Effect and cost-effectiveness of human-centred design-based approaches to increase adolescent uptake of modern contraceptives in Nigeria, Ethiopia and Tanzania: Population-based, quasi-experimental studies

Around half of adolescent pregnancies in low- and middle-income countries are unintended, contributing to millions of unsafe abortions per year. Adolescents 360 (A360), a girl-centred initiative, aimed to increase voluntary uptake of modern contraceptives among adolescents in Nigeria, Ethiopia and Tanzania. We evaluated the effectiveness and cost-effectiveness of A360 in increasing modern contraceptive use in selected geographies. We used before-and-after cross-sectional studies of adolescent girls in four settings. Two Nigerian settings had purposefully selected comparison areas. Baseline and endline household surveys were conducted. The primary study outcome was modern contraceptive prevalence rate (mCPR). Secondary outcomes mapped onto the A360 Theory of Change. Interpretation was aided by a process evaluation along with secular mCPR trends and self-reported A360 exposure data. Incremental design and implementation costs were calculated from implementer systems, site visits, surveys, and interviews. mCPR change was modelled into maternal disability-adjusted life years (DALY) averted to calculate incremental cost-effectiveness ratios. In Oromia, Ethiopia, mCPR increased by 5% points (95% CI 1–10; n = 1,697). In Nigeria, there was no evidence of an effect of A360 on mCPR in Nasarawa (risk ratio: 0·96, 95% CI: 0·76–1·21; n = 5,414) or in Ogun (risk ratio: 1·08, 95% CI: 0·92–1·26; n = 3,230). In Mwanza, Tanzania, mCPR decreased by 9% points (-17 to -0.3; n = 1,973). Incremental cost per DALY averted were $30,855 in Oromia, $111,416 in Nasarawa, $30,114 in Ogun, and $25,579 in Mwanza. Costs per DALY averted were 14–53 times gross domestic product per capita. A360 did not lead to increased adolescent use of modern contraceptives at a population level, except in Oromia, and was not cost-effective. This novel adolescent-centred design approach showed some promise in addressing the reproductive health needs of adolescents, but must be accompanied by efforts to address the contextual drivers of low modern contraceptive use.


Introduction
Around half of adolescent pregnancies in low-and middle-income countries are unintended [1].Ensuring that adolescent girls have access to sexual and reproductive health-care services is critical for achieving universal access (Sustainable Development Goal 3.7).Moreover, being able to control their fertility underpins educational and employment opportunities [2].Although many programs have sought to increase contraceptive use among adolescents, their effectiveness has been limited [3,4].
Adolescents 360 (A360) was a four-year (2016-2020) initiative to increase voluntary uptake of modern contraceptives among adolescent girls in four settings.A360 used human-centred design (HCD) to develop setting-specific interventions through an iterative process of research and prototyping.The program hypothesis was that meaningful engagement of adolescents would catalyse the development of novel, successful approaches.The expectation was that higher design costs would be offset by better-designed and more effective interventions to produce a cost-effective approach.
We describe outcome evaluation and cost-effectiveness studies of A360 in Northern Nigeria, Southern Nigeria, Ethiopia and Tanzania, and draw on process evaluation findings to interpret the results.

Study design
The study design is described in detail elsewhere [5].We used repeat cross-sectional surveys to evaluate the impact of A360 on modern contraceptive use.In Nigeria, the study design included a comparison area purposively selected by the implementers, Society for Family Health, in collaboration with the state Ministry of Health and local government officials.Comparison-intervention pairs (two in Northern Nigeria and one in Southern Nigeria) were selected to be similar with respect to some or all of the following criteria: population density, estimated modern contraceptive prevalence rate (mCPR) among 15 to 49 year olds, number of health facilities and presence of World Bank support for Maternal and Child Health activities.In each setting, eligible girls were identified at the household level.The smallest available administrative unit was used as the primary sampling unit-enumeration area in Nigeria, kebele in Ethiopia and street in Tanzania.Study areas were selected in collaboration with the implementers.The same study design and the same primary sampling units were included at baseline and endline, where possible.Although the design means that it is possible that in each site the same households and individuals may be included in the baseline and endline surveys, no attempt was made to trace individuals or households from baseline to endline.
The outcome and cost-effectiveness studies were conducted in the same locations.In Northern Nigeria, 'Matasa Matan Arewa' ('Adolescent Girls from the North') was evaluated in four local government areas in Nasarawa state; in Southern Nigeria, '9ja Girls' was evaluated in two local government areas in Ogun state.In Ethiopia, 'Smart Start' was evaluated in four woredas (districts) in Oromia state.In Tanzania, 'Kuwa Mjanja' ('Be Smart') was evaluated in Ilemela district in Mwanza region.

