Evaluation of a savings-led family-based economic empowerment intervention for AIDS-affected adolescents in Uganda: A four-year follow-up on efficacy and cost-effectiveness.

BACKGROUND: Children who have lost a parent to HIV/AIDS, known as AIDS orphans, face multiple stressors affecting their health and development. Family economic empowerment (FEE) interventions have the potential to improve these outcomes and mitigate the risks they face. We present efficacy and cost-effectiveness analyses of the Bridges study, a savings-led FEE intervention among AIDS-orphaned adolescents in Uganda at four-year follow-up.
METHODS: Intent-to-treat analyses using multilevel models compared the effects of two savings-led treatment arms: Bridges (1:1 matched incentive) and BridgesPLUS (2:1 matched incentive) to a usual care control group on the following outcomes: self-rated health, sexual health, and mental health functioning. Total per-participant costs for each arm were calculated using the treatment-on-the-treated sample. Intervention effects and per-participant costs were used to calculate incremental cost-effectiveness ratios (ICERs).
FINDINGS: Among 1,383 participants, 55% were female, 20% were double orphans. Mean age was 12 years at baseline. At 48-months, BridgesPLUS significantly improved self-rated health, (0.25, 95% CI 0.06, 0.43), HIV knowledge (0.21, 95% CI 0.01, 0.41), self-concept (0.26, 95% CI 0.09, 0.44), and self-efficacy (0.26, 95% CI 0.09, 0.43) and lowered hopelessness (-0.28, 95% CI -0.43, -0.12); whereas Bridges improved self-rated health (0.26, 95% CI 0.08, 0.43) and HIV knowledge (0.22, 95% CI 0.05, 0.39). ICERs ranged from $224 for hopelessness to $298 for HIV knowledge per 0.2 standard deviation change.
CONCLUSIONS: Most intervention effects were sustained in both treatment arms at two years post-intervention. Higher matching incentives yielded a significant and lasting effect on a greater number of outcomes among adolescents compared to lower matching incentives at a similar incremental cost per unit effect. These findings contribute to the evidence supporting the incorporation of FEE interventions within national social protection frameworks.

RCT Entities:   Population Interventions Outcomes

Introduction

Healthy adolescence is a critical step in transitioning to a healthy adulthood [1]. This developmental period can, however, be thwarted by poverty and lack of access to healthcare, education and social support. Most health risks and challenges, including exploratory sexual behaviors, substance use, and poor mental health functioning, emerge during adolescence [2]. Orphaned adolescents, defined as those who have lost one or both biological parents, face additional economic and psychosocial barriers that further limit their ability to thrive without appropriate support systems [3, 4].

Globally, over 12 million children under age 18 are reported to have lost at least one parent to HIV/AIDS [5]. Of these, 79% live in sub-Saharan Africa (SSA) [5]. Within SSA, Uganda is heavily affected; of the 1.9 orphaned children, those orphaned by AIDS make up 35% (660,000) [6]. Research shows that AIDS orphaned children are more likely than other orphans to initiate sex early [7], report higher rates of transactional sex [8], engage in risky sexual behaviors [9], test positive for HIV infection [10, 11], have higher vulnerability to violence and abuse [3, 8], and report poorer mental health [10, 11]. Poverty mediates all of these effects [8]. Parental loss has also been linked to increased household responsibilities and reduced school attendance and performance in this age group [12].

While social and emotional support promote coping and resilience among adolescent orphans, which is key to leading a healthy and productive life [3, 13], household financial instability is associated with poor educational and health outcomes in this population [14-16]. Theory and evidence suggest that social protection programs can mitigate risks among children affected by HIV/AIDS [14, 17]. In particular, cash assistance may reduce sexual risk taking, including transactional sex, thus reducing new HIV infections, unwanted pregnancies and school dropout rates among adolescents [6, 18]. Grounded in asset theory, family economic empowerment (FEE) interventions go beyond cash transfers by providing capacity building, mentorship, and seed funding and fostering household financial stability through promoting income generating activities and financial literacy [19, 20]. Asset theory posits that orphaned adolescents will experience higher levels of depression, have worse educational outcomes, and engage in higher risk behaviors if they lack financial means to participate in secondary education. FEE interventions that bring financial stability to households act as a protective factor [21]. In fact, these interventions have been shown to increase household financial stability, leading to greater savings for education and improved academic performance and mental health outcomes (lower depressive symptomatology and higher levels of self-esteem) among orphaned adolescences in Uganda [16, 20].

FEE interventions offer promise in addressing several health and developmental needs for poor adolescents impacted by HIV/AIDS in low-resource communities, including SSA [20, 22–25]. To make the case for public investment in these interventions, it is necessary to integrate the evidence on effects with costs. By calculating an incremental cost per unit of benefit, cost-effectiveness analyses make it explicit which interventions will contribute the most relative to their costs and inform resources allocation decisions in the face of competing health priorities and resource constraints. Evidence is extremely limited on the cost-effectiveness of interventions aimed at improving adolescents’ health outcomes, including physical and mental health. This paper contributes to the currently limited scientific body of knowledge on the economic value of an FEE intervention, titled Bridges, that applies a savings-led approach aimed at improving health and developmental outcomes of poor adolescents. Specifically, our earlier study findings showed that the Bridges intervention demonstrated a favorable effect on the critical developmental outcomes of adolescents impacted by HIV/AIDS compared with usual care alone at 24-months post-intervention initiation [26]. A longer-term sustainability question has, however, remained unanswered by those initial short-term findings: how long would the observed outcomes be sustained and at what cost? That is the question addressed in this paper, using data from 48 months post-intervention initiation. If the intervention exhibits sustained efficacy, it is also important to understand the potential changes in the cost-effectiveness of the intervention over time. In summary, this analysis aims to assess the effects and cost-effectiveness of the two arms of a FEE intervention in relation to care-as-usual two years post-intervention.

Materials and methods

Trial population and setting

The current analysis is based on baseline and 48-month follow-up data from the Bridges to the Future intervention (hereinafter, Bridges). Bridges was a five-year (2012–2016) cluster randomized control trial to evaluate the efficacy and cost-effectiveness of a FEE intervention on health, developmental and educational outcomes among adolescents orphaned by HIV/AIDS in southwestern Uganda. All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and national research committee and with 1964 Helsinki declaration and its later amendments. Informed consent and assent were obtained separately from caregivers and adolescents respectively, prior to study participation. The Bridges study received approval from the Columbia University Institutional Review Board (IRB-AAA11950) and the Uganda National Council of Science and Technology (SS2586). The study is registered in the Clinical Trials database (NCT01447615). The clinical trial is described in detail elsewhere [26]. Briefly, a total of 1,410 adolescents orphaned by HIV/AIDS (n = 621 boys, n = 789 girls), between the ages of 10–16 (mean participant age: 12 years at baseline) were recruited from 48 public primary schools (Table A in S1 Appendix). The schools are in four political districts of Rakai, Masaka, Lwengo and Kalungu in southwestern Uganda—a region heavily affected by HIV/AIDS [27]. There were three eligibility criteria: 1) the adolescent had lost one or both parents to HIV/AIDS; 2) the adolescent was enrolled in grades 5 or 6, in a government-aided primary school during recruitment period; and 3) the adolescent was living with a family, not in an institution. A total of 27 adolescents were deemed ineligible after randomization and dropped from the trial, resulting in a final sample of 1,383 adolescents (Fig. A in S1 Appendix). About 20% of the adolescents were double orphans (lost both parents). Schools were randomly assigned to three study arms: control, Bridges, and BridgePLUS (described below).

