Study setting

The Girl Power study is an ongoing study (February 2016–November 2017) being conducted in Malawi and South Africa. These countries represent two prototypical, yet distinct SSA contexts. In Malawi, based on nationally representative household data, first births often occur early, at a median age of 19 years and within the context of marriage.30 In South Africa, sexual activity tends to be outside of the context of marriage, with a somewhat later age of first birth.31 32 Malawi is characterised by extreme poverty with the majority of the population living on less than $1 per day, whereas South Africa has extreme wealth disparities. In spite of these socioeconomic differences, both countries have high HIV prevalence levels—6% of women 20–24 years old are HIV-infected in Malawi and 17% in South Africa, rates considerably higher than their male counterparts.30 Sexual violence is also a serious problem in both countries, with high rates of non-consensual sex reported by young women.33–35 Additionally, both have public sector health facilities with human resource shortages, stock-outs of pharmaceuticals and supplies, long queues and providers who exhibit negative attitudes towards AGYW. It was within these contexts that the Girl Power study was conceptualised.

Four comparable public sector health facilities were selected in each country. In Malawi, the four sites are public sector health centres in Lilongwe. All sites are on a main road, have antenatal volumes >200 clients per month and have antenatal HIV prevalence > 5%. These sites have higher HIV prevalence than the surrounding rural areas where antenatal HIV prevalence is <5%. In South Africa, the four sites are in the Klipfontein and Mitchell’s Plain areas in the Western Cape, which serve primarily black and coloured populations. These are high-density, low socioeconomic periurban townships comprised predominantly of informal dwellings. Antenatal prevalence in the Western Cape is 17%. In both countries, each clinic serves distinct communities. In South Africa, biometric identification is being used to ensure the same people do not enrol in more than one site. In Malawi, all sites are at least 7 km apart.

Prior to the study, all sites in both countries offered free HIV testing and counselling (HTC), contraception and STI services in separate locations within each clinic with separate queues. Condoms were also available for free at all sites—in the pharmacy at the Malawian sites and throughout the clinic at the South African sites. At all sites in both countries, AGYW could receive general health services with the general population. However, there were no distinct YFHS spaces, times or providers, no BIs and no opportunities for conditional cash transfers (CCTs) within the clinic.

Study design and interventions

Girl Power is a quasi-experimental prospective cohort study comparing four different models of service delivery and their impacts on care-seeking and sexual risk behaviours among AGYW (figure 1). The primary outcomes of interest are service uptake and behavioural risks. Participants enrolled at all sites will complete behavioural surveys at baseline, 6  and 12 months and have their clinical data ascertained over a 1-year period. In each country, each clinic was randomly assigned to one of the following models of care:

• Model 1: SOC: The SOC offers HTC, contraception, STI syndromic management and condoms to AGYW. However, healthcare providers have not been trained in YFHS and clinics have not made YFHS modifications, regarding hours, clinical navigation, integration, cost or space.

• Model 2: YFHS: HTC, contraception and STI services are offered by health providers who have been trained in YFHS. YFHS modifications have been made to the clinics, including clinical navigation (both countries), a youth-only space with integrated service provision (Malawi) and longer hours (Malawi).

• Model 3: YFHS+BI: In addition to the YFHS package, participants can attend 12 monthly facilitator-led, small-group interactive sessions. They are based on other evidence-based interventions from the region17 19 21 and are motivated by the theory of gender and power and social cognitive theory. These sessions address sexual health topics (eg, HIV, reproductive health), social issues (eg, partner communication, peer pressure and IPV), financial literacy (eg, budgeting, saving and investing) and general topics (eg, self-esteem, goal-setting and decision-making). Each session includes an activity to do at home after the session in order to encourage behavioural change. Each country adapted curricula in a culturally responsive manner; although sessions were not identical across countries, they explore similar themes. We expect sessions will be delivered to groups of 10–20 girls at once.

• Model 4: YFHS+BI+CCT: In addition to the YFHS package and empowerment sessions, AGYW received a monthly CCT (approximately $6) conditional on attending the monthly empowerment session. Each young woman could receive up to 12 CCTs (1 per month). In Malawi, the CCT is being provided in physical cash and in South Africa the CCT is provided electronically.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Depicts the study design. BI, behavioural intervention; CCT, conditional cash transfer; SOC, standard of care; YFHS, Youth-Friendly Health Services.

Study and clinical personnel, training activities and clinical modifications by country and model are described in greater detail in table 1.

Study population

At each of the eight sites, 250 AGYW will be recruited and followed for 1 year. Eligibility criteria include being female, 15–24 years old, residing in the clinic’s catchment area and willing to be enrolled for a 1-year period. The intention is to recruit AGYW who are already sexually active or likely to become sexually active, although this is not a strict eligibility criterion. In total, 1000 AGYW will be enrolled in each country (n=2000 total).

Recruitment will occur comparably at each site through a combination of community outreach activities, self-referral and referral through invitations from other participants. Outreach workers will visit parts of the catchment area known to be high risk. Through one-on-one conversations, they will build rapport, assess sexual activity and past care-seeking behaviours, promote the services at their site and invite them to participate. AGYW who enrol in the study will be provided with invitations to invite friends who they believe would also benefit.

