Caregiver and Adolescent Mental Health in Ethiopian Kunama Refugees Participating in an Emergency Education Program

Journal of Adolescent Health 51 (2012) 357–365

www.jahonline.org Original article

Caregiver and Adolescent Mental Health in Ethiopian Kunama Refugees Participating in an Emergency Education Program Theresa S. Betancourt, Sc.D., M.A.a,*, Monica Yudron, M.A.b, Wendy Wheaton, M.I.A.c, and Mary C. Smith-Fawzi, Sc.D.d,e a

Department of Global Health and Population, Research Program on Children and Global Adversity, Franåois-Xavier Bagnoud Center for Health and Human Rights at the Harvard School of Public Health, Boston, Massachusetts Department of Human Development and Education, Harvard Graduate School of Education, Boston, Massachusetts c Global Education Cluster, Geneva, Switzerland d Program in Infectious Disease and Social Change, Department of Global Health and Social Medicine, Harvard Medical School, Boston, Massachusetts e Partners In Health, Boston, Massachusetts b

Article history: Received July 14, 2011; Accepted January 3, 2012 Keywords: Refugees; Caregiver distress; Adolescent mental health; War; Ethiopia

A B S T R A C T

Purpose: To examine the role of caregiver mental health and risk and protective factors in influencing levels of internalizing and externalizing emotional and behavioral symptoms over time among a sample of refugee adolescents. Methods: Prospective study of 153 Kunama refugee adolescents receiving an emergency education intervention while living in a camp in Ethiopia. Surveys were collected in 2001 (T1) and 2002 (T2). Adolescent and caregiver mental health were assessed using a Kunamenga adaptation of the Youth Self Report; caregiver mental health was assessed using the Hopkins Symptom Checklist-25. Attitudes toward education, satisfaction with education programming, socioeconomic status, and perceptions of access to services were also explored as variables potentially influencing adolescent mental health at follow-up. Results: Caregiver distress was significantly associated with youth externalizing behavior symptoms (␤ ⫽ 8.34, p ⬍ .001) and internalizing symptoms (␤ ⫽ 4.02, p ⬍ .05). Caregiver perceived access to services had a protective effect on externalizing behaviors (␤ ⫽ ⫺7.54, p ⬍ .05) and internalizing behaviors (␤ ⫽ ⫺13.67, p ⬍ .001). Higher socioeconomic status (␤ ⫽ ⫺1.47, p ⬍ .05) had a protective effect on internalizing symptoms. In terms of modifying effects, among youth with distressed caregivers, those who were satisfied with the International Rescue Committee education intervention had a lower internalizing score (␤ ⫽ ⫺6.34, p ⬍ .001) compared with those who were not satisfied with the program. Conclusions: This study presents a rare prospective investigation of caregiver–adolescent mental health during an active refugee displacement. Results suggest that programs targeting mental health in refugee children should consider children within the larger family system, including caregiver influence on child and adolescent mental health adjustment over time. 䉷 2012 Society for Adolescent Health and Medicine. All rights reserved.

In recent years, research has highlighted that the mental health of caregivers plays an important role in mediating child mental health and behavioral outcomes [1,2]. Although shared

* Address correspondence to: Theresa S. Betancourt, Sc.D., Department of Global Health and Population, Research Program on Children and Global Adversity, FranÈois-Xavier Bagnoud Center for Health and Human Rights at the Harvard School of Public Health. 651 Huntington Avenue, Boston, MA 02115. E-mail address: [email protected] (T.S. Betancourt).

IMPLICATIONS AND CONTRIBUTION

This study presents a rare prospective investigation of adolescent mental health during an active refugee displacement in Ethiopia. Findings suggest that caregiver mental health has a strong association with adolescent outcomes. These results have important implications for future interventions to improve the health of refugee youth.

genetic heritage is thought to contribute to a predisposition to mental health problems in the children of depressed parents [3], shared contextual stressors and the effects of parental symptoms on child mental health are thought to be of considerable importance as well [4 –9]. Specifically, caregiver mental health problems shape the family environment and influence factors like connectedness, access to services, and availability of social support, all of which influence child mental health [10].

1054-139X/$ - see front matter 䉷 2012 Society for Adolescent Health and Medicine. All rights reserved. doi:10.1016/j.jadohealth.2012.01.001

