WUNRN
CONGO - PSYCHOTHERAPY FOR SURVIVORS
OF SEXUAL VIOLENCE - STUDY
N Engl J Med 2013; 368:2182-2191
Survivors of sexual violence have high rates of depression, anxiety,
and post-traumatic stress disorder (PTSD). Although treatment for symptoms
related to sexual violence has been shown to be effective in high-income
countries, evidence is lacking in low-income, conflict-affected countries.
In this trial in the Democratic Republic of Congo, we randomly assigned 16
villages to provide cognitive processing therapy (1 individual session and 11
group sessions) or individual support to female sexual-violence survivors with
high levels of PTSD symptoms and combined depression and anxiety symptoms. One
village was excluded owing to concern about the competency of the psychosocial
assistant, resulting in 7 villages that provided therapy (157 women) and 8
villages that provided individual support (248 women). Assessments of combined
depression and anxiety symptoms (average score on the Hopkins Symptom Checklist
[range, 0 to 3, with higher scores indicating worse symptoms]), PTSD symptoms (average
score on the PTSD Checklist [range, 0 to 3, with higher scores indicating worse
symptoms]), and functional impairment (average score across 20 tasks [range, 0
to 4, with higher scores indicating greater impairment]) were performed at
baseline, at the end of treatment, and 6 months after treatment ended.
A total of 65% of participants in the therapy group and 52% of participants
in the individual-support group completed all three assessments. Mean scores
for combined depression and anxiety improved in the individual-support group
(2.2 at baseline, 1.7 at the end of treatment, and 1.5 at 6 months after
treatment), but improvements were significantly greater in the therapy group
(2.0 at baseline, 0.8 at the end of treatment, and 0.7 at 6 months after
treatment) (P<0.001 for all comparisons). Similar patterns were observed for
PTSD and functional impairment. At 6 months after treatment, 9% of participants
in the therapy group and 42% of participants in the individual-support group
met criteria for probable depression or anxiety (P<0.001), with similar
results for PTSD.
In this study of sexual-violence survivors in a low-income, conflict-affected country, group psychotherapy reduced PTSD symptoms and combined depression and anxiety symptoms and improved functioning. (Funded by the U.S. Agency for International Development Victims of Torture Fund and the World Bank; ClinicalTrials.gov number, NCT01385163.)
Article
Activity
Mental health problems such as depression, anxiety,
and post-traumatic stress disorder (PTSD) are common in survivors of sexual
violence.1-6 In high-income
countries, there are effective treatments for trauma related to sexual violence,7-10 but these
treatments have not been adequately tested in low-income, conflict-affected
countries with few mental health professionals and low literacy rates. The few
studies of effectiveness have had methodologic limitations, including a lack of
controls and high attrition rates.11
Eastern
Democratic Republic of Congo is a low-income, conflict-affected region in which
political and economic instability are ongoing problems and nearly 40% of women
have experienced sexual violence.12 The development
of effective mental health services has important implications for the recovery
of sexual-violence survivors in the Democratic Republic of Congo and similar countries.
We
evaluated an adaptation of group cognitive processing therapy provided by
community-based paraprofessionals (psychosocial assistants), supervised by
psychosocial staff at a nongovernmental organization (NGO) and by clinical
experts based in the United States. Cognitive processing therapy has shown
efficacy in high-income countries, with effects lasting for 5 or more years.13-15 We evaluated
the benefits of adding this therapy to services offered by workers trained only
in case management and individual supportive counseling.
We selected 14 villages in South Kivu province and
2 villages on the border in North Kivu province, from among 23 villages served
by three Congolese NGOs. Selection was based on accessibility, security, and
the availability of psychosocial assistants. All psychosocial assistants had 1
to 9 years of experience providing case management and individual supportive
counseling to survivors of sexual violence and at least 4 years of
post–primary-school education. All underwent a 5-to-6-day training session
conducted by the International Rescue Committee (IRC) in case management and
specific topics, including counseling, family mediation, stress management,
clinical care of survivors, and prevention of human immunodeficiency virus
infection and other sexually transmitted diseases.