Participants
We included girls aged 15 to 19 years who would be eligible for the A360 intervention.In Nasarawa and Oromia, we included girls who were married or living as married; in Ogun, we included girls who were unmarried; and in Mwanza, we included girls who were married or unmarried.
In Nasarawa and Ogun, we collected baseline data between August and September 2017 and endline survey data between November and December 2020.In Oromia, we collected baseline data between September and October 2017 and endline data between November and December 2020.In Mwanza, we collected baseline data between September 2017 and January 2018 and endline data between May and October 2021.
Individual, written consent, was obtained from all participants before conducting the interviews.In Ogun, parental/guardian consent and adolescent girl assent were required for unmarried girls aged up to 17 years.In Mwanza, parental consent waiver was granted for this age group because of the sensitive nature of the survey.

Sample size
The target sample size of 23,481 (Nasarawa 4,555, Ogun 12,020, Oromia 1,926, Mwanza 4,980) was chosen to provide 90% power to detect an intervention effect at each of the study sites.Sample size calculations took into account the design effect (clustering), estimated nonresponse, and the fact that not all girls were sexually active.We assumed A360 would increase mCPR from: 3�0% to 5�1% in Nasarawa, 64�4% to 72�6% in Ogun, 44�0% to 50�8% in Oromia and 26�7% to 32�7% in Mwanza.The effect estimates and baseline mCPR rates were derived from a review of 25 studies as detailed in the protocol paper [5].In Nasarawa, Ogun and Mwanza, baseline mCPR was higher than expected, which led to revised endline sample sizes (S1 Text).

Interventions
The final package of interventions was site-specific and is presented in Table 1, according to template for intervention description and replication (TIDieR) checklist and guide [6].The A360 program tracked other sexual and reproductive health interventions in intervention (and comparison areas in Nigeria), which are also summarized in Table 1 [7].

Objectives
Our primary aim was to evaluate the effectiveness and cost-effectiveness of the A360 approach in increasing modern contraceptive use among sexually active girls aged 15-19 years.Our secondary aims align with the A360 Theory of Change components and are described in S1 Table .We also quantified the association between the respondents' self-reported exposure to A360 and primary and secondary outcomes.

Outcomes
The primary study outcome was the proportion of fecund and sexually active girls who reported using modern contraception at the time of the surveys (mCPR).Modern contraception included male and female sterilisation, contraceptive implants, intrauterine contraceptive  Implementation was conducted in a staggered way across kebeles and by the end of 2020, all kebeles were expected to have receive the intervention. 3,4  In Mwanza, the intervention program was delivered from January 2018 to September 2020; for 32 months.PLOS GLOBAL PUBLIC HEALTH devices, injectables, contraceptive pill/oral contraceptives, emergency contraceptive pill, male condom, female condom, Standard Days Method, Lactational Amenorrhoea Method, diaphragm, spermicides, foams and jelly [5].To better understand the pathways through which the A360 approach could affect mCPR, secondary outcomes were also measured (S1 Table ).One of the secondary outcomes was proportion of current modern contraceptive users who were using a long-acting reversible contraceptive (LARC), which was measured due to the large global emphasis put on this type of methods to offer girls with access to the widest available contraceptive options.Questionnaires were adapted from Demographic and Health Survey and Family Planning 2020 survey instruments and were pre-tested for comprehension, flow, appropriateness and feasibility of implementation.At baseline, all questionnaires were administered face-to-face, whereas at endline, in Nigeria and Ethiopia, the first part of the questionnaire was administered face-to-face while the second part of the questionnaire was administered by phone due to COVID-19 related restrictions on data collection (S1 Text).