Adolescents in the control condition received ‘usual care’ consisting of counseling by community priests and school supplies. Adolescents in Bridges and BridgesPLUS treatment arms received usual care plus an incentivized savings account [Child Development Account (CDA)] with either a 1:1 match rate (Bridges) or 2:1 match rate (BridgesPLUS). In addition, all participants in Bridges and BrigesPLUS received: three sessions on financial literacy and management (FLT), including how to save, budget and support asset accumulation; six sessions on income generating activities; and eight sessions of mentorship by a near peer. The first two activities were conducted by trained research assistants. The only difference between Bridges and BridgesPLUS was the level of incentivized match rate. The participants did not need to attend all the activities to receive incentives. To be eligible, they needed to attend the FLT sessions and open a bank account at a participating financial institution. Matched funds were restricted for use on educational or income-generating activities. The matched funds were contributed by the trial (See Table 1 for program costs, including matched amounts). The attrition rate was 8.8% for the Bridges and the control groups and 11.2% for the BridgesPLUS group at the 48-month follow-up. However, the results from a chi-square test indicate that the attrition rates do not differ significantly by study arms (χ2 = 2.04; p = 0.36) (Table B in S1 Appendix).

Table 1

Itemized total per-participant costs (all costs are in 2012 Ugandan Shillings unless otherwise indicated).

	Usual Care	Bridges	BridgesPLUS
Recurrent costs
School lunches	39,491	39,491	39,491
Educational materials	76,569	76,569	76,569
Counseling	38	38	38
Recruitment of participants	5,847	5,847	5,847
Child savings account
Account opening	-	27,414	27,414
Initial account deposit	-	20,000	20,000
Annual matched savings	-	20,309	42,634
Mentorship	-	55,490	55,490
Financial education and income generating activity training	-	53,710	53,710
Personnel (staff salary and allowances)	100,105	200,210	200,210
Volunteer and donated resources	5,986	429,526	429,525
Capital costs
Furniture, equipment and vehicles	66,285	132,570	132,570
Total per-participant costs	294,321	1,061,174	1,083,498
Total per-participant costs (in 2012 USD)	117	419	428

Measures

Self-rated health is measured by a single 5-item scale ranging from excellent to very poor with higher values indicate better health [28]. Mental health functioning is conceptualized as depression, hopelessness, self-concept and self-efficacy. We used the 27-item Child Depression Inventory (CDI) and the 20-item Beck Hopelessness Scale (BHS) [29] operationalized with higher scores indicating worse mental health. Moreover, the 20-item Tennessee Self-Concept Scale [30], and the 29-item Youth Self-Efficacy Survey [31] are used for measurements of self-concept and self-efficacy with higher values on indicate more positive self-concept and self-efficacy, respectively. Three indictors—sexual risk-taking intentions [20], HIV prevention intention [32], and HIV knowledge [21] are used to capture sexual health. Detailed example questions for each scale are provided in Table C in S1 Appendix with reliability statistics. Briefly, CDI (α = 0.68) and BHS (α = 0.65) show modest internal consistency, whereas Tennessee Self-Concept Scale, the Youth Self-Efficacy Survey, sexual risk-taking intentions, HIV prevention intention, and HIV knowledge all have acceptable internal consistency with Cronbach’s alphas greater than 0.70. All measurements mentioned above are standardized to be comparable across all outcomes.

Evaluation of the intervention

Outcome data were collected by trained research assistants through in-person interviews with adolescents. To examine the intervention effects, multilevel linear regressions were conducted independently for each outcome using the final sample of 1,383 adolescents. We focused on eight key health and mental health outcomes: self-rated health, depression, hopelessness, self-concept, self-efficacy, sexual risk-taking intentions, HIV prevention attitudes, and HIV knowledge (Table C in S1 Appendix). We used a three-level multilevel model that accounted for school and individual clustering-effects by including school ID and child ID as random intercept terms. Further, we allowed the slopes over time to differ for each child. Scores were standardized before being included in the regression models to facilitate comparison of the effects across outcomes with different units. There were no significant differences in distributions across arms in students’ age and sex at baseline. Therefore, these variables were not included in the models. In each model, we included study group dummies (Bridges and BridgesPLUS; usual care as the reference group), a wave dummy (48-month follow-up; baseline as the reference group), and their interactions. The interaction between 48-month follow-up indicator and treatment arms were the key coefficients of interest by demonstrating the effect of the intervention with baseline differences and trends in outcomes controlled for. After each multilevel regression, we used a Wald test to examine whether the coefficients between the two interaction terms were statistically different. All analyses were performed using the mixed command in Stata 15 (StataCorp LP, College Station, TX, USA).

Costs of the intervention

We estimated the costs of the usual care and the two treatment arms from a provider perspective and used a combination of activity-based and ingredients approach to costing where we identified all the activities in each study arm and then measured and valued all the resources used for each activity in each study arm. Activities encompassed recruitment of participants, provision of school lunches and educational materials (uniforms, notebooks and textbooks), counseling, mentorship, financial education and income generating activity trainings, and opening and contributing to child savings accounts. In addition to actual financial expenditures, we also quantified volunteered and donated resources used for all the activities to arrive at the economic costs of the intervention. Costs were carefully recorded throughout the Bridges implementation process and extracted for analysis from the project’s administrative records and CDA-related bank records. The recurrent costs of implementation included the costs of school supplies (students’ lunches, school uniforms and textbooks), bank accounts opening, initial accounts deposits, matching incentive contributions, transportation (fuel, travel allowances and rental fees), maintenance and repair of equipment and vehicles, field office rent, office maintenance (Internet access and phone, utilities, security services), office supplies and printing, field personnel (staff salary and allowances), time costs of volunteers (community leaders and mentors) and other implementing partners (teachers and bank staff), and donated resources (classroom space). Time costs of staff and teachers were apportioned according to time devoted to the activities in each arm. The capital costs included the costs of office furniture and equipment and vehicles. We calculated annual depreciation costs of the capital items assuming an appropriate useful life for each item and apportioned these costs according to their estimated share of use for the activities in each arm. We calculated the per-participant costs for each cost category using the treatment-on-the-treated (TOT) sample and added the costs to estimate the total per-participant cost for each arm over the intervention period. All costs were adjusted for inflation using the Uganda Consumer Price Index [33], discounted at 3% to the first year of the trial [34], and expressed in 2012 US dollars.