358

T.S. Betancourt et al. / Journal of Adolescent Health 51 (2012) 357–365

An emerging body of research examines how wartime stressors affect caregiver mental health, and how caregiver distress may be linked to child outcomes in conflict settings [11–16]. For instance, in a study of war-exposed Central American families, Locke et al [14] observed that posttraumatic stress disorder in children was best predicted by levels of maternal posttraumatic stress disorder, rather than by traumatic events. In a study of Mexican girls living in refugee camps, child depression was best predicted by maternal physical and mental health [16]. Similar dynamics have been observed in Sri Lanka [12], Afghanistan [15], Israel [13], and Bosnia and Herzegovina [11]. Taken together, these studies make important contributions to our understanding of risk factors facing war-affected children and families, and can help inform targeted culturally relevant interventions. However, none of the research to date has examined these associations in sub-Saharan Africa, despite the huge burden of conflict borne by this region. To inform better holistic interventions and culturally appropriate policy responses for refugees and war-affected populations in sub-Saharan Africa, research exploring family relationships and mental health is needed [2,6 –10,17,18]. The present study contributes to the evidence base on African refugee populations by investigating the relationship between caregiver mental health and adolescent outcomes in a prospective study of refugee families displaced by an active conflict. Background The Kunama ethnic group comprises agro-pastoralists traditionally residing in the Eritrean–Ethiopian border region. When war between Eritrea and Ethiopia broke out in 1998, the Kunama people were caught between both sides. In the early stages of the conflict, small numbers of Eritrean Kunama refugees crossed the Gash River border into Ethiopia to avoid conscription, prompting suspicions within the Eritrean government of the group’s disloyalty. During the ensuing years of conflict, fears of persecution spurred further Kunama refugee flights across the Gash River, culminating in a mass migration in 2000 when Eritrea moved to regain control of the border area [19]. During the conflict, Kunama refugees experienced death of family members, loss of homes, disintegration of communities, and separation from extended families [19]. To attend to the immediate needs of refugee children, in 2001, the International Rescue Committee (IRC) initiated an emergency education and psychosocial support program for Kunama refugee children and adolescents residing in the refugee camps in Ethiopia [20]. The project provided broad-based emergency education, recreation, and psychosocial activities. Classes were held in shifts in the refugee camp in large school tents, and other recreation and psychosocial activities, such as soccer and drama clubs, were initiated in the camp when school was not in session. In concert with the launch of the emergency education program, the IRC and the first author initiated a prospective study to assess caregiver and adolescent mental health at baseline and after 8 months of program implementation [20]. For the purposes of this article, “mental health problems” in children and adolescents refer to a range of emotional and behavioral problems as assessed by the internalizing and externalizing subscales of the Achenbach Youth Self Report (YSR) [21]. Mental health problems in adult caregivers refer to symptoms of anxiety and depression as assessed on the Hopkins Symptom Checklist-25

[22]. We define “caregiver(s)” as the person(s) serving as the primary source of care for each child. A unique feature of the present study is its prospective examination of mental health problems in children and adolescents, which allows for investigation of factors influencing mental health over time. In this study, information gathered pre- and postintervention was used to perform robust analyses of patterns of emotional and behavioral adjustment in children and adolescents, looking at both programmatic and family variables. In addition to examining a number of risk factors of interest— including poor caregiver mental health—the study also sought to illuminate protective processes contributing to child and adolescent adjustment to inform ongoing intervention development [4,5,10,23,24]. Methods Sample Baseline assessments were conducted in July and August 2001 (T1). Inclusion criteria for adolescents comprised the following: (1) enrolled in the IRC education program; (2) a Kunama refugee aged ⬍18 years; (3) living in the Walanhiby refugee camp for at least 1 month; and (4) without a severe cognitive disability that limited comprehension of the questionnaire as determined by study social workers (no such cases were identified). Adolescents fulfilling inclusion criteria were screened using a registry compiled by government authorities. We selected all adolescents from the registry for whom complete contact information was available and which met the desired sample size (as determined by a power calculation). At baseline, N ⫽ 168 ethnic Kunama adolescents, aged 11–18 years, and one index caregiver per adolescent (N ⫽ 162) were enrolled (six adolescents were in a group foster situation with no immediate caregiver). When multiple caregivers were present in a household at the time of assessment, they were asked to identify the caregiver who “knows the child best.” This participant became the index caregiver who completed caregiver reports with respect to the adolescent as well as a self-report of caregiver mental health. During the second assessment 1 year later (T2), 6 adolescents and 10 caregivers were lost to follow-up owing to relocation or death, yielding a follow-up sample of 153 adolescents and 152 caregivers. Adolescents completed measures at T1 and T2; caregiver distress was measured at T2. Procedure and ethics The study received approval from the Human Subjects Committee at the Harvard School of Public Health. Because of low literacy in this population, informed consent/assent was obtained from each respondent by reading aloud an explanatory text and then asking for verbal adult consent as well as independent verbal adolescent assent. Approval to conduct the study was given by the IRC and the local Prefecture of the Ethiopian Ministry of Education, as well as the Administration for Refugees and Reternee Affairs (ARRA), the government agency overseeing the Walanhiby camp. Study interviews were conducted by eight trained refugee interviewers (four male, four female) who were fluent and literate in the Kunamenga language. Questionnaires were administered verbally. Supervision was provided in person by IRC field staff. The first author provided supervision both in person and via

