n = 1500) of the sample reported their race as white; 11.1% (n = 221) reported their race as black; 6% (n = 120), as Hispanic; 1.1% (n = 22), Native American; and 6%, (n = 120) Asian, and 0.4% (n = 8) chose not to report their race. For a detailed description of the methods, see the final report to the National Institute of Justice.19
A household probability sample of 3001 adult women also participated in the NWS-R telephone survey (see earlier for description). Whereas the college NWS-R participants were selected using the American Student List, the household-residing NWS-R participants were sampled via random digit dialing methods (see NSA-R sampling description). The use of this method introduces a randomization process in the selection of telephone numbers. Of those contacted (n = 15 982), 76.2% (n = 12 182) were ineligible for participation. Of those eligible for participation (n = 3817), 12.9% (n = 492) refused to participate and 8.5% (n = 324) did not complete the interview. Thus, the cooperation and interview completion rate among eligible participants was 78.6%. Because the majority of women in the general population sample were between the ages of 18 and 34 years (younger women were oversampled to assist comparisons with college women), weightings were created to approximate 2005 US census estimates,20 and the sample had a mean (SD) weighted age of 46.6 (17.87) years and an age range of 18 to 86 years. Approximately 78.2% (n = 2348) of the sample reported their race as white; 11.1% (n = 334) reported their race as black; 5.3% (n = 158), Hispanic; 1.9% (n = 57), Native American; and 1.7% (n = 50) Asian, and 1.7% (n = 54) chose not to report their race. Regarding education, 64.5% of community participants (n = 1936) had attended some college or beyond and 2.7% (n = 82) were enrolled in college at the time of the survey.
Consistent with previous studies (eg, the National Survey of Adolescents21 and the National Women's Study22), SV history was assessed herein using behaviorally specific, dichotomous questions regarding a series of unwanted sexual experiences, including (1) forced anal, vaginal, and/or oral sex; (2) forced digital penetration and/or foreign object penetration; and/or (3) either of the aforementioned events when the individual was voluntarily or involuntarily incapacitated by drugs and/or alcohol. Specifically, participants were asked (1) whether a man or boy ever made you have vaginal, anal, or oral sex (adolescent version = put his private sexual part inside your private sexual part, inside your rear end, or inside your mouth) when you didn't want to by using force or threatening to hurt you or someone close to you, (2) whether a man or boy ever made you have vaginal, anal, or oral sex when you didn't want to after you had taken or been given so much alcohol or drugs that you were very high, drunk, or passed out, (3) whether anyone (male or female) ever inserted fingers or objects into your vagina or rectum when you didn't want them to by using force or threatening to hurt you or someone close to you, or (4) whether anyone (male or female) ever inserted fingers or objects into your vagina or rectum when you didn't want them to after you had taken or been given so much alcohol or drugs that you were very high, drunk, or passed out. The NSA-R also included 2 specific questions about unwanted touching of the respondent's sexual parts (1) by using force or threat of force or (2) after the respondent had taken or was given so much alcohol or drugs that they were very high, drunk, or passed out. Because of NSA-R respondents' ages, even nonpenetrative sexual contact that is unwanted or perpetrated by a family member was considered abusive. For each screening event endorsed, participants also were asked whether the events they had experienced occurred once, twice, or 3 or more times. Participants were classified as revictimized if they endorsed experiencing 2 or more separate incidents.
The PTSD module of the National Survey of Adolescents21 and the the National Women's Study22 was used to assess current PTSD symptoms as well as functional impairment due to PTSD symptoms. This structured diagnostic interview assessed each DSM-IV symptom with a yes/no response indicating the presence of a symptom during the participant's lifetime as well as during the previous 6 months. This measure was validated against the PTSD module of the Structured Clinical Interview for DSM-IV administered by mental health professionals,23 and research provides support for its concurrent validity, temporal stability, internal consistency, and diagnostic reliability.21,24
Among adolescent girls, approximately 11.6% (n = 205) reported sexual victimization, with 5.5% (n = 97) reporting only 1 victimization and 6.1% (n = 108) reporting revictimization. Among victimized adolescents, 52.7% experienced revictimization. Among revictimized adolescents, 20% (n = 21) met criteria for past 6-month PTSD, and 28.8% (n = 30) met criteria for lifetime PTSD. Among single-assault victims, PTSD estimates were lower, with 13.7% (n = 13) of single-assault victims meeting past 6-month criteria for PTSD and 24.2% (n = 23) meeting lifetime criteria for PTSD (Table 1). The χ2 analyses revealed that revictimized adolescents were significantly more likely to report past 6-month PTSD (χ22 [n = 200] = 7.8; P < .05) and lifetime PTSD (χ22 [n = 199] = 11.3; P < .003) when compared with singly victimized adolescents.
