Skip to main content
NCBI home page
PLOS Global Public Health logoLink to PLOS Global Public Health
. 2023 Aug 17;3(8):e0001666. doi: 10.1371/journal.pgph.0001666

Assessing the cost-effectiveness of economic strengthening and parenting support for preventing violence against adolescents in Mpumalanga Province, South Africa: An economic modelling study using non-randomised data

William E Rudgard 1,2,*,#, Sopuruchukwu Obiesie 1,#, Chris Desmond 3, Marisa Casale 1,4, Lucie Cluver 1,5
Editor: Hannah Hogan Leslie6
PMCID: PMC10434898  PMID: 37590179

Abstract

There is limited evidence around the cost-effectiveness of interventions to reduce violence against children in low- and middle-income countries. We used a decision-analytic model to evaluate the cost-effectiveness of three intervention scenarios for reducing adolescent emotional, physical, and sexual abuse in Mpumalanga Province, South Africa. The intervention scenarios were: 1) Community grant outreach to link households to South Africa’s Child Support Grant (CSG) if they are eligible, but not receiving it; 2) Group-based parenting support; and 3) Group-based parenting support ‘plus’ linkage to the CSG. We estimated average cost-effectiveness ratios (ACERs) for intervention scenarios over a ten-year time horizon, and compared them to a South Africa-specific willingness-to-pay (WTP) threshold (USD3390). Health effects were expressed in disability-adjusted life years (DALYs) averted. Our model considered four combinations of routine service versus trial-based costing, and population-average versus high prevalence of violence. Under routine service costing, ACERs for grant outreach and parenting support were below the WTP threshold when considering a population-average prevalence of violence USD2850 (Lower: USD1840-Upper: USD10,500) and USD2620 (USD1520-USD9800) per DALY averted, respectively; and a high prevalence of violence USD1320 (USD908-USD5180) and USD1340 (USD758-USD4910) per DALY averted, respectively. The incremental cost-effectiveness of parenting support plus grant linkage relative to parenting support alone was USD462 (USD346-USD1610) and USD225 (USD150-USD811) per DALY averted at a population-average and high prevalence of violence, respectively. Under trial-based costing, only the ACER for grant outreach was below the WTP threshold when considering a high prevalence of violence USD2580 (USD1640-USD9370) per DALY averted. Confidence intervals for all ACERs crossed the WTP threshold. In conclusion, grant outreach and parenting support are likely to be cost-effective intervention scenarios for reducing violence against adolescents if they apply routine service costing and reach high risk groups. Combining parenting support with grant linkage is likely to be more cost-effective than parenting support alone.

Introduction

Annually, 1.4 billion children are estimated to experience violence or neglect worldwide, leading to more than 5.1 million disability-adjusted life years (DALYs) [1, 2]. The World Health Organization (WHO) and eight other global agencies have endorsed seven evidence-based strategies to address violence against children, known collectively as INSPIRE [3]. These seven strategies include Implementation and enforcement of laws, Norms and values, Safe environments, Parent and caregiver support, Income and economic strengthening, Response and support services, and Education and life skills [3]. Further evidence is needed to help stakeholders prioritise strategies that are likely to have the greatest return on investment [4].

A previous study to guide the choice of ‘best-buy’ INSPIRE strategies in South Africa found that across seven INSPIRE-aligned protective factors, food security, caregiver supervision, and positive caregiving were the most promising targets for addressing multiple forms of violence against adolescents simultaneously [5]. Consistent with positive youth development theories, these three protective factors also combined additively, such that experiencing two or three of them together was associated with a significantly lower probability of experiencing multiple forms of violence, compared to experiencing one of them alone [5]. Research in Tanzania also suggests that parenting support and economic strengthening are likely to combine synergistically to reduce violence against children more than each intervention alone [6].

Evidence on the cost-effectiveness of interventions for reducing violence against children is essential for maximising the health gains from available resources [7]. However, economic evaluations of violence prevention interventions remain uncommon, particularly in low- and middle-income countries [8–11]. Experimental studies are recommended for evaluating cost-effectiveness, but they are expensive and complex, especially when more than one intervention is under consideration [12–14]. Decision analytic modelling is a cheaper alternative approach that uses secondary data sources, including published trials and observational studies, to model the expected costs and consequences of decision options [9, 12].

In this study, we aimed to use secondary data to estimate the cost-effectiveness of a selected list of evidence-based interventions to reduce violence against adolescents [15, 16]. We had three objectives: 1) Select expert- and evidence-based interventions for reducing violence against adolescents via improving food security, caregiver supervision, or positive caregiving in South Africa; 2) Gather together cost and effectiveness data relating to selected interventions from a variety of high-quality secondary data sources; 3) Combine these data in a decision-analytic model to estimate the cost-effectiveness of selected interventions.

Methods

This economic modelling study builds on a non-randomised analysis of seven INSPIRE-aligned protective factors among adolescents in South Africa, which found that food security, caregiver supervision, and positive caregiving were associated with lower odds of multiple forms of adolescent violence victimisation including, emotional abuse, physical abuse, sexual abuse, and community violence victimisation [5]. Our methodological approach involved three steps. First, we consulted with regional experts on preventing violence against adolescents to identify the best candidate interventions for reducing violence by improving food security, caregiver supervision, and positive caregiving in South Africa. Second, we drew together cost and effectiveness data for the selected interventions from published literature, online survey data, and further expert consultation. Third, we combined these data into a decision-analytic model to estimate the cost per DALY averted of our selected interventions. This choice enabled us to generate a single estimate of cost-effectiveness for selected interventions, but also limited us to only considering adolescent violence victimisation outcomes with evidence for their attributable DALYs in the published literature. Of the four kinds of adolescent violence victimisation investigated in the non-randomised analysis in South Africa, we found data on the DALYs attributable to emotional, physical, and sexual abuse, but not community violence victimization [5]. We include full details of data sources and assumptions throughout the analysis and report our study using the Consolidated Health Economic Evaluation Reporting Standard (CHEERS) checklist, S1 Checklist [17, 18].

Study setting

The study setting was Mpumalanga Province, South Africa where intervention costs and adolescents’ experience of violence are expected to lie between the other two provinces where data were collected for the non-randomised analysis of seven INSPIRE-aligned protective factors in South Africa; the poorer Eastern Cape Province, and richer Western Cape Province [5, 19].

Study population

The study population was households with a monthly income per capita below the South African upper poverty line and an adolescent aged between 10 and 19 years [20–22]. In South Africa, the upper poverty line is approximately equal to the means-test threshold used to assess applicant’s eligibility for the national Child Support Grant (CSG).

Hypothesised scenarios for violence prevention

We consulted experts on violence prevention to identify the best candidate interventions for reducing adolescent violence via improving food security, caregiver supervision, and positive caregiving in South Africa. Our focus on these protective factors was based on evidence of their association with a lower probability of multiple forms of violence against adolescents in South Africa [5]. Criteria for candidate interventions were that (i) there should be existing or early implementation within South Africa; and ii) evidence supporting their impact on either food security or positive and supervisory caregiving in Southern Africa. Key details on intervention implementation, staffing, and duration are provided in S1 Table [5].

Scenario 1: Community outreach programme to link households that are eligible but not receiving South Africa’s CSG [Grant outreach]

It is estimated that around 18% of children eligible for anti-poverty social grants in South Africa do not access this support [23, 24]. Regional evidence supports the effectiveness of cash grants in addressing food security [25, 26], and evidence from South Africa supports the potential of community initiatives to link eligible households to social grants [27]. Therefore, in our analysis, we modelled the impact of a community outreach programme implemented over 17-months during which paraprofessional social workers would actively liaise with community leaders and networks to screen for households that may be eligible for the CSG but don’t already receive it. Eligible households not already receiving the CSG would then be supported to access it.

