Sexual Assault in Women Veterans: An Examination of PTSD Risk, Health
Care Utilization, and Cost of Care
ALINA SURı ´S, PHD, LISA LIND, PHD, T. MICHAEL KASHNER, JD, PHD, MPH, PATRICIA D. BORMAN, PHD, AND
FREDERICK PETTY, PHD, MD
Objective: This study examines the differential impact of military, civilian adult, and childhood sexual assault on the likelihood
of developing posttraumatic stress disorder (PTSD). It also examines the relationship of military sexual assault (MSA) to service
utilization and health care costs among women who access services through Veterans Affairs (VA). Methods: A convenience
sample of 270 veteran women receiving medical and/or mental health treatment at the VA North Texas Healthcare System
participated in the study. Participants were interviewed using the Clinician Administered PTSD Scale (CAPS) and categorized into
a sexual assault group using the Interview of Sexual Experiences (ISE). A chart review was also conducted to determine the
frequency of diagnoses among the women. Data regarding health care utilization was obtained from self-report using the Utilization
and Cost Patient Questionnaire (UAC-PQ) and VA administrative records. Results: Compared with those without a history of
sexual assault, women veterans were 9 times more likely to have PTSD if they had a history of MSA, 7 times more likely if they
had childhood sexual assault (CSA) histories, and 5 times more likely if they had civilian sexual assault histories. An investigation
of medical charts revealed that PTSD is diagnosed more often for women with a history of MSA than CSA. CSA was associated
with a significant increase in health care utilization and cost for services, but there was no related increase in use or cost associated
with MSA. Conclusion: Women veterans have differential rates of PTSD due to sexual assault, with higher rates found among
those assaulted while on active duty. Although women with MSA are more likely to have PTSD, results suggest that they are
receiving fewer health care services. Key words: military sexual assault, posttraumatic stress disorder, women veterans, health care
utilization, health care costs, and sexual assault.
PTSD ? posttraumatic stress disorder; MSA ? military sexual
assault; CSA ? civilian sexual assault; ChSA ? childhood sexual
assault; VA ? Veterans Affairs; UAC-PQ ? Utilization and Cost
Patient Questionnaire; ISE ? Interview of Sexual Experiences;
CAPS ? Clinician Administered PTSD Scale; NSA ? no sexual
assault; SD ? standard deviation; ER ? emergency room; SE ?
sexual violence occurs against women of all age groups and
occupations, a review of the literature indicates that the rates
of sexual assault for women veterans is significantly higher
than for civilian women, with about 1 in 4 experiencing sexual
assault during their lifetime (3,4). Recent studies suggest that
rates of posttraumatic stress disorder (PTSD) among female
veterans or active duty members are also higher than those found
in national civilian studies (5–7), with rape being the most fre-
quent traumatic event associated with the onset of PTSD in
women (8). However, no study has specifically examined if
military status is an important risk factor for PTSD.
Research indicates strong relationships between PTSD and
several other mental disorders, including alcohol abuse and
pproximately 1 in 10 women report being sexually as-
saulted at least once in their lifetime (1,2). Although
dependence, major depression, and panic disorder (5,8). Re-
search also indicates that PTSD tends to become chronic and
less amenable to treatment when the symptoms endure for
more than 3 months (9). The sequelae of PTSD have been
linked to a variety of additional problems, such as impair-
ments in physical health and social and occupational function-
ing, and also to multiple costs to society, such as unemploy-
ment, lost work time, and increased health care utilization and
The persistent effects of sexual trauma on psychological
and physical functioning can result in increased rates of men-
tal health and medical service use (12). Women who experi-
ence sexual assault tend to be high health care utilizers (13)
and are more likely than men to talk to their doctor about their
symptoms and to take more medication (14). For example,
Waigandt et al. (14) found that that sexually assaulted women
made significantly more visits to their physicians per year than
did nonvictims. In addition, sexual assault victims are more
likely to see a physician than they are to seek help from legal
aid, mental health services, or victim’s assistance services
(13). In a study examining utilization of a crime victim’s
compensation program, sexual assault and the presence of
PTSD were associated with greater utilization of services
compared with other types of crime or diagnoses (15).