T.S. Betancourt et al. / Journal of Adolescent Health 51 (2012) 357–365

phone and e-mail contact. All interviews were conducted in the camp setting with precautions taken to ensure participants’ privacy. Adolescents and caregivers were most often interviewed in their home compounds away from others. Caregivers were interviewed out of earshot of adolescents and vice versa. Measures Before initiating the baseline assessment, discussions were held with camp leaders and small groups of Kunama caregivers and adolescents to ensure that conceptualizations of locally relevant constructs were well reflected in the measures [25,26]. Key informant interviews with camp leaders and focus groups of youth and caregivers were used to review and suggest adaptations to selected measures. To ensure cultural acceptability and clarity of constructs assessed, all survey questions and consent forms were translated into Kunamenga and back-translated into English following a standard protocol [27]. Demographics. The adolescent survey included demographic items such as age, religion, ethnicity, migration history, gender, family size, education, family separation, economic status, and disability status. The caregiver survey included family demographic items such as caregiver education, family economic status, and displacement history. Emotional and behavioral problems in adolescents. Emotional and behavioral problems in adolescents were measured using adapted self-report and caregiver-report versions of the YSR [21] and Child Behavior Checklist [28]. These measures had been translated and adapted for use with Ethiopian adolescents in previous studies [29]. Because written Kunamenga shares many similarities to Amharic and Tigrinya, these language versions were used as the basis for creating the study’s measures in Kunamenga. The Child Behavior Checklist and YSR instruments comprise statements about the existence of feelings or behaviors experienced by the respondent during the past 6 months. On the YSR, responses to each item were measured on a 3-point Likert scale: “not true,” “somewhat or sometimes true,” or “very or often true.” The YSR assumes a dimensional approach, assessing problems along a continuum whereby internalizing problems refer to inward-directed experiences of distress commonly in depression or anxiety, and externalizing problems refer to outward-directed behaviors such as hostility and aggression [30]. Community feedback on contextual and cultural misalignment led us to remove a few items from the YSR measure. For instance, consultation with staff and community representatives led to the removal of items inquiring about suicidal ideation and self-harm because of their sensitive nature in this predominantly Muslim population. A 29-item internalizing problems subscale was created by summing items pertaining to social withdrawal, somatization, and anxiety/depression. Similarly, a 33-item externalizing problems subscale was created to include all items pertaining to delinquency and aggression. Measures of internal consistency indicated that these subscales functioned well within the sample (Cronbach’s ␣ ⫽ .62 for internalizing problems and .77 for externalizing problems).

359

services assessed how readily adolescents could access medical care, educational programs, or counseling services, if needed. A locally derived checklist of war experiences assessed adolescents’ displacement and family separation histories. Satisfaction with the IRC educational intervention. Adolescents were asked to indicate their perceptions of the IRC education intervention by responding to five items including whether they would like to attend more hours of school and their overall level of satisfaction with the IRC school. Caregiver distress. Caregiver distress was assessed at T2 using the Hopkins Symptom Checklist-25 [22,32]. This scale has been translated and validated in several sub-Saharan African cultures [33–36] and demonstrated very good internal consistency among our sample (Cronbach’s ␣ ⫽ .93). The instrument screens for symptoms experienced in the previous week using a 4-point Likert scale, ranging from “not being bothered at all” by the symptom to being “bothered extremely.” Individual distress severity was measured as the average of these items. A recommended cutoff point used in other war-affected populations (mean symptom score for 25 items exceeding 1.75) was applied to determine scores in the likely clinical range [37]. Roughly 18% of the caregivers in this refugee population scored 1.75 or greater. Statistical analysis Statistical analyses were performed using STATA SE statistical software version 11.1 (Stata Corp., College Station, TX). Sensitivity analyses were conducted comparing the means of key demographic characteristics of the full and analytic sample, and no statistically significant differences were found. Comparisons of means, by gender, for the dependent variables (adolescent externalizing and internalizing emotional and behavioral symptoms) were examined using t tests, and comparisons of frequencies across subgroups were made using ␹2 tests. Bivariate associations between key variables were examined using correlation matrices. Next, a five-stage hierarchical modelbuilding approach was used to construct multivariate linear regression models examining the main outcomes of interest (adolescent internalizing and externalizing problems). First, T1 internalizing or externalizing behavior was added to each analysis before the inclusion of all independent variables. Second, the main effect of caregiver distress was modeled. Third, a set of family- and adolescent-level socioeconomic and demographic characteristics including caregiving arrangement was added to the models in sequence. Fourth, a series of covariates indicating adolescent perception of education and community was included in the model. At each of these first four steps, interactions between each covariate and caregiver distress were added to the models to test for moderating effects. Type of caregiver (i.e., biological parents vs. extended family member) was investigated as a possible correlate of adolescent internalizing and externalizing symptoms, and as a possible moderator of the impact of having a caregiver struggling with distress. Results

Informal and formal supports and services for adolescents. Perceived social support from peers, family, and significant others was assessed using a translation of the multidimensional scale of perceived social support [31]. A locally derived scale of access to

Table 1 contains descriptive statistics on key sociodemographic adolescent and family characteristics. Most adolescents in the sample (average age 14 years at T1) lived with at least one

360

T.S. Betancourt et al. / Journal of Adolescent Health 51 (2012) 357–365

Table 1 Adolescent and family sociodemographic characteristics and mental health outcomes of the full sample by gender Characteristic