Scenario 2: Parenting support programme based on WHO/UNICEF’s Parenting for Lifelong Health (PLH) programme [Parenting support]

Across South Africa, parenting support programmes are offered by a variety of non-profit organisations, including in Mpumalanga Province, South Africa as part of the Mothers2Mothers (M2M) Children and Adolescents are My Priority (CHAMP) project [28, 29]. Evidence also supports the potential of parenting support programmes to improve caregiver supervision in South Africa [30]. In our analysis, we thus modelled the roll-out of a parenting support programme akin to the Parenting for Lifelong Health Teen (PLH Teen), which was co-founded with WHO and UNICEF, and is structured as a 14-session intervention for small family groups. Participants are recruited into the programme by self or community-referral using the two screening questions: ‘do you and your teen argue and shout a lot?’ and ‘do you sometimes end up hitting your teen when things are really stressful?’ [30, 31].

Scenario 3: Integrated parenting support programme plus component to link households that are eligible but not receiving South Africa’s CSG [Parenting support plus grant linkage]

In this combined scenario, we sought to explore the additive benefits of an intervention that acts on two protective factors. We modelled the roll-out of a parenting support plus grant linkage programme, structured similarly to the 14-session PLH Teen programme but including one additional session during which facilitators would assess families for their eligibility for inclusion in South Africa’s CSG. Families found to be eligible but not receiving the CSG would be linked to the relevant social services to support them in accessing the grant. This parenting support plus grant linkage intervention would be distinct from the community grant outreach intervention, as it would focus on families that attend the parenting support intervention rather than the broader community.

Choice of model

To estimate the cost-effectiveness of selected scenarios, we constructed a probability tree model that modelled i) the effect of selected interventions on food security and/or caregiver supervision, and ii) the corresponding effect of the estimated improvement in these intermediary protective factors on adolescent violence victimisation, Fig 1. This type of model was a simple and logical way for modelling the effect of selected interventions on violence outcomes via intermediary protective factors. We did not model parenting support or parenting support ‘plus’ grant linkage acting via positive caregiving as there is mixed evidence for whether group-based parenting support programmes improve this intermediary protective factor. Also, we did not model parenting support as reducing sexual abuse as there was no evidence that higher caregiver supervision was associated with this violence outcome in the observational analysis of seven INSPIRE-aligned protective factors that informed this study [5]. With these considerations, we modelled grant outreach acting on emotional, physical, and sexual abuse via food security, parenting support acting on emotional and physical abuse via caregiver supervision, and parenting support plus grant linkage acting on emotional, physical, and sexual abuse via both food security and caregiver supervision.

Fig 1. Probability tree model for estimating the effect of intervention scenarios on adolescent violence outcomes via intermediary protective factors.

Fig 1

Open in a new tab

1Estimate synthesized across two studies examining the impact of grants on household food security in sub-Saharan Africa and crosschecked using data from South Africa’s 2018 General Household Survey. 2Estimate obtained from the primary analysis of a randomised evaluation of PLH Teen in South Africa. 3This tree applies only to a subset of adolescents from households in the parenting plus grant linkage scenario who, having been identified as eligible and not currently receiving the CSG. These adolescents would benefit from enhanced caregiver supervision and food security. The remaining adolescents would only benefit from enhanced caregiver supervision as in scenario two.

Data sources

Model inputs were drawn from published literature, online survey data, and expert consultation, Table 1, S2 Table. Estimates of the percentage of adolescents in Mpumalanga living below the upper poverty line were derived from South Africa’s 2018 General Household Survey [32]. Estimates of the prevalence of emotional, physical, and sexual abuse among adolescents were obtained from the non-randomised analysis of seven INSPIRE-aligned protective factors for adolescent violence in South Africa [5].

Table 1. Summary of key model inputs, their description, and source.

  Value Description Sources
Mpumalanga Province  
    Total population 4,523,000 Mid-year population estimate 2018, Stats SA
    Adolescent population 18% Mid-year population estimate 2018, Stats SA
    Adolescents living below the UPL 35% UPL = Monthly household income per capita < USD 79 South Africa GHS 2018, Stats SA
    Number of adolescents per household 1.03 South Africa GHS 2018, Stats SA
Number of children per household 2.56 South Africa GHS 2018, Stats SA
Intervention scenario 1: Grant outreach  
    Eligible children receiving a CSG 1,105,791 Children eligible for the CSG are <18 years SASSA, 2019
    Eligible children not receiving CSG 18% UNICEF, 2016; South Africa GHS 2018, Stats SA
    Number of household visits needed to identify eligible adolescent 3 Expert consultation
    Success rate in linking eligible adolescents to CSG 70% Expert consultation; Thurman et al. 2015
    Number of household visits, per paraprofessional social worker, per day 3 Travel by public transport/ walking Expert consultation
    Intervention duration 17 months Implementation is ongoing until all eligible adolescents not receiving CSG have been visited Derived based on intervention staffing and number of household visits needed to identify eligible adolescent not receiving CSG
Staff  
    Paraprofessional social workers 1000 Human resources needed to link 6% of eligible households per month Staffing needs were based on a target of linking 6% of eligible adolescents to the CSG per month, and a hierarchical organisational structure similar to PLH Teens. Data from the PLH Teen trial were obtained through personal communication with Dr Jamie Lachmann
    Supervisors 50 1 per 20 paraprofessional social workers
    District coordinators 3 1 per district
Provincial representatives 1 1 per province
Effectiveness
    Effect on food security1 1.22 (1.06; 1.34) RR (95% CI) Meta-analysis of cash transfer effects on household food security Handa et al. 2022 and Bhalla et al. 2018
    Effect on adolescent violence  
        Emotional abuse 0.96 (0.93; 0.99) RR (95% CI) Observational analysis of protective factors and adolescent violence by Cluver & Rudgard et al. 2020
        Physical abuse 0.97 (0.95; 0.99)
        Sexual abuse 0.94 (0.89; 0.98)
Cost, USD  
    Staff training, per staff 38; 62 Routine service; Trial-based estimate Unit costs were based on data from the PLH Teen trial obtained through personal communication with Dr Jamie Lachmann, and unit quantities were based on expert consultation
    Community outreach, per household visit 62; 113
Intervention scenario 2: Parenting support  
    Eligible families targeted, per year 5% Based on PLH Teen scale-up in Thailand and Tanzania PLH Teen data obtained through personal communication with Dr Jamie Lachmann
    Family uptake success rate 90% We expect 5% lower compliance in a real-world setting than in a trial Randomised evaluation of PLH Teen by Cluver et al. 2018
    Number of families, per family group 15; 25 Routine service; Trial-based estimate PLH Teen data obtained through personal communication with Dr Jamie Lachmann
    Duration of implementation 10 years Implementation is continuous over the study time horizon  
Staff  
    Facilitators 91; 151 Routine service; Trial-based estimate
6; 6 family groups per facilitator per year
10; 10 facilitators per coach
20; 20 facilitators per coordinator
5; 10 facilitators per assistant coordinator
1 district per district coordinator
1 province per provincial representative		 Staffing needs were based on the hierarchical organisational structure used by PLH Teen. Data on PLH Teen staffing were obtained through personal communication with Dr Jamie Lachmann
    Coaches 9; 15
    Coordinators 5; 8
    Assistant coordinators 9; 30
    District coordinators 3; 3
Provincial representatives 1; 1
Effectiveness  
    Effect on caregiver supervision 1.21 (2.09; 0.32) Effect size (95% CI) Randomised evaluation of PLH Teen by Cluver et al. 2018
    Effect on adolescent violence  
        Emotional abuse 0.95 (0.91; 0.99) RR (95% CI) Observational analysis of protective factors and adolescent violence by Cluver & Rudgard et al. 2020
        Physical abuse 0.97 (0.94; 0.99)
        Sexual abuse1 -
Cost, USD  
    Staff training, per staff 112; 365 Routine service; Trial-based estimate PLH Teen data obtained through personal communication with Dr Jamie Lachmann
    Delivery cost, per family group 33; 201
    Office space and equipment, per year 6300; 38,000
Intervention scenario 3: Parenting support plus grant linkage
    Eligible families targeted, per year 5% Based on PLH Teen scale-up in Philippines and Thailand PLH Teen data obtained through personal communication with Dr Jamie Lachmann
    Family uptake success rate 90% We expect 5% lower compliance in a real-world setting than in a trial Randomised evaluation of PLH Teen by Cluver et al. 2018
    Number of families, per family group 15; 25 Routine service; Trial-based estimate PLH Teen data obtained through personal communication with Dr Jamie Lachmann
    Duration of implementation 10 years Implementation is continuous over the study time horizon  
    Eligible children not receiving CSG 18% UNICEF, 2016; South Africa GHS 2018, Stats SA
    Success rate in linking eligible adolescents to CSG 70% Expert consultation; Thurman et al. 2015
Effectiveness  
    Effect on caregiver supervision 1.21 (0.32; 2.09) Effect size (95% CI) Randomised evaluation of PLH Teen by Cluver et al. 2018
    Effect on food security 1.22 (1.06; 1.34) RR (95% CI) Meta-analysis of cash transfer effects on household food security Handa et al. 2022 and Bhalla et al. 2018
    Effect on adolescent violence  
        Emotional abuse 0.95 (0.84; 0.97) RR (95% CI) Observational analysis of protective factors and adolescent violence by Cluver & Rudgard et al. 2020
        Physical abuse 0.94(0.89; 0.98)
        Sexual abuse 0.89 (0.82; 0.97)
Staff  
    Facilitators 91; 151 Routine service; Trial-based estimate
6; 6 family groups per facilitator per year
10; 10 facilitators per coach
20; 20 facilitators per coordinator
5; 10 facilitators per assistant coordinator
1 district per district coordinator
1 province per provincial representative		 Staffing needs were based on the hierarchical organisational structure used by PLH Teen. Data on PLH Teen staffing were obtained through personal communication with Dr Jamie Lachmann
    Coaches 9; 15
    Coordinators 5; 8
    Assistant coordinators 9; 30
    District coordinators 3; 3
    Provincial representatives 1; 1
Cost, USD  
    Staff training, per staff 190; 370 Routine service; Trial-based estimate PLH Teen data obtained through personal communication with Dr Jamie Lachmann
    Delivery cost, per family group 35; 208
    Office space and equipment, per year 6600; 39,400
Open in a new tab

