Moderators of Treatment Effectiveness for War-Affected Youth With Depression in Northern Uganda

Journal of Adolescent Health 51 (2012) 544 –550

www.jahonline.org Original article

Moderators of Treatment Effectiveness for War-Affected Youth With Depression in Northern Uganda Theresa S. Betancourt, Sc.D., M.A.a,b,*, Elizabeth A. Newnham, Ph.D.b,c, Robert T. Brennan, Ed.D.b, Helen Verdeli, Ph.D.d,e, Ivelina Borisova, Ph.D.f, Richard Neugebauer, Ph.D.g, Judith Bass, Ph.D.h, and Paul Bolton, M.B.B.S.i,j a

Department of Global Health and Population, Harvard School of Public Health, Boston, Massachusetts Franåois-Xavier Bagnoud Center for Health and Human Rights, Harvard School of Public Health, Boston, Massachusetts c School of Psychology, The University of Western Australia, Perth, Australia d Department of Clinical Psychology, Teachers College, Columbia University, New York, New York e New York State Psychiatric Institute, Columbia University, New York, New York f Department of Education and Child Development, Save the Children, Washington, DC g Department of Psychiatry, Faculty of Medicine, Columbia University, New York, New York h Department of Mental Health, Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland i Center for Refugee and Disaster Response, Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland j Department of International Health, Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland b

Article history: Received September 9, 2011; Accepted February 9, 2012 Keywords: War; Depression; Treatment moderators; Interpersonal therapy; Child soldiers

A B S T R A C T

Purpose: As we build the evidence base of interventions for depression among war-affected youth, it is critical to understand factors moderating treatment outcomes. The current study investigated how gender and history of abduction by Lord’s Resistance Army rebels moderated treatment outcomes for war-affected youth. Methods: The study—a three-armed, randomized, controlled trial—was conducted with internally displaced war-affected adolescents in northern Uganda. Participants with significant depression symptoms (N ⫽ 304; 57% female; 14 –17 years of age) were randomly assigned to an interpersonal psychotherapy group (IPT-G), a creative play/recreation group, or a wait-list control condition. Secondary analyses were conducted on data from this randomized controlled trial. Results: A history of abduction by Lord’s Resistance Army rebels was reported by 42% of the sample. Gender and abduction history interacted to moderate the effectiveness of IPT-G for the treatment of depression. In the IPT-G intervention arm, treatment effectiveness was greatest among female subjects without an abduction history, with effect size ⫽ 1.06. IPT-G was effective for the treatment of depression for both male and female subjects with a history of abduction (effect size ⫽ .92 and .50, respectively). Male subjects with no abduction history in IPT-G showed no significant improvement compared with those in the control conditions. Conclusions: Abduction history and gender are potentially important moderators of treatment effects, suggesting that these factors need to be considered when providing interventions for war-affected youth. IPT-G may be an effective intervention for female subjects without an abduction history, as well as for both male and female former child soldiers, but less so for male subjects without an abduction history. 䉷 2012 Society for Adolescent Health and Medicine. All rights reserved.

* Address correspondence to: Theresa Betancourt, Sc.D., M.A., FranÈois-Xavier Bagnoud Center for Health and Human Rights, Department of Global Health and Population, Harvard School of Public Health, 651 Huntington Avenue, 7th Floor, Boston, MA 02115. E-mail address: [email protected].

IMPLICATIONS AND CONTRIBUTION

Limited information is available on effective interventions for war-affected children. Secondary analyses of data from an RCT of IPT-G for depression among waraffected youth in northern Uganda indicated that in addition to its effectiveness for war-affected girls, IPT-G was effective for treating depression among male and female former child soldiers.

The conflict in northern Uganda, spanning ⬎20 years, remains one of history’s deadliest complex humanitarian emergencies. The war had particularly dire consequences for children and adolescents. The Lord’s Resistance Army (LRA), the main rebel movement behind the conflict, was notorious for the abduction of civilians, including ⬎25,000 children, over the course of the

1054-139X/$ - see front matter 䉷 2012 Society for Adolescent Health and Medicine. All rights reserved. doi:10.1016/j.jadohealth.2012.02.010