Sociodemographic characteristics Age of adolescents (years) Age of caregivers (years) Adolescent reports of access to services Caregiver reports of access to services Caregiver at time of study Biological mother Biological father Both biological mother and father Siblings Extended family Family SES Both a radio and cattle A radio or cattle Neither a radio nor cattle Adolescent views education as important Mental health outcomes Caregiver distress Externalizing behavior T2 Internalizing symptoms T2

Full sample (N ⫽ 153) Mean (SD)

Males (N ⫽ 68) Mean (SD)

Females (N ⫽ 85) Mean (SD)

p*

14.02 (1.97) 42.35 (11.58) 1.69 (.46) 1.79 (.37)

13.8 (2.0) 41.26 (11.22) 1.69 (.41) 1.79 (.33)

14.19 (1.94) 43.21 (11.86) 1.69 (.50) 1.78 (.40)

.41 .61 .99 .92

N (%)

N (%)

N (%)

24 (15.7) 5 (3.2) 97 (63.3) 6 (3.5) 17 (10.0)

10 (14.7) 1 (1.5) 56 (82.3) 3(3.8) 4 (5.0)

14 (16.5) 4 (4.7) 41 (48.2) 3 (3.3) 13 (14.4)

.36 .36 .04 .88 .04

14 (8.2) 53 (31.2) 103 (60.6) 149 (97.4)

3 (3.8) 26 (32.5) 51 (63.8) 66 (97.1)

11 (12.2) 27 (30.0) 52 (57.8) 83 (97.6)

.05 .73 .43 .82

Mean (SD)

Mean (SD) 95% CI

Mean (SD) 95% CI

35.83 (7.76) 8.02 (6.35) 14.52 (5.55)

35.99 (6.94) 1–2.73 7.23 (5.88) 1–23 14.25 (5.09) 6–22

34.51 (8.33) 1–2.87 8.73 (6.70) 2–22 14.76 (6.70) 6–24

.23 .13 .55

SD ⫽ standard deviation; SES ⫽ socioeconomic status; CI ⫽ confidence interval. * t tests were used to evaluate mean differences between male and female participants for continuous variables and ␹2 test used for categorical data.

biological parent. Variables examined did not differ by gender of adolescents, with the exception of family socioeconomic status (SES): female adolescents were significantly more likely to live in a family who owned cattle, a radio, and so forth, when compared with males; females were also less likely to live with both biological parents. Externalizing emotional and behavioral symptoms among Kunama adolescents Table 2 contains the results of a series of bivariate correlations between externalizing behavior and a set of social and educational variables. This table reveals several potential predictors of externalizing behavior. For example, as expected, adolescent perceptions of community connectedness and access to services within the refugee community generally had an inverse statistically significant correlation with adolescent externalizing behavior. In addition, caregiver distress had a strong, positive, and statistically significant correlation with externalizing behavior at T2. When models were fit to determine more precisely the nature of the relationship between adolescent externalizing behavior and these variables, caregiver distress remained a robust predictor in adjusted models. Table 4 contains three models illustrating this relationship. Model 1 shows the baseline relationship between T1 and T2 adolescent externalizing behavior. Model 2 shows the main effect of caregiver distress after controlling for T1 adolescent externalizing behavior. Model 3 is the final parsimonious model fit to illustrate the key predictors of adolescent externalizing behavior. In this model, caregiver distress had a statistically significant and positive relationship with externalizing behaviors (␤ ⫽ 8.34, p ⬍ .001), such that, on average, adolescents with a caregiver scoring at or greater than the cutoff point in the distress measure had externalizing behavior that was 8.34 units higher than adolescents with a care-

giver scoring below the cutoff point. In the final parsimonious model, caregiver reports of access to services had an inverse and statistically significant relationship with adolescent externalizing behavior (␤ ⫽ ⫺7.54, p ⬍ .05). Yet, caregiver reports of access to services moderated the impact of distress, such that caregivers who scored greater than or at the distress cutoff point but who also reported having access to services had adolescents with externalizing behaviors 4.26 units lower than adolescents with distressed caregivers who did not report having access to health care services; however, this did not reach statistical significance (p ⬍ .10). No other sociodemographic nor community or education variables were found to have a statistically significant main or moderated effect on adolescent externalizing behavior. Internalizing emotional and behavioral symptoms among Kunama adolescents When adolescent internalizing behavior problems were considered, a similar robust relationship was found between internalizing emotional and behavioral symptoms and caregiver distress. The bivariate correlations displayed in Table 3 indicate inverse and significant correlations between adolescent internalizing symptoms, family SES, and perceived access to services. The correlation between internalizing symptoms and attitudes about education was also inverse, but did not reach statistical significance (p ⫽ .07). Table 4 displays the results of a set of regression models to determine more precisely the relationship between caregiver distress and adolescent internalizing behavior. Model 1 displays the relationship between T1 and T2 internalizing symptoms. In model 2, the main effect of caregiver distress after controlling for T1 internalizing behavior is shown. In the final model 3, adolescent internalizing behavior at T2 is predicted by caregiver distress (␤ ⫽ 4.02, p ⬍ .05). Having a caregiver score in the clinical range for distress was associated