1Violence outcome was not considered for the parenting support scenario, as there was no evidence that it was significantly associated with caregiver supervision. Abbreviations: UPL, upper poverty line; Stats SA, Statistics South Africa; CSG, Child Support Grant; RR, relative risk; CI, confidence interval; GHS, General Household Survey; USD, United States dollar; PLH, Parenting for Lifelong Health

Intervention effects

Estimates of intervention effectiveness were drawn from the published literature. For community grant outreach, we generated a pooled estimate for the effect of cash transfers on food security by meta-analysis of two studies identified using a rigorous literature review, S1 Text [25, 26]. For parenting support, we used estimates of the effect of the PLH Teen intervention on caregiver supervision taken from the original randomised evaluation in South Africa [30]. For parenting support plus grant linkage, we used our pooled estimate for the effect of cash transfers on food security and our estimate of the effect of PLH Teen on caregiver supervision. Finally, for the effect of enhanced levels of food security and caregiver supervision on adolescent violence victimisation, we used estimates from the non-randomised analysis of seven INSPIRE-aligned protective factors for reducing adolescent violence in South Africa [5]. We modelled food security and caregiver monitoring as combining additively to reduce adolescent violence victimisation, since there was no evidence in the original analysis that they combine multiplicatively with respect to violence against adolescents [5].

Intervention costs

All three scenarios were costed from the provider’s perspective using the ingredients method. Unit costs and quantities for parenting support and parenting support plus grant linkage interventions were based on data from the PLH Teen trial in South Africa and PLH Teen implementation in Thailand and Tanzania [6, 30, 33]. We were unable to find cost data on an intervention similar to the modelled grant outreach intervention, so unit costs were based on the PLH Teen trial and unit quantities based on expert consultation. We did not consider the operational costs of delivering the CSG in our analysis (i.e. the value of grants, overheads) as it was agreed with experts that these costs are already budgeted for by the South African government in its annual projection for the expected number of households that are eligible to receive the grant.

Data management

All costs captured before 2021 were adjusted for inflation and converted to USD using the conversion rate of USD 1 = South African Rand 14.93.

Data analysis

We modelled the costs and DALYs due to violence victimisation averted for the three intervention scenarios over a time horizon of ten years, as recommended by WHO-CHOICE guidelines.

Modelling the effectiveness of interventions

For each intervention scenario, first, we used our probability tree model to estimate the relative risk of emotional, physical, and sexual abuse comparing the probability of these outcomes in the presence and absence of cash transfers and/ or parenting support, Table 1. We used the 95% confidence intervals around relative risk ratios to estimate upper and lower bounds around our main effects. Second, we estimated the cumulative number of cases averted due to intervention activities, accounting for an annual dropout rate from grant outreach (2%), an attrition rate from parenting support (3%), an annual ageing out effect for grant outreach (13%) and parenting support (11%), and an annual decay in the intervention’s effects on food security (3%) and caregiver monitoring (33%) [34–37]. Effects of receiving the CSG were modelled to cease when adolescents turned 18 and became ineligible to receive this service. Third, we converted estimated subtotals of cases of violence averted into DALYS averted and summed across them to estimate the total DALYS averted for each intervention. DALYs attributable to emotional, physical, and sexual abuse were calculated from the only known evaluation of the economic consequences of violence against children in South Africa, and estimates of the prevalence of violence against adolescents in South Africa, S3 Table [11, 38, 39]. For full details on this method, see Redfern et al., 2019 [11].

Modelling the cost of interventions

For each intervention scenario, we estimated the total cost of modelled intervention activities over ten years. For grant outreach, this was the cost of a one-off 17-month community outreach initiative to link households that are eligible to receive South Africa’s CSG but do not currently receive it. The grant outreach programme would end once all eligible but excluded households had been visited. For parenting support and parenting support plus grant linkage, this was the total cost of sequential rounds of 14-week parenting support programmes over ten years without and with grant linkage, respectively. For all three scenarios, we applied an annual discount rate of 4% to costs, as recommended for economic evaluations in upper middle-income countries [40]. We also estimated the cost of interventions using ‘routine service’ and ‘trial-based’ ingredient costing. These two costings targeted the same number of families but varied in staff salaries, with the latter providing significantly higher salaries, and budgeting for venue hire, food for participants, printed training materials, and communication with participants via mobile phones. Routine service costing was based on daily salary rates equivalent to South Africa’s average monthly earnings in March 2021, while trial-based costing was based on daily salary rates three times higher than this [41].