T.S. Betancourt et al. / Journal of Adolescent Health 51 (2012) 544 –550

war [1]. Children associated with armed forces and armed groups, commonly referred to as “child soldiers,” are recruited or forced into a range of roles, including, but not limited to, frontline soldiers, spies, cooks, porters, and sexual slaves [2]. They often experience a range of violence exposures, including forced separation from loved ones and involvement in atrocities [3], as well as high rates of domestic and community violence after release or escape from the fighting forces [4,5]. Previous research has demonstrated a great deal of unmet mental health need in this subgroup, including high levels of posttraumatic stress symptoms, depression, and anxiety [6,7]. Clearly, war-affected children in general and former child soldiers in particular require a range of services to address their psychological needs; yet, for mental health interventions, rigorous data informing the appropriate design of intervention models are sparse. A handful of studies have begun to document promising therapeutic techniques used to treat children affected by armed conflict and displacement. The use of interpersonal psychotherapy [8], trauma/grief-focused therapy [9], cognitive behavioral therapy [10], narrative exposure therapy [11], recreational and connectivity activities [12], and school-based interventions [13] have all demonstrated partial efficacy in controlled studies among war-affected populations of youth. However, few studies have examined how at-risk subgroups, such as former child soldiers, fare in such interventions. Many features of a child’s war-related and postconflict experience may affect how he or she fares in mental health services. In previous research, the strength of association between exposure to war violence and psychological difficulties has varied among children affected by war [14,15]. Some children will experience distress and dysfunction as a result of violence exposure, whereas others will demonstrate resilience and establish productive roles in the community [16]. Even within the same conflict, adolescents may respond differently to traumatic exposure. For example, Slone [17] found that Israeli adolescents (Jewish and Arab) reported a positive dose–response relationship between exposure to violence and levels of distress, whereas Palestinian youth showed a reverse dose–response relationship, with more distress at low levels of exposure and significantly less distress at high levels of exposure. Similarly, Betancourt et al demonstrated great variability in trajectories of internalizing symptoms among war-affected youth in Sierra Leone [18]. Worsening or maintenance of high symptoms was associated with death of a caregiver during war, postconflict family abuse and neglect, and community stigma. It is thus possible that some interventions may be effective for some subgroups of children but ineffective (or even harmful) for others [13,19]. A recent cluster-randomized trial of trauma-focused treatment for youth affected by violence revealed both regional and gender differences in treatment outcomes. Intervention effects were strongest in Asian populations in Nepal and Indonesia but less effective in Burundi, where the conflict was characterized by ongoing insecurity [20]. Further examination of treatment effects in this school-based mental health intervention for children revealed that gender, household size, sibling support, and social support moderated changes in functional impairment [21]. The role of potential treatment moderators clearly requires further exploration in systematic evaluations. Secondary analyses of treatment moderators may assist in identifying the conditions under which interventions are most effective. A moderator is a baseline characteristic that has an interactive effect with treatment on the outcome of interest [22].

545

The critical role of moderators has been widely investigated but rarely tested in treatments available for war-affected youth. The randomized controlled trial (RCT) examined here was originally conducted in northern Uganda to investigate whether a psychotherapeutic intervention (interpersonal psychotherapy for groups [IPT-G]) and an activity-based intervention (creative play [CP]) were effective for reducing depression symptoms among adolescents living in internally displaced persons (IDP) camps [8]. Previous research with this population identified five salient mental health syndromes, of which three featured depressionlike symptoms (two tam, par, and kumu), one reflected mixed anxiety and depression symptoms (ma lwor), and the other reflected conduct difficulties (kwo maraco) [15]. Depression symptoms had demonstrated amenability to treatment in a previous RCT conducted with adults [23], which prompted investigation of the intervention in an adolescent sample. In a previous article [8], we reported that the CP intervention did not significantly improve depression symptoms for either gender compared with control subjects. IPT-G demonstrated effectiveness among girls, but no significant reduction in the burden of depression was evident for boys, and no significant effect was seen for additional outcomes such as conduct problems or functional impairment. The present analysis examines whether male and female subjects reporting LRA abduction fared better or worse in interventions. Trauma history is likely to have a significant impact on treatment outcome, impeding response to depression treatment [24]. In the present study, many children in the IDP camps had experienced abduction by the LRA rebels. In this setting, former child soldier status (defined as LRA involvement) was not only a risk indicator for exposure to particularly toxic war traumas, but may have also been associated with stigma in the postconflict setting [25]. Given debates in the field that trauma history may impede the effectiveness of depression treatment [24], and our previous finding that girls experienced greater benefit of treatment than boys in IPT-G, it was hypothesized that treatment would be less effective with former child soldiers of either gender than with children who had not been child soldiers, and, in both of these groups, female subjects would show greater benefit than male subjects. Methods Participants The RCT was conducted in two IDP camps near the town of Gulu, northern Uganda. Study inclusion criteria consisted of ability to understand and speak Acholi Luo and having lived in either of the two camps for at least 1 month before the baseline screening interview. Eligible adolescents were aged 14 –17 years, scored above a predetermined threshold of depression symptom severity based on the Acholi Psychosocial Assessment Instrument (APAI) Depression Problems Subscale [26], and reported some functional impairment over the previous month [8]. Exclusion criteria included significant suicidal ideation or a severe cognitive or physical disability leading to an inability to answer the survey questions (no such cases were identified). Full details of the trial are available elsewhere [8]. The original study was approved by the Boston University Institutional Review Board and by the Ugandan National Council of Science and Technology in Kampala, Uganda. A total of 667 Acholi youth were screened for participation in the trial; 314 participants were recruited and stratified by camp