Using
a mixed-methods approach described previously,16 we selected,
adapted, and tested measures. We conducted qualitative studies in three
linguistically different communities to identify salient mental health problems
of sexual-violence survivors. Abandonment and rejection by family and friends,
concerns about providing for self and family, fear, and stigma were major
issues. Informants described psychological symptoms that were consistent with
depression, anxiety, and PTSD.
On the
basis of these findings, we selected the Hopkins Symptom Checklist (HSCL-25)17,18 to assess
depression (15 items) and anxiety (10 items) and the PTSD Checklist — Civilian
Version19 to assess PTSD
symptoms (16 items). The checklists were adapted and pilot-tested in each
language group. Both the HSCL-25 and the PTSD Checklist have been used
internationally with sexual-violence survivors20 and have solid
psychometric properties with conflict-affected samples21-23; locally derived
psychometric properties of both measures are presented in the Supplementary Appendix, available with the full text of
this article at NEJM.org. Participants rated the frequency of each symptom in
the prior 4 weeks on a four-point Likert scale (with 0 denoting not at all, 1 a
little bit, 2 a moderate amount, and 3 a lot). Average per-item scores were
generated for each measure, with scores ranging from 0 to 3 and higher scores
indicating greater severity. An average HSCL-25 score of 1.75 or higher and an
average PTSD Checklist score of 1.75 or higher were considered to be predictive
of clinically significant depression or anxiety and PTSD, respectively, on the
basis of data from other conflict-affected populations.24,25
Assessment of functional impairment was based on the degree of difficulty in performing important tasks of daily living that were identified on the basis of qualitative data from the study villages, with the use of methods described elsewhere.16,26 For each of 20 tasks, participants were asked to rate the degree of difficulty in performing the task on a five-point Likert scale (with 0 denoting no difficulty, 1 little difficulty, 2 a moderate amount of difficulty, 3 a lot of difficulty, and 4 often unable to perform the task). An average per-item score was generated for each participant, with scores ranging from 0 to 4 and higher scores indicating greater impairment. The items used to assess depression, anxiety, and PTSD symptoms as well as functional impairment are shown in Table S3 in the Supplementary Appendix.
Women who had experienced or witnessed sexual
violence (translated as “rape” locally) were eligible for the study if they had
a total symptom score of at least 55 (i.e., an average score of 1 for each of
55 symptoms, comprising the HSCL-25 items, the PTSD Checklist items, and
additional locally relevant symptoms) and a functional-impairment score of at
least 10 (i.e., dysfunction on at least half the activities). Suicidality that
was judged by clinical staff to require immediate treatment was a criterion for
exclusion. Study measures were translated into five local languages: Kibembe,
Kifuliro, Kihavu, Mashi, and Swahili.
The 16 study villages, each with one psychosocial assistant, were grouped into blocks of 2 to 4 villages on the basis of proximity and shared language and were randomly assigned to provide cognitive processing therapy or individual support. After therapy training, one psychosocial assistant was excluded because training-based quizzes and skill observation raised competency concerns; therefore, the village in which she worked was excluded. The trial included 15 study villages (7 that provided therapy and 8 that provided individual support). The study protocol and statistical analysis plan are available at NEJM.org.
Recruitment occurred in December 2010. Psychosocial assistants reviewed their files of current and prior clients to identify women with clinically significant psychological problems. They invited the women to their offices, where research assistants, who were unaware of the village treatment assignments, obtained informed consent and administered the study questionnaires. Interviews were continued until 28 to 30 eligible women had been identified per village. Questionnaires were reviewed to confirm eligibility, and lists of eligible women were given to the psychosocial assistants, who invited the women to participate. In villages that provided therapy, psychosocial assistants recruited up to 24 participants (with a maximum of 8 women per treatment group). Psychosocial assistants ordered the list of eligible women in their village by proximity to the NGO office and selected those who were closest, continuing down the list until treatment groups were filled. In villages that provided individual support, where there was no limit to the number of participants, psychosocial assistants invited all eligible women.