Modelling the cost-effectiveness of interventions

We calculated average cost-effectiveness ratios (ACERs) for intervention scenarios by dividing intervention total costs by total DALYs averted. We calculated incremental cost-effectiveness ratios (ICERs) for parenting support plus grant linkage relative to standalone parenting support by dividing the two scenarios difference in cost by the difference in DALYs due to violence victimisation averted [12, 18]. Estimated cost-effectiveness ratios and their upper and lower bounds were evaluated against a South African-specific willingness-to-pay (WTP) threshold estimated by Edoka and Stacey, equivalent to USD 3390 [42].

We calculated cost-effectiveness ratios for all three of our interventions under routine service and trial-based costings, and at two estimates of the prevalence of adolescent violence outcomes. These were ‘population-average’ prevalence, which was equal to the rates observed in our source reference [5]; and ‘high’ prevalence, which was equal to double the rates observed in our source reference [5]. The rationale for modelling a high prevalence of violence was that adolescents targeted by our grant-outreach intervention (i.e., eligible but not receiving the CSG), and our parenting support and parenting support plus grant linkage interventions (i.e., screened for regular arguments and/or physical abuse against children) are likely to experience higher than average levels of vulnerability and adolescent violence victimisation.

Robustness checks

First, we evaluated the robustness of our findings to using South Africa’s GDP per capita for 2021 as our WTP threshold [18, 43]. Second, in the absence of a published evaluation of the effect of South Africa’s CSG on food security, we also checked the robustness of our findings for the cost-effectiveness of community grant outreach, and parenting support plus grant linkage to a country-specific estimate for the relationship between income and food security estimated using South Africa’s 2018 General Household Survey [44]. Further details of this secondary analysis are included in S2 Text.

Ethics

This study used secondary data published in the public domain throughout the analysis.

Results

Population estimates for Mpumalanga Province and the expected number of beneficiaries for the three intervention scenarios are summarized in Table 2. We estimated that approximately 434,000 adolescents are living below the poverty line in Mpumalanga. Under the grant outreach intervention scenario, 57,100 adolescents living in 55,500 households would be linked to the CSG. Under the parenting support intervention scenario, 66,900 adolescents living in 65,300 households would attend parenting support. Under the parenting support plus grant linkage scenario, 66,900 adolescents living in 65,300 households would attend parenting support, and 8200 of these would also be linked to the CSG.

Table 2. Summary of provincial population estimates and the expected number of beneficiaries for each intervention scenario based on model inputs.

Estimate
Mpumalanga Province
Adolescents 667,000
Adolescents living below the UPL 434,000
Adolescents receiving the CSG 374,000
Intervention scenario 1: Grant outreach
Eligible adolescents not receiving the CSG 77,000
Eligible adolescents linked to CSG 57,100
Eligible households linked to CSG 55,500
Intervention scenario 2: Parenting support
Adolescents attending parenting support 66,900
Households attending parenting support 65,300
Intervention scenario 3: Parenting plus grant linkage
Adolescents attending parenting support 66,900
Households attending parenting support 65,300
Eligible adolescents not receiving the CSG 11,700
Eligible adolescents linked to CSG 8200
Open in a new tab

All estimates are rounded to three significant figures.

Abbreviations: UPL, upper poverty line; CSG, Child Support Grant.

Estimated intervention effects

Estimates for the number of averted cases of violence victimisation and averted DALYS from the grant outreach, parenting support, and parenting support plus grant linkage scenarios are reported in Table 3. Our model estimated that through improvements in food security, the grant outreach scenario would avert 1180 (Lower: 320; Upper: 1830) and 2360 (649; 3700) DALYs attributable to emotional, physical, and sexual abuse, at population-average and high prevalence of violence, respectively. Through improvements in caregiver monitoring, the parenting support scenario would avert 995 (254; 1640) and 1910 (509; 3300) DALYs attributable to emotional and physical abuse, at population-average and high prevalence of violence, respectively. Through improvements in caregiver supervision and food security, the parenting support and grant linkage would avert 1150 (310; 1900) and 2310 (620; 3900) DALYs attributable to emotional, physical, and sexual abuse, at population-average and high prevalence of violence, respectively.

Table 3. Summary of the estimated cases of violence victimisation, DALYs averted, costs, and cost-effectiveness ratios of intervention scenarios over a ten-year time horizon.

Averted cases of violence1 (Lower; Upper) DALYs averted (Lower; Upper) Total cost per adolescent, USD Total cost provincial, USD ACER (Lower; Upper) ICER (Lower; Upper)
Routine service costing
    Population-average prevalence of violence
        Grant outreach 4540 (1250; 6970) 1180 (320; 1830) 59 3,360,000 2850 (1840; 10,500) 2850 (1840; 10,500)
        Parenting support 3700 (987; 6350) 955 (254; 1640) 36 2,500,000 2620 (1520; 9800) 2620 (1520; 9800)
        Parenting support plus grant linkage 4470 (1200; 7500) 1150 (310; 1900) 38 2,590,000 2250 (1363; 8350) 462 (346; 1610)
    High prevalence of violence
        Grant outreach 9100 (2500; 13,920) 2360 (649; 3700) 59 3,360,000 1420 (908; 5180) 1420 (908; 5180)
        Parenting support 7410 (1970; 12,700) 1910 (509; 3300) 36 2,500,000 1320 (758; 4910) 1320 (758; 4910)
        Parenting support plus grant linkage 8900 (2400; 15,000) 2310 (620; 3900) 38 2,590,000 1120 (664; 4180) 225 (150; 811)
Trial-based costing
    Population-average prevalence of violence
        Grant outreach 4540 (1250; 6970) 1180 (320; 1830) 106 6,080,000 5150 (3320; 19,000) 5150 (3320; 19,000)
        Parenting support 3700 (987; 6350) 955 (254; 1640) 215 14,900,000 15,600 (9090; 58,700) 15,600 (9090; 58,700)
        Parenting support plus grant linkage 4470 (1200; 7500) 1150 (310; 1900) 223 15,400,000 13,400 (8110; 49,700) -
    High prevalence of violence
        Grant outreach 9100 (2500; 13,920) 2360 (649; 3700) 106 6,080,000 2580 (1640; 9370) 2580 (1640; 9370)
        Parenting support 7410 (1970; 12,700) 1910 (509; 3300) 215 14,900,000 7800 (4520; 29,300) 7800 (4520; 29,300)
        Parenting support plus grant linkage 8900 (2400; 15,000) 2310 (620; 3900) 223 15,400,000 6670 (3950; 24,800) -
Open in a new tab

All estimates are rounded to three significant figures. We did not estimate ICERs for parenting support plus grant linkage when the ACER for parenting support alone was above our WTP threshold of USD3390.

1Forms of violence for grant outreach and parenting support plus grant outreach include emotional, physical, and sexual abuse, and for parenting support include emotional and physical abuse.

Abbreviations: DALY, disability-adjusted life years; USD, United States dollar; ACER, average cost-effectiveness ratio; ICER, incremental cost-effectiveness ratio.