546

T.S. Betancourt et al. / Journal of Adolescent Health 51 (2012) 544 –550

667 Participants recruited and screened

329 Not eligible

338 Met eligibility criteria

314 Randomized

105 Group Interpersonal Therapy 103 Enrolled 1 Refused consent 1 Lost to follow up prior to consent

105 Creative Play 99 Enrolled 3 Refused Consent 2 Lost to follow-up prior to consent 1 Already enrolled in another creative play program

104 Waitlist Control 102 Enrolled 1 Refused consent 1 Lost to follow-up prior to consent

7 Lost to follow-up at end 3 Could not be found 1 Away from camp at assessment 3 Moved to another camp

11 Lost to follow-up at end 6 Could not be found 3 Away from camp at assessment 2 Moved to another camp

14 Lost to follow-up at end 9 Could not be found 2 Away from camp at assessment 2 Moved to another camp 1 Died

103 Included in analysis

99 Included in analysis

102 Included in analysis

Figure 1. Participant flow diagram for the randomized controlled trial.

and gender and then randomized and enrolled in one of the three study arms: IPT-G (n ⫽ 103), CP (n ⫽ 99), and wait-list control (n ⫽ 102). Ten participants declined or dropped out before consent. A flowchart outlining participant recruitment, screening, group assignment, and retention is outlined in Figure 1. Measures APAI (Depression Problems Subscale). The APAI was created based on qualitative data collected at the same sites [15,26]. The 35-item Depression Problems Subscale includes most of the cardinal Diagnostic and Statistical Manual of Mental Disorders, fourth edition, symptoms for major depressive disorder, as well as additional terms in the Acholi Luo language for depressionlike symptoms. Analyses indicated that the Depression Symptoms Subscale has good internal consistency (␣ ⫽ .92) as well as inter-rater (r ⫽ .84) and test–retest reliability (r ⫽ .84) [8,26]. Sociodemographic items. Questions were included on sociodemographics and duration of camp residence. Abduction history. A history of abduction by the LRA was assessed with a yes/no question. Interventions Interpersonal psychotherapy is an evidence-based treatment for depression. It has been adapted for use with adolescents and

has demonstrated effectiveness in multiple RCTs (e.g., [27]). IPT-G [28] was adapted and tested in a previous RCT among adults in rural southeastern Uganda [29]. Based on this experience and our preliminary qualitative work in northern Uganda [15], IPT-G’s focus on interpersonal triggers and group relationship-building was considered compatible with Acholi culture and the experiences of war-affected youth. Therapy groups consisted of six to eight adolescent boys or girls and a gendermatched facilitator. The CP intervention for war-affected youth was developed by War Child Holland as a manualized intervention drawing on verbal and nonverbal expression of thoughts and feelings to interpret life lessons and personal difficulties. Four mixed-gender groups were formed, with 25–30 adolescents in each, consistent with its routine application. Two group leaders (one male, one female) co-facilitated all groups. Training, supervision, and intervention content for both conditions have been outlined elsewhere [8,30]. Statistical analyses Baseline characteristics comparisons of the IPT-G, CP, and wait-list control groups were tested using analysis of variance with planned contrasts. The impact of treatment on APAI depression scores, with LRA abduction history and gender considered as potential moderators, was examined using multilevel modeling. Analyses of functional impairment did not result in identifying treatment effects in the RCT [8] and have therefore been ex-

T.S. Betancourt et al. / Journal of Adolescent Health 51 (2012) 544 –550

547

Table 1 Sociodemographic characteristics of the IPT-G, CP, and wait-list control groups Characteristic

IPT-G (n ⫽ 103)

CP (n ⫽ 99)