Table 2 Bivariate correlation matrix for externalizing behavior at T2 and social and educational covariates Adolescent externalizing behavior T2

Adolescent Externalizing Behavior T2 Caregiver distress (anxiety and depression) Female

1.00

Age of adolescent Adolescent externalizing behavior T1 Family SES Highest grade completed by caregiver Highest grade completed by teen Teen expects education to be practical Adolescent perception of community connection Caregiver access to services for self and adolescent Adolescent access to services for self Adolescent believes education is important

.56 .00

Caregiver distress (anxiety and depression)

Female

Age of adolescent

Adolescent externalizing behavior T1

Family SES

1

Highest grade completed by caregiver

Highest grade completed by teen

Teen expects education to be practical

Adolescent perception of community connection

Caregiver access to services for self and adolescent

Adolescent access to services for self

Adolescent believes education is important T.S. Betancourt et al. / Journal of Adolescent Health 51 (2012) 357–365

Variable

1

.1 .24 ⫺.05 .57 .13 .12 ⫺.1 .2 .08 .34 ⫺.03 .69 ⫺.22 .01 ⫺.38 .00

.1 .23 ⫺.11 .18 .03 .73 ⫺.07 .37 .04 .59 ⫺.04 .65 ⫺.3 .00 ⫺.4 .00

1 .13 .11 .14 .1 .08 .35 .02 .82 ⫺.3 .00 .05 .55 .00 .96

.08 .35 .07 .37 .08 .32 .16 .05 .08 .3 ⫺.06 .45

.01 .91 ⫺.15 .07 .09 .25 .09 .29 ⫺.03 .71

.14 .09 .12 .14 .08 .32 ⫺.02 .84

⫺.49 .00

⫺.68 .00

⫺.05 .56

.02 .82

⫺.13 .11

.07 .4

.00 1

⫺.02 .85

.36 .00

.49 .00

⫺.51 .00 ⫺.32 .00

⫺.62 .00 ⫺.23 .00

.00 .99 .00 .97

.07 .37 .18 .02

.00 .99 .08 .33

.1 .23 .09 .27

.02 .8 ⫺.03 .72

⫺.02 .85 .03 .68

.36 .00 .15 .06

.49 .00 .23 .00

1

1 .37 .00 .02 .81 ⫺.05 .52

1 .11 .16 .04 .64

1 .27 .00

1

1

.69 .00 .26 .00

1 .21 .01

1

SES ⫽ socioeconomic status.

361

362

Table 3 Bivariate correlation matrix for internalizing symptoms at T2 and social and educational covariates Adolescent internalizing behavior T2

Adolescent Internalizing Behavior Time 2 Caregiver distress (anxiety and depression) Female

1.00

Age of adolescent Adolescent internalizing behavior T1 Family SES Highest grade completed by caregiver Highest grade completed by teen Teen expects education to be practical Adolescent perception of community connection Caregiver access to services for self and adolescent Adolescent access to services for self Adolescent believes education is important SES ⫽ socioeconomic status.

Caregiver distress (anxiety and depression)

Female

Age of adolescent

Adolescent internalizing behavior T1

Family SES

Highest grade completed by caregiver

Highest grade completed by teen

.5 .00 .01 .87 ⫺.08 .33 .15 .07 ⫺.21 .01 .00 .95 .00 .96 ⫺.12 .15 ⫺.23 .00 ⫺.4 .00

.1 .23 ⫺.11 .18 .03 .73 ⫺.07 .37 .04 .59 ⫺.04 .65 ⫺.3 .00 ⫺.4 .00 ⫺.68 .00

.13 .11 .14 .1 .08 .35 .02 .82 ⫺.3 .00 .05 .55 .00 .96 ⫺.05 .56

.08 .35 .07 .37 .08 .32 .16 .05 .08 .3 ⫺.06 .45 .02 .82

.01 .91 ⫺.15 .07 .09 .25 .09 .29 ⫺.03 .71 ⫺.13 .11

.14 .09 .12 .14 .08 .32 ⫺.02 .84 .07 .4

.37 .00 .02 .81 ⫺.05 .52 .00 1

.11 .16 .04 .64 ⫺.02 .85

⫺.34 .00 ⫺.15 .07

⫺.62 .00 ⫺.23 .00

.00 .99 .00 .97

.07 .37 .18 .02

.00 .99 .08 .33

.1 .23 .09 .27

.02 .8 ⫺.03 .72

⫺.02 .85 .03 .68

Teen expects education to be practical

Adolescent perception of community connection

Caregiver access to services for self and adolescent

Adolescent access to services for self

Adolescent believes education is important T.S. Betancourt et al. / Journal of Adolescent Health 51 (2012) 357–365

