Skip to main content An official website of the United States government Here's how you know Here's how you know Official websites use .gov A .gov website belongs to an official government organization in the United States. Secure .gov websites use HTTPS A lock ( Lock Locked padlock icon ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites. Search Log in Dashboard Publications Account settings Log out Search… Search NCBI Primary site navigation Search Logged in as: Dashboard Publications Account settings Log in Search PMC Full-Text Archive Search in PMC Journal List User Guide PERMALINK Copy As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: PMC Disclaimer | PMC Copyright Notice J Adolesc Health . Author manuscript; available in PMC: 2026 Apr 11. Published in final edited form as: J Adolesc Health. 2026 Jan 5;78(3):427–436. doi: 10.1016/j.jadohealth.2025.11.010 Search in PMC Search in PubMed View in NLM Catalog Add to search Exploring Mediators of an Economic Empowerment Intervention (Suubi4Her) on Hopelessness and Depressive Symptoms Among Adolescent Girls in Uganda Rachel Brathwaite Rachel Brathwaite , Ph.D. a International Center for Child Health and Development, Brown School, Washington University in St. Louis, Missouri Find articles by Rachel Brathwaite a , Samuel Kizito Samuel Kizito , M.D., Ph.D. b Silver School of Social Work, New York University, New York, New York Find articles by Samuel Kizito b , Torsten B Neilands Torsten B Neilands , Ph.D. c Division of Prevention Science, University of California, San Francisco, California Find articles by Torsten B Neilands c , Vicent Ssentumbwe Vicent Ssentumbwe , M.P.H. a International Center for Child Health and Development, Brown School, Washington University in St. Louis, Missouri Find articles by Vicent Ssentumbwe a , Flavia Namuwonge Flavia Namuwonge , M.B.A. a International Center for Child Health and Development, Brown School, Washington University in St. Louis, Missouri Find articles by Flavia Namuwonge a , Phionah Namatovu Phionah Namatovu , M.P.H. d International Center for Child Health and Development, Masaka, Uganda Find articles by Phionah Namatovu d , Proscovia Nabunya Proscovia Nabunya , Ph.D. a International Center for Child Health and Development, Brown School, Washington University in St. Louis, Missouri Find articles by Proscovia Nabunya a , Fred M Ssewamala Fred M Ssewamala , Ph.D. b Silver School of Social Work, New York University, New York, New York Find articles by Fred M Ssewamala b, * Author information Article notes Copyright and License information a International Center for Child Health and Development, Brown School, Washington University in St. Louis, Missouri b Silver School of Social Work, New York University, New York, New York c Division of Prevention Science, University of California, San Francisco, California d International Center for Child Health and Development, Masaka, Uganda * Address correspondence to: Fred M. Ssewamala, Ph.D., Silver School of Social Work, New York University, New York, NY 10003. [email protected] (F.M. Ssewamala). Issue date 2026 Mar. PMC Copyright notice PMCID: PMC13068019 NIHMSID: NIHMS2159173 PMID: 41489586 The publisher's version of this article is available at J Adolesc Health Abstract Purpose: A community-based economic empowerment and family strengthening intervention (Suubi4Her) has shown positive effects in decreasing depressive symptoms and hopelessness among adolescent girls. However, the specific mechanisms underlying these positive outcomes were unclear. We examined the intervention effects of Suubi4Her on key mediators (including family cohesion, social support, and confidence in saving) and their subsequent influence on two critical psychological outcomes—depressive symptoms and hopelessness—over 24 months. Methods: We analyzed data from 1,260 Ugandan adolescent girls and employed the Mitchell and Maxwell’s cross-lagged auto-regressive approach for mediation assessment. This involved regressing outcome scores at each wave against intervention group assignment and scores from the previous wave. Correlations between residuals for mediators and outcomes at each wave as well as regression pathways from each covariate to the mediators and outcomes at subsequent waves were included in the model as random variables. Results: The intervention significantly reduced depressive symptoms and hopelessness indirectly. Crucially, family cohesion and social support, but not confidence in saving at 12 months, were identified as significant mediators. Depressive symptoms and hopelessness at 12-month follow-up mediated the intervention’s effect on depressive symptoms and hopelessness at 24 months. Additionally, a significant direct effect of the intervention on these outcomes suggested partial mediation. Discussion: Our findings endorse community-based family group interventions as effective in enhancing mental health among adolescent girls, emphasizing the importance of strong family relationships and social support networks. Keywords: Poverty, Low-resource setting, Mediation analysis, Adolescent girls and young women, Sub-Saharan Africa In many low- and middle-income countries (LMICs) where greater than 90% of the world’s adolescent population reside [ 1 ], mental health is recognized as a significant public health concern, accounting for a sizeable proportion of overall disease burden [ 2 ]. Among adolescents and youths residing in LMICs, girls are disproportionately affected by mental health problems [ 1 , 3 , 4 ]. However, due to limited financial resources, many LMICs are left with inadequate resources to address the mental health burden [ 5 ]. In Uganda—the focus country of this paper—for example, less than one percent of the health-care budget is dedicated to mental health care [ 6 ]. As a result, a vast majority of the population lacks adequate access to mental health care and treatment. To bridge this treatment gap, several LMICs have engaged in task-shifting approaches in an attempt to improve and ensure early access to much-needed preventive interventions [ 7 - 9 ]. Moreover, considering the numerous poverty-related challenges and stressors that contribute to poor mental health in resource-limited environments, combined approaches like economic empowerment and family strengthening have been highlighted as potential strategies that can enhance mental health wellbeing [ 9 ]. The same strategies have been found to decrease risky behaviors among young people in these settings [ 9 ]. In the resource-limited region of southern Uganda, researchers have culturally adapted and tested community-based economic empowerment and family strengthening interventions in the Suubi4Her study [ 10 ]. Suubi4Her combines economic empowerment and family strengthening interventions to tackle the various poverty-driven challenges and stressors impacting mental health of vulnerable girls, especially in resource-limited environments. The economic empowerment intervention is in the form of a one-to-one matched savings youth development account (YDA) opened in the name of the adolescent for the purpose of paying for school fees and investing in a family-based income-generating activity (IGA) [ 10 ]. The family strengthening component was delivered using multiple family group (MFG) sessions designed to strengthen family relationships and build social support in a community setting [ 10 ]. Favorably, we have observed the beneficial impact of the intervention on mental wellbeing of adolescent girls in the intervention groups compared to the control group. In particular, after a 24-month follow-up period, adolescent girls in the combined YDA + MFG group displayed fewer feelings of hopelessness than adolescent girls in the control group [ 11 ], and girls in both YDA and YDA + MFG groups had lower levels of depressive symptoms than their counterparts in the control group [ 11 ]. Given that these interventions were effective yet consisted of varying components, it is crucial to determine the specific mechanisms that drove these positive outcomes. By pinpointing precisely what precisely has caused the observed effects, we can gain a clearer understanding of how and why the intervention works, providing opportunities to refine, modify, or eliminate certain components for a more effective intervention. Several potential mediators can explain the positive impact of the MFG and YDA intervention on the mental wellbeing of adolescent girls. Social support theory posits that the presence of strong social support systems, such as cohesive families, and supportive peer groups can have a buffering effect on mental health outcomes [ 12 ]. While family cohesion reflects the emotional bonds, shared activities, and mutual support within the family [ 13 ], social support, on the other hand, involves broader relationships beyond the family—such as peers, teachers, and the broader community. Both mechanisms improve mental health, but via distinct pathways. Family cohesion mitigates internalized distress through secure attachment [ 13 ], whereas social support builds resilience through collective problem-solving and resource sharing [ 14 ]. In the context of Suubi4Her, the family strengthening intervention delivered via multiple family groups [ 15 ], as well as the financial literacy group training sessions, a component of economic empowerment, may likely enhance family cohesion and improve social support networks, which can act as protective factors against depressive symptoms and hopelessness. Given the group delivery format and the principles upon which the interventions were based—namely to strengthen families and improve social support—we hypothesize that family cohesion and social support would mediate the relationship between the intervention and depressive symptoms and feelings of hopelessness in adolescent girls. With respect to the economic empowerment intervention, according to the asset theory [ 16 , 17 ], the accumulation of assets, including financial resources, can enhance one’s self of security and self-worth, ultimately influencing mental health outcomes. The economic empowerment intervention in Suubi4Her focused on improving girls’ financial literacy and ability to save. This could potentially boost girls’ confidence in saving and managing their financial futures, which may in turn lead to reduced symptoms of depression and reduced feelings of hopelessness. In other words, confidence in saving acts as a mediator by building a sense of control over one’s future, which in turn reduces hopelessness and depressive symptoms (Sherraden, 1991) [ 17 ]. Indeed, empirical studies suggest that financial self-efficacy acts as a buffer against stress [ 18 ]. Moreover, recent evidence shows financial knowledge exerts its influence in reducing financial stress through the mediating effect of confidence in saving [ 19 ]. Given that the YDA intervention focused on building participants’ efficacy in saving for the future, we hypothesize that confidence to save would mediate the relationship between the intervention and depressive symptoms and feelings of hopelessness among participants. Methods Study setting The Suubi4Her study was a 5-year (2017–2022) three-arm cluster randomized control trial ( ClinicalTrials.gov Identifier: NCT03307226 ). The primary aim was to reduce HIV risk among adolescent girls and young women in Uganda. The study was conducted in 47 secondary schools across five geopolitical districts (Masaka, Kyotera, Lwengo, Kalungu, and Rakai) in the greater Masaka region of Southwestern Uganda. Only secondary schools with having students in their first year of secondary school were included. Details are elaborated in the Suubi4Her study protocol publication [ 10 ]. Recruitment, study population, and retention Potentially eligible participants were identified in collaboration with the school administration and the Masaka Diocese. To be eligible for participation in the study, adolescents had to meet specific criteria, which included being [ 1 ] female, [ 2 ] aged 14–17 years, and [ 3 ] in their first or second year of secondary school (an equivalent of high school in the United States of America education system) [ 10 ]. Importantly, individuals residing in institutions or orphanages were considered ineligible, as the intervention focused on family-based components. From a list of 111 potential secondary schools, 47 schools that met the inclusion criteria were selected. In the study, 1,260 adolescent girls from selected schools were recruited and randomized at the school level into three groups: control (408 girls from 16 schools), YDA (471 girls from 16 schools), and YDA + MFG (381 girls from 15 schools). Baseline interviews were conducted prior to the interventions, with follow-ups at 12 and 24 months. The study had high retention rates, approximately 96% (n = 1,219) at 12 months and 92% (n = 1,165) at 24 months ( Figure 1 for Consort diagram). Figure 1. Open in a new tab Consort flow diagram for the Suubi4Her study. Data collection procedures Participants underwent a 90-minute interview at their homes or schools, conducted by trained assistants trained in good clinical practice, data collection techniques, and human subjects protection. Surveys, translated into the local language, and Luganda, back-translated for accuracy, were overseen by language experts from Makerere University, Uganda. Local research assistants fluent in