Wait-list controls (n ⫽ 102)

Females, n (%) Age at time of survey (years), mean (SD; range) History of abduction, n (%) Currently enrolled in school, n (%) Current level of school (years), mean (SD; range) Highest level of education completed (years), mean (SD; range) Participant has a child, n (%) Baseline depression symptom score, mean (SD; range)

58 (56.3) 14.94 (1.08; 14–17) 41 (39.8) 68 (66) 5.79 (1.0; 2–7) 5.01 (1.55; 0–8) 10 (9.7) 43.38 (10.19; 27–78)

56 (56.6) 14.73 (.95; 14–17) 46 (46.5) 68 (68.7) 5.84 (.82; 4–7) 5.11 (1.44; 0–8) 11 (11.1) 43.92 (11.25; 22–87)

59 (57.8) 15.18 (1.18; 14–17) 40 (39.2) 71 (69.6) 5.99 (1.12; 3–9) 5.25 (1.28; 2–9) 12 (11.8) 44.02 (10.84; 27–80)

IPT-G ⫽ interpersonal psychotherapy for groups; CP ⫽ creative play.

cluded for the current analysis. To accommodate the clustering of subjects within intervention groups, preliminary analyses using a null model (see later in the text) demonstrated that the grouping of subjects was a significant source of variation [31]. Multilevel models were estimated to examine our specific hypothesis that a combination of gender and abduction history predicted treatment outcome. Accordingly, six models were created. All models were adjusted for clustering within intervention groups by controlling for within-group correlated observations and between-group variability [8,32]. Additionally, all models were adjusted for baseline depression scores. The first model (M1) was a null model (no predictors), and the second model (M2) was a null model with individuals’ baseline scores, to provide an estimate of the variance accounted for by baseline values. The third model (M3) outlined the main effects of treatment, without considering any interactions. The fourth (M4) and fifth (M5) models were main effects and two-way interaction models, which examined the roles of gender and LRA abduction, first independently and then in relation to IPT-G and one another. In the final model (M6), we introduced the three-way interaction effect between IPT-G, gender, and LRA abduction. Statistical significance was set at p ⬍ .05 for main effects and interaction terms. Adjusted posttest means were calculated based on the final three-way interaction model. Analyses are presented for the intent-to-treat sample (N ⫽ 304), which includes all eligible individuals randomized to one of the three trial conditions regardless of whether they participated. Missing data were addressed using multiple imputation. This approach considers the relationship of missing data to other observed characteristics in the data set, thereby reducing bias, in

addition to accounting for sampling variability across imputations by introducing an error term for each imputed value. Twenty multiple imputed data sets were analyzed in HLM v7.0 (Scientific Software International, Inc., Lincolnwood, IL) [33], and descriptive analyses were conducted in PASW Statistics 18 (SPSS Inc., Chicago, IL) [34]. Results The IPT-G, CP, and wait-list control groups had similar mean age and gender distributions (Table 1). The trial arms did not differ significantly in baseline characteristics, including rates of LRA abduction; however, male subjects reported a significantly higher rate of abduction (n ⫽ 68, 52%) compared with female subjects (n ⫽ 59, 34%) (p ⬍ .01). There were no significant differences in baseline depression scores across groups. Further sociodemographic details are provided elsewhere [8]. Multilevel models were estimated to determine the interaction between gender, abduction, and treatment outcomes [31] (Table 2). Gender did not significantly predict depression outcomes in a model containing baseline depression scores and the interventions (model 4). In contrast, abduction significantly predicted depression outcomes. The final model included all terms necessary to test the three-way interaction between the IPT-G intervention, abduction, and gender (model 6). In the final model, both the abduction ⫻ IPT-G interaction term and the gender ⫻ IPT-G interaction term reached significance. In addition, the three-way interaction was statistically significant. Table 3 depicts the post-treatment means adjusted for pretreatment depression score for all combinations of treatment,

Table 2 Taxonomy of multilevel models (M) describing the relationship between depression score at outcome and treatment condition Characteristic

M1

M2

M3

M4

M5

M6

Intercept IPT-G CP Baseline depression Female Abduction Female ⫻ IPT-G Abduction ⫻ IPT-G Female ⫻ abduction Female ⫻ abduction ⫻ IPT-G Level-one variance, ␴2 Level-two variance, ␶