Estimated intervention costs

Estimates for the costs associated with intervention scenarios are reported in Table 3. A more detailed breakdown of costs is also provided in S4 Table. Grant outreach was estimated to cost USD 59 and USD 106 per adolescent, and USD 3,360,000 and USD 6,080,000 at the provincial level, under routine service and trial-based costing, respectively. Parenting support was estimated to cost USD 36 and USD 215 per adolescent beneficiary, and USD 2,500,000 and USD 14,900,000 at the provincial level, under routine service and trial-based costing, respectively. The major cost drivers were staff salaries (80% and 54% of total costs for routine service and trial-based, respectively), and food for participants and facilitators (0% and 21% of total costs under the rountine service and trial-based, respectively). Parenting support plus grant linkage was estimated to cost USD 38 and USD 223 per adolescent, and USD 2,590,000 and USD 15,400,000 at the provincial level, under routine service and trial-based costing, respectively.

Intervention cost-effectiveness

ACERs and ICERS for intervention scenarios are summarised in Table 3. ACERS are also summarised against a South African-specific WTP threshold in Fig 2. An intervention is considered cost-effective if it’s ACER lies below the WTP threshold of USD 3390. The further an ACER lies to the right of the plane while remaining below the WTP threshold the higher its indicated cost-effectiveness.

Fig 2.

Fig 2

Open in a new tab

Cost-effectiveness plane scatter plots of costs per adolescent over effectiveness (in DALYs averted) for four scenarios at A) Routine service costing and population-average prevalence of violence, B) Routine service costing and high prevalence of violence, C) Trial-based costing and population-average prevalence of violence, and D) Trial-based costing and population-high prevalence of violence. Diagonal lines indicate an evidence-based SA-specific willingness to pay threshold for health interventions estimated by Edoka and Stacey, 2020. Strategies become increasingly cost-effective as they lie to the north easternmost part of the cost-effectiveness plane below the willingness to pay threshold. Strategies above the line are not cost-effective at the stated threshold. Abbreviations: DALYs, disability-adjusted life years; USD, United States dollar.

Under routine service costing

At a population-average prevalence of violence, ACERs for grant outreach, parenting support, and parenting support plus grant linkage scenarios were all below the WTP threshold at USD 2850 (1840; 10,500), USD 2620 (1520; 9800), and USD 2250 (1363; 8350) per DALY averted, respectively. For all three scenarios, upper estimates of ACERs crossed the WTP threshold. The parenting support plus grant linkage scenario had an ICER of 225 (150; 811) per DALY averted compared to the parenting support alone scenario.

At a high prevalence of violence, ACERs for grant outreach, parenting support, and parenting support plus grant linkage scenarios were all below the WTP threshold at USD 1420 (908; 5180), USD 1320 (758; 4910), and USD 1120 (664; 4180) per DALY averted, respectively. For all three scenarios, upper estimates of ACERs crossed the WTP threshold. The parenting support plus grant linkage scenario had an ICER of USD 225 (150; 811) per DALY averted compared to the parenting support alone scenario.

Under trial-based costing

At a population-average prevalence of violence, ACERs for grant outreach, parenting support, and parenting support plus grant linkage scenarios were all above the WTP threshold at USD 5150 (3320; 19,000), USD 15,600 (9090; 58,700), and USD 13,400 (8110; 49,700) per DALY averted, respectively. For the grant outreach scenario, the lower ACER estimate crossed the WTP threshold. Since the ACER for parenting support alone was above our WTP threshold, we did not estimate the ICER for parenting support plus grant linkage.

At a high prevalence of violence, the ACER for grant outreach scenario was below the WTP threshold at USD 2580 (1640; 9370) per DALY averted, but the ACERs for parenting support and parenting support plus grant linkage scenarios were above the WTP threshold at USD 7800 (4520; 29,300) and USD 6670 (3950; 24,800) per DALY averted. For the grant outreach scenario, the upper ACER estimate crossed the WTP threshold. Since the ACER for parenting support alone was above our WTP threshold, we did not estimate the ICER for parenting support plus grant linkage.

Sensitivity analysis and robustness check

When using the WHO CHOICE WTP threshold based on South African GDP per capita instead of a South Africa-specific WTP threshold, results for routine service costing were the same except that at a high prevalence of violence, the lower estimates for ACERs did not cross the WTP threshold. Results for trial-based costing were also the same except that at a population-average prevalence of violence, the ACER for grant outreach was below the WTP threshold, S1 Fig. Substituting our pooled estimate for the effect of cash transfers on food security with an estimate of the relationship between income and food security from South Africa’s 2018 General Household Survey, the findings for parenting support plus grant linkage remain consistent, but grant outreach ceases to be cost-effective, S2 Fig.

Discussion

We find that decision-analytic modelling is a valuable and low-cost approach for assessing the cost-effectiveness of three expert- and evidence-informed interventions aimed at reducing adolescent violence victimisation in Mpumplanga, South Africa. Our model indicates that so long as routine service costing is used, investments in grant outreach, and parenting support may be cost-effective for reducing adolescent violence victimisation. Wide confidence intervals around our effect estimates highlight a need for further research before any strong conclusions are drawn. We also find that under routine service costing, adding a grant linkage component to parenting support is likely to be more cost-effective than parenting support alone. Across the three interventions, the most cost-effective option is likely to be the community grant outreach intervention. We find that if trial-based costing is used for interventions, parenting support and parenting support plus grant linkage are unlikely to be cost-effective at either a population-average or high prevalence of violence, and community grant outreach may only be cost-effective when the prevalence of violence is high among beneficiary households.

The findings from our analysis provide early evidence around the cost-effectiveness of a community outreach intervention to support households’ access to social grants if they are eligible but not receiving them. The cost of the proposed grant outreach intervention per adolescent is equivalent to the value of three child support grants, and falls on the lower end of costs for home-visiting interventions in high-income settings [45]. Our findings around the cost-effectiveness of parenting support implemented using routine service costing match a previous economic evaluation of the original PLH Teen intervention in South Africa [11]. In that study, under routine service costing, the cost per DALY averted of parenting support was estimated to be USD 2650 [11]. Our assumption that the modelled routine service intervention is just as effective as the trial-based intervention, is supported by evidence from a pre- and post-test evaluation of PLH Teens implemented with routine service costing [46]. The increasing roll-out and scale-up of parenting support programmes across sub-Saharan Africa will present further opportunities to validate this [47]. The previous economic evaluation of PLH Teen found that parenting support implemented using trial-based costing may be more cost-effective than we estimated in our model [11]. Our study is likely to have underestimated the effects of parenting support by only modelling one impact pathway, caregiver supervision. Other evidence-based impact pathways include improved caregiver mental health, caregiver alcohol/drug avoidance, and improved economic welfare [48]. We also modelled a more conservative prevalence of adolescent violence among families than the previous economic evaluation [11].

Evidence of the enhanced effectiveness of combining multiple interventions with a ‘plus’ approach has grown substantially in recent years. However, there have been few attempts to evaluate the cost-effectiveness of these interventions [49, 50]. Our findings of the incremental cost-effectiveness of combining an additional grant linkage session with parenting support further validates the added value of integrated services for reducing violence against adolescents. Building on this, future research should consider other possible approaches for combining parenting support with an economic strengthening intervention [51–53].