Statistical methods
The impact of A360 interventions was assessed by quantifying change between baseline and endline, guided by a pre-specified analysis plan (S1 Text).In Nasarawa and Ogun, we used a difference in difference approach using Poisson regression (binary outcomes) or linear regression (continuous outcomes).The model included area (intervention versus comparison), time, and an interaction between area and time.The interaction term coefficient, reflecting the effect of A360 beyond the time trend, was used to assess impact.Models additionally included age, education level, number of living children, religion and wealth quintile.Clustering at the level of the primary sampling unit was accounted for by using cluster-robust standard errors to calculate p-values and confidence intervals.Differences between intervention and comparison areas at baseline were tested using t-test or Pearson χ 2 test, as appropriate.The validity of the difference in difference approach depends on the assumption that, if the intervention had not occurred, the trend in mCPR would have been the same in the intervention and comparison areas [17].To assess these trends, we used Health Management Information System service data.
In Oromia and Mwanza, linear regression models were fitted to data from the baseline and endline surveys aggregated at the level of the primary sampling unit.Matching was accounted for by including the primary sampling unit in the regression model as a categorical variable.The models also included time and the confounders variables listed above.Observed changes in mCPR could be due to secular trends [5].Therefore, we examined trends over time in study settings, using secondary datasets.The details of this analysis are presented in S1 Text.
The association between self-reported exposure to A360 and mCPR was evaluated using data from intervention areas at endline.A girl was considered exposed if she reported hearing about A360 interventions available in the place where she lived (S1 Text).Poisson (Nasarawa and Ogun) or logistic regression (Oromia and Mwanza) were used for binary outcomes and linear regression was used for continuous outcomes.The models included a variable denoting exposure status, and the confounder variables listed above.Cluster-robust standard errors were used to account for the clustering at the primary sampling unit level.

Costing and cost-effectiveness
The cost-effectiveness analysis defined the comparator for A360 as the status quo for design and implementation of adolescent programming.The design comparator was Population Services International's (PSI) DELTA design methodology [18], the standard used at the time A360 initiated.The implementation comparator was the existing contraceptive programming available to adolescents in the A360 study geographies.We estimated incremental cost-effectiveness ratios for each study setting.Incremental costs were A360 design and implementation costs minus the comparator cost.See S1 Text for additional details.

Role of the funding source
The final study designs were discussed and agreed upon with the funders.The funders had no role in the data collection, data analysis, or writing of the manuscript.The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.

Ethics approvals
The

A360 outcome evaluation
Survey response rates were between 69% and 100% (Fig 1).Selected sociodemographic characteristics of girls in the baseline sample are presented in Table 2. S2 Table illustrates differences in number of children, religion, wealth and mobile phone access in intervention and comparison areas at baseline.