35.13* — — — — — — — — — 257.12 46.94

35.07* — — .24** — — — — — — 251.13 42.91

37.01* ⫺9.52** 3.16 .25** — — — — — — 251.63 13.16

33.41* ⫺9.81* 3.11 .23*** .99 6.62** — — — — 251.37 2.73

31.87* ⫺5.67 3.18 .22*** 3.20 7.55*** ⫺6.07 ⫺2.16 ⫺.64 — 251.22 .71

29.77* .28 3.57 .23*** 6.26*** 11.27** ⫺15.01** ⫺13.11*** ⫺7.53 20.83*** 246.32 .028

M1 ⫽ null model; M2 ⫽ model adjusted for baseline depression score; M3 ⫽ treatment model; M4 ⫽ treatment model adjusted for gender and abduction status; M5 ⫽ treatment model with treatment by gender and abduction two-way interaction; M6 ⫽ treatment model with treatment by gender and abduction three-way interactions. * p ⬍ .001, ** p ⬍ .01, *** p ⬍ .05.

548

T.S. Betancourt et al. / Journal of Adolescent Health 51 (2012) 544 –550

Table 3 Adjusted post-treatment means for APAI depression problems among abducted and nonabducted male and female subjects Group

Abducted female subjects Nonabducted female subjects Abducted male subjects Nonabducted male subjects

Group interpersonal therapy

Creative play

n

M (ES)

n

M

Wait-list control n

M

17 41 24 21

32.76 (.50) 21.30 (1.06) 28.21 (.92) 30.05 (⫺.02)

25 31 21 22

43.34 39.60 44.61 33.34

17 42 23 20

39.77 36.03 41.04 29.77

Adjusted means were calculated as the predicted values from a regression equation including pretreatment depression score, gender, treatment condition, and interactions among pretreatment depression, gender, and IPT-G treatment (M6). ES estimates IPT-G compared with wait-list control condition. Cohen d effect size: small (.15–.40), medium (.40 –.75), large (⬎.75). APAI ⫽ Acholi Psychosocial Assessment Instrument; ES ⫽ effect size.

sion symptoms [11]. This study is the first to demonstrate the effectiveness of psychotherapy for depression symptoms among children with a history of abduction by rebels in northern Uganda. Several factors that may have contributed to the differential effects of treatment by abduction history and gender are relevant to other studies of war-affected youth. Many former child soldiers experienced stigma on release from the LRA [35], and the opportunity for therapeutic support in a social setting may have offset postconflict difficulties. Further, there is substantial evidence that former abductees were at risk for continued domestic and community violence and poor economic opportunities after escape or release from the LRA, which might compound the effects of stigma [4,5,35]. Future adaptations of IPT-G for waraffected children might consider adding modules that specifically address dealing with traumatic experiences related to abduction, and anticipating and navigating community stigma and reintegration challenges for former child soldiers. Given concerns about exacerbating stigma toward former child soldiers [25], in situations like in northern Uganda, where most youth have had broad exposure to war, the feasibility of intervention models involving both former child soldiers and those who were not abducted should be explored. Gender differences also require ongoing investigation in research on war-affected youth. Our observations and a previous qualitative study in these same IDP camps suggest that, overall, the experiences of boys and girls in these camps are qualitatively different. Interviews with local people indicated that it was common for boys to “roam about” the camp (lak atata) [15]. In comparison, girls had more limited mobility, with much of their social contact restricted to the immediate household or interac-

gender, and abduction history based on the final model (M6). For nonabducted female subjects, a lower adjusted post-treatment mean was evident for IPT-G (21.3) than CP (39.6) or wait-list control (36.03). This pattern was consistent for both male and female subjects with an abduction history (former child soldiers) (see Figure 2). In contrast, there was little difference in posttreatment means for nonabducted male subjects. Because the interaction terms must be interpreted in combination, we present effect size (ES) estimates in combination with mean estimates in Table 3, wherein the ES estimates for the IPT-G treatment group are in contrast to the untreated wait-list control group. IPT-G yielded large ESs for female subjects without a history of abduction and male former child soldiers (1.06 and .92, respectively). A moderate ES of IPT-G was also indicated for female former child soldiers (.50). Discussion We found that the effect of IPT-G on depression symptoms was different among youth with a history of abduction compared with never abducted youth. Importantly, both male and female former child soldiers benefited from IPT-G. The intervention was most effective among war-affected adolescent girls without a history of abduction by rebel forces; however, in contrast to our hypotheses, those least likely to benefit from IPT-G were male subjects with no history of abduction. These findings are compelling, given that few interventions have been evaluated for use with former child soldiers [11]. Individualized trauma-focused treatment has demonstrated effectiveness for reducing posttraumatic stress disorder symptoms among former child soldiers in northern Uganda, but has shown little specific effect for depres-