Our modelling study enabled us to consider important questions around the design of three interventions for reducing violence against adolescents. These included how best to build on existing social policies such as South Africa’s CSG, the resources needed for implementing interventions at scale, and approaches for combining two interventions to simultaneously promote food security and caregiver supervision [3]. Study strengths included its significantly lower cost compared to running a randomised evaluation of the three evaluated interventions, and also our use of a widely applicable methodology that could be used to generate similar evidence for other interventions and/or settings. Our study also had limitations. While many of our model parameters were drawn from high-quality randomised studies, estimates of the association between protective factors and violence outcomes were based on observational research and may be affected by sources of bias associated with this type of research design, including unmeasured confounding [5]. Our focus on the impact of interventions via food security and caregiver supervision alone did not consider the full complement of protective factors via which selected interventions could act on adolescent violence victimisation. There is evidence that parenting support is also likely to act on adolescent violence via improving caregiver mental health, caregiver alcohol/drug avoidance, and improved economic welfare [48, 54, 55]. Gaps in the academic literature also meant that we had to make assumptions about some of the parameters in our model, for example, the post-intervention decay in the effectiveness of cash grants and parenting support programmes. Several methodological choices also mean that our estimates of cost-effectiveness are likely to be conservative. These include our focus on DALYs averted for measuring cost-effectiveness, which while allowing us to generate a single estimate of cost-effectiveness for selected interventions, also meant we could not consider community violence victimisation in our model. Due to a lack of evidence in the wider literature, we were also unable to consider the wider societal benefits of preventing adolescent violence. Such benefits might include reduced health service use, social service use, and court case time. Finally, all three of the interventions considered in this study are likely to have benefits for adolescent development beyond reducing violence victimisation, for example by boosting school enrolment or promoting mental health. Impacts across these additional domains of development should be considered for accurately valuing intervention cost-effectiveness in the future. Doing so may also support cross-sectoral buy-in from multiple government departments, including health, education, and social development [56].

In South Africa, 40% of young people are estimated to experience some form of emotional, physical, or sexual abuse in their lives [57]. Reviewing and analysing the literature on violence against adolescents in South Africa, our study provides novel and policy-relevant evidence for much-needed action to reduce these forms of abuse against adolescents. Our findings suggest that all three of our selected interventions may be cost-effective so long as they are implemented using routine service costing and reach adolescents with the highest risk of violence victimisation. The study also provides a comprehensive summary of evidence gaps that should be investigated as a priority for further informing the scale-up of interventions in this field. These include limited estimates of the prevalence of adolescent violence victimisation among the most vulnerable groups, the effects of cash transfers on adolescent violence victimisation [58, 59], the mechanisms via which cash transfers act on this outcome, and the wider benefits of reducing violence victimisation to communities and society. Future attempts to quantify the cost-effectiveness of social interventions should also account for their effects on other areas of adolescent development, including education and health.

Conclusion

Investments in grant outreach, and parenting support are most likely to be cost-effective for reducing adolescent violence victimisation if they are provided using routine service costing and they reach adolescents at high risk of violence victimisation. We also find that under routine service costing, adding a grant linkage component to parenting support is likely to be more cost-effective than parenting support alone. Further research to account for all of the pathways via which interventions act to reduce adolescent violence victimisation is necessary to strengthen our confidence in the findings.

Supporting information

S1 Checklist. CHEERS checklist.

(DOCX)

Click here for additional data file. (28KB, docx)
S1 Fig

Cost-effectiveness plane scatter plots of costs per adolescent over effectiveness (in DALYs averted) for four scenarios with the SA GDP per capita as the willingness-to-pay threshold at: A) Routine service costing and population-average prevalence of violence, B) Routine service costing and high prevalence of violence, C) Trial-based costing and population-average prevalence of violence, and D) Trial-based costing and population-high prevalence of violence.

(DOCX)

Click here for additional data file. (170.7KB, docx)
S2 Fig. Cost-effectiveness plane scatter plots of costs per adolescent over effectiveness (in DALYs averted) for four scenarios using estimates for the relationship between household income and food security drawn from secondary analysis of the South Africa’s 2018 GHS.

The SA-specific threshold per capita is used as the willingness-to-pay threshold at: A) Routine service costing and population-average prevalence of violence, B) Routine service costing and high prevalence of violence, C) Trial-based costing and population-average prevalence of violence, and D) Trial-based costing and population-high prevalence of violence.

(DOCX)

Click here for additional data file. (212.5KB, docx)
S1 Table. Description of hypothesised interventions, their duration, and staffing structure.

(DOCX)

Click here for additional data file. (24.2KB, docx)
S2 Table. Summary of data sources used to estimate the effectiveness and cost of grant outreach, parenting support, and parenting support plus grant linkage.

(DOCX)

Click here for additional data file. (29.4KB, docx)
S3 Table. Estimated DALY per case of physical, emotional, and sexual abuse in South Africa.

(DOCX)

Click here for additional data file. (23.3KB, docx)
S4 Table. Detailed breakdown of the cost of grant outreach, parenting support, and parenting support plus grant linkage in United States dollars.

(DOCX)

Click here for additional data file. (25.1KB, docx)
S1 Text. Synthesis of the effect of cash grants on household food insecurity in sub-Saharan Africa.

(DOCX)

Click here for additional data file. (32KB, docx)
S2 Text. Analysis of the relationship between household income and food security in South Africa using data from the 2018 General Household Survey.

(DOCX)

Click here for additional data file. (28.5KB, docx)

Acknowledgments

We thank Dr Jamie Lachman for providing us with costing data from past implementations of parenting programmes. We thank Dr Yulia Shenderovich for providing data from the parenting for Lifelog Health (PLH Teen) trial. We also thank Gloria Khoza (UNICEF South Africa) and Mpume Danisa (Clowns without Borders South Africa), whose experience in scaling up parenting programmes in South Africa provided valuable insight into the intervention scenarios during the early phases of this work.

Data Availability

All relevant data are within the paper and its supporting information files.

Funding Statement

Research reported in this publication was supported by the UK Research and Innovation Global Challenges Research Fund (UKRI GCRF) Accelerate Hub [ES/S008101/1 to LC]; the Oak Foundation [OFIL-20-057 to LC]; Wellspring Philanthropic Fund [Grant No. 16204 to LC]; and Oak Foundation/GCRF "Accelerating Violence Prevention in Africa" [R46194/AA001 to LC]. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. For the purpose of Open Access, the author has applied a CC BY public copyright licence to any Author Accepted Manuscript version arising from this submission.

References

PLOS Glob Public Health. doi: 10.1371/journal.pgph.0001666.r001

Decision Letter 0

Hannah Hogan Leslie

6 Mar 2023

PGPH-D-23-00203

Assessing the cost-effectiveness of economic strengthening and parenting support for preventing violence against adolescents in South Africa: An economic modelling study using non-randomised data.

PLOS Global Public Health

Dear Dr. Rudgard,

Thank you for submitting your manuscript to PLOS Global Public Health. After careful consideration, we feel that it has merit but does not fully meet PLOS Global Public Health’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, ensuring that all modeling decisions are carefully justified and fully described.

Please submit your revised manuscript by Apr 20 2023 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at globalpubhealth@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pgph/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

  • A rebuttal letter that responds to each point raised by the editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

  • A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

  • An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

We look forward to receivi

ng your revised manuscript.

Kind regards,

Hannah Hogan Leslie, PhD

Academic Editor

PLOS Global Public Health

Journal Requirements:

1. Please insert an Ethics Statement at the beginning of your Methods section, under a subheading 'Ethics Statement'. It must include:

1) The name(s) of the Institutional Review Board(s) or Ethics Committee(s)

2) The approval number(s), or a statement that approval was granted by the named board(s) 

3) (for human participants/donors) - A statement that formal consent was obtained (must state whether verbal/written) OR the reason consent was not obtained (e.g. anonymity). NOTE: If child participants, the statement must declare that formal consent was obtained from the parent/guardian.