Cost-effectiveness methodology

This cost-effectiveness analysis used the costing data collected during the Bridges study planning phase, baseline, through study implementation to closeout. The outcome data comes from data collected at baseline and at 48-month follow-up. The analysis centered on incremental cost-effectiveness ratios (ICERs), where the numerator represented the cost difference between the treatment arms and the usual care, and the denominator represents the difference in average treatment effects. To that end, the cost-effectiveness analysis of the Bridges intervention involved examining how much Bridges or BridgesPLUS costs to achieve a unit of effect relative to usual care. First, we calculated the total per-participant costs in each study arm. Because the intent-to-treat (ITT) sample is larger than those who actually received the intervention, we calculated the per-participant costs conservatively based on the TOT sample. The use of ITT sample does not affect the relative difference in incremental costs across the two intervention arms and hence the relative difference in the cost-effectiveness ratios. Second, we estimated the effects of the intervention on our outcomes of interest as the standardized mean difference between each of the two treatment arms and the usual care arm, known as effect sizes, using an ITT approach. For the outcomes, we chose 0.2 SD change as a threshold, which corresponds to a small effect size as per guidelines provided by Cohen [35]. Third, we calculated the ICERs using the formulation above and computed the per-participant cost per 0.2 SD change for each outcome. Reporting of this analysis followed the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) [36].

Sensitivity analysis

In sensitivity analysis, we used the 95% confidence interval (CI) of the effect sizes and varied the total per-participant costs of the intervention from 80% to 120% and re-calculated the ICERs over these cautious ranges so as to explore the pessimistic (high cost/low effectiveness) and optimistic (low cost/high effectiveness) scenarios for Bridges and BridgesPLUS [37].

Results

The effects of the intervention

Table 2 present the intervention effects on the key health and mental health outcomes from multilevel linear regression analyses. The intervention increased self-rated health among participants in both treatment arms as compared to those receiving usual care (Bridges: 0.26 SD, 95% CI 0.08, 0.43; BridgesPLUS: 0.25 SD, 95% CI 0.06, 0.43). Participants in both Bridges and BridgesPLUS gained in HIV knowledge over participants in the control arm (0.22 SD, 95% CI 0.05, 0.38 and 0.21 SD, 95% CI 0.01, 0.41, respectively). Only BridgesPLUS resulted in statistically significant lower levels of hopelessness (-0.28 SD, 95% CI -0.43, -0.12), and higher levels of self-concept (0.26 SD, 95% CI 0.09, 0.44), and self-efficacy (0.26 SD, 95% CI 0.09, 0.43) as compared to the usual care arm. There are no observable statistically significant intervention effects on levels of depression, sexual risk-taking intentions, and HIV prevention attitudes at four-year follow-up. Wald tests point to statistically significant differences in efficacy of the treatment arms over time on self-concept (χ2 = 7.10, p = 0.008) and self-efficacy (χ2 = 7.06, p = 0.008), and a moderate though not statistically significant difference on hopelessness (χ2 = 3.71, p = 0.054).

Table 2

Intervention effects on self-rated health, mental health functioning, and sexual health.

Outcome	Self-rated health(n = 1,383)		Depression(n = 1,382)		Hopelessness(n = 1,382)		Self-concept(n = 1,383)		Self-efficacy(n = 1,383)		Sexual risk-taking intention(n = 1,383)		HIV prevention attitudes(n = 1,383)		HIV knowledge(n = 1,383)
	Effect size(95% CI)	p-value	Effect size(95% CI)	p-value	Effect size(95% CI)	p-value	Effect size(95% CI)	p-value	Effect size(95% CI)	p-value	Effect size(95% CI)	p-value	Effect size(95% CI)	p-value	Effect size(95% CI)	p-value
Group (ref: usual care)
Bridges	-0.19(-0.34 to -0.04)	0.011	-0.01(-0.17 to 0.16)	0.942	-0.08(-0.21 to 0.04)	0.177	0.002(-0.14 to 0.14)	0.982	-0.08(-0.24 to 0.07)	0.293	0.02(-0.14 to 0.17)	0.849	0.09(-0.11 to 0.29)	0.381	0.01(-0.20 to 0.22)	0.948
Bridges PLUS	-0.16(-.32 to -0.01)	0.041	0.07(-0.12 to 0.25)	0.496	0.04(-0.11 to 0.19)	0.628	-0.20(-0.34 to -0.06)	0.006	-0.20(-0.41 to 0.0001)	0.050	0.01(-0.14 to 0.15)	0.948	0.00(-0.17 to 0.18)	0.979	-0.03(-0.26 to 0.21)	0.828
Time (ref: baseline)
48 months	-0.15(-0.28 to -0.02)	0.027	-0.22(-0.33 to -0.10)	<0.001	-0.51(-0.59 to -0.43)	<0.001	0.39(0.25 to 0.52)	<0.001	0.37(0.25 to 0.49)	<0.001	-0.07(-0.20 to 0.06)	0.297	0.81(0.64 to 0.99)	<0.001	0.77(0.63 to 0.91)	<0.001
Group X time
Bridges X 48 months	0.26(0.08 to 0.43)	0.01	-0.10(-0.31 to 0.11)	0.337	-0.10(-0.25 to 0.06)	0.220	0.03(-0.16 to 0.22)	0.752	0.02(-0.16 to 0.20)	0.818	-0.17(-0.38 to 0.03)	0.097	-0.12(-0.34 to 0.10)	0.273	0.22(0.05 to 0.38)	0.012
Bridges PLUS X 48 months	0.25(0.06 to 0.43)	0.03	-0.14(-0.32 to 0.03)	0.098	-0.28(-0.43 to 0.13)	<0.001	0.26(0.09 to 0.44)	0.003	0.26(0.09 to 0.43)	0.003	-0.06(-0.22 to 0.10)	0.465	-0.04(-0.25 to 0.16)	0.679	0.21(0.01 to 0.41)	0.04
Wald test: Bridges X 48 months = Bridges PLUS X 48 months (χ2)	0.01	0.92	0.16	0.69	3.71	0.054	7.1	0.008	7.06	0.008	1.5	0.221	0.92	0.336	0.01	0.934

Note: One child in the control group, and one child in the Bridges PLUS were found to have missing values for both baseline and 48-months, in term of depression and hopelessness, respectively, thus they were excluded in the analysis for the corresponding outcome.

The costs and cost-effectiveness of the intervention

Table 1 presents the total per-participant costs for the three study arms. The total per-participant cost was $117 for the usual care arm, $419 for the Bridges arm, and $428 for the BridgesPLUS arm. The observed very small difference in cost between Bridges and BridgesPLUS was a result of the match rate difference.