The intervention period lasted from April through July 2011. Follow-up data were collected within 1 month after treatment ended and 6 months later. For both follow-up assessments, two research assistants, who were unaware of the village treatment assignments, spent 1 week in each village assessing the study participants. Scores for PTSD symptoms and combined depression and anxiety symptoms were the primary outcomes, and the score for functional impairment was a secondary outcome.
Institutional review boards at the Johns Hopkins
Bloomberg School of Public Health and Kinshasa School of Public Health approved
the protocol. Study participants provided oral informed consent, and none
received compensation. The research was sponsored by the U.S. Agency for
International Development Victims of Torture Fund and the World Bank. The
sponsors had no role in the trial design or conduct; the collection,
management, analysis, or interpretation of data; or manuscript preparation,
review, or approval. All authors vouch for the completeness and accuracy of the
data and analysis and the fidelity of the study to the protocol.
Psychosocial assistants in the comparison villages provided
access to individual support. When women were informed of their eligibility,
psychosocial assistants invited them to receive individual support services as
desired, including psychosocial support and economic, medical, and legal
referrals. Psychosocial assistants were available throughout the treatment
period for women who sought their services. IRC supervisors monitored the
services provided by means of monthly visits and reviews of interim monitoring
forms.
Cognitive
processing therapy is a protocol-based therapy for treating depression,
anxiety, and PTSD in sexual-violence survivors.9,27-29 The group
format was chosen to reach large numbers of women. We used the cognitive-only
model (i.e., without a trauma narrative) because its efficacy is similar to
that of the full version of the therapy,14 while providing
greater ease of administration in groups and greater retention by participants.13,27,28 The
treatment included 1 individual session (1 hour) and 11 sessions with six to eight
women per group (2 hours each). Each psychosocial assistant concurrently led
three groups. Participants in the therapy group had access to the psychosocial
assistants as desired outside the therapy.
Psychosocial
assistants who provided therapy underwent 2 weeks of in-person training with
trainers based in the United States (the fourth and fifth authors), with the
use of a manual30 that was adapted
and translated locally. Ongoing supervision was provided through a multitiered
supervision system that was first used in Uganda31 and subsequently
expanded.32 Congolese
psychosocial supervisors who were employees of the IRC provided direct
supervision to psychosocial assistants through weekly telephone or in-person
meetings; a bilingual clinical social worker trained in the United States
provided in-country supervision and communicated with the U.S. trainers through
weekly calls for supervision and quality assurance. Fidelity to the therapy
protocol was assessed with the use of checklists of key treatment elements and
global ratings of treatment knowledge and skills, as observed by supervisors
during group sessions.
The therapy was adapted for
illiterate participants and those potentially exposed to ongoing violence. The
adaptations included an initial individual psychoeducational session, oral
completion of assignments during group sessions, and simplification of
materials to facilitate understanding and memorization. Further details about
modifications are presented in the Supplementary Appendix.
Assuming a 20% dropout rate, we calculated that enrollment of 180 participants in each study group would provide 80% power to detect at least a 0.5-point difference between groups with respect to the reduction in average symptom scores, adjusting for a variance-inflation factor of 2.0. After excluding one village that was to provide therapy, we expected fewer than 180 participants in the therapy group.
Baseline characteristics were compared between
study groups with the use of the chi-square test and Student's t-test. Factors
associated with loss to follow-up were identified with the use of logistic
regression; those at a significance level of less than 0.20 were used to
generate weighting estimates to adjust for loss to follow-up (data were missing
for 135 participants [33%] at the end of treatment and for 92 participants
[23%] at 6 months after treatment). Indicators of treatment effect were derived
by comparing mean changes in HSCL-25, PTSD Checklist, and functional-impairment
scores between groups from baseline to each follow-up assessment. Analyses were
performed on data from all participants, regardless of the level of
participation. Item-level missing data were imputed on the basis of mean values
for other items in the scale.