Nasarawa, Nigeria
In Nasarawa, mCPR increased from 16% to 38% in the intervention areas and from 13% to 27% in comparison areas.After accounting for the change in mCPR in comparison areas via the difference in differences analysis, we found no evidence of an impact of A360 (risk ratio: 0�96; 95% CI: 0�76-1�21; p = 0�734; Table 3).
We observed intervention effects in two out of 17 secondary outcomes.For instance, girls in intervention areas reported a greater increase over time in positive attitudes to couples using modern contraceptives compared to those in comparison areas (S3 Table ).
Out of 17 secondary outcomes, the only effect observed in the hypothesised direction was on the proportion of LARC users among all modern contraceptive users which increased from 0�3% to 2�2% at the intervention area and dropped from 1�4% to 1�0% at the comparison area (S3 Table ).
Self-reported exposure to 9ja Girls was 8% (95% CI: 6-10) in the intervention area.There was no relationship between self-reported exposure and primary and secondary outcomes, except for a weak evidence of a positive relationship between being exposed and being aware of contraceptive products (S4 Table ).
This increase in mCPR was accompanied by an increase in four out of 17 secondary outcomes (S4 Table ).For instance, there was a 10% absolute increase in the proportion of current modern contraceptive users using a LARC (95% CI: 3-17; p = 0�004), and an 11% increase (95% CI: 1-21; p = 0�025) in awareness of contraceptive products.There was also evidence of an in increase in the proportion of adolescent with positive attitudes towards the use of modern contraceptives (S3 Table ).
This decrease in mCPR was accompanied by declines in several secondary outcomes.For instance, there was a 5% decrease (95% CI: -9 to -3; p<0�001) in the proportion of adolescent girls agreeing that contraception can help adolescent woman/girl to complete their education, find a better job and have a better life.There was also a 14% decrease (95% CI: -24 to -4; p = 0�008) on the intention to use a modern method, and a 9% increase (95% CI: 2-16; p = 0.019) in the proportion of modern contraceptive users using a LARC over time (S3 Table ).

Analysis of trend using other sources of data
In S1 Text, we present the results of the analysis of trends in modern contraceptive use for all three sites, using secondary data sources.In Nigeria, the data provided some evidence on the validity of the parallel trend assumption in Nasarawa and only weak evidence in Ogun.In Ethiopia, no clear trend was observed in mCPR, but in Tanzania, we observed an upward trend (S1 Text).

Cost effectiveness
A360 design costs were seven to nine times higher than the comparator DELTA approach.A360 implementation costs were also substantially higher than maintaining existing contraceptive programming available to adolescents in the A360 study geographies.Incremental cost for the study geographies was $484,900 for Nasarawa, $513,220 for Ogun, $970,667 for Oromia, and $120,479 for Mwanza.The annual per capita (total population) spending on A360 design and implementation was between $0.22 in Mwanza and $0.67 in Oromia.As a percent of total health spending per capita (using WHO estimates of per capita health spending, adjusted to 2020 USD) [19], spending on A360 represented between 0�3% in Ogun and 2�5% in Oromia.Spending per eligible girl (sexually active, fecund, married/unmarried as per geography) per year of implementation ranged between $13 in Mwanza to $102 in Nasarawa (Table 4).The mCPR changes translated to 4�4 maternal disability-adjusted life years (DALY) averted in Nasarawa, 17�0 in Ogun, 31�5 averted in Oromia, and 4�7 in Mwanza.Despite the lack of change in Nasarawa and the declines in modern contraceptive prevalence in Mwanza seen in the outcome evaluation, DALY impacts were positive due to increases in the number of eligible adolescents that result in positive additional users over the life of the project.
A360, adolescents 360 approach, mPCR, proportion of fecund and sexually active girls who reported using modern contraception at the time of the surveys, RR, risk ratio, RD, risk difference, OR, odds ratio 1 This is a girl-level calculation 2 In Nasarawa and Ogun, these correspond to adjusted risk ratios (RR) resulting from difference in difference analysis (Poisson regression) at girl-level; In Oromia and Mwanza, these are adjusted risk differences resulting from a pre-post comparison (linear regression) at kebele-level in Oromia and at street-level in Mwanza 3 In Nasarawa and Ogun, these correspond to adjusted risk ratios (RR) resulting from Poisson regression models at girl-level; In Oromia and Mwanza, these are odds ratios (OR) resulting from logistic regression models at girl-level https://doi.org/10.1371/journal.pgph.0002347.t003