Among college women, 12.5% (n = 250) reported at least 1 sexual victimization, with 6.3% (n = 125) reporting a single victimization and 6.3% (n = 125) reporting revictimization. Approximately 50% of victimized college women experienced revictimization. Among revictimized college women (n = 125), 40% (n = 50) met criteria for past 6-month PTSD and 58.4% (n = 73) met criteria for lifetime PTSD. In comparison, 21.6% (n = 27) of single-assault victims met criteria for past 6-month PTSD and 32% (n = 40) met criteria for lifetime PTSD (Table 1). Single-assault victims reported significantly lower estimates of PTSD. The χ2 analyses revealed that revictimized college women were significantly more likely to report past 6-month PTSD (χ21 [n = 252] = 10.4; P < .001) and lifetime PTSD (χ21 [n = 252] = 18.4; P < .001) when compared with singly victimized college women.
Among adult household-residing women, 20% (n = 600) reported at least 1 sexual victimization, with 8.2% (n = 247) reporting a single victimization and 11.8% (n = 353) reporting revictimization. Approximately 58.8% of victimized women experienced revictimization. Among revictimized women (n = 353), 27.2% (n = 96) met criteria for past 6-month PTSD and 45.6% (n = 161) met criteria for lifetime PTSD. In comparison, single-assault victims reported significantly lower estimates of PTSD, with 13.7% (n = 34) meeting criteria for past 6-month PTSD and 25.1% (n = 62) meeting criteria for lifetime PTSD (Table 1). The χ2 analyses revealed that revictimized household-residing women were significantly more likely to report past 6-month (χ21 [n = 600] = 13.9; P < .001) and lifetime (χ21 [n = 600] = 24.5; P < .001) PTSD when compared with singly victimized women.
Logistic regression analyses that included revictimization, single victimization, and age of participant revealed that the adjusted odds of meeting criteria for past 6-month PTSD were 5.1 for revictimized adolescents (β = 1.6; SE = 0.28; P < .001), 3.3 for singly victimized adolescents (β = 1.2; SE = 0.33; P < .001), and 1.3 for older adolescents (β = 0.21; SE = 0.07; P < .001) (Table 2). When logistic regression analyses predicting lifetime PTSD were conducted, a similar pattern of findings emerged such that the adjusted odds of meeting lifetime PTSD criteria were 4.3 for revictimized adolescents (β = 1.5; SE = 0.24; P < .001) and 3.5 for singly victimized adolescents (β = 1.2; SE = 0.26; P < .001). Age also was positively predictive such that older adolescents were more likely to meet lifetime criteria for PTSD (odds ratio = 1.2; β = 0.22; SE = 0.05; P < .001).
Logistic regression analyses revealed that the adjusted odds of meeting criteria for past 6-month PTSD were 6.7 for revictimized college women (β = 1.9; SE = 0.20; P < .001) and 2.8 for singly victimized college women (β = 1.0; SE = 0.23; P < .001). Age (β = 0.02; SE = 0.02; P = .31) was not associated with likelihood of meeting past 6-month PTSD criteria. When logistic regression analyses predicting lifetime PTSD were conducted, a similar pattern of findings emerged such that the odds of meeting lifetime PTSD were 8.2 for revictimized college women (β = 2.1; SE = 0.19; P < .001) and 2.8 for singly victimized college women (β = 1.0; SE = 0.20; P < .001). Age (β = 0.02; SE = 0.02; P = .24) was not associated with likelihood of meeting lifetime PTSD criteria.