Table 3 presents the ICERs for the outcomes on which the effects of Bridges and BridgesPLUS were found to be statistically significant relative to usual care. The ICER for self-rated health for Bridges was US$236 (95% CI 139–766) per 0.2 SD and lower than for BridgesPLUS at US$252 (95% CI 144–1014) per 0.2 SD, reflecting the higher mean effect size given the very small difference in the total per-participant costs of the interventions relative to usual care. Precisely for the same reasons, the ICER for HIV knowledge was lower at US$279 (95% CI 157–1,235) per 0.2 SD for Bridges compared to US$298 (95% CI 153–6,933) per 0.2 SD for BridgesPLUS. The intervention effects on hopelessness, self-concept and self-efficacy were statistically significant only for the BridgesPLUS arm, and the ICERs were computed as US$224 (95% CI 145–499), US$236 (95% CI 143–686), and US$242 (95% CI 145–734) per 0.2 SD, respectively. Overall, BridgesPLUS had a statistically significant effect on a higher number of health and mental health outcomes compared to Bridges, with ICERs ranging between US$224–298 per 0.2 SD change.

Table 3

Incremental cost-effectiveness ratios (ICER) relative to usual care (in 2012 US dollars).

Outcome	BridgesICER (95% CI)	BridgesPLUSICER (95% CI)
Self-rated health	236 (139–766)	252 (144–1,040)
Hopelessness	624 (n.s.)	224 (145–499)
Self-concept	1,952 (n.s.)	236 (143–686)
Self-efficacy	288 (n.s)	241 (145–734)
HIV knowledge	279 (157–1,235)	298 (153–6,933)

Note: Unit represents a 0.2 standard deviation change in mean outcome. Italics indicate non-significance.

Table 4 shows the results of the sensitivity analysis and presents the ICERs for the same set of mental health and health related outcomes under the optimistic and pessimistic scenarios, corroborating our findings that Bridges cost less than BridgesPLUS to achieve a 0.2 SD change in the outcomes for which the effects were statistically significant for both treatment arms at four-year follow-up.

Table 4

Sensitivity analysis of incremental cost-effectiveness ratios (ICERs) (in 2012 US dollars).

	Bridges		BridgesPLUS
Outcome	Optimistic scenario(Low cost/High effectiveness)	Pessimistic scenario(High cost/Low effectiveness)	Optimistic scenario(Low cost/High effectiveness)	Pessimistic scenario(High cost/Low effectiveness)
Self-rated health	112	919	115	1,246
Hopelessness	-	-	116	598
Self-concept	-	-	114	822
Self-efficacy	-	-	116	880
HIV knowledge	126	1,482	122	8,307

Note: Unit represents a 0.2 standard deviation change in mean outcome.

Discussion

In this study, we examined the efficacy and cost-effectiveness of a savings-led FEE intervention targeted at adolescents orphaned by HIV/AIDS in Uganda. We focused on eight different health and mental health outcomes, namely self-rated health, depression, hopelessness, self-concept, self-efficacy, sexual risk-taking intentions, HIV prevention attitudes, and HIV knowledge. At 24 months post intervention initiation [26], adolescents receiving Bridges (lower saving incentive) and BridgesPLUS (higher saving incentive) reported similar outcomes in regards to health and mental health, self-concept, self-efficacy, and HIV knowledge. Specifically, when compared to those in usual care, adolescents in both treatment conditions did better, although a higher savings incentive did not lead to differential treatment effects and cost-effectiveness ratios between the two treatment conditions [26]. At 48-month follow-up, the effects of Bridges and BridgesPLUS on self-rated health and HIV knowledge remained statistically significant relative to usual care, while Bridges proved to be on average more cost-effective compared to BridgesPLUS with lower ICERs for these outcomes to attain a 0.2 SD change relative to usual care. From 24-month to 48-months (the post-intervention period), the effects on hopelessness, self-concept and self-efficacy became statistically insignificant for adolescents receiving Bridges (lower saving incentive), while BridgesPLUS (higher saving incentive) significantly improved these outcomes, with ICERs ranging between US$224–298 per 0.2 SD change. These ICERs are overall higher than those calculated at 24-month follow-up (US$166–263 per 0.2 SD change) [26] because of the reduced effects of the intervention at 48-month follow-up. Both treatment conditions had no observable effects on HIV prevention attitudes, depression and sexual risk-taking intentions at 48-month follow-up. Thus, these specific outcomes were not included in the cost-effectiveness analysis. However, it is noteworthy to point out that, even though not statistically significant, the changes in depression and sexual risk-taking intentions were in the expected direction in both treatment conditions.

We calculated the per-participant costs conservatively based on the TOT sample. As a comparison, the total per-participant costs based on the ITT sample would be lower at $103, $363, and $372, for the usual care, Bridges and BridgesPLUS arms, respectively (Table D in S1 Appendix) and would result in lower ICERs per 0.2 SD change relative to usual care. Given the resource-intensive implementation strategy used in delivering the Bridges FEE intervention, and the fact that the cost analysis presented here used a conservative costing approach using the TOT sample, the cost per participant is likely to decrease when the intervention is integrated into a broader healthcare system. It is also likely that there may be economies of scale that the intervention would benefit from when it is delivered at scale.

The Bridges study targeted multiple outcomes for adolescents, and we presented the ICERs individually for each outcome, highlighting the important differences across the two treatment arms over time. Our findings indicate that Bridges and BridgesPLUS continued to have a positive effect on several health and mental outcomes, including self-rated health and HIV knowledge relative to usual care at 48-month follow-up, with ICERs ranging between US$236–298 per 0.2 SD change. An especially noteworthy finding of this follow-up is that the higher matching rate in BridgesPLUS yielded a significant and lasting effect on multiple outcomes in the long term, namely adolescents’ self-rated health, hopelessness, self-concept, and self-efficacy compared to Bridges at a similar incremental cost per unit effect, whereas the effects on adolescents who received the lower matching rate faded over the same period.

Cash transfer programs, both conditional and unconditional, targeted at most vulnerable households have been shown to impact a multiplicity of education and health outcomes in low resource-settings [38]. In Malawi, the government-run cash transfer program targeting ultra-poor, labor-constrained households increased school enrollment and reduced drop-out rates and improved mental health outcomes in children after one year [39]. In Kenya, the government-led program for orphans and vulnerable children, aimed at reducing HIV risk through unconditional cash transfers, reduced the odds of sexual debut by 31% [40]. Structural interventions to reduce HIV incidence have been found to be the most cost-effective [41], particularly when targeting multiple outcomes, such as education, in addition to HIV/AIDS incidence [42]. A privately funded program in Malawi conditioned on continued school enrollment of adolescent girls reduced the prevalence of HIV, Herpes simplex virus and depression and improved school attendance in this population. At 18 months, the intervention was estimated to avert 209 disability-adjusted life years (DALYs) at a cost of $297 per DALY averted [43]. Despite the growing literature on cash transfer programs, few studies have focused on interventions that include an economic empowerment component [41].

