Differential roles of childhood adversities and stressful war experiences in the development of mental health symptoms in post-war adolescents in northern Uganda

As previously reported elsewhere [25],[26], the study was conducted in Gulu district, northern Uganda, which until mid-2006 remained the epicentre of over 20 years of armed conflict between the Ugandan government and the rebel group the Lord’s Resistance Army (LRA). Data are derived from self-report questionnaires completed by 551 school-going adolescents in August-September 2010. Out of the 14 secondary schools in Gulu district, we purposively selected and contacted seven government and private-operated schools, with additional stratification by single (2 government, 2 private)-/mixed (2 government, 1 private)-sex schools. The study, its recruitment and consent/assent procedures were approved by the institutional review boards at Ghent and Gulu University respectively, and the Uganda national council of science and technology. Six hundred questionnaires were distributed post-consent. Overall, 24 adolescents declined or withdrew participation and 25 questionnaire sets with significant portions of missing responses were excluded, resulting in a response rate of 92% (551/600). Researchers remained present during the completion of the questionnaires, to address participants’ questions and to make sure respondents understood all items.

Self-report questionnaires were reviewed with local mental health staff and teachers and later pretested with non-participating adolescents to ensure local validity. All questionnaires were administered in English, the official language used in all Ugandan schools. A socio-demographic questionnaire was used to determine participants’ age, gender, ethnicity, living situation, marital status of parents and whether the adolescent’s parents were still alive or not.

Childhood adversity was measured using a modified version of the Adverse Childhood Exposure (ACE) questionnaire used by Bruffaert and colleagues [27]. Six context-specific intra-familial childhood adversities: not growing up with both parents, financial adversity (i.e. dependence on food aid for at least six months), physical abuse by parents and/or other adults, sexual abuse, and witnessing fights between the parents before the age of 13 years were assessed. Although parental psychopathology, substance use and criminality are also common elements in adverse childhood experiences, we did not include them as the local experts review team advised they would be too intrusive.

Stressful war experiences were measured using a Stressful War Events (SWE) questionnaire designed specifically for the study population [28]. The SWE consists of 17 yes/no questions (including a final question inquiring `Any other stressful war event experienced?’), each referring to a stressful war event typically encountered during active conflict in northern Uganda, such as “Did you experience death of close family or friends?” Because of our interest in the experience of exposure to war events prior to 2006, when the conflict in Northern Uganda was still intense, participants were asked to limit their responses to that specific period. Childhood adversities and stressful war experiences were modelled as continuous scores in our analyses.

PTSD symptoms were assessed with the Impact of Event Scale-Revised (IES-R) [29], a 22-item scale consisting of three subscales of intrusion (8 items), avoidance (8 items) and hyperarousal (6 items). The IES-R has been administered in a number of war-affected adolescent populations in Africa, and has demonstrated good reliability and validity [4],[30]-[32]. Respondents were asked to identify a specific event on either the ACE or SWE questionnaire as a reference point for completing the IES-R, and indicate how much it distressed them in the past month by rating each item on a 5-point scale from 0 (not at all) to 4 (extremely). Cronbach’s alpha values demonstrated good reliability (intrusion: .88, avoidance: .83, hyperarousal: .82, total IES-R score: .93).

Depression and anxiety symptoms were assessed using the Hopkins Symptom Checklist-37 for Adolescents (HSCL-37A) [33], a self-report questionnaire which had been previously used in war-affected and refugee populations [32],[34]. Thirty-seven items, using a 4-point scale from 1 (never) to 4 (always), inquire about the severity of Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) based symptoms. In this study, we use the subscales measuring symptoms of depression (15 items) and anxiety (10 items), both subscales showing good Cronbach’s alphas (respectively .83 and .78). For both IES-R and HSCL-37A, we used continuous scores for analyses to retain greater variability in the data.

Five separate hierarchical linear regression analyses for each mental health outcome were conducted. All necessary quality criteria for regression analyses were met. In the first step, demographic variables (age, gender, mother and father still alive or not, and parent marital status) were entered. In the second step, adverse childhood experiences (ACE) and stressful war events (SWE) were added, with both covariates mean centred. In the third step, the interaction term SWE x ACE was included, followed by simple slope analysis to further explore this relationship.

Detailed baseline characteristics of the study population and gender differences in childhood adversities, stressful war experiences and mental health symptoms are presented in Table 1. Tables 2 and 3 shows the nature and frequency of childhood adversities and stressful war experiences respectively. More girls than boys reported experiencing sexual violence and witnessing inter-parental violence and war-related experiences of lack food and water, getting wounded or disabled and not feeling accepted by family or community, although differences were relatively small.

Hierarchical regression models revealed a main effect of gender on all mental health outcomes except avoidance symptoms, with girls reporting higher scores than boys (Table 4). Analyses showed that stressful war experiences predicted all symptoms, whereas childhood adversity only predicted depression symptoms. Lastly, a statistically significant interaction between childhood adversity and stressful war experiences was found only for avoidance symptoms. Simple slopes for the association between avoidance and war-related traumatic experience were tested for low (-1.5 SD below the mean) and high (+1.5 SD above the mean) levels of childhood adversity (Figure 1). Each of the simple slope tests revealed a significant positive association between avoidance symptoms and stressful war experiences. However, stressful war experiences were more strongly related to avoidance symptoms at low levels of childhood adversity (b = .95, SE = .14, t(438) = 6.66, p < .001) than at high levels of childhood adversity (b = .47, SE = .14, t(438) = 3.30, p < .01).

The data were retrospective and self-reported, and consequently subject to recall and social desirability biases. Although recall bias is expected to be less extreme in a survey of adolescents than adults or younger children, recall bias is still a possibility in retrospective reports of childhood adversities and stressful war experiences. The cross-sectional design precludes both causal inference (as event reporting may be confounded by current psychological functioning and age) and the longitudinal analysis of adjustment trajectories [38]. The generalizability of our findings is limited to only school-going adolescents. Furthermore, given that active war ended in 2006, the extended time frame (four years post-war) created a situation in which post-war factors, such as post-war trauma could occur and compete for explained variance. The time frame also sampled youths 9 to 17 years old at the end of the war - a group that was possibly less severely exposed than older adolescents. This may have led to low endorsement rates for some event types (e.g., physical violence) that prevented their predictive effects from being adequately tested [39]. As a consequence, factors such as daily stressors and more recent traumatizing events, or parental and peer attachment, might have influenced mental health symptoms [37]. Fifth, our coverage of childhood adversities was not exhaustive. For purposes of this study, and to avoid participant fatigue, we focused specifically on family-related adversities, consistent with previous work [27], but numerous other adversities are associated with elevated risk for mental health symptoms. Non-violent [40] and stressful war events such deprivation of food and water and being forced to perform rituals [2] have been linked to PTSD, but they may not constitute DSM-IV qualifying trauma. Further, the psychometric properties of the Stressful War Events Questionnaire are unknown. The measure was developed with the assistance of local expert teams familiar with the types of war events children experienced in Uganda. However, it is possible that some important traumatic events (e.g., witnessing (re)burial of loved ones) were not listed on the questionnaire. Lastly, using count scores summing up conceptually diverse childhood adversities and stressful war experiences assumes an arithmetic relationship between these experiences, an assumption which may not be valid. Although this is in accordance with earlier studies [41] stating that both low- and high-threat events increased the risk for mental health symptoms, and that counts of events (without regard to type of trauma) predicted outcomes, no consensus exists yet on methods of weighting aspects of traumatic exposure [39].