Logistic regression analyses revealed that the adjusted odds of meeting criteria for past 6-month PTSD were 5.8 for revictimized women (β = 1.7; SE = 0.15; P < .001) and 2.5 for singly victimized women (β = 0.93; SE = 0.21; P < .001). Younger age also was significantly predictive of past 6-month PTSD (odds ratio = 0.96; β = −0.04; SE = 0.01; P < .001). When logistic regression analyses predicting lifetime PTSD were conducted, a similar pattern of findings emerged such that the odds of meeting lifetime PTSD were 5.9 for revictimized (β = 1.8; SE = 0.13; P < .001) and 2.4 for singly victimized (β = 0.86; SE = 0.16; P < .001) women. Younger age also was associated with lifetime PTSD (odds ratio = 0.98; β = −0.02; SE = 0.003; P < .001).
The sensitivity, specificity, positive predictive power, and negative predictive power of any victimization vs no victimization and revictimization vs single victimization among each of the 3 samples are summarized in Table 3. Relative to no victimization, any victimization was associated with sensitivity of 0.31 to 0.49, specificity of 0.83 or higher, positive predictive power of 0.17 to 0.45, and negative predictive power of 0.86 to 0.96. However, among victims, revictimization (relative to single victimization) was associated with sensitivity of 0.57 to 0.74, specificity of 0.45 to 0.62, positive predictive power of 0.20 to 0.58, and negative predictive power or 0.75 to 0.86.
To better understand whether age (older adolescence/early adulthood) or environment (college) are associated with increased risk for sexual victimization/revictimization and the development of PTSD, we examined associations between revictimization and PTSD among women from the household-residing sample who were within the ages of 18 to 34 years (n = 879). Among this subgroup, 10.7% (n = 94) had experienced a single victimization and 13.1% (n = 115) had been revictimized. Compared with singly victimized women, revictimized women were significantly more likely to meet criteria for past 6-month (χ21 [n = 879] = 9.8; P < .01) and lifetime (χ21 [n = 879] = 9.4; P < .01) PTSD. Specifically, 40% and 55% of revictimized women met criteria for past 6-month and lifetime PTSD, respectively, compared with 19% and 34% of singly victimized women. Approximately 9% (n = 78) of the women in this subgroup were currently in college. Although power to detect significant effects may have been low, there were no significant differences between women currently in college and those not in college in terms of victimization (χ21 [n = 879] = 0.50; P = .48), revictimization (χ21 [n = 879] = 0.6; P = .44), past 6-month PTSD (χ21 [n = 879] = 0.16; P = .67), or lifetime PTSD (χ21 [n = 879] = 0.005; P = .94).
Although sexual revictimization and PTSD have been the topic of myriad research articles during the last 3 decades, to our knowledge, no published studies to date have used national probability samples of adolescent girls and women to better understand the scope of the problem of sexual revictimization across the life span. Further, although many of these studies using samples of convenience have suggested strong linkages between sexual revictimization and PTSD symptoms, none have explored the prevalence of PTSD in relation to sexual revictimization among representative samples of female adolescents and women. Thus, the present study examined PTSD prevalence among sexually revictimized women using 3 national epidemiological samples: adolescent, college, and household-residing women. Approximately 11% of adolescents, nearly 13% of college women, and 20% of household-residing women reported any sexual victimization. Although these estimates are somewhat lower than those found in other studies,12 definitions used herein were more restrictive than those used in previous studies (eg, unwanted fondling, kissing, and verbally coercive tactics were excluded herein). In fact, with the exception of a small number of adolescent sexual assault cases, the vast majority of these incidents involved rape. Consistent with hypotheses, revictimization estimates among victims ranged from 52.7% for adolescents to 50% for college women and 58.8% for household-residing women. Estimates of revictimization appear to increase as a function of age, with older household-residing women having had the greatest opportunity for revictimization to occur when compared with adolescent or college respondents. However, samples are not directly comparable as the mean ages and age ranges of each sample vary. Nonetheless, high estimates of revictimization reported by adolescent, college, and household-residing participants highlight the importance of studying this topic across the life span.