Discussion
Our evaluation suggested evidence that A360 did not lead to increased adolescent use of modern contraceptives at a population level, except in Oromia, Ethiopia.Self-reported exposure to A360 was low, ranging from 5-24%, and in three settings was positively associated with mCPR.We observed a positive intervention impact on some secondary outcomes linked to the A360 Theory of Change.External data sources suggested an upward trend in mCPR among women aged 15-49 years.A360 was not cost-effective, with design costs higher than the comparator design approach.The absolute amount spent in relationship to population size and overall health spending was substantial.The outcome evaluation had many strengths including accounting for clustering when calculating required sample sizes, the collection of comparable data before and after intervention implementation, use of multiple data sources to track population level mCPR and, in Nasarawa and Ogun, collection of data from populations both exposed and not exposed to the intervention.Participants were representative of A360 target populations increasing internal and external validity of the study.Data on self-reported exposure to the A360 programmes at endline allowed examination of the association between individual-level engagement with A360 and modern contraception use.
An important limitation was the lack of comparison areas in Mwanza and Oromia as a change in mCPR may reflect a time trend rather than an intervention effect [17,[20][21][22].Our findings may not be generalizable to other areas of the countries where A360 was implemented.We relied on respondent self-reporting to measure modern contraceptive use, sexual activity and exposure to the programs; these behaviours may have been subject to reporting bias.The COVID-19 pandemic led to changes to endline survey procedures including the use of face masks and the administration of the second section of the questionnaire by phone in Nasarawa, Ogun and Oromia.In Nasarawa and Ogun, participants in intervention and comparison areas were not entirely comparable in terms of sociodemographic factors.While we adjusted for changes in sociodemographic factors there may be some residual confounding.Finally, the validity of the difference in difference approach used in Nasarawa and Ogun depends on the mCPR time trend being the same in both intervention and comparison areas [17].The only data available to evaluate similarity in time trends were Health Management Information System data from female clients aged 15-49 years using modern contraceptives at health facilities between early 2016 and mid-2020.These data reflect a broader age range than was included in the survey population and provided only weak evidence on the validity of the parallel trend assumption.
The costing and cost effectiveness study used a consistent approach in all three countries, repeated measures, and drew much of its information from accounting systems.Limitations are detailed in S1 Text.One-way and multi-way cost sensitivity analysis addressed many of these limitations, producing plausible lower and upper ranges to total cost used in sensitivity analysis.
Human-centred design (HCD) was a key element of the A360 approach.As defined by Giacomin [23], HCD 'is based on the use of techniques which communicate, interact, empathize and stimulate the people involved, obtaining an understanding of their needs', and it incorporates several features of social, behavioural and community engagement interventions.Deploying HCD in combination with youth engagement, insights from different disciplines, and working adaptively, helped the interventions resonate with girls, communities and government stakeholders [7].HCD helped A360 integrate aspirational content, which attracted girls to events, built government buy-in, and allowed the program to operate in the context of high levels of stigma.In Oromia, messaging built around couples' counselling and financial planning, resonated strongly with married couples.Change was enabled by the integration of Smart Start into the Ethiopian government's Health Extension Program with delivery by Health Extension Workers, who are known and trusted in communities [24].Among married girls aged 15-19 years in Nasarawa and Oromia, A360 implementation was associated with a population-level change in girls' belief in the benefits of modern contraception and in the proportion of girls with positive attitudes toward modern contraception.
Sociocultural factors remain a barrier to adolescent contraceptive use in Nasarawa, Ogun and Mwanza.For instance, in Nasarawa, one of the main reasons participants said they were not using contraception was the desire to bear children [25], reflecting established social norms [26].In Ogun and Mwanza, A360 appeared less effective among unmarried girls perhaps because of insufficient impact on the stigma associated with premarital sex [27][28][29].Unmarried girls need youth-friendly service delivery with ensured confidentiality and provider discretion.In both Ogun and Mwanza the majority of unmarried girls reported obtaining contraceptives at settings other than local health facilities.Process evaluation suggested A360 was not designed or resourced to address social norms [7].A360 instead used light-touch approaches to engage communities, including: parents' sessions in Mwanza and couples' counselling in Oromia; working with community leaders, local government officials and trusted community structures; and, developing messaging that tapped into existing community concerns.Other studies suggest that community-level interventions need to be intensive and sustained to have long term impacts on knowledge, attitudes, practices and behaviors [4,30].
In Mwanza, we observed a population level decrease in mCPR among adolescent girls.This might be explained by three main factors.First, during Kuwa Mjanja implementation in 2018, the Tanzanian former president made negative statements about contraception [31], and advertisements on contraception were later banned [32].This at one point led to a halt to A360 outreach activities.Second, we saw an increase in the number of adolescent girls residing in study areas (which led to changes in sampling strategy at endline; see S1 Text), and a higher level of education at endline, both of which may have affected results.