Variable

1 1 1 1 1 1 1 1 .27 .00 .36 .00 .36 .00 .15 .06

1 .49 .00 .49 .00 .23 .00

1

.69 .00 .26 .00

1 .21 .01

1

T.S. Betancourt et al. / Journal of Adolescent Health 51 (2012) 357–365

363

Table 4 Results of fitting a hierarchy of linear regression models for two adolescent mental health outcomes (N ⫽ 153) Variable

Intercept Main effects Externalizing behavior T1 Internalizing symptoms T1 Caregiver distress Female Family SES Highest level of education of caregiver Adolescent views education as important Adolescent values experience in IRC education intervention Caregiver has access to health services Moderated effects Caregiver education by caregiver distress Adolescent values IRC education intervention by caregiver distress Caregiver access to health services by caregiver distress R2 SSmodel SSerror dferror

Externalizing

Internalizing

Model 1 ␤ (SE)

Model 2 ␤ (SE)

5.41* (1.04)

3.64* (.86)

.26** (.08)

.27* (.07)

Model 3 ␤ (SE) 12.63** (4.00)

Model 1 ␤ (SE)

Model 2 ␤ (SE)

11.46* (1.75)

10.54* (1.82)

.06 412.96 6,100.34 151

.38 2,458.93 4,054.42 150

20.07* (3.62)

.27* (.07) .17*** (.09)

9.83* (1.09)

Model 3 ␤ (SE)

8.34* (1.99) 1.50 (.83) ⫺1.23*** (.65) .40 (.41) ⫺3.49 (2.62) 3.93 (2.60)

.15*** (.08) 10.54* (.50)

.12 (.08) 4.02**** (.20) ⫺.10 (.75) ⫺1.47**** (.57) 1.19* (.37) ⫺2.28 (2.24) 7.16* (1.86)

⫺7.54**** (3.31)

⫺13.67* (2.01)

⫺.23 (.26) ⫺3.86*** (2.10)

⫺.88* (.23) ⫺6.34* (1.89)

4.26*** (2.27)

8.94* (2.01)

.46 2,963.35 3,549.99 141

.02 103.65 4,645.88 151

.27 1279. 61 3,469.92 150

.41 1927. 65 2,821.88 141

SES ⫽ socioeconomic status. Values are reported as ␤ standard error (SE). * p ⬍ .001; ** p ⬍ .01; *** p ⬍ .10; **** p ⬍ .05.

with an internalizing problem score 4.02 points higher, on average, after adjusting for baseline, which is equal to an effect size of .72 standard deviation (SD). Family SES had a sustained predictive relationship with adolescent internalizing symptoms, such that higher average SES was associated with lower average levels of internalizing problems at T2 (␤ ⫽ ⫺1.47, p ⬍ .05), adjusting for all other factors. In addition, caregiver-perceived access to health services had an inverse statistically significant relationship with internalizing symptoms (␤ ⫽ ⫺13.67, p ⬍ .001). The effect of caregiver distress was moderated by adolescent satisfaction with the IRC intervention (␤ ⫽ ⫺6.34, p ⬍ .001), such that adolescents with distressed caregivers who were satisfied with the IRC program had a lower internalizing score by 6.34 points compared with children of distressed caregivers who were not satisfied with the IRC program. Discussion Evidence from this study suggests that caregiver distress plays an important role in the emotional and behavioral adjustment of Kunama refugee adolescents over time. In our study, caregiver distress was a robust predictor of externalizing and internalizing emotional and behavior problems in adolescents at follow-up, adjusting for all other factors, including baseline adolescent distress. This finding is consistent with a number of other studies that have examined the association between caregiver mental health and children’s psychological distress in war-affected settings [11–16]. In the present study, the association between caregiver distress and adolescent externalizing symptoms remained independent of any of the educational interventions underway in the camp at the time of the study, whereas adolescent satisfaction with the emergency education intervention moderated the effect of caregiver distress on internalizing symptoms. In other words,

our data indicate that participation alone in the education program did not significantly influence adolescents’ externalizing behavior outcomes, but positive attitudes toward education and access to other support services did have protective effects on internalizing symptoms. Of particular interest is the interaction observed whereby children with distressed caregivers who were satisfied with the IRC program had lower average levels of internalizing symptoms compared with children of distressed caregivers who were not satisfied with the IRC program. This finding suggests that educational interventions and family-focused programming may serve to complement one another. This study’s findings support recent theory on mental health of refugee children, which argues that the family and broader social support systems play a critical role in children’s recovery from war- and displacement-related stressors [2,4 – 6,23,24]. As early as World War II, Burlingham and Freud observed effects of caregiver response to war stressors on children’s psychological distress [38]. The present study is the first documentation of such dynamics among a population of refugees in sub-Saharan Africa. In the Walanhiby refugee camp context, we observed that despite programmatic efforts to embed psychosocial supports within educational services (e.g., group discussions about stresses and worries), adolescents enrolled in these interventions did not flourish as much, on average, when caregiver mental health was poor. These results suggest that more targeted attention must be paid to the larger family environment to maximize the potential of youth-focused educational and psychosocial interventions. In addition, a more ecological perspective that takes into account diverse family structures is warranted; a close look at the sample shows that 10% of caregivers were “extended family” members, 3.5% were “siblings,” and more than 13% were single parents. These types of caregivers may have experienced loss