Across all 3 representative samples, prevalence estimates of PTSD were higher among sexually revictimized participants when compared with singly victimized participants. Interestingly, PTSD estimates appear to be highest among revictimized college women, with 40% meeting criteria for PTSD in the preceding 6 months. Revictimized college women may have experienced more recent sexual victimization, and thus, these higher estimates may reflect the fact that they are still in the midst of coping with this acute trauma. However, 60% of revictimized college women met criteria for lifetime PTSD, perhaps suggesting a heightened vulnerability to experiencing psychopathology in the face of multiple stressors. Among revictimized adolescent girls and household-residing women, PTSD estimates also were notable, ranging from 20.0% to 45.3% depending on the assessment time frame. Estimates of past 6-month or lifetime PTSD among single-assault victims were substantially lower, ranging from 13.4% for adolescent and household-residing victims to 32.0% for college victims. These findings have clinical implications in that screening for multiple victimization experiences, as opposed to only asking about the presence or absence of sexual victimization, may enhance practitioners' abilities to focus on women most likely to be at risk for psychiatric problems. However, even among revictimized women, a substantial proportion about 40% of revictimized college women, about 55% of revictimized household-residing women, and about 72% of revictimized adolescents) do not report lifetime PTSD symptoms and an even greater proportion do not report current (past 6-month) PTSD. Thus, resiliency to PTSD in the wake of multiple sexual assaults is not uncommon, and studying factors that promote resilience among these groups may better inform PTSD treatment and revictimization prevention programs.
Revictimization consistently emerged as the strongest predictor of past 6-month and lifetime PTSD across all 3 samples when examined in relation to single victimization and participant age. Indeed, odds ratios suggested that revictimized adolescent, college, and household-residing women were between 5.1 and 6.7 times more likely to meet current PTSD criteria in comparison with non–sexually victimized women whereas singly victimized women were 2.5 to 3.3 times more likely to meet current PTSD criteria. Revictimized participants were 4.3 to 8.2 times more likely than nonvictims to develop lifetime PTSD whereas single-assault victims were only 2.4 to 3.5 times more likely to report lifetime PTSD when compared with nonvictims. Older adolescents and younger household-residing women were more likely than younger adolescents and older household-residing women to endorse both lifetime and past 6-month PTSD. These findings suggest that older adolescence/early adulthood is a period associated with the greatest risk for PTSD, likely because of heightened risk of sexual victimization and revictimization faced by women in this age range.25 However, age did not significantly predict lifetime or past 6-month PTSD for college women, perhaps because of the homogeneous and narrow age range of college participants in this sample. Interestingly, exploratory analyses suggest that age, rather than the college environment itself, may be the more salient predictor of revictimization. More specifically, early adulthood appears to be a developmental period during which women are at especially high risk for sexual revictimization, perhaps because of more frequent dating and social experiences that may not be unique to the college experience.
Sensitivity and specificity analyses suggested that the “true-positive” rate of PTSD in association with any rape or sexual assault ranged from 30% to approaching 50% within the household probability sample, disproportionate to the prevalence of SV within each population, which ranged from nearly 12% to 20%. Positive predictive power, or percentage with PTSD among those who would be predicted based on a positive screen, ranged from nearly 1 in 5 for current PTSD among those with any sexual violence within adolescent and household probability samples and 1 in 3 within the college sample to 1 in 4 within the adolescent sample to 45% within the household probability sample for lifetime PTSD. Among those who had experienced SV, revictimization provided additional useful information. While half of adolescent and college women and 58% of household probability sample women had experienced revictimization, sensitivity or true-positive rate of PTSD was 57% and 62% for lifetime and current PTSD, respectively, in the adolescent sample and ranged from 65% to 74% in the other 2 samples, indicating that two-thirds to three-fourths of PTSD cases occurred within the revictimized subgroup. Among victims, the sensitivity associated with revictimization appears to increase with age. Similarly, associated positive predictive power was relatively higher within college and household probability samples. These findings are consistent with the notion that these women may be more likely to have passed through the age ranges at highest risk for experiencing rape and revictimization if they are going to be exposed, and thus, more information is available regarding revictimization and risk of PTSD within the older samples.
Overall, findings suggest that there may be predictive incremental value in screening women for rape experiences that are most likely to be associated with PTSD by (1) assessing for any sexual victimization and (2) following affirmative responses with a query about revictimization. Although treatment approaches for PTSD symptoms emanating from single or multiple sexual assaults are not likely to be markedly different, a longer course of treatment may be useful in the case of revictimization, particularly if significant self-blame or safety concerns are present because of experiencing repeated victimization. Further, although responsibility for the assault lies solely with the perpetrator, victimized women may benefit from additional risk reduction strategies designed to help them identify risky situations, cope with distress in such situations, and respond assertively to risk cues to avoid further victimization. Indeed, PTSD symptoms have been suggested as a potential mechanism through which revictimization occurs26; thus, identifying and treating these symptoms may be one means of reducing risk for additional victimization among those most at risk.