English and Luganda collected the data using the interviewer-administered surveys. Study arms The control arm. Participants in the control arm received the usual health education offered to adolescents in the region. The sexual health education included information from the Adolescent Sexual and Reproductive Health curriculum, a sex and health education handbook, as well as information on gender equality. Youth development account (YDA) intervention arm. Participants in the YDA arm, in addition to control arm components, opened incentivized Youth Development Accounts (YDAs) at a bank, cosigned with a primary caregiver. They were encouraged to save, with their savings matched 1:1 for school fees or family income activities, such as farming or sewing. Participants and caregivers also attended four financial literacy sessions, led by trained assistants, and received homework handouts. Youth development accounts and multiple family group (YDA + MFG) intervention arm. The YDA + MFG arm included all components of the control arm and YDA intervention, plus an additional component aimed at strengthening family relationships and supporting the psychosocial wellbeing of adolescents. The MFG component provided a safe setting for multiple families to gather and directly discuss family challenges, shared experiences, adolescent mental health challenges, and potential strategies for mitigating these challenges. The MFG is a manualized 16-week intervention facilitated by trained parent peers and Community Health Workers (CHWs). It is structured around the principles of the "Four Rs" (Rules, Responsibility, Relationships, and Respectful Communication) and the "2 Ss" (Stress and Social Support), targeting essential family skills and processes [ 20 ]. The MFG intervention aimed to reduce stigma by normalizing the sharing of experiences. It acknowledged poverty as a potential stressor affecting parenting and tackled contextual issues like elevated poverty levels, instances of violence, and family losses due to health risks, notably HIV [ 21 ]. Ethics Participation was voluntary, ensuring that caregivers and adolescent girls had the choice to participate or withdraw at any time. Written consent was obtained from caregivers who were willing and interested in the study, while written assent was obtained from the adolescent girls once consent from their caregivers was received. The consent and assent activities were conducted separately to avoid potential coercion. All procedures in the study were approved by the Institutional Review Board at Washington University in St. Louis (IRB- #201703102), the Uganda Virus Research Institute (GC/127/17/07/619), and the Uganda National Council of Science and Technology (SS4406). Outcome measures Depression was measured using Beck’s Depression Inventory [ 22 ], which has been validated in a similar setting in neighboring Kenya [ 23 ]. The scale measures characteristic attitudes and symptoms of depression, including pessimism, self-dissatisfaction, guilt, self-dislike, self-accusation, suicidal ideas, crying, irritability, social withdrawal, indecisiveness, body image change, work difficulty, and insomnia. The scale consists of 21 sets of statements, ranked based on severity on a four-point continuum (0 = least, 3 = most). The theoretical range for the Beck’s Depression Inventory is 0–63, with higher scores indicating higher levels of depressive symptoms (Cronbach’s alpha = 0.83). Hopelessness was measured using the Beck hopelessness scale, a 20-item inventory that assesses a subject’s motivation and expectations about the future [ 24 ]. The scale has been validated for use in the East-African context [ 25 ] and has been widely used in previous studies [ 11 , 26 ]. Sample items include: “All I can see ahead is unpleasant rather that pleasant” and “I can’t imagine what my life will be like in 10 years” (1 = true vs. 0 = False). Items in the opposite were reverse-coded, and the theoretical scores range from 0 to 20 (Cronbach’s alpha = 0.71), with higher scores indicating greater hopelessness. Mediators. The mediators were carefully selected, informed by the social support theory and asset theory, and findings from our prior work, which showed the positive effect of economic strengthening on improving family dynamics [ 27 ]. While the YDA alone may not directly target family cohesion or social support, its financial empowerment component (e.g., caregiver involvement in savings accounts, financial literacy training) indirectly strengthens family dynamics by reducing economic strain—a known stressor for relational wellbeing. Family cohesion was measured using a seven-item scale that assessed family members’ commitment and support for each other. Sample scale items include: “Do your family members ask each other for help before asking nonfamily members for help?”, “We do things together as a family?” [ 21 ]. The scale items were rated on a Likert scale from 1 = never to 5 = always. The theoretical range was seven–35, with higher scores indicating greater levels of family cohesion (Cronbach’s alpha = 0.72). Social support from multiple sources, i.e., caregivers, teachers, classmates, and peers, was measured using a 30-item scale adapted from the Friendship Qualities Scale [ 28 ]. The items assess the impressions and the quality of adolescents’ friendships and relationships. Sample items included: “Some youth have parent(s) or guardian(s) who care about their feelings and “Some youth have a close friend who they can tell problems to. How often does this apply to you?” The scale items were rated on a Likert scale from 1 = never to 5 = always with a theoretical range of 30–150. Items in the opposite direction were reverse-coded, and total scores were computed for higher scores to indicate greater levels of social support (Cronbach’s alpha = 0.80). Confidence in saving money. Saving confidence was assessed using five questions that assessed participants’ confidence in saving money. Sample items include: “Save money for a family business”. Scale Items were rated on a five-point Likert scale from 1(not confident at all) to 5 (extremely confident), with a theoretical range of 5–25 and higher scores indicating greater confidence in saving (Cronbach’s alpha = 0.68). Independent variable: The primary independent variable was participation in the intervention. Participants in the control group were coded as 0. Participants in both YDA and YDA + MFG were grouped in one category coded as one since previous analyses discovered no significant differences in depressive symptoms and hopelessness between the two intervention groups [ 11 , 29 ]. Statistical analysis Data management and descriptive statistics were performed