This study utilized baseline data from the Suubi4Her study, a longitudinal randomized clinical trial (2017-2022), funded by the National Institute of Mental Health (Grant # R01MH113486). A total of 1260 adolescent girls between 14–17 years of age at study initiation, were enrolled in the study. Adolescents were eligible to participate if they met the following inclusion criteria: 1) female, 2) age 14–17 years, 3) enrolled in the first or second year of secondary school, and 4) living within a family (broadly defined and not an institution or orphanage, as those in institutions have different familial needs).
Adolescents were identified and recruited from 47 secondary schools in five geopolitical districts of Rakai, Kyotera, Masaka, Lwengo and Kalungu, in southern Uganda. The schools included in the study were matched on the following characteristics: socioeconomic status of the students attending these schools, school size (total number of students enrolled), location (urban vs. rural), and overall performance based on the Uganda Certificate of Education (UCE) examinations, administered by the Uganda Government’s Ministry of Education and Sports. Potential participants and their parents/caregivers were identified with the help of school administrations. Parents/caregivers were given flyers notifying them of the study and inviting them to contact the school headteacher for further details. In addition, community development officers and implementing partners distributed flyers during their frequent community visits to inform parents/caregivers whose children met the inclusion criteria but may not yet have reported to school. Caregivers and adolescents who expressed interest were later invited to meet with the in-country project coordinator for a one-on-one informational meeting. During the meeting, parents/caregivers and adolescents were informed verbally and in writing, the purpose of the study, voluntary participation, extent of their participation, risk and benefits, as well as protection and confidentiality issues. Interested caregivers/parents signed the informed consent and adolescent girls signed the assent forms. Detailed information on participants recruitment and selection process, power analysis, as well as the intervention is described in the study protocol and in our other publications [75,76].
Data were collected using a 90-minute interviewer administered survey. Survey instruments were translated into Luganda – the most widely spoken language in the study region – and back translated into English to ensure accuracy. This process was overseen by certified language experts at the Makerere University in Uganda. Each interviewer received Good Clinical Practice training and obtained the Collaborative Institutional Training Initiative (CITI) Certificate before interacting with study participants.
All measures utilized in this study were tested in our previous studies in Uganda among children and adolescents affected by HIV/AIDS in the study area [ 62-65, 77, 78].
Depressive symptoms were measured by the Beck’s Depression Inventory (BDI) . The scale measures characteristic attitudes and symptoms of depression including mood, pessimism, and sense of failure, self-dissatisfaction, guilt, punishment, self-dislike, self-accusation, suicidal ideas, crying, irritability, social withdrawal, indecisiveness, body image change, work difficulty, insomnia, fatigability, loss of appetite, weight loss, somatic preoccupation, and loss of libido. The scale consists of 21 sets of statements, ranked based on severity on a 4-point continuum (0=least, 3=most). The theoretical range for the BDI is 0-63, with higher scores indicating higher levels of depressive symptoms. The scale demonstrated a high internal consistency (Cronbach’s alpha =0.83).
Measures of Psychological Wellbeing
These were assessed by participants’ self-concept, self-esteem and hopelessness. Self-concept was measured using the Tennessee Self-Concept Scale . The 20-item scale measures children’s perception of identity and self-satisfaction. Each of the 20 items was rated on a 5-point scale, with 1= always false and 5= always true. The theoretical range for TSCS is 20-100, with higher scores indicating higher levels of child self-concept. A high internal consistency (Cronbach alpha =0.83) was reported for this scale. Participants’ self-esteem was measured using the Rosenberg Self-Esteem Scale . The 10-item scale measures individual self-esteem on a 4-point Likert scale, with 1=strongly disagree to 4=strongly agree. The theoretical range for the RSES is 10-40, with high scores indicating high levels of self-esteem. The scale demonstrated a moderate internal consistency (Cronbach's alpha = 0.66). Hopelessness was measured using the Beck Hopelessness Scale . The 20-item scale measures children’s hopelessness and pessimistic attitudes toward the future. Items have a “true” or “false” response rating, coded as “1” or “0” respectively. The theoretical range for the BHS is 0-20, with higher scores indicating a high level of hopelessness and pessimistic attitudes. The scale demonstrated a moderate internal consistency (Cronbach’s alpha =0.71).
Family and Social Support Factors
Family support factors were measured by three indicators: 1) family cohesion, 2) perceived child–caregiver support, all adapted from the Family Environment Scale  and the Family Assessment Measure , and 3) family care and relationships. 1) Family cohesion was measured using 7 items that assess the degree of commitment, help, and support that family members provide to one another. Respondents were asked to rate how often each item occurred in their family, on a 5-point Likert scale, with 1=never and 5=always. The theoretical range for this scale is 7-35, with high scores indicating higher levels of family cohesion (Cronbach alpha = 0.72); 2) Perceived child-caregiver support was assessed using 17 items adapted from the Social Support Behaviors Scale (SS-B) scale . Respondents were asked to rate the adults they live with, on a 5-point Likert scale, with 1= never and 5= always. The theoretical range for this scale is 17-85, with high summated scores indicating high levels of perceived support from caregivers (Cronbach alpha =0.78); 3) Family care and relationship was measured using 6 items related to things that parents sometimes do with their children. Respondents were asked to rate how often each item occurred in their family, on a 5-point Likert scale, with 1=never and 5=always. The theoretical range for this scale is 6-30, with high scores indicating higher levels of family care and relationships (Cronbach alpha = .60).
Social support from multiple sources was measured using 30 items adapted from the Friendship Qualities Scale . The scale assesses the impressions of the quality of children’s friendships and relationships with their classmates, peers, teachers and parents. Respondents were asked to rate how each statement applied to them, on a 5- point Likert scale, with 1=never and 5= always. The theoretical range for this scale is 30-150, with high scores indicating higher levels of social support and relationships (Cronbach’s alpha =0.81).
Finally, participant’s sociodemographic and household characteristics included in the analysis as control variables were: 1) participants’ age categorized into 14-15 years versus 16-17 years, 2) orphanhood status (orphan versus non-orphan), 3) primary caregiver, 4) household size (i.e. number of people in the household, total number of children in the household), and 5) household assets.
Data Analysis Procedures
Data was analyzed using Stata software SE. 12.1. We analyzed sociodemographic and household characteristics of the sample, followed by bivariate analyses of predictors of depressive symptoms (sociodemographic and household characteristics, family and social support factors and psychological wellbeing) across participants’ age groups. We estimated the chi- square or t-test values for each of the variables. To address research questions 1 and 2 (i.e. determine the prevalence of depressive symptoms and variation by age groups), BDI scores were divided into 4 categories based on the scoring guidelines in non-clinical populations , minimal (0- 9), mild (10-18) moderate (19-29) and severe (30+) symptoms. We then conducted chi square tests by age groups i.e. younger adolescents (14-15years) versus older adolescents (16-17 years). Older adolescents tend to have high levels of depressive symptoms compared to young ones [10,11]. To address research question 3, hierarchical regression models were conducted to determine the predictors of depressive symptoms. We conducted three models, with each model controlling for a block of predictors. Model 1 controlled for socio-demographics and household characteristics; model 2 controlled for family and social support factors, and model 3 controlled for psychological wellbeing. We compared the adjusted R squares to determine the strength of each model.