The findings presented herein should be considered in the context of study limitations. First, respondents were asked whether they had experienced each type of unwanted sexual act once, twice, or 3 or more times and whether each incident was separate or part of an ongoing series of events. Although questions permitted women reporting separate acts that occurred twice or 3 or more times to be classified as revictimized, this approach did not permit analysis of the continuous number of victimizations experienced. Similarly, PTSD was assessed as a function of meeting criteria for the disorder either in the past 6 months or over the life span. Although this approach provides clinically useful data (particularly because functional impairment was included), our ability to explore the severity of PTSD symptoms was limited. Finally, although random digit dialing facilitates collection of representative data, participants without landlines may have been excluded; however, this concern is lessened by US Census Bureau reports that 91% of participants in the age ranges of interest had landlines in 2005 and 2006. Nonetheless, future efforts should attempt to contact potential participants without telephones or with cellular phones. Further, a proportion of girls and women deemed eligible for each study chose not to participate, and as such, we lack data to compare those who enrolled with those who declined participation on key study variables. However, the procedures used herein are commonly used to generate representative data with the understanding that participant consent introduces an element of self-selection bias. Nonetheless, future efforts should be made to increase the response rates across studies of this kind.
Despite these limitations, the present study illuminated the role that sexual revictimization plays in predicting a debilitating psychiatric disorder among women from a variety of backgrounds. Revictimized college women, derived from a population typically considered relatively high functioning, evidenced high prevalence of PTSD with functional impairment. These findings suggest the importance of developing and instituting evidence-based revictimization risk reduction programming in school and higher-education settings. Further, results point to a strong need for practitioners working in student health clinics to assess revictimization experiences and treat psychiatric problems emanating from these incidents. Early detection and treatment of psychiatric problems resulting from sexual revictimization may help to alleviate the public health burden of such experiences by preventing additional victimization experiences and improving college adjustment and performance among victims experiencing functional impairment. A substantial proportion of revictimized adolescent and household-residing women also reported PTSD symptoms with functional impairment, suggesting the need to screen for victimization and assess PTSD symptoms among these groups as well. Medical and mental health providers working in school-based and community clinics, particularly those serving older adolescents and young adult women (eg, Planned Parenthood clinics), may wish to consider adding a small number of screening questions to standard assessment procedures to ensure that girls and women in need of services receive appropriate assessment and referrals. The use of sensitive, behaviorally specific screening questions similar to those used in the current studies is recommended to assess incidents consistent with definitions of sexual violence and rape related to multiple tactics including drug or alcohol facilitation and incapacitation.27 Such behaviorally specific approaches are more likely to accurately identify those who have had such experiences as opposed to use of labels such as rape that may be unclear in terms of meaning and that may detect cases that are most consistent with stereotypic characteristics (eg, stranger assailant and presence of weapon) and that are not representative of all such cases.
Correspondence: Kate Walsh, PhD, National Crime Victims Research and Treatment Center, Medical University of South Carolina, 67 President St, 2 South, MSC861, Charleston, SC 29455 (firstname.lastname@example.org).
Submitted for Publication: August 2, 2011; final revision received February 6, 2012; accepted February 8, 2012.
Published Online: May 7, 2012. doi:10.1001 /archgenpsychiatry.2012.132
Author Contributions: Drs Walsh and Resnick had full access to the data and take responsibility for the integrity of the data and the accuracy of the data analysis.
Financial Disclosure: None reported.
Funding/Support: This study was supported by grant 5R01HD046830 from the National Institute of Child Health and Human Development (principal investigator [PI]: Dr Kilpatrick) and grant 2005-WG-BX-0006 from the National Institute of Justice (PI: Dr Kilpatrick). Manuscript preparation was partially supported by T-32 institutional training fellowship MH018869 (PI: Dr Kilpatrick) and National Institute on Drug Abuse grants R01DA023099 (PI: Dr Resnick) and K23018686 (PI: Dr Danielson).
Disclaimer: Views expressed herein are those of the authors and do not necessarily reflect those of the National Institute of Child Health and Human Development, National Institute of Justice, National Institute of Mental Health, National Institute on Drug Abuse, or other institutions.