Women veterans with military sexual assault (MSA) report
significantly more physical symptoms and have poorer overall
health functioning compared with women veterans with no
MSA (16,17). Poorer overall health functioning, depressive
symptomatology, and alcohol problems in women reporting
MSA compared with veterans with no MSA history have been
found, suggesting that there are significant differences be-
tween veterans with MSA and no MSA (18). However, a
limitation of this study is that PTSD was not assessed, and
assumptions about differential rates of PTSD and its relation-
ship to health outcomes cannot be made based on the study
design and data collected.
From the Department of Veterans Affairs (A.S., L.L., T.M.K.), Research
Service, North Texas Health Care System, Dallas, Texas; the Department of
Psychiatry (A.S., T.M.K., P.D.B.), University of Texas Southwestern Medical
Center, Dallas, Texas; the Department of Psychiatry (F.P.), Creighton Uni-
versity, Omaha, Nebraska; and the Nebraska/Western Iowa Health Care
System (F.P.), Omaha, Nebraska.
Address correspondence and reprint requests to Alina M. Suris, PhD,
Veterans Affairs North Texas Healthcare System, Mental Health, 4500 South
Lancaster Road (151), Dallas, TX 75216. E-mail: email@example.com
Received for publication January 23, 2004; revision received May 7, 2004.
Portions of this manuscript were presented at the 2001 International Society
for Traumatic Stress Studies, December 6–9, New Orleans, Louisiana.
This study was partially funded by a grant from the Department of Veterans
Affairs Veterans Integrated Service Network –17 (Protocol #98-125 to A.S.)
and an HSR&D Career Scientist Award (RCS 92-403 to T.M.K.).
749 Psychosomatic Medicine 66:749–756 (2004)
Copyright © 2004 by the American Psychosomatic Society
Although sexual assault in any environment can be asso-
ciated with detrimental effects, one can hypothesize that the
consequences of MSA may differ from nonmilitary sexual
assault. The nature of the perpetrator–victim relationship has
been found to be associated with the severity of subsequent
symptoms (19) and in MSA, the woman’s perpetrator may be
a coworker, supervisor, or personnel with higher rank. The
woman veteran may be required to continue working with her
perpetrator, which is less likely to occur in many civilian
situations. Although today’s military is an all-volunteer force,
military personnel are not considered to be volunteers in the
usual sense of the word. They are unable to leave their duty
stations without permission and they are subject to disciplin-
ary action, including court marshal, if they attempt to leave.
Consequently, women who are sexually assaulted are not able
to transfer easily to another duty station or to quit their jobs.
Therefore, they are often forced to have repeated contact with
the perpetrator. This is not the case in the civilian work
environment. In addition, the unit cohesion that usually pro-
vides a protective barrier in the military setting may not be
available to a woman who has been assaulted by another
member of the unit (20). It has been hypothesized that such
unique aspects of the military system might intensify the
severity of symptoms seen after sexual assault (16), especially
given that military personnel have no down time to process the
trauma because they are always considered to be on duty.
Sexual trauma program practitioners report clinical differ-
ences in diagnosing and treating different types of sexual
trauma (MSA, civilian sexual assault [CSA], and child sexual
assault [ChSA]), but there is a paucity of scientific literature
examining these differences. Specifically, there are no known
systematic clinical studies investigating the consequences of
MSA verses CSA and how survivors of these assault types
may differentially develop PTSD. In addition, there are no
known published studies that examine the effect of MSA on
health care utilization and costs. Examining the relationships
between MSA and mental health variables is especially im-
portant given the large number of women currently serving in
Afghanistan and Iraq.
This study assesses the differential impact of MSA, CSA,
and ChSA on the likelihood of developing PTSD. A secondary
aim is to examine the relationship of trauma history on health
care utilization and cost of care among women veterans who
access care through Veterans Affairs (VA). Primary hypoth-
eses include the following: (1) sexual assault among women
veterans is a risk factor for PTSD; (2) women veterans who
have experienced sexual assault are more likely to have PTSD
if they were sexually assaulted while on active duty; and (3)
MSA is associated with increased levels of health care utili-
zation and cost.