using Stata version 17.0 [ 30 ]. We investigated mediation using the Mitchell and Maxwell cross-lagged autoregressive approach [ 31 ], as illustrated in Figure 2A and B . All models were fitted using M plus 8.10, and estimation was performed using maximum likelihood with cluster-adjusted robust standard errors (M plus estimator MLR). Mediators were included at all time points, and their autoregressive associations were examined in an attempt to reduce bias. Separate models were fitted for each outcome and mediator. In each mediation model, we regressed the outcomes (depressive symptoms or hopelessness) at each time point onto the intervention group variable (0 = control; 1 = intervention) and onto the outcomes (depressive symptom or hopelessness score) at the preceding time point. Similarly, we regressed each mediator onto the intervention group variable and the mediator score at the preceding time point. All residuals for mediators and the outcomes (depressive symptoms and hopelessness) were allowed to correlate at each time point. Regression pathways were included from the covariate age to mediators and outcomes at subsequent waves. For example, age at time one predicted the mediator and outcome at times two and 3. Age was specified as a random variable to ensure that participants with partial data were still included in the analysis. Full-information maximum likelihood (FIML) estimation enabled that the full sample was included in the analysis. Age was allowed to correlate with the intervention group variable and the baseline mediator and outcome variables. Following Mitchell and Maxwell’s parameterization recommendations, the regression coefficients, residual variances, and correlations were set to be equal across time. In all the analyses, we combined both intervention groups given the comparability of both YDA and YDA + MFG interventions on mental health outcomes [ 29 , 32 ] to simplify the cross-lagged Structural Equation Modeling analysis for parsimony. Figure 2. Open in a new tab (A) and (B): Mitchell and Maxwell cross-lagged Structural Equation Models for the mediators of the effect of the Suubi4Her intervention on depression and hopelessness, respectively. Global model fit was evaluated using Hu and Bentler’s cutoff criteria, which requires meeting at least two out of three fit indices for satisfactory global model fit: (1) Comparative Fit Index≥ 0.95; (2); Root Mean Square Error of Approximation ≤ 0.06 or; (3) Standardized Root Mean Square Residual ≤ 0.08 [ 33 ]. For all unmodified models, we first examined the fit indices and reviewed the modification indices for improvement in model fit via reduction in chi-square model fit if certain parameter constraints were released. We then added the following modification pathways connecting the mediator at time one to the mediator at time three and the outcome at time one to the outcome at time 3. We then estimated the direct effect of the intervention on depressive symptoms and feelings of hopelessness and the total and specific indirect effects of the intervention on depressive symptoms and feelings of hopelessness via each mediator. For each effect, we report the unstandardized estimate (B), the 95% confidence interval (CI) of B, and the corresponding standardized coefficient (β). The 95% confidence intervals were estimated using 5,000 bias-corrected bootstrap resamples [ 34 ]. Results At baseline, we enrolled 1,260 school-going adolescent girls, with a mean age of 15.4 years ( Table 1 ). On average, the girls resided in households with an average of seven occupants. In terms of family structure, the majority of the girls were non-orphans (82.9%) ( Table 1 ). Table 1. Baseline characteristics of study population: Suubi4her study, 2017–2022 Characteristics Total (n = 1,260) mean (SD) or n (%) Usual care (n = 408) mean (SD) or n (%) YDA and YDA + MFG (n = 852), mean (SD) or n (%) Mean age in years Orphanhood status a 15.4 ± 0.9 15.2 ± 0.9 15.5 ± 0.9 Double orphan 24 (1.9) 7 (1.7) 17 (2.0) Single orphan 191 (15.2) 59 (14.5) 132 (15.5) Nonorphan 1045 (82.9) 342 (83.8) 703 (82.5) Mean household size 7.0 ± 2.7 6.8 ± 2.6 7.09 ± 2.8 Open in a new tab This column contains participants from the two treatment groups (YDA and YDA + MFG). a A single orphan lost one parent while a double orphan is one who lost both parents. In this paper, we investigated whether: (1) family cohesion, social support, and confidence in saving mediated the impact of the Suubi4Her intervention on depressive symptoms at 24 months and (2) whether family cohesion, social support, and confidence in saving mediated the impact of the intervention on feelings of hopelessness at 24 months postintervention initiation among adolescent girls in Uganda. We hypothesized that the Suubi4Her intervention would lead to an increase in family cohesion, social support, and confidence in saving and thus reduce depressive symptoms and feelings of hopelessness among adolescent girls. We summarized the outcomes by study group and time point using means and standard deviations in Table 2 . Table 2. Summary of mean (SD) scores for the outcome and mediators by study group and timepoint Variable Timepoint Total Usual care Suubi4Her intervention N Mean SD N Mean SD N Mean SD Outcome 1: Depressive symptoms Baseline 1260 18.47 10.19 408 19.18 10.29 852 18.13 10.13 12 months 1219 14.93 9.38 396 16.62 9.98 823 14.11 8.98 24 months 1165 13.47 8.93 380 14.77 9.33 785 12.84 8.67 Outcome 2: Hopelessness Baseline 1260 4.20 2.97 408 4.11 2.93 852 4.25 3.00 12 months 1219 3.38 2.71 396 3.64 2.92 823 3.25 2.60 24 months 1165 3.48 2.89 380 3.77 3.10 785 3.33 2.77 Mediator 1: Family cohesion Baseline 1260 26.58 5.70 408 26.76 5.74 852 26.49 5.68 12 months 1219 26.96 5.54 396 26.54 5.59 823 27.17 5.51 24 months 1165 26.51 5.30 380 25.77 5.45 785 26.87 5.19 Mediator 2: Social Support Baseline 1260 115.92 14.81 408 115.51 14.73 852 116.12 14.85 12 months 1122 118.46 15.42 379 116.34 15.67 743 119.53 15.19 24 months 1027 122.07 14.34 347 120.36 14.07 680 122.94 14.42 Mediator 3: Confidence in saving Baseline 1260 19.88 3.89 408 19.58 4.03 852 20.02 3.81 12 months 1219 19.54 4.05 396 19.08 4.05 823 19.76 4.04 24 months 1165 19.95 3.85 380 19.41 4.15 785 20.21 3.68 Open in a new tab For social support, the number is less than the interviewed participants because cases for out-of-school participants were automatically dropped (questions related to classmates were not applicable to these participants). According to Hu and Bentler, all models had satisfactory global model fit since it met at least two of the following cutoff criteria: Comparative Fit Index ≥0.95, Root Mean Square Error of Approximation ≤0.06, and Standardized Root Mean