Mean APAI Depression Problems

Group Inter personal Therapy 50 45

Creative Play

Waitlist

50 45

● ●

50

● ●

● ●

●

40

35

35

35

30

30

30

25

25

25

●

20 Time 1

Time 2

●

45

40

●

●

20

40

●

● Abducted Females ● Non−Abducted Females Abducted Males Non−Abducted Males

20 Time 1

Time 2

Time 1

Time 2

Figure 2. Observed pre- and post-treatment means of Acholi Psychosocial Assessment Instrument depression problems scores by gender, abduction status, and treatment condition.

T.S. Betancourt et al. / Journal of Adolescent Health 51 (2012) 544 –550

tion with others while collecting water from local wells or harvesting family plots. For boys with an abduction history, having an intimate group setting to share their concerns and get advice from other young men, and having a male role model in the group facilitator may have offered an important supplement to the social contacts that many male subjects had in the camps. Our qualitative data collected in this setting illustrated that seeking guidance from others and “gathering” around a child who presents with symptoms of the depression-like problems of kumu, par, or two tam was seen as helpful from a local cultural perspective [15]. For this reason, it could be hypothesized that the availability of social supports and guidance in the IPT-G groups may have been helpful for formerly abducted male youth who otherwise experienced social isolation and stigma on return to home. Our review of the IPT-G facilitator records indicated that formerly abducted boys appeared to benefit in noticeable ways from the opportunity to discuss their war-related experiences and from efforts to reintegrate into life in the IDP camps in the group sessions [30]. This same opportunity may not have been meaningful for boys without an abduction history. These findings augment previous research suggesting that male former child soldiers demonstrate proactive efforts to reintegrate. They have been shown in some research to be more likely to engage in community and civic responsibilities (such as voting) after war than male subjects without a history of abduction [36]. Our findings indicate a need for further examination of the differential effects of depression treatment among war-affected youth. In previous studies, more activity-based or skills-oriented therapies have shown greater effectiveness for boys, particularly younger boys [37]. For older boys without an abduction history, other features of evidence-based treatments for depression, such as behavioral activation and sequential problem solving, may augment skill building and therapeutic content offered by standard IPT-G. Further study of the differential effects of IPT-G treatment by trauma history, gender, and other potential treatment modifiers, such as age and cultural background, is necessary in other populations of youth affected by violent conflict. Such “unpacking” of trauma histories and moderators of treatment effects will inform determinations about the fit of potential treatment models to different cultures and settings. Overall, given the constellation of risk factors characterizing child soldiers and their risk for poor psychological adjustment and social reintegration [3], more research is needed that focuses on interventions specifically designed and tested for this subgroup of war-affected youth. Limitations This exploratory study indicates that gender and having a history of abduction by rebel forces may moderate the effects of IPT-G treatment for depression in this setting in northern Uganda. Without a more targeted design, it is difficult to isolate the specific components or mechanisms of IPT-G contributing to the therapeutic effects observed (e.g., psychoeducation vs. interpersonal problem solving) in former child soldiers. Moreover, we were not able to determine how long the benefits will endure, as no follow-up study of the RCT was conducted. While our experience among adults in southwest Uganda suggests that improvements may be comparatively long-lasting [38], the differences in setting and developmental age of study subjects at this site must be considered. In addition, there could also be intervention effects that were not immediately apparent at the time of these