Both use of VA and non-VA sources of care were consid-
ered. Use of VA care was measured from provider-based VA
files. Use of non-VA sources of care was measured by the
Utilization and Cost Patient Questionnaire (UAC-PQ) (21,22).
The reliability of the UAC-PQ applied to women veterans
with trauma was assessed by comparing VA provider records
with UAC-PQ self-reports on VA care. We also examined
medical chart diagnoses among the three groups to examine
diagnoses and corroborate our findings.
Eligible participants were female veterans enrolled in a medical and/or
mental health clinic within the VA North Texas Health Care System who were
seen for at least one outpatient appointment during the 5 years before contact.
We did not include non-VA users because we were interested in the health
care utilization and cost of care for women currently seeking VA care.
Participants were recruited by the research coordinator, who was a doc-
toral level psychologist, between 1997 and 2000. Recruitment procedures
involved advertising for the study via fliers at the Dallas VA Medical Center
and approaching women veterans at all medical and mental health clinics
within the medical center. Those women veterans who became aware of the
study from the advertised fliers were informed to call the study coordinator for
more information. When these women called, the study coordinator read a
description of the study, answered questions about enrollment, and screened
the women to ensure that they were female veterans enrolled at the VA North
Texas Health Care System. Those women approached in one of the clinics
within the Dallas VA Medical Center were given a description of the study
and had enrollment questions answered by the research coordinator.
With both recruitment methods, participants were told that we were
examining health, quality of life, use of health care services, and trauma
history including sexual trauma (results of health and quality-of-life data are
reported elsewhere). If the woman veteran was interested in participating in
the study, an appointment was made to obtain informed consent and complete
the study measures.
A total of 385 women veterans were recruited. Of this number, 77 (20%)
declined to participate, 31 (8%) scheduled appointments to participate but did
not show up, and 7 (2%) cancelled appointments and did not reschedule,
leaving a final sample of 270 women (70%).
After obtaining informed consent, subjects were interviewed about their
sexual assault histories. Data regarding utilization of care was obtained from
both administrative records and participant self-report. Assault history was
obtained before the standardized assessment for PTSD. Due to the nature of
this study, the research coordinator and the two interviewers were females
with master’s level or doctoral degrees in clinical/counseling psychology or
A review of the literature indicated a lack of instruments that included
MSA as one of the types used to categorize subjects into groups according to
sexual assault history. Previously published instruments reviewed were found
to differentiate only between adult sexual assault and ChSA. A frequently
used measure, the Sexual Experiences Survey (23), did not allow the ability
to capture qualitative information that differentiated trauma types, which was
needed for this study. As a result, we designed Interview of Sexual Experi-
ences (ISE). The ISE provides the patient with specific definitions and
examples of sexual harassment and assault before asking questions about
exposure in each of the following three settings: (1) as a child younger than
the age of 14, (2) as a civilian adult, not serving in the military, and (3) as an
adult on active duty in the military. The definitions used come from the
National Women’s Study (24) and are as follows.
Uninvited and unwanted sexual advances, physical contact, verbal com-
ments, and/or similar behavior of a sexual nature. Some examples are de-
mands for sexual favors, jokes, references to body parts, innuendoes, and
A. SURI´S et al.
750 Psychosomatic Medicine 66:749–756 (2004)
Any type of sexual conduct including vaginal, anal, or oral sex, achieved
or attempted without the person’s consent and with the use of threat or force.
Unwanted sexual attention or talk includes things such as demands or
suggestions for sexual favors, unwanted phone calls, being followed, or
whistles, jokes, looks, and gestures.
Unwanted sexual touching includes things such as being patted on the
bottom, being rubbed up against, and being fondled.
Unwanted vaginal sex includes sexual intercourse (penis inserted in
vagina) or having items (or fingers) inserted into the vagina. Unwanted oral
sex includes being forced to take a man’s penis in the mouth or being forced
to submit to him performing oral sex on oneself. Unwanted anal sex includes
having a penis or any other object (including fingers) inserted in the anus.