Additional Editor Comments (if provided):

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

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Does this manuscript meet PLOS Global Public Health’s publication criteria? Is the manuscript technically sound, and do the data support the conclusions? The manuscript must describe methodologically and ethically rigorous research with conclusions that are appropriately drawn based on the data presented.

Reviewer #1: Partly

Reviewer #2: No

**********

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

Reviewer #1: N/A

Reviewer #2: No

**********

3. Have the authors made all data underlying the findings in their manuscript fully available (please refer to the Data Availability Statement at the start of the manuscript PDF file)?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception. 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: No

Reviewer #2: Yes

**********

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

PLOS Global Public Health 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

**********

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: Reviewer #: Article Number: PGPH-D-23-00203.

Article Title: Assessing the cost-effectiveness of economic strengthening and parenting support for preventing violence against adolescents in South Africa: An economic modelling study using non-randomised data.

In this article, the authors used a decision-analytic model to assess the cost-effectiveness of three interventions (linking eligible households to anti-poverty cash grants, group-based parenting support and a combination of the two interventions - group-based parenting support ‘plus’ linkage of eligible households to anti-poverty cash grants) to prevent violence against children in low- and middle-income countries. Using modelling to assess the cost-effectiveness of interventions to prevent violence against children is of great importance as experimental studies, though recommended are costly and complex especially if more than one intervention is being studied. The evaluations of costs and effects presented by the authors can assist public health authorities.

The model that authors use is suitable for the issue addressed, however the authors did not give full details of the model which makes it difficult to understand how the results were obtained. They mention model parameters in the document, but the parameters are not listed in one table which makes it difficult for the reader.

My main considerations are as follows:

Title: The title indicates that the study is on cost-effectiveness of economic strengthening and parenting support for preventing violence against adolescents in South Africa. However, the study is based on data from Mpumalanga province one of the nine provinces in South Africa. Though the authors indicate that Mpumalanga province lie between (in terms of its economic characteristics) the other two provinces Eastern and Western Cape, I feel that the population and economic characteristics in the Mpumalanga province is not a representative of the population in South Africa as a country hence results obtained may not be generalizable to South Africa. I suggest editing the title to reflect that this study is for Mpumalanga province in South Africa.

Introduction: The introduction is quite long (I suggest editing this section). Lines 52-62 can be in the discussion section.

Line 50 to 51 – I suggest deleting this sentence in the introduction. Rather have it in the discussion section as a limitation to this study.

Method section:

1. The use the word intervention (Lines 112, 122 and 132) is confusing. I suggest using scenario instead of intervention. Each scenario will focus on an intervention or a combination of interventions.

2. Fig 1. Graphical representation of probability tree model for estimating the effectiveness of interventions on adolescent violence outcomes – more information is needed in this figure. The structure of a tree is not evident. No probabilities are shown in the tree diagram. To enhance understanding of the methods section all necessary information must appear in the tree diagram or must be well explained in the main text. As it stands now it is not easy to replicate this study. This section is the backbone of this research hence it is necessary to show all the information. Alternatively share references where detailed information about the probability tree model can be found.

3. It might be necessary to justify the discount rate used in this analysis.

Results section:

1. Line 241 – authors mention that the estimates are for 10 years of intervention. However, there is no period stated for results presented before this line. Does this mean that grant outreach intervention was done over 10 years?

2. Line 248 – authors say that grant outreach intervention would be expected to reduce the probability of sexual abuse by -0.54ppts (-0.86; -0.16). If the word reduce is used, the negative sign must be deleted. Also edit the sentences after line 248 (all instances where the word reduce/decrease is used).

3. There are values quoted in the text which do not appear in the main article tables (see lines 293 to 294). I suggest adding all values discussed in the manuscript in tables appearing in the main text.

Discussion section:

1. The low intervention scenario is only mentioned in the discussion section. No mention of such scenario is in the methods section. I suggest adding information on all scenarios in the methods section.

Minor Essential Revisions

1. Page 3 Affiliations: The 4th affiliation is not linked to any author. Please edit

2. Line 25 Abstract: delete “only”

3. Line 27 to 28 Abstract: Last sentence is confusing – cost-effectiveness analysis show that combining the two interventions is second to linking households to grants cost-effective. How does adding a grant linkage component to parenting support enhance cost-effectiveness.

4. Line 95 Study setting: Add “South Africa” after “Mpumalanga province”

5. Line 164 Fig 1: I suggest making the + sign a superscript.

6. Line 272 Parenting support: It should be USD36 (as in Table 3) instead of USD37.

7. Line273-274 Parenting support: It’s not clear where the following values are coming from “(USD 275,700 and 1,640,000 annually)”.

8. Line 280: Value (2 617 000) is different from what is in Table 3.

Clear labelling for all figures is required. Generally, grammar is good despite minor grammatical errors picked in the document. Reading through the document after final revision is recommended.

Supplementary file

Authors are commended for submitting a comprehensive supplementary file. I have a few comments to make

1. CHEERS list:

� Choice of model – this point is not fully addressed in the manuscript. See Method section comments above.

� Study parameters – not all parameters are in Table 1.

2. Page 11 Use ‘INSPIRE’ instead of ‘Inspire’

3. Since the study is based on Mpumalanga province. Is it necessary to list provincial aOR’s for all provinces?

4. S8: Which one is the lean and ideal scenario? In the main text authors use trial and routine service delivery. The Low effect scenario is not described in the manuscript.

Reviewer #2: The authors conducted a cost-effectiveness analysis of the individual and joint effects of two interventions (outreach+linkage to cash transfer and parenting support) on violence experiences among South African adolescents. I enjoyed reading the paper and feel it makes interesting contributions to support violence prevention policy and prioritization. However, I identified several weaknesses, some fairly major, and points of clarification that should be addressed, as outlined below:

1. The violence domains the authors explore in this paper could be better and more explicitly defined in the introduction/methods. The introduction mentions adolescents ‘affected by violence,’ but the methods and discussion section more fluidly mixes a bunch of violence outcomes: victimization (e.g. physical and sexual abuse), perpetration (e.g. youth lawbreaking), and an item that is unclear in its mapping to violence (transactional sex). It becomes clearer later on that the authors are only looking at emotional, physical, and sexual abuse, but this seems to be a practical restriction rather than one made based on domains/modes of intervention efficacy. Suggest more clearly defining area of outcome focus throughout. Specifically the mention of perpetration outcomes and transactional sex in the methods and discussion is confusing.

2. Importantly, Interventions 1 and 3 (grant outreach) are only able to assist households that do not already access the Child Support Grant, but this is only a small percentage of the target population. It also seems like the models are parameterized with intervention effects of the cash transfer. So there is a mixture of interventions and parameterization here that is confusing. I think the CSG is actually the intervention you want to model, in target populations with varying levels of pre-existing uptake. It seems that this is also the cost that is most relevant: the cost of the actual CSG program, not just the cost to link people to the program.

3. Adolescents age 18 and 19 are not eligible for the Child Support Grant (coverage only goes until age 18), but the target population includes adolescents age 11-19?

4. The Child Support Grant is modeled as acting on violence outcomes via food security, which seems rather narrow. The CSG could also plausibly work through, for example, education, mental health, and stress pathways, among others. Similar point is probably true for parenting interventions. Do you allow any of these interventions to work through other pathways besides the limited protective factors specified (food security and positive parenting)? It seems a major underestimate to restrict these holistic interventions to only their effects through narrow intermediate pathways and not their broader set of direct and indirect effects.