Square Residual ≤0.08 ( Table 3 ). Table 3. Global model fit indexes of the three mediator models Outcome Depressive symptoms Hopelessness Global model fit indexes Family cohesion Social support Confidence in saving Family cohesion Social support Confidence in saving χ 2 66.54 52.32 45.98 66.30 53.57 35.71 DF 13 13 13 131 13 13 p value <.001 <.001 <.001 <.001 <.001 .0007 CFI 0.95 0.96 0.95 0.93 0.96 0.94 RMSEA 0.06 0.05 0.05 0.06 0.05 0.04 SRMR 0.08 0.11 0.06 0.07 0.11 0.06 Open in a new tab Global model fit tests were computed in M plus 8.10 using full information maximum likelihood estimation with robust test statistics (M plus estimator MLR). Bolded values met criteria. DF = degrees of freedom. Indirect and direct effects Family cohesion. Regarding depressive symptoms, we found a total indirect effect that was both negative and statistically significant (B = −0.57; 95% CI: −0.80, −0.36; β = −0.028). Specifically, the Suubi4Her intervention had an indirect impact on depressive symptoms at the 24-month follow-up through its influence on family cohesion at the 12-month follow-up (B = −0.15; 95% CI: −0.26, −0.07; β = −0.007) ( Table 4 ). This meant that for every one-unit change in the assignment to the intervention group (being part of the Suubi4Her intervention group as opposed to the control group), there was a corresponding decrease of −0.15 units in depressive symptom scores, and this change was attributed to the improvement in family cohesion resulting from the intervention. We also observed an indirect effect of the intervention on depressive symptoms at 24 months through the levels of depressive symptoms at 12 months (B = −0.42; 95% CI: −0.61, −0.25; β = −0.021). In addition to these indirect effects, there was a direct effect of the intervention on depressive symptoms at 24 months (B = −1.86; 95% CI: −2.63, −1.80; β = −0.093). This direct effect represents the portion of the variance in depressive symptoms at 24 months that can be attributed to the intervention but is not explained by the mediation structure of our model, indicating that there was partial mediation by family cohesion and depressive symptoms at 12 months. Table 4. Unstandardized total, total indirect, specific indirect, and direct effects of the Suubi4Her intervention on depressive symptoms and hopelessness at 24 months, N = 1260 Outcome Mediator Effect B (95% CI) Depressive symptoms Family cohesion Total effect −2.42 (−3.40, −1.47) Total indirect −0.57 (−0.80, −0.36) Specific indirect: Suubi4Her intervention → family cohesion at 12 months → depressive symptoms at 24 months (a 1 → a 3 ) −0.15 (−0.26, −0.07) Specific indirect: Suubi4Her intervention → family cohesion→ depressive symptoms at 12 months → depressive symptoms at 24 months (a 1 →a 2 →a 4 ) −0.42 (−0.61, −0.25) Direct effect: Suubi4Her intervention → depressive symptoms at 24 months −1.86 (−2.63, −1.80) Depressive symptoms Social support Total effect −2.32 (−3.32, −1.38) Total indirect −0.52 (−0.74, −0.32) Specific indirect: Suubi4Her intervention → social support at 12 months → depressive symptoms at 24 months (d 1 → d 3 ) −0.12 (−0.22, −0.06) Specific indirect: Suubi4Her intervention → social support → depressive symptoms at 12 months → depressive symptoms at 24 months (d 1 →d 2 →a 4 ) −0.40 (−0.59, −0.22) Direct effect: Suubi4Her intervention → depressive symptoms at 24 months −1.81 (−2.60, −1.04) Depressive symptoms Confidence in saving Total effect −2.32 (−3.33, −1.37) Total indirect −0.48 (−0.70, −0.29) Specific indirect: Suubi4Her intervention → confidence in saving at 12 months → depressive symptoms at 24 months (e 1 → e 3 ) −0.04 (−0.11, 0.01) Specific indirect: Suubi4Her intervention → confidence in saving → depressive symptoms at 12 months → depressive symptoms at 24 months (e 1 →e 2 →a 4 ) −0.45 (−0.66, −0.29) Direct effect: Suubi4Her intervention → depressive symptoms at 24 months −1.84 (−2.64, −1.07) Hopelessness Family cohesion Total effect −0.57 (−0.87, −0.28) Total indirect −0.11 (−0.16, −0.06) Specific indirect: Suubi4Her intervention → family cohesion at 12 months → hopelessness at 24 months (w 1 → w 3 ) −0.04 (−0.08, −0.02) Specific indirect: Suubi4Her intervention → family Cohesion→ hopelessness at 12 months → hopelessness at 24 months (w 1 →w 2 →w 4 ) −0.07 (−0.11, −0.03) Direct effect: Suubi4Her intervention -→ hopelessness at 24 months −0.46 (−0.72, −0.22) Hopelessness Social support Total effect −0.54 (−0.93, −0.25) Total indirect −0.10 (−0.15, −0.05) Specific indirect: Suubi4Her intervention → social support at 12 months → hopelessness at 24 months (y 1 → y 3 ) −0.04 (−0.07, −0.02) Specific indirect: Suubi4Her intervention → social support →hopelessness at 12 months → hopelessness at 24 months (y 1 →y 2 →w 4 ) −0.06 (−0.10, −0.03) Direct effect: Suubi4Her intervention → hopelessness at 24 months −0.45 (−0.70, −0.20) Hopelessness Confidence in saving Total effect −0.55 (−0.86, −0.26) Total indirect −0.08 (−0.14, −0.04) Specific indirect: Suubi4Her intervention → confidence in saving at 12 months → hopelessness at 24 months (z 1 → z 3 ) −0.01 (−0.03, 0.01) Specific indirect: Suubi4Her intervention → confidence in Saving → hopelessness at 12 months → Hopelessness at 24 months (z 1 →z 2 →w 4 ) −0.07 (−0.13, −0.04 ) Direct effect: Suubi4Her intervention → hopelessness at 24 months −0.47 (−0.72, −0.22) Open in a new tab Regression coefficients (B) were estimated via full-information maximum likelihood using M plus 8.10; 95% confidence intervals were estimated via the bias-corrected bootstrap based on 5,000 replicate samples. Confidence intervals which do not include zero are bolded and are significant at p < 0.05. For hopelessness, the analysis revealed an indirect effect of the intervention on hopelessness mediated by family cohesion, indicating that within the intervention group (Suubi4Her vs. Control), a one-unit change corresponded to a decrease of −0.04 (−0.08, −0.02), β = −0.006 in hopelessness scores through the influence of family cohesion. Additionally, the impact of the intervention on hopelessness at 12 months was found to mediate the hopelessness levels at 24 months, with a coefficient of −0.07 (95% CI: −0.11, −0.03); β = −0.011. We also observed a direct effect of the intervention on hopelessness at 24 months, represented by a coefficient of −0.46 (95% CI: −0.72, −0.22); β = −0.077. These findings collectively suggest evidence of partial mediation by family cohesion and hopelessness at 12 months, highlighting the presence of unaccounted variation in hopelessness scores. Social support. In the context of depressive symptoms, the influence of social support at the 12-month follow-up was found to mediate the impact of the intervention on depressive