Our findings make a unique contribution to the existing literature on incentivized savings as previous studies on the effectiveness of savings incentives have primarily focused on economic outcomes [44-46]. Given the extremely limited cost-effectiveness evidence base on savings-led FEE interventions, there are no benchmarks associated with these intermediate health and mental outcomes, which limits our interpretation of the cost-effectiveness results. Generally speaking, the issue of what an additional unit of outcome is worth needs to be addressed for these interventions. As we wait for the empirical evidence that describes the longer-run, final outcomes such as the incidence of HIV or other sexually transmitted diseases becomes available, this issue can be best addressed with accumulating evidence from cost-effectiveness analyses of these interventions in similar settings and populations by examining if the same effect can be achieved at a lower cost, presuming that the effects are measured in the same units [47, 48]. In this pioneering study, we based and compared and contrasted our findings on a rigorous analysis of the efficacy and cost data relating to the two treatments arms with differing incentivized match rates at 24- and 48-month post-intervention initiation.

Overall, our findings suggest that this multifaceted intervention has the potential to positively contribute, both in the short-term and long-term, to the health and overall development of adolescents impacted by HIV/AIDS by mitigating household financial instability. Further, the rate of incentive seems to matter to sustain the effects of the intervention in the long run. We believe that the significant effects sustained on ‘multiple’ critical outcomes in this high-risk population justify investments in savings-led economic empowerment interventions in resource limited-settings.

A limitation of this study is that multiple hypotheses were tested for multiple outcomes using the same exposure and covariate data, potentially leading to inflated Type 1 error rates. Future studies seeking to replicate our findings might consider concentrating on the subset outcomes found to be statistically significant in this study and employing multiplicity adjustments to further control Type 1 error rates. Another limitation is that because our study was quantitative by design, we did not have qualitative data on adolescent perceptions of the program, which could have added nuance and context to better elucidate the mechanisms of the effectiveness of the program on youth. Future qualitative research is needed to understand the causal pathways of effectiveness of such family economic strengthening programs. Use of self-reported measures is the most common approach to assessment in low resource settings, but this approach is limited by self-report bias [49]. This study also relied on self-reported measures for physical and mental health and sexual risk-taking behavior. Therefore, our findings are limited to self-reported changes in these outcomes rather than a clinical diagnosis, for instance, for mental health outcomes. Future studies should consider incorporating biomarkers and other objective measures. Further research is also warranted to replicate and extend these findings in other similar settings and help establish cost-effectiveness benchmarks that can be useful for researchers in intervention development and for policymakers in decision-making. Accumulating evidence will further strengthen confidence in the benefits and economic value of savings-led and incentivized economic empowerment interventions for vulnerable adolescents living in countries with limited resources.

Fig. A. CONSORT Flow Diagram: The Bridges to the Future Study (2011–2017) Table A. Descriptive statistics of adolescent characteristics at baseline Table B. Descriptive statistics on characteristics of attrited sample Table C. Description of outcome measures Table D. Itemized total per-participant costs based on intent-to-treat sample (all costs are in 2012 Ugandan Shillings unless otherwise indicated).

(DOCX)

Click here for additional data file.

5 Sep 2019

PONE-D-19-20904

Evaluation of a savings-led family-based economic empowerment intervention for AIDS-affected adolescents in Uganda: A four-year follow-up on efficacy and cost-effectiveness

PLOS ONE

Dear Dr. Ssewamala,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

The reviewers provide useful comment and suggestion for improvement. Please ensure you address all comments related to the methods and interpretation of results.

We would appreciate receiving your revised manuscript by Oct 20 2019 11:59PM. When you are ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter.

To enhance the reproducibility of your results, we recommend that if applicable you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). This letter should be uploaded as separate file and labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. This file should be uploaded as separate file and labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. This file should be uploaded as separate file and labeled 'Manuscript'.

Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.

We look forward to receiving your revised manuscript.

Kind regards,

Christopher M Doran, BEc (Hons) PhD

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

http://www.journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and http://www.journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

1. Please update your Ethics Statement on the online submission form to include the information on consent provided in the methods section of your manuscript.

2. There is some discrepancy between your manuscript and the details registered at https://clinicaltrials.gov/ct2/show/NCT01447615. Please clarify whether the ages eligible for study were 11-17 as stated at www.clinicaltrials.gov or 10-16 years as stated in your manuscript.

3. Data availability issue. In your statement you say "All relevant data are within the paper and its Supporting Information files", but as we explain in http://journals.plos.org/plosone/s/data-availability#loc-faqs-for-data-policy you should provide the individual data points behind means, medians and variance measures presented in the results, tables and figures, and not just those summary statistics. Please provide these underlying participant-level data in a supporting information file or public repository, taking care not to include identifying information (see http://www.bmj.com/content/340/bmj.c181.long); if these data cannot be publicly deposited or included in the supporting information, e.g. due to patient privacy, legal reasons, or being provided by a third party, please explain why and explain how researchers may access them. Note that authors should not be the sole named individuals responsible for ensuring data access.

4. We note that you have included the phrase “data not shown” in your manuscript. Unfortunately, this does not meet our data sharing requirements. PLOS does not permit references to inaccessible data. We require that authors provide all relevant data within the paper, Supporting Information files, or in an acceptable, public repository. Please add a citation to support this phrase or upload the data that corresponds with these findings to a stable repository (such as Figshare or Dryad) and provide and URLs, DOIs, or accession numbers that may be used to access these data. Or, if the data are not a core part of the research being presented in your study, we ask that you remove the phrase that refers to these data.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: I Don't Know

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: No

Reviewer #3: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: No

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: The manuscript is generally well written and presented. However there some minor limitations for publishing.

1. The core question being addressed is not specifically stated or clearly linked to the extensive analyses provided. This could be due to the fact that the authors assume the audience is clear in regards to the intent of the trail, as well as of he actual project itself. The authors would greatly improve this issue by providing the readers with both a slightly elaborated description of the overall project, more on the underlying rationale for a focus on cost-effectiveness of such projects in global low income and rural settings, especially those with a history of social and cultural upheaval.

2. The retrospective focus of the analyses does not take into account the contextual and process variables that have also influenced the observed outcomes. This is one limitation of a single paradigm (quantitative) focus and self-reported behavioral outcomes. The authors need to better address the potential influence of these factors - what was provided was highly inadequate.

3. In such project the personnel and setting are important to any analyses since both are critical to the functional environments (e.g, relationship with youth, physical settings, so forth). While recognized as not a part of the present study, some mention of contextual variables was both expected and needed.

4. Lastly, operational definitions of the key variables of health and others were not provided. This needs to be addressed.

Overall, a good study. If the limitations are addressed this manuscript will provide a means to contribute our knowledge of such projects as suggested.