Random-effects models were used for hypothesis testing and to predict the relative risk of meeting the criteria for probable depression or anxiety or probable PTSD at each follow-up assessment.33 Four random effects were evaluated: randomization block, village, assigned treatment group, and participant. Models with and those without the randomization block were not significantly different (P=0.99) according to the Hausman test,34 so the three-level model was used. Time and treatment condition (therapy or individual support) were included as fixed effects. Effect sizes reflecting regression adjustments were calculated with the use of Cohen's d statistic,35 which represents the mean between-group differences standardized with the use of the baseline pooled standard deviation. Effect sizes are equivalent to a z score of a standard normal distribution (i.e., an effect size of 1.0 would mean that the average symptom score for participants in the therapy group was 1.0 SD above the average symptom score for participants in the individual-support group).
All comparisons were prespecified, and all tests
were two-sided. A P value of less than 0.05 was considered to indicate
statistical significance. Analyses were conducted with the use of Stata
software, version 12 (StataCorp).
A total of 494 women were
screened for eligibility (Figure 1Figure 1
). Of these women, 434 (88%) met the inclusion criteria; 7 women with severe suicidality requiring immediate assistance by psychosocial assistants and IRC staff were not deemed eligible. Of the 434 eligible women, 402 (93%) agreed to participate. An additional 4 women who did not meet the inclusion criterion of a total symptom score of at least 55 were mistakenly recruited in one village that provided therapy; these women were included in the analysis. After the exclusion of 1 participant who received therapy owing to paperwork errors, the study included 405 women.
A total of 231 women (57%) completed all three
assessments; 352 (87%) completed the baseline assessment and at least one
follow-up assessment. Factors that were significantly associated with loss to
follow-up were older age (with age as a continuous variable), assignment to
individual support, pregnancy at baseline, experience with or witnessing of a
wider range of traumas, and language spoken. Problems with security and cases
in which the wrong women were interviewed reduced the number of follow-up
assessments. Participants in the therapy group attended an average of 8.5 of 12
sessions offered, with 141 participants (90%) completing at least 9 sessions
(defined as treatment completion). Women in the therapy group who missed a
session were visited by psychosocial assistants to identify the reason for
their absence and were encouraged to rejoin. A total of 182 participants in the
individual-support group (73%) attended at least 1 session with the
psychosocial assistant. Among these women, the average number of sessions
attended during the treatment period was 5.
Despite regional instability,
80% of the women were living in their territory of origin. As compared with
participants in the therapy group, those in the individual-support group were
younger and less likely to be married, and they lived with fewer people (Table 1Table 1
Table 2Table 2), with participants in the individual-support group
having higher symptom scores at baseline than those in the therapy group across
all measures.
Both the individual-support
and therapy groups had significant improvements during treatment, with effects
maintained at 6 months (Figure 2Figure 2). For PTSD symptoms (PTSD Checklist) and
combined depression and anxiety symptoms (HSCL-25), participants in the therapy
group had significantly greater improvements than those in the
individual-support group (Table 2) at both follow-up
assessments, with all treatment-effect sizes greater than 1.0.
Approximately 70% of
participants in the therapy group met our criteria for probable depression or
anxiety at baseline, with 10% or less meeting the criteria at either follow-up
assessment (Table 2). In the
individual-support group, the proportions of participants who met the criteria
were as follows: 83% at baseline, 53% at the end of treatment, and 42% at 6
months after treatment. The relative risks of meeting the criteria for
depression or anxiety and PTSD were significantly greater with individual
support than with therapy at the end of treatment and 6 months after treatment
(P<0.001 for all comparisons).
In our
study, cognitive processing therapy, as compared with individual support alone,
was effective in reducing PTSD symptoms and combined depression and anxiety
symptoms and improving functioning in female survivors of sexual violence in
eastern Democratic Republic of Congo. The benefits were large and were
maintained 6 months after treatment ended. Participants who received therapy
were significantly less likely to meet the criteria for probable depression or
anxiety or probable PTSD. Our findings are consistent with the results of
trials conducted in high-income countries, both for cognitive behavioral
interventions in general36 and for
cognitive processing therapy specifically,14,27 with effect
sizes of 0.62 and 2.7 in the latter two studies.
