Participants
The examined children live in a non-governmental orphanage in the Southern Highlands of Tanzania, situated near a small village in a rural area. The orphanage consists of four houses with nine to twelve children of different ages and sexes with two caretakers for each house. The caretakers had mostly no preparatory qualification for their jobs as caretakers and only primary school education. Children were either full or partial orphans or had been severely abused or neglected by their families and were therefore taken into orphan care. Children, who were seven years or older, were interviewed for two hours on average at time point 1 (t1) and six months later at time point 2 (t2). The younger children could only answer part of the questions. Further qualitative information concerning mental ill-health, especially of the younger children, was gained through behavioral observation by the investigators who lived five weeks (during t1) and three weeks (during t2) with the children. In general, the analyses included all children (N = 38; 53% boys) who were in the orphanage during both assessment periods. The mean age was M = 8.64 years (range 3 - 16) at t1 and M = 9.16 years (range 3 - 16) at t2. The Tanzanian and German board of the organization managing the orphanage gave their consent and ethical approval.
Materials
The interview sets were basically identical for both assessments. All instruments were applied as a structured interview by clinicians with extensive working experience including an East African context. This experience and the application through an interview allowed the interviewers to complete the interview with many children of seven years or older.
Socio-demographic data: The first part of the interview consisted of socio-demographic information, in which the children were also asked about their parents, the reason for death of the parents and about relationship to relatives.
Physical health: The children were interviewed about their physical health in the past four weeks based on a checklist (concerning cough, stomach pain, tuberculosis, headache, malaria, flu, pain, diarrhoea, fever/shivering, skin rush/scabies, and vomiting) [
32].
Stressful and traumatic experiences: In the subsequent section of the interview, the children were asked about their experiences of violence. This included physical, psychological and sexual violence as well as neglect and witnessed violence. The children were asked 41 questions about violence (following C. Catani at
http://www.vivo.org). At t1 they were asked about the experienced violence at home, in school or neighborhood, and in the orphanage during their whole lifetime. At t2 they were only interviewed about experienced violence in neighborhood or school and the orphanage in the last six months.
Mental health: Concerning the mental health of the children, internalizing and externalizing problems, PTSD, and depression were assessed.
Internalizing and externalizing problems: The self-evaluation of strengths and difficulties was assessed with the Strengths and Difficulties Questionnaire (SDQ) [
33]. The SDQ comes with good psychometric properties and is internationally implemented [
34]. This study uses the self-report version for children from 11 to 17 years. It consists of 25 statements with the possible responses that the statement is
not true, somewhat true or
certainly true for themselves. Each of the five subscales (conduct problems, hyperactivity, emotional symptoms, peer problems and prosocial behavior) consists of five items. The total difficulties score is generated by summing the scores of all items, except the items for prosocial behavior, and ranges from 0 to 40. A score over 20 indicates an abnormal amount of internalizing and externalizing problems. The total difficulties score is a good measure for a general impression of internalizing and externalizing problems and is, therefore, a sufficient measure for this study.
Post-traumatic stress disorder: The UCLA PTSD Index for Children DSM IV [
35] was used to screen for exposure to traumatic events and for symptoms of PTSD. This instrument was originally constructed as a self-report and assesses the severity of symptoms based on the frequency of symptoms reported by the child. The occurrence of each DSM-IV symptom within the last month is scored on a scale ranging from
none of the time to
most of the time. Thus, an overall PTSD severity score can be calculated by summing the scores for each question, which results in a maximum possible score of 68. The UCLA PTSD Index shows good psychometric properties and has been successfully utilized in non-western settings [
21,
23].
Depression and suicidality: Depression and suicidality were assessed with the Mini-International Neuropsychiatric Interview kid for children and adolescents (M.I.N.I.; Section A and C) [
36]. Additionally, the severity of depressive symptoms was assessed by means of the Children's Depression Inventory (CDI) [
37]. The CDI is a reliable and well-tested clinical research instrument designed for school-aged children and adolescents. It has been successfully implemented in Tanzanian settings [
3,
38]. Originally it is administered as a self-report instrument and evaluates the severity of specific depressive symptoms. It contains 27 items with three statements each and the child has to choose which statement fits best. For each item, the points range from 0 to 2, where higher values represent more clinically severe symptoms. Thus, the possible maximum score is 54.
Aggression: Aggressive behavior was assessed at t2 with the Reactive-Proactive Questionnaire [
39]. The children were asked how often they have exhibited a specific aggressive behavior, in which they have to choose between
never, sometimes and
often. One item of originally 23 items was removed, because it was not appropriate for the conditions in rural Tanzania (Item 18:
Made obscene phone calls for fun) and two items were slightly rephrased for a better understanding (Item 4:
students replaced with
children and Item 9:
gang fight replaced with
fight). The sum of the points assigned to the answer represents the total aggression and ranges from 0 to 44.
Procedure
The first assessment in March 2010 was carried out by four of the authors. They worked together with trained translators and stayed for five weeks in the orphanage. The second assessment was carried out in September 2010, six month after the first assessment, by the two other authors (KH and TH) again with trained, but now different translators. This second team of interviewers was blind with respect to any information gathered during the first assessment and did not know who had received psychotherapeutic assistance. The second assessment was completed after three weeks. The translators were trained before both assessments and the interviewers had standardized the form of assessment by practicing in joint interviews to achieve a high inter-rater reliability. All instruments were translated word-by-word into Kiswahili and the translation was intensely discussed to guarantee a precise translation.
Every child of seven years or older was interviewed alone in a quiet place by one interviewer and one translator. To provide a trustworthy environment, the girls were interviewed by at least one woman. The interview took two hours on average. Children were assured that the whole interview was confidential and that there would be no punishment for whatever information was given. The amount of breaks varied with the child's ability to concentrate. Children received drinking water and a fixed number of sweets during the interview to help them to stay focused. Children were encouraged to draw a picture or to play their favorite game at the end of the interview. In addition, the behavior of all children was observed in their typical daily surrounding. During the periods of assessment, interviewers and translators stayed in the orphanage and shared the meals with the children and played with them in their free time.
Analyses
All variables except one met the preconditions for the analyses. The sum of depressive symptoms at t1 was not distributed normally. Therefore, the Spearman coefficient was computed for correlations using the sum of depressive symptoms at t1. The Pearson coefficient was calculated for all other correlations. The Bonferroni correction was used in cases of multiple testing to prevent alpha-inflation. All hypotheses about mental health were subdivided in specific hypotheses for PTSD, depression, and internalizing and externalizing problems. Due to the directional hypotheses, analyses were computed one-tailed. According to the age of the children, n = 22 children could be included in the analyses of the severity of PTSD symptoms, whereas n = 33 children were included concerning the severity of depressive symptoms and internalizing and externalizing problems. The analysis of the relation between experienced violence in the orphanage and aggression included n = 29 children.