Reviewer #2: I have made comments and raised queries in the attached track-changed word documents.

However I would like to add two specific things here that I'd expect the authors to address 1) They need to note that as shown in the supplementary tables specifically Table S1 there was a marked higher attrition by 48 months for the BridgesPlus group (11.2%) as opposed to both Bridges and Control (8.8%) groups compared to the numbers at baseline. 2) There should be some mention of the fact that they are using the same exposure and co-variate data to test multiple hypotheses and some readers may wonder whether there is a multiplicity problem here requiring a lower P-value to be used. Both of these should be highlighted in a small limitations sub-section within the discussion.

Reviewer #3: The investigators have presented the long-term (48 months) health-related and cost outcomes for an economic intervention with HIV/AIDs related orphans. The findings have the potential to contribute significantly to the evidence on savings-led programs with adolescents and sustained impact and cost. I have several suggestions for improvement. The methods section is lacking in important details related to setting, sample, self-report measures (psychometrics, reliability of measures) used with adolescents, including data collection procedures, and brief description (perhaps a table) to detail the differences between Bridges and BridgesPlus, including the savings match ($) contributed - was this provided by the research or was in part of a government protection scheme? Details on retention of the sample over the 48 month and any difference by gender would have been useful to the findings. Also, was their a dose response - did the youth need to attend all the activities in both intervention approaches to receive the incentives? I know the original study has been reported elsewhere (24 months) but context is important for understanding the findings and the impact. For example, it is not clear how hopelessness is defined for the study. Further, the analysis section and results should be reviewed by a health economist or someone with stronger skills than mine for interpreting the results. I found the result as presented difficult to understand - mainly because it is not clear what the BridgesPlus received in comparison to Bridges -therefore, for me it is difficult to determine if the additional cost of BridgesPlus worth the noted impact in knowledge and hopelessness. Qualitative interviews from the participants would be very useful for a nuanced understanding of the impact on youth - would also be interesting to have a better understanding of the income-generation or educational impacts given the savings could only be used in those two area - why are these results not presented? I do think it is important that the investigators positioned the research and noting the lack of cost data for comparison. Additionally, it would have been interesting to include in discussion the use of cash-transfer or savings as government programs for protection of vulnerable youth, etc.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: Yes: Greg Fegan

Reviewer #3: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files to be viewed.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at figures@plos.org. Please note that Supporting Information files do not need this step.

Submitted filename: PLOS Manuscript_07.24.2019_GFcomments.docx

Click here for additional data file.

Submitted filename: PLOS_Supplementary Material_07.24.2019_GF.docx

Click here for additional data file.

14 Oct 2019

Reviewer #1: The manuscript is generally well written and presented. However there some minor limitations for publishing.

1. The core question being addressed is not specifically stated or clearly linked to the extensive analyses provided. This could be due to the fact that the authors assume the audience is clear in regards to the intent of the trail, as well as of he actual project itself. The authors would greatly improve this issue by providing the readers with both a slightly elaborated description of the overall project, more on the underlying rationale for a focus on cost-effectiveness of such projects in global low income and rural settings, especially those with a history of social and cultural upheaval.

We thank the reviewer for this comment. On Pages 5, Lines 24-31, we added the following explanation on the rationale of the trial and the uniqueness of FEE interventions: “Grounded in asset theory, family economic empowerment (FEE) interventions go beyond cash transfers by providing capacity building, mentorship, and seed funding and fostering household financial stability through promoting income generating activities and financial literacy. Asset theory posits that orphaned adolescents will experience higher levels of depression, have worse education outcomes, and engage in higher risk behaviors if they lack financial means to participate in secondary education. FEE interventions that bring financial stability to households act as a protective factor.”

In terms of contextualizing the importance of cost-effectiveness analysis, on Page 5, Lines 37-41, we clarified its rationale as follows: “To make the case for public investment in these interventions, it is necessary to integrate the evidence on effects with costs. By calculating an incremental cost per unit of benefit, cost-effectiveness analyses make it explicit which interventions will contribute the most relative to their costs and inform resources allocation decisions in the face of competing health priorities and resource constraints.”

Additionally, we provided more detail on the trial population, the eligibility criteria, the study setting and the study measures on Page 7 Lines 78-83; Page 7 Lines 87-88, Page 7 Lines 93-95 and Pages 8, Lines 114-127

2. The retrospective focus of the analyses does not take into account the contextual and process variables that have also influenced the observed outcomes. This is one limitation of a single paradigm (quantitative) focus and self-reported behavioral outcomes. The authors need to better address the potential influence of these factors - what was provided was highly inadequate.

Thanks for your comment. We agree with the reviewer that qualitative data would be helpful to understand contextual and process variables that have also influenced the observed outcomes, as well as the limitation in self-reported behavioral outcomes. Please refer to limitation section Page 22, Lines 369-375 where we highlighted this as “Use of self-reported measures is the most common approach to assessment in low resource settings, but this approach is limited by self-report bias. This study also relied on self-reported measures for physical and mental health and sexual risk-taking behavior. Therefore, our findings are limited to self-reported changes in these outcomes rather than a clinical diagnosis, for instance, for mental health outcomes. Future studies should consider incorporating biomarkers and other objective measures.” Also, on Page 22, Lines 365-369 we further added that “Another limitation is that because our study was quantitative by design, we did not have qualitative data on adolescent perceptions of the program, which could have added nuance and context to better elucidate the mechanisms of the effectiveness of the program on youth. Future qualitative research is needed to understand the causal pathways of effectiveness of such family economic strengthening programs.” In addition, we also want to clarify this paper presents a cost-effectiveness analysis where costs are not retrospective, but reflect the actual expenditure captured in real time.

3. In such project the personnel and setting are important to any analyses since both are critical to the functional environments (e.g, relationship with youth, physical settings, so forth). While recognized as not a part of the present study, some mention of contextual variables was both expected and needed.

Thank you for this comment. We have added details to clarify the study setting, including the eligibility criteria, that read” Those schools are in four political districts of Rakai, Masaka, Lwengo and Kalungu in South Western Uganda—a region heavily affected by HIV/AIDS. There were three eligibility criteria: 1) the adolescent had lost one or both parents to HIV/AIDS, 2) the adolescent was enrolled in grades 5 or 6, in a government-aided primary school during recruitment period, 3) the adolescent was living with a family, not an institution.” Please refer to Page 7, Lines 78-83. We also clarified the personnel for various activities on Page 7, Line 88 and Lines 93-94.

4. Lastly, operational definitions of the key variables of health and others were not provided. This needs to be addressed.