The endline surveys of our outcome evaluation were conducted in late 2020, approximately one year after the start of COVID-19 pandemic.In Nigeria, data from Performance Monitoring for Action 2020 [33,34] and from Krubiner and colleagues [35] indicate that family planning service and product availability for females aged 15-49 years are unlikely to have been impaired due to COVID-19, but it is difficult to know if this was also true among adolescent girls aged 15-19 years in our study geographies.In a State of Northern Nigeria (Kano), women aged 15-24 years who changed their contraceptive use status were more likely to adopt (4%) than to discontinue a method (<1%) [33].On the other hand, in a State of Southern Nigeria (Lagos), women aged 15-24 years who changed their contraceptive use status were more likely to discontinue (11%) than to adopt a method (5%) [34], but only 2% of women aged 15-49 years stopped or interrupted their contraceptive method use due to COVID-19 restrictions [36].In Ethiopia, even though some studies showed limited availability of family planning services and products due to COVID-19 [37][38][39], endline survey data, A360 monitoring data (collected by the implementers) and Health Management Information System service data, indicated that the effects of the pandemic were minimal [40].Finally, in Tanzania, the COVID-19 pandemic stopped A360 service delivery between March and May 2020 [7], after which Kuwa Mjanja implementation was limited to door-to-door visits and short discussions, to avoid mass gatherings [7].Nevertheless, the effects of COVID-19 on the supply of contraceptive commodities also seem limited in Tanzania, with the country following the World Health Organization guidance during the pandemic, by relaxing contraceptive prescription requirements and recommending that emergency contraception be available at pharmacies [35].
The low levels of reported exposure to A360 suggests that implementation intensity may have been lower than anticipated in the evaluation geographies and insufficient to achieve population-level change in mCPR.When the evaluation study was designed, the intervention implementation plans had not yet been finalised but the intention was for A360 to be implemented widely in the selected study geographies.As noted previously, HCD-based initiatives may require a phased evaluation approach with an outcome evaluation designed only when the programme and implementation strategy have been finalised [41].Other alternative explanations for the low levels of reported exposure to A360 are: the tool was inappropriate to capture exposure to the program, although this is unlikely as the tool was validated with the implementers; high levels of migration, although this is also unlikely as in a secondary analysis we found very small proportion of girls having migrated for more than three months in the 12 months previous to the surveys.
The more intensive design effort in A360 was not cost effective in relation to the size of health outcomes achieved.Incremental cost-effectiveness ratios were far above the three times per capita GDP threshold for a cost-effective health intervention, per WHO-CHOICE standards [42].They were also much higher than the $225 per DALY averted proposed as a cut-off for inclusion of interventions in Universal Health Care package, and far above the cost per DALY averted reported for other family planning interventions (between $235 and $587) [43].Further analysis suggested that in Oromia, no level of mCPR increase would have led to costeffective results, while in Nasarawa and Ogun large increases in mCPR would have been needed to reach a benchmark of three times GPD per capita.Efforts to reduce implementation costs will be needed to produce more cost-effective models.Ethiopia and Nigeria are shifting management and service delivery responsibilities for the A360 legacy interventions to governments, which may lower costs.In Mwanza, A360 costs were more in line with potential impact, and even just maintaining baseline mCPR could have led to cost-effective results.
Our results highlight the challenges associated with identifying cost-effective approaches to increase the voluntary use of modern contraceptives among adolescents.A360 is being scaledup and modified in Ethiopia and Nigeria, and expanded to Kenya [44].According to the A360 implementers (PSI), the decision to scale-up was driven largely by the governments of the implementing countries.We recommend increased efforts to address social norms that prevent girls from accessing or using contraceptives as well as additional evaluation of A360 among both married and unmarried adolescents.Accordingly, A360's second phase aims to have a more consistent and rigorous approach to improving the enabling environment [44].The modified A360 approaches are going to be evaluated through new cost-effectiveness studies.
outcomes measured for the Adolescents 360 outcome evaluation, pre-and post-intervention, by site.This table also presents the impact of the Adolescents 360 approach in each site.(DOCX) S4 Table .Exposed versus non-exposed comparison of primary and secondary outcomes at endline, by site.A360, Adolescents 360 approach, mCPR, modern contraceptive prevalence rate, LARC, long-acting reversible contraceptive, Data are n (%) or mean (SE). 1 Girls who agreed with the sentence 'Using modern contraception can allow an adolescent woman girl to complete her education, find a better job and have a better life' 2 Girls who agreed with the sentence 'Using modern contraception can allow a girl to achieve her life goals' 3 The impact of the A360 exposure is defined as the risk of mCPR in the exposed compared to girls not exposed to A360.S4 Table presents a description of primary and secondary outcomes measured for the Adolescents 360 outcome evaluation, by girls who reported being exposed to the intervention and girls who did not report being exposed to the intervention, by site.(DOCX)