To determine the validity of the ISE, we administered the sexual assault
items from the event history assessment used in the National Women’s Study
(24). This allowed for a comparison of the frequency of unwanted sexual
touching and unwanted sex across instruments. Agreement occurred in 96%
of cases. If adjusted for chance, the agreement was 93% (? ? 0.934
[?0.023]), with a sensitivity of 0.97 and specificity of 0.97. This reflects very
high agreement and suggests the presence of concurrent validity.
For the purposes of this study, veterans were classified as having ChSA if
they reported at least one sexual assault while they were less than 14 years old
and had no MSA or CSA histories. If they reported at least one sexual assault
that occurred while they were adults and not on active duty, and they had no
ChSA history, they were classified as CSA. Veterans were classified as
having MSA if they reported at least one sexual assault that occurred while on
active duty. Women veterans with both ChSA and CSA histories were
categorized in the ChSA group. Patients reporting ChSA and CSA in addition
to MSA were still included in the MSA group. The contributions of the ChSA
and CSA assaults were statistically controlled for in analyses for the MSA
group. For patients in the CSA group who had both ChSA and CSA, the
contribution of CSA was controlled for in the analyses.
The Clinician Administered PTSD Scale (CAPS), which measures the
frequency and intensity of PTSD symptoms according to Diagnostic and
Statistical Manual of Mental Disorders, 4th Edition, criteria, was used to
clinically determine a diagnosis of current PTSD (25). The CAPS was
administered up to three times per subject, with different events used for
criterion A: MSA, CSA, ChSA, and/or other traumatic experience. If they did
not indicate any traumatic events, then the CAPS was not administered. The
CAPS has good psychometric properties (test–retest reliability for all 17 items
ranges from 0.90 to 0.98, with internal consistency of 0.94), and the total
severity score of the CAPS strongly correlates with other measures of PTSD
(Mississippi Scale for Combat-related PTSD, r ? 0.91; PK scale of the
Minnesota Multiphasic Personality Scale-2, r ? 0.77) (25).
The two interviewers used in this study were trained during their doctoral
internship on the use of the CAPS by the primary author. Training included
face-to-face sessions in which reliable administration of the CAPS was
presented with published data and manuals on administering the CAPS
available from the National Center for Post-traumatic Stress Disorder. Train-
ing included observation of the primary author administering the CAPS to
multiple patients. The interviewers were then observed administering the
CAPS and given feedback. Both interviewers used the instrument successfully
throughout their year of training. Because they were initially trained in a
clinical setting, inter-rater reliability was not calculated for the present study.
The UAC-PQ (21) is a face-to-face, structured interview with care quan-
tified into units (visit, days, encounters) and classified by setting and provider.
Psychometrically, UAC-Q has been tested on public mental health outpa-
tients, with intraclass correlations of 45% (single raters) and 70% (group) and
an average bias of 7% (21).
Inpatient care and outpatient care files were accessed through the VA’s
Austin Automation Center. These databases permitted the measurement of
patient use of care at any VA Medical Center.
Direct Health Care Cost
VA utilization was taken from all services provided by any VA facility to
a study patient. Unit costs were obtained from VA’s Reasonable Charges
program for FY2001, representing the 80th percentile of national transaction
charges. These same transaction charge rates were applied to non-VA sources
of care so that VA and non-VA comparisons would reflect differences in
utilization rather than differences in cost schedules. Non-VA use of care was
measured with the UAC-PQ. Patient responses were translated into unit costs
by regressing VA care measured with the UAC-PQ with VA costs calculated
with VA administrative databases (22). These “cost regressions” were applied
to UAC-PQ responses for non-VA care to compute the cost of non-VA care.