Thank you for this comment. We have added a measurement paragraph that reads “Self-rated health is measured by a single 5-item scale ranging from excellent to very poor with higher values indicate better health. Mental health functioning is conceptualized as depression, hopelessness, self-concept and self-efficacy. We used 27-item Child Depression Inventory (CDI) and the 20-item Beck Hopelessness Scale (BHS) as operationalization with higher scores indicate worse mental health. Moreover, 20-item Tennessee Self-Concept Scale, and 29-item Youth Self-Efficacy Survey are used for measurements of self-concept and self-efficacy with higher values on indicate more positive self-concept and self-efficacy. Three indictors— sexual risk-taking intentions, HIV prevention intention, and HIV knowledge are used to capture sexual health. Detailed example questions for each scale are provided in table S2 with reliability statistics. Briefly, CDI (α = .68) and BHS (α = .65) show modest internal consistency, whereas Tennessee Self-Concept Scale, Youth Self-Efficacy, sexual risk-taking intentions, HIV prevention intention, and HIV knowledge all have acceptable internal consistency with Cronbach’s alpha greater than .70. All measurements mentioned above are standardized to be comparable across all outcomes.” Please refer to Pages 8, Lines 114-127. We also added references to S2 Table available in supplementary material.

Overall, a good study. If the limitations are addressed this manuscript will provide a means to contribute our knowledge of such projects as suggested.

Reviewer #2: I have made comments and raised queries in the attached track-changed word documents.

However I would like to add two specific things here that I'd expect the authors to address 1) They need to note that as shown in the supplementary tables specifically Table S1 there was a marked higher attrition by 48 months for the BridgesPlus group (11.2%) as opposed to both Bridges and Control (8.8%) groups compared to the numbers at baseline.

Thank you for this comment. We did a chi-square test and the attrition rate did not differ by study arms (χ2(2) = 2.04; p = 0.36). We have added language to the manuscript that reads “The attrition rate was 8.8% for Bridges and control group and 11.2% for the BridgesPlus group at the 48-month follow-up. However, the results from a chi-square test suggests that the attrition rates do not differ by study arms (χ2(2) = 2.04; p = 0.36). Please refer to Pages 7-8, Lines 99-112. Further, we did analyses to compare if the retained and attritted subsamples differ by demographic factors. Please refer to S2 Table, Descriptive statistics on characteristics of attritted sample in supplementary material.

2) There should be some mention of the fact that they are using the same exposure and co-variate data to test multiple hypotheses and some readers may wonder whether there is a multiplicity problem here requiring a lower P-value to be used. Both of these should be highlighted in a small limitations sub-section within the discussion.

Thank you for this comment. We added text to the limitations paragraph addressing this specific issue that reads “A limitation of this study is that multiple hypotheses were tested for multiple outcomes using the same exposure and covariate data, potentially leading to inflated Type 1 error rates. Future studies seeking to replicate our findings might consider concentrating on the subset outcomes found to be statistically significant in this study and employing multiplicity adjustments to further control Type 1 error rates.” Please refer to Page 22, Lines 361-369.

Reviewer #3: The investigators have presented the long-term (48 months) health-related and cost outcomes for an economic intervention with HIV/AIDs related orphans. The findings have the potential to contribute significantly to the evidence on savings-led programs with adolescents and sustained impact and cost. I have several suggestions for improvement.

● The methods section is lacking in important details related to setting, sample, self-report measures (psychometrics, reliability of measures) used with adolescents, including data collection procedures,

Thank you for this comment. We have added details regarding the setting and the sample that read “Those schools are in four political districts of Rakai, Masaka, Lwengo and Kalungu in southwestern Uganda—a region heavily affected by HIV/AIDS. There were three eligibility criteria: 1) the adolescent had lost one or both parents to HIV/AIDS, 2) the adolescent was enrolled in grades 5 or 6, in a government-aided primary school during recruitment period, 3) the adolescent was living with a family, not an institution.” (Page 7, Lines 80-83). We also added a new sub-section on Measures that reads: “Self-rated health is measured by a single 5-item scale ranging from excellent to very poor with higher values indicate better health. Mental health functioning is conceptualized as depression, hopelessness, self-concept and self-efficacy. We used the 27-item Child Depression Inventory (CDI) and the 20-item Beck Hopelessness Scale (BHS) operationalized with higher scores indicating worse mental health. Moreover, the 20-item Tennessee Self-Concept Scale, and 29-item Youth Self-Efficacy Survey are used for measurements of self-concept and self-efficacy with higher values on indicate more positive self-concept and self-efficacy. Three indictors— sexual risk-taking intentions, HIV prevention intention, and HIV knowledge are used to capture sexual health. Detailed example questions for each scale are provided in table S2 with reliability statistics. Briefly, CDI (α = 0.68) and BHS (α = 0.65) show modest internal consistency, whereas Tennessee Self-Concept Scale, Youth Self-Efficacy, sexual risk-taking intentions, HIV prevention intention, and HIV knowledge all have acceptable internal consistency with Cronbach’s alpha greater than 0.7. All measurements mentioned above are standardized to be comparable across all outcomes.” (Pages 8, Lines 114-127). Also, please refer to S2 Table (Appendix 1, Page 5) for sample questions, psychometrics and reliability of all the measurements and Page 7, Lines 95-97 for more detail on the data collection procedures.

● and brief description (perhaps a table) to detail the differences between Bridges and BridgesPlus, including the savings match ($) contributed - was this provided by the research or was in part of a government protection scheme?

Thank you for this comment. Table 2, on Page 16, presents the breakdown of intervention costs in all three study arms, including the matched contribution amounts in the two intervention arms. Table 2 also specifies the activities carried out in each arm and the total costs per participant. For example, it is possible to see that participants in the three arms received counselling, but only participants in the two treatment arms took part in income generating activities.

On Page 7, Lines 98-99, we added the following clarification: “The matched funds were contributed by the trial (See Table 2 for program costs, including matched amounts per intervention arm).”

● Details on retention of the sample over the 48 month and any difference by gender would have been useful to the findings.

Thanks for your comment. We further did analyses to compare if the retained and attritted samples differ by demographic variables. Please refer to S2 Table in Appendix 1. Descriptive statistics on characteristics of the attritted sample are given in the supplementary material.

● Also, was their a dose response - did the youth need to attend all the activities in both intervention approaches to receive the incentives?

Thank you for this question. The participants do not need to attend all the activities to receive the incentives. To receive the incentives, they need to 1) attend the Financial Literacy Training 2) open an account at a participating financial institution. We added these details on Page 7, Lines 95-97.

● I know the original study has been reported elsewhere (24 months) but context is important for understanding the findings and the impact. For example, it is not clear how hopelessness is defined for the study.