549

assessments, such as the potential for changes in functioning several months after intervention. Such possibilities should be explored further in future research. It must also be acknowledged that our measure of child soldier status was limited. Previous research on former child soldiers indicates a great deal of variation in the experience of abduction, including the nature and severity of violence exposure [3,35]. Given the broad range of tasks and roles common among children associated with armed forces and armed groups (child soldiers; “A child associated with an armed force or armed group refers to any person below 18 years of age who is, or who has been, recruited or used by an armed force or armed group in any capacity, including but not limited to children, boys and girls, used as fighters, cooks, porters, spies or for sexual purposes. It does not only refer to a child who is taking, or has taken, a direct part in hostilities.” [2]), use of more detailed abduction and war exposure measures would have provided us with greater ability to examine the role of specific war-related and postconflict variables (e.g., stigma) to further understand their association with treatment outcomes. There is great heterogeneity within the groups assessed (male subjects, female subjects, abducted and nonabducted youth), and the results may have been further informed by the measurement of other symptom areas and potential protective factors across the social ecology (including presence or absence of family support). Finally, small sample sizes in the subgroups limited the statistical power of these exploratory analyses. Conclusions These findings contribute to a growing, yet nascent, evidence base on mental health interventions for war-affected youth, including former child soldiers [39]. The present study indicates that the effectiveness of IPT-G for war-affected youth may vary according to the joint effects of abduction status and gender. These findings raise the possibility that the effectiveness of other treatments might also vary by similar factors, and that such treatment moderators should be explicitly investigated in future studies [21]. Overall, research in higher resource settings is moving beyond the investigation of efficacy and effectiveness to identify both treatment moderators as well as specific mechanisms of treatment that are associated with optimal outcomes [40]. It is time to apply this same lens to interventions for waraffected youth [39]. Acknowledgments The authors are grateful to the project staff, local leaders, and the communities for their warm welcome and professional engagement with the study. They thank Ista Zahn and Sidney Atwood for their assistance with data management and analysis. This study was supported by funding from World Vision, War Child Holland, grant 1K01MH077246-01A2 from the National Institute of Mental Health, grant P60 MD002261 from the National Center for Minority Health and Disparities, the American Australian Association and Morgan Stanley. References [1] Coalition to Stop the Use of Child Soldiers. Child soldiers: Global report 2004. London, UK: Coalition to Stop the Use of Child Soldiers, 2004. [2] UNICEF. The Paris Principles: Principles and guidelines on children associated with armed forces or armed conflict. Paris, France: UNICEF, 2007. [3] Betancourt TS, Borisova I, Rubin-Smith JE, et al. Psychosocial adjustment and social reintegration of children associated with armed forces and armed

550

[4] [5] [6] [7]

[8]

[9]

[10]

[11]

[12]

[13]

[14]

[15]

[16]

[17]

[18]

[19]

[20]

[21]

T.S. Betancourt et al. / Journal of Adolescent Health 51 (2012) 544 –550

groups: The state of the field and future directions. Austin, TX: Psychology Beyond Borders, 2008. Klasen F, Oettingen G, Daniels J, Adam H. Multiple trauma and mental health in former Ugandan child soldiers. J Trauma Stress 2010;23:573– 81. Annan J, Brier M. The risk of return: Intimate partner violence in Northern Uganda’s armed conflict. Soc Sci Med 2010;70:152–9. Derluyn I, Broekaert E, Schuyten G, De Temmerman E. Post-traumatic stress in former Ugandan child soldiers. Lancet 2004;363:861–3. Betancourt TS, Borisova I, de la SoudiÉre M, et al. Sierra Leone’s child soldiers: War exposures and mental health problems by gender. J Adolesc Health 49:21– 8. Bolton P, Bass J, Betancourt TS, et al. Interventions for depression symptoms among adolescent survivors of war and displacement in northern Uganda: A randomized controlled trial. JAMA 2007;298:519 –27. Layne CM, Saltzman WR, Poppleton L, et al. Effectiveness of a school-based group psychotherapy program for war-exposed adolescents: A randomized controlled trial. J Am Acad Child Adolesc Psychiatry 2008;47:1048 – 62. Ehntholt KA, Smith PA, Yule W. School-based cognitive-behavioural therapy group intervention for refugee children who have experienced warrelated trauma. Clin Child Psychol Psychiatry 2005;10:235–50. Ertl V, Pfeiffer A, Schauer E, et al. Community-implemented trauma therapy for former child soldiers in northern Uganda: A randomized controlled trial. JAMA 2011;306:503–12. Loughry M, Ager A, Flouri E, et al. The impact of structured activities among Palestinian children in a time of conflict. J Child Psychol Psychiatry 2006; 47:1211– 8. Tol WA, Komproe IH, Susanty D, et al. School-based mental health intervention for children affected by political violence in Indonesia: A cluster randomized trial. JAMA 2008;300:655– 62. Barber BK. Glimpsing the complexity of youth and political violence. In: Barber BK, ed. Adolescents and War: How Youth Deal With Political Violence. New York, NY: Oxford University Press, 2009. Betancourt TS, Speelman L, Onyango G, Bolton P. A qualitative study of mental health problems among children displaced by war in northern Uganda. Transcult Psychiatry 2009;46:238 –56. Betancourt TS, Brennan RT, Rubin-Smith J, et al. Sierra Leone’s former child soldiers: A longitudinal study of risk, protective factors, and mental health. J Am Acad Child Adolesc Psychiatry 2010;49:606 –15. Slone M. Growing up in Israel: Lessons on understanding the effects on political violence on children, In: Barber BK, ed. Adolescents and War: How Youth Deal With Political Violence. New York, NY: Oxford University Press, 2009. Betancourt TS, McBain R, Newnham EA, et al. Trajectories of internalizing problems in war-affected Sierra Leonean youth: Examining conflict and post-conflict factors. Child Dev (in press). Betancourt TS, Williams T. Building an evidence base on mental health interventions for children affected by armed conflict. Intervention 2008;6: 39 –56. Tol WA, Jordans MJ, Komproe IH, et al. Interventions for children affected by war: Synthesis of cluster randomized trials in four countries. Florence, Italy: International Congress for Transcultural Psychiatry, 2009. Tol WA, Komproe IH, Jordans MJD, et al. Mediators and moderators of a psychosocial intervention for children affected by political violence. J Consult Clin Psychol 2010;78:818 –28.