5. On page 7: “For parenting support plus grant linkage, we used our pooled estimate for the effect of cash transfers on food security and our effect estimate for caregiver supervision from the PLH Teen evaluation.” It’s unclear to me whether this pooled estimate incorporates any potential interaction between the two interventions or just adds their effect together?

6. The 10 year implementation timeline strikes me as a bit odd – people in your target population would age in and out of adolescence and out of CSG eligibility over that timeframe.

7. How is it possible that the grant outreach alone prevents more cases of violence than the grant outreach + parenting support (Table 3)?

8. The conversion of ‘Averted cases of violence’ to DALYs is not presented clearly in the results section.

9. The conclusions in the discussion “Our model indicates that implementing each of our three selected interventions using a routine service delivery model is likely to be cost-effective for reducing emotional, physical, and sexual abuse against adolescents” and the abstract “Findings indicate that investments in community grant outreach, and parenting support interventions are likely to be cost-effective for preventing adolescent violence” don’t seem to align with the findings as reported in the results section. The results seemed to indicate that, by and large, the interventions were not cost-effective (below the WTP threshold for all except under assumptions of high violence)? So I’m not clear on how these conclusions were arrived at.

Minor:

10. The language around ‘incurring’ DALYs in the intro seems off. My understanding is that they are a measure of years of healthy life lost – suggest updating the language to better align with its definition.

11. Small typos and missing words throughout – needs copyediting.

**********

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.

Do you want your identity to be public for this peer review? If you choose “no”, your identity will remain anonymous but your review may still be made public.

For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: 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.]

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 PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

PLOS Glob Public Health. doi: 10.1371/journal.pgph.0001666.r003

Decision Letter 1

Hannah Hogan Leslie

10 Jul 2023

Assessing the cost-effectiveness of economic strengthening and parenting support for preventing violence against adolescents in Mpumalanga Province, South Africa: An economic modelling study using non-randomised data.

PGPH-D-23-00203R1

Dear Mr. Rudgard,

We are pleased to inform you that your manuscript 'Assessing the cost-effectiveness of economic strengthening and parenting support for preventing violence against adolescents in Mpumalanga Province, South Africa: An economic modelling study using non-randomised data.' has been provisionally accepted for publication in PLOS Global Public Health. Please note two minor requests for clarification below as you prepare for publication.

Before your manuscript can be formally accepted you will need to complete some formatting changes, which you will receive in a follow up email. A member of our team will be in touch with a set of requests.

Please note that your manuscript will not be scheduled for publication until you have made the required changes, so a swift response is appreciated.

IMPORTANT: The editorial review process is now complete. PLOS will only permit corrections to spelling, formatting or significant scientific errors from this point onwards. Requests for major changes, or any which affect the scientific understanding of your work, will cause delays to the publication date of your manuscript.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they'll be preparing press materials, please inform our press team as soon as possible -- 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 globalpubhealth@plos.org.

Thank you again for supporting Open Access publishing; we are looking forward to publishing your work in PLOS Global Public Health.

Best regards,

Hannah Hogan Leslie, PhD

Academic Editor

PLOS Global Public Health

***********************************************************

Please address the minor points raised on re-review in the process of preparing the manuscript for publication.

Reviewer Comments (if any, and for reference):

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. Does this manuscript meet PLOS Global Public Health’s publication criteria? Is the manuscript technically sound, and do the data support the conclusions? The manuscript must describe methodologically and ethically rigorous research with conclusions that are appropriately drawn based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

**********

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

Reviewer #1: N/A

Reviewer #2: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available (please refer to the Data Availability Statement at the start of the manuscript PDF file)?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception. 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 Global Public Health 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

**********

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: The authors are commended for addressing all the comments and making detailed revisions to the manuscript. The clarity and organisation of the manuscript has significantly improved. The introduction provides a clearer background and rationale for the study. The Methods section has been revised to include more details on the model and the probability tree. Tables and figures in the manuscript and supplementary file are clear and appropriately labelled, aiding in the understanding of the analysis. The authors have added information on the limitations of the study. I recommend acceptance of manuscript pending two minor comments/revisions outlined below:

1. Line 86 to 88: “Of the four violence outcomes investigated in the non-randomised analysis in South Africa, this included emotional, physical, and sexual abuse, but excluded community violence victimization (5).” This sentence needs revision. I think there is something missing after the word “this” maybe “…this study included….”

2. Table 1 – Authors have 1,105,791 in the value column against Eligible children receiving a CSG but 18% of the Total population is 814 140 – (first two rows in the table which should be the total Adolescent population in Mpumalanga). How is it then possible to have 1,105,791 eligible children receiving a CSG. May the authors have a second look on these data. I might be interpreting the data incorrectly.

Other than the minor comments above I recommend acceptance of manuscript

Reviewer #2: This response was a pleasure to read. The authors have thoroughly responded to all prior comments and I have no further concerns.

**********

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.

Do you want your identity to be public for this peer review? If you choose “no”, your identity will remain anonymous but your review may still be made public.

For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: Yes: Molly Rosenberg

**********

Associated Data

    This section collects any data citations, data availability statements, or supplementary materials included in this article.

    Supplementary Materials

    S1 Checklist. CHEERS checklist.

    (DOCX)

    Click here for additional data file. (28KB, docx)
    S1 Fig

    Cost-effectiveness plane scatter plots of costs per adolescent over effectiveness (in DALYs averted) for four scenarios with the SA GDP per capita as the willingness-to-pay threshold at: A) Routine service costing and population-average prevalence of violence, B) Routine service costing and high prevalence of violence, C) Trial-based costing and population-average prevalence of violence, and D) Trial-based costing and population-high prevalence of violence.

    (DOCX)

    Click here for additional data file. (170.7KB, docx)
    S2 Fig. Cost-effectiveness plane scatter plots of costs per adolescent over effectiveness (in DALYs averted) for four scenarios using estimates for the relationship between household income and food security drawn from secondary analysis of the South Africa’s 2018 GHS.

    The SA-specific threshold per capita is used as the willingness-to-pay threshold at: A) Routine service costing and population-average prevalence of violence, B) Routine service costing and high prevalence of violence, C) Trial-based costing and population-average prevalence of violence, and D) Trial-based costing and population-high prevalence of violence.

    (DOCX)

    Click here for additional data file. (212.5KB, docx)
    S1 Table. Description of hypothesised interventions, their duration, and staffing structure.

    (DOCX)

    Click here for additional data file. (24.2KB, docx)
    S2 Table. Summary of data sources used to estimate the effectiveness and cost of grant outreach, parenting support, and parenting support plus grant linkage.

    (DOCX)

    Click here for additional data file. (29.4KB, docx)
    S3 Table. Estimated DALY per case of physical, emotional, and sexual abuse in South Africa.

    (DOCX)

    Click here for additional data file. (23.3KB, docx)
    S4 Table. Detailed breakdown of the cost of grant outreach, parenting support, and parenting support plus grant linkage in United States dollars.

    (DOCX)

    Click here for additional data file. (25.1KB, docx)
    S1 Text. Synthesis of the effect of cash grants on household food insecurity in sub-Saharan Africa.

    (DOCX)

    Click here for additional data file. (32KB, docx)
    S2 Text. Analysis of the relationship between household income and food security in South Africa using data from the 2018 General Household Survey.

    (DOCX)

    Click here for additional data file. (28.5KB, docx)
    Attachment

    Submitted filename: Response to reviewers.docx

    Click here for additional data file. (63.3KB, docx)

    Data Availability Statement

    All relevant data are within the paper and its supporting information files.

    Articles from PLOS Global Public Health are provided here courtesy of PLOS

    RESOURCES