364

T.S. Betancourt et al. / Journal of Adolescent Health 51 (2012) 357–365

or separation from children, parents, or spouses during displacement, thus putting them at risk for higher levels of mental health problems and contributing to poorer outcomes in the children they care for [39]. High levels of mental health problems among caregivers can also impede children’s access to other support services and protective resources, as emotional distress and despair may cloud a caregiver’s ability to navigate the few resources available in the camp setting and may impair his or her ability to provide a “protective shield” for children [1]. To promote better mental health outcomes in children and adolescents living in refugee camps, we need to learn more about holistic family- and community-based approaches for promoting mental health in populations exposed to war. For example, programs that incorporate strategies to promote positive caregiver mental health, bolster caregiver– child relationships, and improve access to services deserve increased attention. Several study limitations must be noted. In the present study, caregiver mental health was only assessed at T2; thus, the directionality of relationships between caregiver distress and adolescent outcomes is difficult to determine. Although the results presented here suggest that family-centered approaches may have tremendous impact at multiple ecological levels, our assessments were not able to capture the full scope of potentially important risk and protective processes among refugee adolescents and caregivers. For instance, additional information on caregiver monitoring of adolescents in the camp would have broadened our ability to understand sources of support and informal social control on life in the camps. Such constructs remain important for consideration by future research. In this population and more broadly, the application of ethnographic and qualitative methods to explore mental health, healing, and trauma from the “ground up” has the potential to inform measures selection and adaptation to diverse cultural contexts [40]. In future research, multiple points of assessment of caregiver and adolescent mental health as well as risk and protective factors may contribute to a more robust understanding of the long-term impact of interventions in refugee populations and trajectories of mental health adjustment. Finally, although our analyses found no statistically significant differences by caregiver type, analyses were limited by sample size. Future research that is powered to examine potential differences by type of caregiver would be useful. Overall, by strengthening the local relevance and cultural validity of assessments through the integration of qualitative data, promoting prospective and longitudinal study designs, and retaining a focus on addressing children and adolescents in the context of their communities and families, this and future research can help to ensure more relevant and sustainable supports to the thousands of war-affected children and families living in refugee camps today.

Acknowledgments The authors thank Asfaw Yitna (United Nations Economic Commission for Africa) for his assistance in data management and analysis and Marie de la Soudiere (UNICEF) for her unswerving dedication to applied research in war-affected settings. This study was funded by the International Rescue Committee. This