[22] Kraemer HC, Wilson T, Fairburn CG, Agras WS. Mediators and moderators of treatment effects in randomized clinical trials. Arch Gen Psychiatry 2002; 59:877– 83. [23] Bolton P, Bass J, Neugebauer R, et al. Results of a clinical trial of a group intervention for depression in rural Uganda. JAMA 2003;289:3117–24. [24] Bas¸og˘lu M. Treatment for depression symptoms in Ugandan adolescent survivors of war and displacement. JAMA 2007;298:2138. [25] Betancourt TS, Agnew-Blais J, Gilman SE, et al. Past horrors, present struggles: The role of stigma in the association between war experiences and psychosocial adjustment among former child soldiers in Sierra Leone. Soc Sci Med 2010;70:17–26. [26] Betancourt TS, Bass J, Borisova I, et al. Assessing local instrument reliability and validity: A field-based example from northern Uganda. Soc Psychiatry Psychiatr Epidemiol 2009;44:685–92. [27] Mufson L, Dorta KP, Wickramaratne P, et al. A randomized effectiveness trial of interpersonal psychotherapy for depressed adolescents. Arch Gen Psychiatry 2004;61:577– 84. [28] Mufson L, Gallagher T, Dorta KP, Young JF. A group adaptation of interpersonal psychotherapy for depressed adolescents. Am J Psychother 2004;58: 220 –37. [29] Bolton P, Bass J, Neugebauer R, et al. Group interpersonal psychotherapy for depression in rural Uganda: A randomized controlled trial. JAMA 2003;289: 3117–24. [30] Verdeli H, Clougherty K, Onyango G, et al. Group interpersonal psychotherapy for depressed youth in IDP camps in Northern Uganda: Adaptation and training. Child Adolesc Psychiatr Clin N Am 2008;17:605–24. [31] Raudenbush SW, Bryk AS. Hierarchical linear models: Applications and data analysis methods. London, UK: Sage Publications, 2002. [32] Omar RZ, Thompson SG. Analysis of a cluster randomized trial with binary outcome data using a multilevel model. Statist Med 2000;19:2675– 88. [33] Raudenbush SW, Bryk AS, Congdon RT. HLM 7: Hierarchical linear and nonlinear modeling. Lincolnwood, IL: Scientific Software International, Inc., 2011. [34] SPSS, Inc. PASW Statistics 18 core system user’s guide. Chicago, IL: SPSS Inc., 2007. [35] Annan J, Blattman C, Horton R. The state of youth and youth protection in Northern Uganda: Findings from the survey of war affected youth. New York, NY: UNICEF, 2006. [36] Blattman C. From violence to voting: War and political participation in Uganda. Am Polit Sci Rev 2009;103:231– 47. [37] Gordon JS, Staples JK, Blyta A, Bytyqi M. Treatment of posttraumatic stress disorder in postwar Kosovo high school students using mind-body skills groups: A pilot study. J Trauma Stress 2004;17:143–7. [38] Bass J, Neugebauer R, Clougherty KF, et al. Group interpersonal psychotherapy for depression in rural Uganda: 6-month outcomes: Randomised controlled trial. Br J Psychiatry 2006;188:567–73. [39] Betancourt TS. Attending to the mental health of war-affected children: The need for longitudinal and developmental research perspectives. J Am Acad Child Adolesc Psychiatry 2011;50:323–5. [40] Newnham EA, Page AC. Bridging the gap between best evidence and best practice in mental health. Clin Psychol Rev 2010;30:127– 42.