Prior research has suggested that short-term therapies may not be effective for populations exposed to ongoing trauma or multiple severe traumas.37,38 In our study, all villages reported at least one major security incident during the trial, including attacks, displacement due to fighting, and robbery by armed groups. In addition, there was concern that providing therapy to illiterate persons would be challenging. Our findings suggest that despite illiteracy and ongoing conflict, this evidence-based treatment can be appropriately implemented and effective.
Limitations of the study include baseline differences in symptom severity between study groups that may limit comparability. Randomization was performed within blocks of two to four villages grouped on the basis of language and proximity, with the assumption that villages close to one another would be similar; however, this assumption was not empirically confirmed. The small number of village clusters (six) made randomization less likely to result in comparability. There may also have been biases in recruitment that resulted in higher average symptom scores in villages that provided individual support because psychosocial assistants recruiting patients knew ahead of time whether they would be providing therapy or individual support. To assess whether higher baseline scores in the individual-support group biased the results, we performed sensitivity analyses restricted to women with baseline HSCL-25 scores higher than 2.0 (84 women in the therapy group and 171 in the individual-support group) and found that effect sizes remained greater than 1.0.
An additional limitation is the use of measures of
unknown validity for identifying clinical cases of PTSD and combined depression
and anxiety. Because symptoms of these disorders could be nonpathologic
reactions to extreme circumstances, it is unclear what proportions of
participants actually met clinical criteria. Although the clinical meaning of
standard cutoff scores is therefore uncertain, the score itself can still be
meaningfully interpreted: 1.75 indicates that women are reporting that symptoms
occur, on average, nearly a “moderate” amount of the time (a score of 2.0). The
symptom scores of participants in the therapy group decreased, on average, to
less than “a little bit” (a score of 1), whereas the scores of participants in
the individual-support group remained closer to a “moderate amount.”
Finally, differences in how therapy and individual support were provided may affect our conclusions. Because therapy, but not individual support, was provided in groups, it is unclear how much of the treatment effect was due to the group context. Psychosocial assistants who provided therapy received greater supervision than those who provided individual support. Prior studies have shown that increases in the quality and quantity of clinical supervision can explain some treatment effects.39,40 Participants in the therapy group attended, on average, a greater number of treatment sessions than did those in the individual-support group. Participants in the therapy group were also told that they could have additional sessions with the psychosocial assistant. The time commitment required to receive therapy and the lack of reports of additional therapy sessions by the counselors suggest that their use of additional services was much less than that of participants in the individual-support group. The overall therapy effects must therefore be viewed as program effects, which include the therapy itself, the number of sessions, the group process, supervision systems, and possibly some additional counseling sessions.
This trial provides evidence of effectiveness of a mental health intervention for sexual-violence survivors in a low-income, conflict-affected setting. The results indicate that with appropriate training and supervision, psychotherapeutic treatments such as cognitive processing therapy can be successfully implemented and can have an effect in settings with few mental health professionals. This therapy holds promise as a community-based service for sexual-violence survivors in similar contexts and warrants confirmatory studies and scale-up evaluations.
Supported by the U.S. Agency for International Development Victims of Torture Fund and the World Bank.
From the Departments of Mental Health (J.K.B., S.M.M., L.K.M., P.A.B.) and International Health (P.A.B.), Johns Hopkins Bloomberg School of Public Health, Baltimore; the Research and Evaluation Unit, International Rescue Committee, New York (J.A.); the Departments of Psychiatry and Behavioral Sciences and Global Health (D.K.), and the Harborview Center for Sexual Assault and Traumatic Stress (S.G.), University of Washington, Seattle; the International Rescue Committee, Kinshasa, Democratic Republic of Congo (T.C.); and the School of Social Work, University of Texas, Austin (K.W.).