As shown in Table 1, the mean age of the 270 female
veterans in the sample was 46.69 years (standard deviation
[SD] ? 11.52). More than half of the study participants were
Caucasian (n ? 173, 64.1%), 32.2% were African American
(n ? 87), and 4% (n ? 8) were Hispanic. The mean number
of years of education for the sample was 14.5 ? 2.0 years
(range ? 10–20 years). One third of the sample was married
(n ? 88, 32.6%), whereas 58 participants (21.5%) were never
married, and 92 participants (34.1%) were divorced. Just over
half (n ? 144, 53.3%) indicated that they were unemployed at
the time of their participation in the research interview. The
participants came from all service branches, with the majority
from the Army (n ? 126, 46.7%), followed by the Air Force
(n ? 82, 30.4%), Navy (n ? 47, 17.4%), Marines (n ? 14,
5.2%), and Coast Guard (n ? 1, 0.4%).
Women veterans with a positive history for sexual assault
TABLE 1.Demographic Characteristics of Sample by Sexual Assault (SA) Type
Type of Sexual Assault
None MilitaryCivilian Adult
Mean age (SD)
High school education
48.8 yrs (13.4)
45.1 yrs (8.1)
45.4 yrs (8.9)
45.4 yrs (9.8)
45.5 yrs (9.8)
46.7 yrs (11.5)
Note. Military, Civilian Adult, and Childhood SA categories are not mutually exclusive. Of the veterans with Civilian SA histories, a total of 84 had no history
of Military SA.
SEXUAL ASSAULT IN WOMEN VETERANS
751Psychosomatic Medicine 66:749–756 (2004)
were significantly younger than those veterans with no history
of sexual assault (NSA) (mean ? 45.55 vs. 48.77, p ? .05).
Subsequent analyses comparing MSA veterans to CSA veter-
ans revealed no significant differences between the groups on
any of the demographic characteristics.
Comparison of Present Sample With Previous
An examination of the demographic profile of our sample
reveals many similarities with a previous random national
sample reported by Skinner et al. (26). For example, the mean
age and education levels are comparable between samples for
those who experienced MSA (age ? 42.6 vs. 45.1 years;
education ? 14 vs. 14.5 years) in Skinner’s sample and ours,
respectively. Similarly, marital status, pattern of military ser-
vice, and employment status are also comparable between
samples. The only recognizable dissimilarity, although slight,
is that our sample consisted of more diversity in terms of
racial backgrounds (32.6% non-Caucasians experiencing
MSA) compared with Skinner’s sample (23.2% non-Cauca-
sians experiencing MSA).
Experience of Sexual Assault
A participant was considered to have had an experience of
sexual assault if she made an affirmative response to the
question, “Did you experience unwanted vaginal, oral, or anal
sex. . . ” for the type of sexual assault specified (i.e., military,
civilian adult, or childhood). Of the 270 participants, 173
(64.1%) reported a positive history for at least one of the three
types of sexual assault assessed: 89 (33.0%) reported MSA,
105 (38.9%) reported CSA, and 73 (27.0%) reported ChSA.
The numbers provided exceed 173 because of the endorse-
ment of multiple types of sexual assault by many of the
veterans. For example, 27 (10.0%) reported both MSA and
CSA, 18 (7.0%) reported both CSA and ChSA, 15 (6.0%)
reported both MSA and ChSA, and 17 (6.0%) reported a
positive history for all three types of sexual assault assessed.
Of those veterans reporting CSA, 84 indicated no history of
MSA. Of the women veterans who reported MSA, only 5.6%
reported that their perpetrator was a civilian. There were no
significant differences between branches of military service
and experience of sexual assault, regardless of the type of
sexual assault reported.
Risk of PTSD
When considering the results of the CAPS, 94 women (35%)
reported symptoms that met the criteria for a current diagnosis of
PTSD. Those diagnosed with PTSD were significantly younger
(mean ? 44.27 vs. 47.26, p ? .05), typically reported a positive
history of sexual assault (89.4% vs. 62.7%, p ? .001), and more
frequently noted a current status of unemployment (64.5% vs.
50%, p ? .05). Of those women who did not have a history of
sexual assault, 10 (10.4%) met the criteria for a current diagnosis
of PTSD for a nonsexual assault event.