Table 1 . (Continued) Item no. Item Nasarawa, Nigeria Ogun, Nigeria Oromia, Ethiopia Mwanza, Tanzania 7 Description of each of the procedures, activities, and/ or processes used in the intervention.
(a) Female mentors who moderate LFH sessions are trained by the A360 approach.(b) Contraceptive counselling is delivered by A360 Young Providers, who are recruited by A360, and work full time in public health clinics.It may also be delivered by government health workers.All providers are trained in youth-friendly service provision and use counselling protocols that focus on issues that are of most concern to girls.

face-to- face) of the intervention and whether it was provided individually or in a group.
(a) LFH sessions are conducted through face-to-face sessions, in groups of 12.(b) Contraceptive counselling is provided individually after each session if a girl does not 'opt-out'.(c) Girls can also access walk-in appointments with trained providers at any time at their nearest clinic (a) LLH sessions are conducted face-to-face; provided in groups of 15-20 girls.(b) Contraceptive counselling is provided individually after each session if a girl does not 'opt-out'.(c) Girls can also access walk-in appointments with trained providers at any time at their nearest clinic (d) Mothers' sessions are conducted in groups.The session is conducted face-toface, to the girl (individually), or to the couple.(a) Sessions are implemented face-to-face and events are conducted in groups.(b) Contraceptive counselling is provided individually unless a girl wishes to 'opt-out'

Table 1 .
(Continued) Start dates were defined using PSI reports and monitoring data.PSI revised and confirmed these start dates with the on 09/02/2021. 3'Specifically, all kebeles in Fentale received the interventions by May 2019, in Ada'a by January 2020, in Lome by December 2020 and in Wara Jarso by May 2020'.: PSI led the overall design and implementation of Adolescents 360 and worked in partnership with IDEO.org and the Center for the Developing Adolescent at the University of California at Berkeley.Country work was led by PSI Ethiopia, PSI Tanzania, and Society for Family Health, a non-governmental organisation in Nigeria.Implementation was accompanied by a monitoring and evaluation component to examine process level indicators such as the number of girls reached.MMA, Matasa Matan Arewa, LGA, local government area, LFH, Life Family and Health, LLH, Life Love and Health, A360, Adolescents 360 intervention https://doi.org/10.1371/journal.pgph.0002347.t001 1 Source of information: Population Services International (PSI), 03/03/2021. 2 4 Source of information: Document entitled "A360 OE site mapping" shared by PSI, on 12/11/2020.Note study components were approved by the National Health Research Ethics Committee of

Table 2 . Demographic characteristics of adolescent girls in intervention areas at baseline a .
a Data are n (%) or mean (SD) N/A, not applicable https://doi.org/10.1371/journal.pgph.0002347.t002