symptoms at 24 months. Participants within the intervention group displayed lower depressive symptom scores compared to participants in the control group, and this difference was mediated by social support (B = −0.12; 95% CI: −0.22, −0.06; β = −0.006). Additionally, the depressive scores at 12 months mediated the intervention’s effect on depressive scores at 24 months, as indicated by a coefficient of −0.40 (95% CI: −0.59, −0.22); β = −0.020. These findings also revealed the presence of a direct effect of the intervention on depressive symptoms at the 24-month mark, thus signifying partial mediation (B = −1.81; 95% CI: −2.60, −1.04; β = −0.091. Regarding hopelessness, the analysis revealed that the intervention exhibited a statistically significant negative total indirect effect (B = −0.10: 95% CI: −0.15, −0.05; β = −0.016). This suggests that the intervention had an impact on hopelessness through indirect pathways. At the 12-month follow-up, social support emerged as a mediating factor, partially explaining the mechanism of change between the intervention and hopelessness at 24 months. Participants in the intervention group displayed lower hopelessness scores compared to their counterparts in the control group, and this effect was mediated by social support (B = −0.04: 95% CI: −0.07, −0.02; β = −0.01). Moreover, hopelessness scores at 12 months also played a mediating role in the association between the intervention and hopelessness scores at 24 months (B = −0.06: 95% CI: −0.10, −0.03; β = −0.010). Additionally, there was evidence of a direct effect of the intervention on hopelessness at the 24-month follow-up, indicating partial mediation (B = −0.45; 95% CI: −0.70, −0.20; β = −0.074). Confidence in saving. Regarding confidence in saving, the total indirect effect was both negative and statistically significant (B = −0.48: 95% CI: −0.70, −0.29; β = −0.024). Interestingly, there was no evidence of confidence in saving mediating the relationship between the intervention and depressive symptoms at the 24-month follow-up (B = −0.04: 95% CI: −0.11, 0.01; β = −0.002). However, depressive symptoms at the 12-month time point mediated the impact of the intervention on depressive symptoms at 24 months (B = −0.45: 95% CI: −0.66, −0.29; β = −0.022). Furthermore, we observed a significant negative direct effect of the intervention, suggesting partial mediation (B = −1.84: 95% CI: −2.64, −1.07; β = −0.092). A similar pattern was identified concerning hopelessness and confidence in saving. Specifically, confidence in saving did not appear to mediate the effect of the intervention on hopelessness (B = −0.01: 95% CI: −0.03, 0.01; β = −0.001). Nevertheless, the total indirect effect remained significant and negative (B = −0.08: 95% CI: −0.14, −0.04; β = −0.014). Hopelessness at the 12-month follow-up did play a mediating role, influencing hopelessness at the 24-month follow-up as a result of the intervention (B = −0.07: 95% CI: −0.13, −0.04; β = −0.013). Furthermore, the total direct effect was significant and negative, signifying a state of partial mediation (B = −0.47: 95% CI: −0.72, −0.22; β = −0.078). Discussion In this study, we conducted an in-depth investigation into the effects of an intervention on key mediators, including family cohesion, social support, and confidence in saving, and their subsequent influence on two critical psychological outcomes, depressive symptoms and hopelessness over 24 months. Our findings provide insights into the mediating pathways between the intervention and depressive symptoms and hopelessness outcomes over time. First, the intervention exerted a significant negative total indirect effect on both depressive symptoms and hopelessness, underscoring its impact on these outcomes through indirect pathways. While confidence in saving did not emerge as a mediator for either outcome, family cohesion and social support at the 12-month follow-up played a significant mediating role. They mediated the effect of the intervention on depressive symptoms and hopelessness at the 24-month follow-up, indicating their crucial contribution to participants’ psychological well-being. Additionally, our analysis revealed a significant negative direct effect of the intervention on both outcomes, indicating partial mediation. These findings illuminate the complex interplay between the intervention, family cohesion, social support, and confidence in saving, underscoring their collective influence on psychological well-being. These insights have important implications for future research and the development of effective intervention strategies. The partial mediation by family cohesion is consistent with previous research, which has consistently shown that strong family bonds can serve as a protective factor against depressive symptoms and hopelessness in adolescents [ 35 - 38 ]. The theory-based family strengthening intervention, delivered through multiple family groups, likely played a crucial role in enhancing family cohesion, which in turn contributed to better mental health outcomes among the adolescent girls. Moreover, the role of social support as a mediator aligns with the Social Support Theory and previous research [ 14 , 39 , 40 ], which observed that strong social support systems can buffer against negative mental health outcomes. In the Suubi4Her intervention, the collaborative nature of the family groups as well as the financial literacy training sessions and family contributions for the YDA accounts likely fostered supportive relationships among participants, which subsequently had a positive impact on adolescent psychological wellbeing. Although confidence in saving did not emerge as a direct mediator in our study, it is essential to acknowledge that financial empowerment can still be a valuable component of overall wellbeing, even if it did not specifically mediate the intervention’s effects on depression and hopelessness in this instance [ 41 ]. For confidence in saving, it may take longer to build up over time than can be captured within the study’s data collection window. For instance, if participants are saving but mostly forget about what they are saving, they might not think much about saving. We anticipate that confidence may emerge later when they make their first withdrawal to start/grow an existing business or pay for higher education. Our study had numerous strengths. First, we utilized a longitudinal design, which allowed us to track changes over 24 months. This approach provides a more comprehensive understanding of how the intervention and mediating variables unfold over time and their impact on depressive symptoms and hopelessness. We also included autoregressive paths to and from earlier