Between-group differences in PTSD were computed as
odds ratios (with 95% confidence intervals [CI] in parenthe-
ses) calculated from logistic regressions. Estimates were ad-
justed with covariates to account for differences in age (in
years at time of interview assessing PTSD symptoms), eth-
nicity (African American vs. other), education (in years),
marital status (divorced/never married vs. other), and employ-
ment status (past year employed full or part time vs. not
employed). Measured associations are invariant to the pres-
ence or absence of covariates. Results are reported in Table 2.
Female veterans with a positive history of any type of
sexual assault were 5 times more likely to meet the CAPS
criteria for PTSD than were veterans without a history of
sexual assault (adjusted odds ratio [OR] ? 5.24, 95% CI ?
[2.39–11.47], Wald  ? 17.14, p ? .0001). When the type
of sexual assault was examined, MSA was associated with a
more than 9-fold increased risk of PTSD (adjusted OR ? 9.27,
95% CI ? [3.75–22.95], Wald  ? 23.18, p ? .0001), and
ChSA was associated with a 7-fold increased risk of PTSD
(adjusted OR ? 7.26, 95% CI ? [2.75–19.17], Wald  ?
16.02, p ? .0001) compared with a 5-fold increased risk of
PTSD (adjusted OR ? 4.64, 95% CI ? [2.04–10.54], Wald
 ? 13.45, p ? .0001) associated with CSA. After adjusting
for demographic differences and using ChSA and CSA histo-
ries as covariates, MSA was associated with an almost 4-fold
increased risk of PTSD diagnosis for all women veterans
within the sample (adjusted OR ? 3.87, 95% CI ? [2.09–
7.17], Wald  ? 18.5, p ? .0001).
We also examined MSA as a risk factor for PTSD by
subgroup. Focusing on the subgroup of female veterans with
any positive history for sexual assault, an experience of MSA
TABLE 2.Adjusted Odds Ratios for Posttraumatic Stress Disorder
(PTSD) Diagnosis by Clinical Interview
Type of Sexual Assault
All Veterans in Sample
Regardless of Assault History
(n ? 173)
Adjusted OR 95% CI
Civilian Adult SA
Type of Sexual Assault
Veterans with Sexual Assault
Histories Only (n ? 173)
Adjusted OR†95% CI
Civilian Adult SA
* PTSD risk for veterans with positive histories of SA as compared to
veterans with no history of SA.
† PTSD risk for veterans with a positive history of Military SA by patient
¶ p ? 0.05; § p ? 0.01; ‡ p ? 0.0001.
Analyses adjusted for demographic differences in age, ethnicity, marital
status, education, and employment status. SA ? sexual assault; OR ? odds
ratio; CI ? confidence interval.
A. SURI´S et al.
752 Psychosomatic Medicine 66:749–756 (2004)
was associated with a 3-fold increased risk of PTSD, even
after adjusting for the influence of demographic variables
(adjusted OR ? 2.73, 95% CI ? [1.41–5.25], Wald  ?
8.97, p ? .01). Focusing on veterans with a positive history of
ChSA, the experience of MSA resulted in a more than 3-fold
increased risk for developing PTSD (adjusted OR ? 3.26,
95% CI ? [1.04–10.20], Wald  ? 4.11, p ? .05). In
contrast, focusing on veterans without a history of ChSA, the
experience of MSA resulted in a 5-fold increased risk of
PTSD (adjusted OR ? 5.06, 95% CI ? [2.34–10.93], Wald
 ? 17.03, p ? .0001). Similarly, for veterans who experi-
enced CSA (with or without ChSA history), an experience of
MSA conferred a 4-fold increased risk for PTSD (adjusted
OR ? 4.15, 95% CI ? [1.72–10.01], Wald  ? 9.98, p ?
.01). Interestingly, the risk of PTSD was only slightly lower
for veterans with the experience of MSA but without the
experience of CSA (n ? 123) (adjusted OR ? 3.88, 95% CI ?
[1.64–9.23], Wald  ? 9.47, p ? .005). The reader is
referred to Clayton and Hills (27) for further information
regarding the Wald test.
Medical Record Diagnosis of Psychiatric Disorders
Further analyses were performed to investigate the impact
of the type of rape on clinical diagnoses obtained from patient
medical records for the 5 years prior to enrollment into the
study. Specifically, Yates corrected ?2tests were performed to
determine the relationship between medical record diagnosis
of psychiatric disorders and type of sexual assault experience.