Thanks for this comment. We have added a new subsection on Measures on Page 8, Lines 114-127 that reads “Self-rated health is measured by a single 5-item scale ranging from excellent to very poor with higher values indicate better health. Mental health functioning is conceptualized as depression, hopelessness, self-concept and self-efficacy. We used the 27-item Child Depression Inventory (CDI) and the 20-item Beck Hopelessness Scale (BHS) operationalized with higher scores indicating worse mental health. Moreover, the 20-item Tennessee Self-Concept Scale, and the 29-item Youth Self-Efficacy Survey are used for measurements of self-concept and self-efficacy with higher values on indicate more positive self-concept and self-efficacy. Three indictors— sexual risk-taking intentions, HIV prevention intention, and HIV knowledge are used to capture sexual health. Detailed example questions for each scale are provided in table S2 with reliability statistics. Briefly, CDI (α = 0.68) and BHS (α = 0.65) show modest internal consistency, whereas the Tennessee Self-Concept Scale, the Youth Self-Efficacy Survey, sexual risk-taking intentions, HIV prevention intention, and HIV knowledge all have acceptable internal consistency with Cronbach’s alpha greater than .70. All measurements mentioned above are standardized to be comparable across all outcomes.” We also added references to S3 Table in Appendix 1 of the supplementary material.

● Further, the analysis section and results should be reviewed by a health economist or someone with stronger skills than mine for interpreting the results. I found the result as presented difficult to understand - mainly because it is not clear what the BridgesPlus received in comparison to Bridges -therefore, for me it is difficult to determine if the additional cost of BridgesPlus worth the noted impact in knowledge and hopelessness.

Table 2, on Page 16, details what participants received in each of the three study arms, the breakdown of the costs per participant, including the difference in matched contributions received by Bridges and BridgesPLUS participants, as well as the total cost per participant. Specifically, Table 2 details that the program contributed a total of 20,309 Ugandan Shillings to Bridges participants and 42,634 Ugandan Shillings to BridgesPlus participants. The intervention arms, Bridges and BridgesPLUS, are described in more detail on Page 7, Lines 93-97.

● Qualitative interviews from the participants would be very useful for a nuanced understanding of the impact on youth - would also be interesting to have a better understanding of the income-generation or educational impacts given the savings could only be used in those two area - why are these results not presented?

Thank you for your comment. We agree with the reviewer that qualitative interviews from the participants would be very useful. Unfortunately, our study was quantitative by design. We acknowledge this as a limitation on Page 22, Lines 365-369 that reads: “Another limitation is that because our study was quantitative by design, we did not have qualitative data on adolescent perceptions of the program, which could have added nuance and context to better elucidate the mechanisms of the effectiveness of the program on youth. Future qualitative research is needed to understand the causal pathways of effectiveness of such family economic strengthening programs.”

Regarding income-generating and educational impacts, we investigated in other papers: (1) Ssewamala, F. M., Wang, J. S. H., Neilands, T. B., Bermudez, L. G., Garfinkel, I., Waldfogel, J., ... & Kirkbride, G. (2018). Cost-effectiveness of a savings-led economic empowerment intervention for AIDS-affected adolescents in Uganda: Implications for scale-up in low-resource communities. Journal of Adolescent Health, 62(1), S29-S36; (2) Wang, J. S. H., Ssewamala, F. M., Neilands, T. B., Bermudez, L. G., Garfinkel, I., Waldfogel, J., ... & You, J. (2018). Effects of Financial Incentives on Saving Outcomes and Material Well‐Being: Evidence From a Randomized Controlled Trial in Uganda. Journal of Policy Analysis and Management, 37(3), 602-629. This paper is focused on health-related outcomes and a cost-effectiveness analysis.

● I do think it is important that the investigators positioned the research and noting the lack of cost data for comparison.

Thank you for this comment. The lack of costing and cost-effectiveness studies on FEE interventions are highlighted in the manuscript on Page 5, Lines 37-41, and Pages 21, Lines 342-346.

● Additionally, it would have been interesting to include in discussion the use of cash-transfer or savings as government programs for protection of vulnerable youth, etc.

We agree with the reviewer. We added a full paragraph on conditional cash transfers in the discussion section on Pages 20-21 (Lines 325-339) as follows: “Cash transfer programs, both conditional and unconditional, targeted at most vulnerable households have been shown to impact a multiplicity of education and health outcomes in low resource-settings. In Malawi, the government-run cash transfer program targeting ultra-poor, labor-constrained households increased school enrollment and reduced drop-out rates and improved mental health outcomes in children after one year. In Kenya, the government-led program for orphans and vulnerable children, aimed at reducing HIV risk through unconditional cash transfers, reduced the odds of sexual debut by 31%. Structural interventions to reduce HIV incidence have been found to be the most cost-effective, particularly when targeting multiple outcomes, such as education, in addition to HIV/AIDS incidence. A privately funded program in Malawi conditioned on continued school enrollment of adolescent girls reduced the prevalence of HIV, Herpes simplex virus and depression and improved school attendance in this population. At 18 months, the intervention was estimated to avert 209 disability-adjusted life years (DALYs) at a cost of $297 per DALY averted. Despite the growing literature on cash transfer programs, few studies have focused on interventions that include an economic empowerment component.”

Submitted filename: Response to Reviewers_101019_final.docx

Click here for additional data file.

9 Dec 2019

Evaluation of a savings-led family-based economic empowerment intervention for AIDS-affected adolescents in Uganda: A four-year follow-up on efficacy and cost-effectiveness

PONE-D-19-20904R1

Dear Dr. Ssewamala,

We are pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it complies with all outstanding technical requirements.

Within one week, you will receive an e-mail containing information on the amendments required prior to publication. When all required modifications have been addressed, you will receive a formal acceptance letter and your manuscript will proceed to our production department and be scheduled for publication.

Shortly after the formal acceptance letter is sent, an invoice for payment will follow. To ensure an efficient production and billing process, please log into Editorial Manager at https://www.editorialmanager.com/pone/, click the "Update My Information" link at the top of the page, and update your user information. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to enable them to help maximize its impact. If they will be preparing press materials for this manuscript, you must inform our press team as soon as possible and no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

With kind regards,

Christopher M Doran, BEc (Hons) PhD

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

Reviewer #2: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: (No Response)

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: (No Response)

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: (No Response)

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: (No Response)

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: I have read both the original submission and the updated submission with track changes. In the latter the authors provided extensive and detailed responses to each comment. The responses were clear and specific - this was expected, allowing the authors to be very targeted and strengthened the manuscript.

The oversall result was a much imporved and substantively informative manuscripr that will contrbute to our understanding of the challenges of developing and implementing programs deisgned to address wicked problems faced by children and their families in LICs and similar.

Reviewer #2: (No Response)

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: Yes: Greg Fegan

12 Dec 2019

PONE-D-19-20904R1

Evaluation of a savings-led family-based economic empowerment intervention for AIDS-affected adolescents in Uganda: A four-year follow-up on efficacy and cost-effectiveness

Dear Dr. Ssewamala:

I am pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please notify them about your upcoming paper at this point, to enable them to help maximize its impact. If they will be preparing press materials for this manuscript, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

For any other questions or concerns, please email plosone@plos.org.

Thank you for submitting your work to PLOS ONE.

With kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Professor Christopher M Doran

Academic Editor

PLOS ONE