and later measures of the same mediator and outcome variables, which accounts for the fact that earlier measurements of a mediator or outcome can influence their subsequent measurements, potentially reducing bias that may result from ignoring these temporal dependencies. However, although our mediation analysis included temporal precedence, there may still be factors outside the scope of our study that could influence the results. While our findings provide valuable insights for interventions that can improve mental health among adolescent girls in Uganda, it may not be directly generalizable to other cultural and or geographical contexts such as among girls in high-income and more developed settings which may observe different effectiveness of the intervention and different mediating pathways. As such the intervention should be tailored to the specific context and target population first to ensure effectiveness. Furthermore, we recommend a qualitative investigation of the intervention process and participants’ lived experiences, as this could help unpack more of the "why" behind the mediation that occurred (for depressive symptoms and hopelessness) versus the mediation that did not occur(for saving confidence). Lastly, our study relied on self-reported measures, which introduces the possibility of response bias and does not represent clinical levels of depression. We mitigated this bias by relying on scales that were validated in settings similar to the Ugandan context, such as Kenya. Overall, our findings contribute to the literature in support of combination economic strengthening interventions that promote stronger family relationships and build social relationships as beneficial for mental health of adolescent girls. Policymakers and organizations working on adolescent wellbeing should consider implementing comprehensive family-strengthening interventions that foster positive family dynamics and provide social support for networks for adolescents. IMPLICATIONS AND CONTRIBUTION. The Suubi4Her intervention indirectly reduced depressive symptoms and hopelessness through improvements in family cohesion and social support. Community-based family group interventions that promote stronger family relationships can go a long way toward the protection of child and adolescent mental health in low-income settings. Acknowledgments The study outlined in this protocol is supported by the National Institute of Mental Health (NIMH) under award 1R01MH113486-01 (PI: Fred M. Ssewamala, PhD). We are grateful to Abel Mwebembezi at Reach the Youth–Uganda, Joseph Kato Bakulu at Masaka Catholic Diocese, Gertrude Nakigozi and Godfrey Kigozi at Rakai Health Sciences Program in Uganda, and Flavia Namuwonge, and Sarah Namutebi at the International Center for Child Health and Development, Masaka Office, for their respective contributions to the study design and implementation. In addition, we are grateful to the financial institutions that agreed to work with the adolescent girls in opening savings accounts and the extension workers who have committed time to train the adolescent girls in conducting income-generating activities. Our thanks also go to the Ugandan Government Ministry of Education and the 47 secondary schools that have agreed to participate in the Suubi4Her study. Note. The content is solely the responsibility of the authors and does not necessarily represent the official views of NIMH or the National Institutes of Health. Footnotes Conflicts of interest: The authors have no conflicts of interest to declare that are relevant to the content of this article. References [1]. Kieling C, Baker-Henningham H, Belfer M, et al. Child and adolescent mental health worldwide: Evidence for action. Lancet 2011;378:1515–25. [ DOI ] [ PubMed ] [ Google Scholar ] [2]. Patel V. Mental health in low- and middle-income countries. Br Med Bull 2007;81-82:81–96. [ DOI ] [ PubMed ] [ Google Scholar ] [3]. Jörns-Presentati A, Napp AK, Dessauvagie AS, et al. The prevalence of mental health problems in sub-saharan adolescents: A systematic review. PLoS One 2021;16:e0251689. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [4]. Erskine H, Baxter A, Patton G, et al. The global coverage of prevalence data for mental disorders in children and adolescents. Epidemiol Psychiatr Sci 2017;26:395–402. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [5]. World health Organization. Mental health Atlas 2017. Geneva, Switzerland: World Health Organization; 2018. [ Google Scholar ] [6]. Kigozi F, Ssebunnya J, Kizza D, et al. An overview of Uganda’s mental health care system: Results from an assessment using the world health organization’s assessment instrument for mental health systems (WHO-AIMS). Int J Ment Health Syst 2010;4:1–9. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [7]. Fulton BD, Scheffler RM, Sparkes SP, et al. Health workforce skill mix and task shifting in low income countries: A review of recent evidence. Hum Resour Health 2011;9:1. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [8]. Hoeft TJ, Fortney JC, Patel V, Unützer J. Task-sharing approaches to improve mental health care in rural and other low-resource settings: A systematic review. J Rural Health 2018;34:48–62. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [9]. Mabrouk A, Mbithi G, Chongwo E, et al. Mental health interventions for adolescents in Sub-Saharan Africa: A scoping review. Front Psychiatry 2022;13:937723. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [10]. Ssewamala FM, Bermudez LG, Neilands TB, et al. Suubi4Her: A study protocol to examine the impact and cost associated with a combination intervention to prevent HIV risk behavior and improve mental health functioning among adolescent girls in Uganda. BMC Public Health 2018;18:693. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [11]. Ssewamala FM, Brathwaite R, Neilands TB. Economic empowerment, HIV risk behavior, and mental health among school-going adolescent girls in Uganda: Longitudinal cluster-randomized controlled trial, 2017-2022. Am J Public Health 2023;113:306–15. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [12]. Kort-Butler LA, ed. Social support theory. The encyclopedia of juvenile delinquency and justice. Hoboken, NJ: John Wiley & Sons. Inc; 2017:1–4. [ Google Scholar ] [13]. Van Eickels RL, Tsarpalis-Fragkoulidis A, Zemp M. Family cohesion, shame-proneness, expressive suppression, and adolescent mental health—A path model approach. Front Psychol 2022;13:921250. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [14]. Ozbay F, Johnson DC, Dimoulas E, et al. Social support and resilience to stress: From neurobiology to clinical practice. Psychiatry (Edgmont) 2007;4:35–40. [ PMC free article ] [ PubMed ] [ Google Scholar ] [15]. McKay MM, Gonzales JJ, Stone S, et al. Multiple family therapy groups: A responsive intervention model for inner city families. Social Work Groups 1995;18:41–56. [ Google Scholar ] [16]. Sherraden M. Stakeholding: Notes on a theory of welfare based on assets. Social Serv Rev 1990;64:580–601. [ Google Scholar ] [17]. Sherraden M. Assets and the poor: A new American welfare policy. New York: ME Sharpe; 1991:344. [ Google Scholar ] [18]. Kim YK, Lim Y, Maleku A, et al. Financial stress and depression among African refugees in the southern United States: A serial multiple mediation analysis of financial self-efficacy and financial anxiety. J Immigrant Refugee Stud 2020;18:151–71. [ Google Scholar ] [19]. Lee JM, Rabbani A, Heo W. Examining financial anxiety focusing on interactions between financial knowledge and financial self-efficacy. J financial Ther 2023;14:2. [ Google Scholar ] [20]. Sensoy Bahar O, Byansi W, Kivumbi A, et al. From "4Rs and 2Ss" to "Amaka Amasanyufu" (happy families): Adapting a U.S.-based evidence-based intervention to the Uganda context. Fam Process 2020;59:1928–45. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [21]. Skinner HA, Steinhauer PD, Santa-Barbara J. The family assessment measure. Can J Commun Ment Health 2009;2:91–103. [ Google Scholar ] [22]. Beck AT, Ward CH, Mendelson M, et al. An inventory for measuring depression. Arch Gen Psychiatry 1961;4:561–71. [ DOI ] [ PubMed ] [ Google Scholar ] [23]. Abubakar A, Kalu RB, Katana K, et al. Adaptation and latent structure of the Swahili version of beck depression Inventory-II in a low literacy population in the context of HIV. PLoS one 2016;11:e0151030. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [24]. Beck AT, Weissman A, Lester D, Trexler L. The measurement of pessimism: The hopelessness scale. J consulting Clin Psychol 1974;42:861–5. [ DOI ] [ PubMed ] [ Google Scholar ] [25]. Kagotho N, Bowen NK, Ssewamala FM, et al. Cross-national validity of the beck hopelessness scale for children and adolescents: Findings from the YouthSave-Impact study Kenya. Int J Cult Ment Health 2018;11:457–69. [ Google Scholar ] [26]. Ssewamala FM, Shu-Huah Wang J, Brathwaite R, et al. Impact of a family economic intervention (Bridges) on health functioning of adolescents orphaned by HIV/AIDS: A 5-year (2012–2017) cluster randomized controlled trial in Uganda. Am J Public Health 2021;111:504–13. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [27]. Karimli L, Nabayinda J, Nartey PB, Ssewamala FM. Poverty reduction and family functioning: Results from an experimental study in Sub-Saharan Africa. J Child Fam Stud 2024;33:3104–18. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [28]. Bukowski WM, Hoza B, Boivin M. Measuring friendship quality during Pre- and early adolescence: The development and psychometric properties of the friendship qualities scale. J Social Personal Relationships 1994;11:471–84. [ Google Scholar ] [29]. Byansi W, Ssewamala FM, Neilands TB, et al. The short-term impact of a combination intervention on depressive symptoms among school-going adolescent girls in Southwestern Uganda: The Suubi4Her cluster randomized trial. J Adolesc Health 2022;71:301–7. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [30]. StataCorp. Stata statistical software: Release 17. College Station, TX: StataCorp LLC; 2021. [ Google Scholar ] [31]. Mitchell MA, Maxwell SE. A comparison of the cross-sectional and sequential designs when assessing longitudinal mediation. Multivariate Behav Res 2013;48:301–39. [ DOI ] [ PubMed ] [ Google Scholar ] [32]. Filiatreau LM, Tutlam N, Brathwaite R, et al. Effects of a combination economic empowerment and family strengthening intervention on psychosocial Well-being among Ugandan adolescent girls and young women: Analysis of a cluster randomized controlled trial (Suubi4Her). J Adolesc Health 2023;72:S33–40. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [33]. Hu LT, Bentler PM. Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Struct Equation Model A Multidisciplinary J 1999;6:1–55. [ Google Scholar ] [34]. Mackinnon DP, Lockwood CM, Williams J. Confidence limits for the indirect effect: Distribution of the product and resampling methods. Multivariate Behav Res 2004;39:99. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [35]. Chen P, Harris KM. Association of positive family relationships with mental health trajectories from adolescence to midlife. JAMA Pediatr 2019;173:e193336. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [36]. Askeland KG, Bøe T, Breivik K, et al. Life events and adolescent depressive symptoms: Protective factors associated with resilience. PLoS One 2020;15:e0234109. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [37]. Nabayinda J, Kizito S, Ssentumbwe V, et al. The relationship between family cohesion and depression among school-going children with elevated symptoms of behavioral challenges in southern Uganda. J Adolesc Health 2023;72:S11–7. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [38]. Nyoni T, Nabunya P, Ssewamala FM. Perceived social support and psychological wellbeing of children orphaned by HIV/AIDS in Southwestern Uganda. Vulnerable Child Youth Stud 2019;14:351–63. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [39]. Ioannou M, Kassianos AP, Symeou M. Coping with depressive symptoms in young adults: Perceived social support protects against depressive symptoms only under moderate levels of stress. Front Psychol 2019;9:2780. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [40]. Bjørlykhaug KI, Karlsson B, Hesook SK, Kleppe LC. Social support and recovery from mental health problems: A scoping review. Nordic Social Work Res 2022;12:666–97. [ Google Scholar ] [41]. Ettman CK, Adam GP, Clark MA, et al. Wealth and depression: A scoping review. Brain Behav 2022;12:e2486. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] ACTIONS View on publisher site PDF (781.9 KB) Cite Collections Permalink PERMALINK Copy RESOURCES Similar articles Cited by other articles Links to NCBI Databases Cite Copy Download .nbib .nbib Format: AMA APA MLA NLM Add to Collections Create a new collection Add to an existing collection Name your collection * Choose a collection Unable to load your collection due to an error Please try again Add Cancel Follow NCBI NCBI on X (formerly known as Twitter) NCBI on Facebook NCBI on LinkedIn NCBI on GitHub NCBI RSS feed Connect with NLM NLM on X (formerly known as Twitter) NLM on Facebook NLM on YouTube National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 Web Policies FOIA HHS Vulnerability Disclosure Help Accessibility Careers NLM NIH HHS USA.gov Back to Top