As shown in Table 3, there were significant differences in the
likelihood of receiving certain psychiatric diagnoses based on
the history of sexual assault. In this table, “Yes” refers to the
experience of the specified assault, whereas “No” refers to the
absence of an assault history for the specified assault. Com-
pared with veterans with no assault history, those with any
history of sexual assault had significantly higher rates of all
psychiatric diagnoses with the exception of other anxiety
disorders. Furthermore, a strong trend emerged for increased
diagnostic prevalence of other anxiety disorders (p ? .05).
For veterans with a positive history of MSA, there were
significantly higher rates of PTSD and depressive disorder
diagnoses compared with veterans without a history of MSA.
The most common diagnosis for those with MSA was depres-
sion (56.2%), followed closely by PTSD (41.6%). The expe-
rience of CSA conferred a higher risk for the psychiatric
diagnoses of PTSD, other anxiety disorders, depressive disor-
ders, substance use disorders, and personality disorders. Inter-
estingly, no significant associations were found between the
experience of ChSA and medical record psychiatric diagnoses.
Utilization and Cost
Utilization and cost data were available on 230 participants.
This subsample did not differ significantly from excluded pa-
tients (N ? 40) with respect to mean age (46.7 vs. 46.7, t 
? 0.02, p ? .99), African-American status (33.0% vs. 27.5%, ?2
 ? 2.58, p ? .63), mean years of education (14.5 vs. 14.3,
t  ? 0.72, p ? .47), high school graduate status (87.0% vs.
90.0%, ?2 ? 0.29, p ? .59), living alone (67.7% vs. 65.0%,
?2 ? 0.11, p ? .74), or past year employment status (62.4%
vs. 52.5%, ?2 ? 1.40, p ? .24).
There were 127 of 230 patients (54%) who used non-VA
sources of care. Only 8 of 230 patients (3%) were getting
psychiatric services from non-VA sources of care, a sample
too small statistically to analyze. However, 104 of 230 pa-
tients (45%) were getting general medical care from non-VA
sources. There were 228 of 230 (99%) who made at least one
psychiatric visit, 228 of 230 (99%) who made at least one
general medical visit, and 216 of 230 (94%) who had at least
one general medical encounter with the VA. There were 112
of 230 subjects (49%) who made at least one trip to the
emergency room (ER) from any source, 74 of 230 subjects
(32%) used the VA ER, and 61 of 230 (27%) used the ER
from non-VA sources.
VA administrative files and UAC-PQ self-reports were
compared to assess the reliability of the UAC-PQ to measure
use of health care for the study population. For combined
annual emergency and outpatient care (VA reported mean
annual visits ? 20.9, SD ? 24.9), the intraclass correlation
was 0.40, with patients underreporting 8.9% of visits (? ?
?1.9, standard error [SE] ? 1.6, t  ? 1.18, p ? .24, 95%
CI ? [?5.0, 1.2]). Although VA records revealed that 99% of
the sample had at least one outpatient psychiatric encounter
(VA reported mean annual visits ? 6.1, SD ? 12.8), patients
tended to under-report only 4.6% of their VA recorded psy-
chiatric visits (? ? 0.3 visits, SE ? 0.8, t  ? 0.33, p ?
TABLE 3. Psychiatric Medical Record Diagnoses According to Experience of Sexual Assault (SA)
Percentage of Veterans Diagnosed with Psychiatric Disorders According to Experience of Sexual Assaulta
(n ? 89)
(n ? 105)
Civilian Adult SA
(n ? 73)
(n ? 173)
YesNo YesNo YesNo
Other anxiety disorders
Substance use disorders
* p ? 0.05; ** p ? 0.001.
aThese 2 ? 2 comparisons are tested using Yates corrected ?2(continuity corrected).
PTSD ? posttraumatic stress disorder.
SEXUAL ASSAULT IN WOMEN VETERANS
753 Psychosomatic Medicine 66:749–756 (2004)