work was also supported by the FranÈois-Xavier Bagnoud Center for Health and Human Rights at Harvard University, Boston, MA. References [1] Beardslee WR, Versage EM, Gladstone TR. Children of affectively ill parents: A review of the past 10 years. J Am Acad Child Adolesc Psychiatry 1998;37: 1134 – 41. [2] 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. [3] Sullivan PF, Neale MC, Kendler KS. Genetic epidemiology of major depression: Review and meta-analysis. Am J Psychiatry 2000;157:1552– 62. [4] Overstreet S, Dempsey M, Graham D, Moely B. Availability of family support as a moderator of exposure to community violence. J Clin Child Psychol 1999;28:151–9. [5] Roberts RE, Bengtson VL. Relationships with parents, self-esteem, and psychological well-being in young adulthood. Soc Psychol Q 1993;56:263–77. [6] Kliewer W, Murrelle L, Mejia R, et al. Exposure to violence against a family member and internalizing symptoms in Colombian adolescents: The protective effects of family support. J Consult Clin Psychol 2001;69:971– 82. [7] Farhood LF. Testing a model of family stress and coping based on war and non-war stressors, family resources and coping among Lebanese families. Arch Psychiatr Nurs 1999;13:192–203. [8] Bronfenbrenner U. The ecology of human development: Experiments by nature and design. Cambridge, MA: Harvard University Press, 1979. [9] Earls F, Carlson M. The social ecology of child health and well-being. Annu Rev Public Health 2001;22:143– 66. [10] Elbedour S, Bensel R, Bastien DT. Ecological integrated model of children of war: Individual and social psychology. Child Abus Negl 1993;17:805–19. [11] Dybdahl R. Children and mothers in war: An outcome study of a psychosocial intervention program. Child Dev 2001;72:1214 –30. [12] Catani C, Jacob N, Schauer E, et al. Family violence, war, and natural disasters: A study of the effect of extreme stress on children’s mental health in Sri Lanka. BMC Psychiatry 2008;8:33. [13] Laor N, Wolmer L, Cohen DJ. Mothers’ functioning and children’s symptoms 5 years after a SCUD missile attack. Am J Psychiatry 2001;158:1020 – 6. [14] Locke CJ, Southwick K, McCloskey LA, FernÂndez-Esquer ME. The psychological and medical sequelae of war in Central American refugee mothers and children. Arch Pediatr Adolesc Med 1996;150:822– 8. [15] Mghir R, Freed W, Raskin A, Katon W. Depression and posttraumatic stress disorder among a community sample of adolescent and young adult Afghan refugees. J Nerv Ment Dis 1995;183:24 –30. [16] Miller KE. The effects of state terrorism and exile on indigenous Guatemalan refugee children: A mental health assessment and an analysis of children’s narratives. Child Dev 1996;67:89 –106. [17] Patel V, Araya R, Bolton P. Treating depression in the developing world. Trop Med Int Health 2004;9:539 – 41. [18] Sideris T. War, gender and culture: Mozambican women refugees. Soc Sci Med 2003;56:713–24. [19] Cultural Orientation Resource Center. Refugee backgrounder No. 3: The Kunama. 2007. Available at: http://www.cal.org/co/pdffiles/backgrounder_ kunama.pdf. Accessed December 13, 2011. [20] Stichick T. Stressors, risk, resilience and psychosocial adjustment among Eritrean Kunama refugee adolescents in Ethiopia: report of baseline data collection. Addis Ababa, Ethiopia: International Rescue Committee (IRC), 2001. [21] Achenbach TM. Manual for the youth self report and 1991 profile. Burlington, VT: Department of Psychiatry, University of Vermont, 1991. [22] Mollica RF, Wyshak G, De Marneffe D, et al. Hopkins symptom checklist-25 manual: Cambodian, Laotian, and Vietnamese versions. In: US Office of Refugee Resettlement edition. Washington, D.C: US Office of Refugee Resettlement, 1986. [23] Punamaki R-L, Qouta S, El-Sarraj E. Resiliency factors predicting psychological adjustment after political violence among Palestinian children. Int J Behav Dev 2001;25:256 – 67. [24] Kerestes G. Children’s aggressive and prosocial behavior in relation to war exposure: Testing the role of perceived parenting and child’s gender. Int J Behav Dev 2006;30:227–39. [25] Roberts RE, Alegria M, Roberts CR, Chen IG. Concordance of reports of mental health functioning by adolescents and their caregivers: A comparison of European, African and Latino Americans. J Nerv Ment Dis 2005;193:528 –34. [26] Weisz JR, Sigman W, Weiss B, Mosk J. Parent reports of behavioral and emotional problems among children in Kenya, Thailand, and the United States. Child Dev 1993;64:98 –109. [27] van Ommeren M, Sharma B, Thapa S, et al. Preparing instruments for transcultural research: Use of the translation monitoring form with Nepalispeaking Bhutanese refugees. Transcult Psychiatry 1999;36:285. [28] Achenbach TM. Manual for the child behavior checklist and 1991 profile. Burlington, VT: Department of Psychiatry, University of Vermont, 1991.

T.S. Betancourt et al. / Journal of Adolescent Health 51 (2012) 357–365

[29] Mulatu MS. Prevalence and risk factors of psychopathology in Ethiopian children. J Am Acad Child Adolesc Psychiatry 1995;34:100 –9. [30] Achenbach TM, Rescorla LA. Multicultural understanding of child and adolescent psychopathology: Implications for mental health assessment. New York, NY: Guilford Publications, 2006. [31] Zimet GD, Dahlem NW, Zimet SG, Farley GK. The multidimensional scale of perceived social support. J Pers Assess 1988;52:30 – 41. [32] Derogatis LR, Lipman RS, Rickels K, et al. The Hopkins symptom checklist (HSCL): A self-report symptom inventory. Behav Sci 1974;19:1–15. [33] Onyut LP, Neuner F, Ertl V, et al. Trauma, poverty and mental health among Somali and Rwandese refugees living in an African refugee settlement - An epidemiological study. Confl Health 2009;3:6. [34] Kaaya SF, Fawzi MC, Mbwambo JK, et al. Validity of the Hopkins symptom checklist-25 amongst HIV-positive pregnant women in Tanzania. Acta Psychiatr Scand 2002;106:9 –19. [35] Bolton P. Cross-cultural validity and reliability testing of a standard psychiatric assessment instrument without a gold standard. J Nerv Ment Dis 2001;189:238 – 42.

365

[36] Ertl V, Pfeiffer A, Saile R, et al. Validation of a mental health assessment in an African conflict population. Psychol Assess 2010;22:318 –24. [37] Bolton P, Neugebauer R, Ndogoni L. Prevalence of depression in rural Rwanda based on symptom and functional criteria. J Nerv Ment Dis 2002;190:631–7. [38] Burlingham D, Freud A. Children and war. New York, NY: Ernst Willard, 1943. [39] Karen A, Joy DO. Parenting after trauma: Supporting parents and caregivers in the treatment of children impacted by violence. Infant Ment Health J 2003;24:111–25. [40] Betancourt TS, Meyers-Ohki SE, Stulac S, et al. Using mixed methods to plan, assess, and evaluate international mental health programs for children and families facing adversity: The development of the Family Strengthening Intervention (FSI) in Rwanda. In: Corbin JN, ed. Children and Families Affected by Armed Conflicts in Africa: Implications for Helping Professionals in the United States. Washington, DC: National Association of Social Workers Press (in press).