Follow-up phone and physical tracing will be conducted for girls who miss their 6-month or 12-month visits. We will make comparable attempts to trace participants at all four clinics to avoid differential loss-to-follow-up and provide the same transport reimbursement for the three research visits across all four sites. These efforts are designed to minimise differential loss-to-follow-up.

Data collection and management

The two primary sources of data collection are a detailed behavioural survey and clinical service records. The behavioural survey is administered at three time points—at study enrolment, month 6 and month 12. The behavioural survey contains questions about demographics, socioeconomic status, past and current care-seeking behaviours, sexual history, depression (10-item Center for Epidemiologic Studies of Depression Scale), IPV (Modified Conflict Tactic Scale), alcohol consumption (National Institute on Alcohol Abuse and Alcoholism brief alcohol screening tool). The 12-month assessment also includes assessments of the BI and the CCT. The behavioural survey will be self-administered in South Africa, except among illiterate participants; for these participants, it will be interviewer-administered. In Malawi, it will be interviewer-administered to all participants by young female research assistants. All surveys will be administered on encrypted password-protected Android tablets using Open Data Kit software and stored on secure servers.

In both countries, we will document delivery of HTC, family planning and STI services. Due to the unique clinical contexts, data collection will occur differently in the two countries. In Malawi, a clinic card that contains all clinical information has been developed for the study and will be used by clinical staff at each patient encounter and housed at the clinic. In South Africa, clinical staff will record clinical activities in existing clinical records and study staff will transcribe this information onto a study-specific form. In both countries, these data will then be entered into tablets in Open Data Kit by trained research assistants. In all clinics, study staff will systematically examine clinical records to ensure consistent ascertainment.

In both countries, each participant will be identified through a unique identifier. In South Africa, participants’ identities will be verified through biometric fingerprinting.

Study outcomes, analytical methods and sample size

The primary clinical outcomes are care-seeking behaviours using clinical data. Within each quarter, we will compare the proportion of participants who received HTC, male and/or female condoms, hormonal or long-acting contraception and both condoms and another form of contraception. We will also explore the proportion of participants who receive STI services. For all of these services, we will assess the proportion of participants who use each method ever, in each quarter and in all quarters. These data are available through clinical abstraction, the primary measure of clinical effectiveness. Details are reported in table 2.

HIV risk behaviours collected on the behavioural survey are primary behavioural outcomes of interest. Questions of particular interest are the proportion of AGYW who have multiple sexual partners, have a considerably older male partner and experience IPV. We will compare these indicators between participants in each country in each model.

For all primary clinical outcomes, we will conduct an intention to treat analysis, analysing each participant in her assigned clinic and assessing whether she received services in that clinic. We will use generalised estimating equations to account for correlated records between each participant at each time point. A log or identity link, binomial distribution and robust variance estimates will be used to estimate risk ratios and risk differences and 95% CIs. We will compare absolute differences at each time point, as well as changes over time.

In Stata V.12.0, we conducted two-sample tests of proportions to determine how much statistical power would be available to detect differences between any two arms at any time point within each country. Using an alpha level of 0.05 and a sample size of 250 participants per model, we have >80% power to detect differences >13% between any two models at any time point.

Qualitative substudy

A qualitative substudy will be conducted to better understand participant experiences with each service delivery model. Out of 15 in-depth interviews (IDIs), 12 will be conducted at each site in each country (n=96–120 IDIs total). IDIs will focus on individual experiences with SOC, YFHS, BI and CCT. At each clinic, we will purposively select a mixture of good care-seekers, poor care-seekers and HIV-infected young women. Focus group discussions (FGDs) (n=2–3/site) will address norms surrounding these same topics and strategies for improving or enhancing these services. At each site, we will have at least one FGD with adolescent girls 15–19 years old and one with young women 20–24 years old. IDIs and FGDs will be conducted in Chichewa in Malawi and in isiXhosa or English in South Africa. All IDIs and FGDs will be transcribed and translated into English. They will be coded and analysed using a thematic approach.

Ethics and dissemination

In South Africa, all AGYW 15–24 years could decide whether to provide informed consent for themselves; those 15–17 years could seek optional parental consent. We requested that minors 15–17 years be able to consent for themselves because they are able to receive all of these clinical services without parental consent. In a study designed to reduce barriers to care-seeking, obtaining parental consent could pose an undue barrier; the ethics committee agreed. In Malawi, AGYW 18–24 years will provide informed consent for themselves. AGYW 15–17 years will provide assent and have a parent, guardian or authorised representative provide informed consent. We requested that minors be able to consent as adults, but this provision was denied. At each site, a community club of individuals >18 years who could serve as authorised representatives will be established, a suggestion made by the National Health Sciences Research Committee (NHSRC).

Dissemination activities will occur at all stages of the study: prior to implementation, at intermediate points during implementation and at study culmination. Prior to study implementation, meetings were held with key local stakeholders, such as local health leadership (eg, district, province or city managers), clinical supervisors and clinical staff. These meetings were designed to seek permission for study implementation, orient these stakeholders to study goals and work together on implementation questions. Sensitisation activities were also conducted with key community stakeholders, such as local chiefs and religious leaders, school headmasters and library managers. These sensitisations were designed to inform key local leaders about the study, elicit buy-in and support and facilitate recruitment. During study implementation, these same stakeholders will be engaged to report on study progress and discuss challenges. At the end of the study, results will be disseminated so that stakeholders and participants are aware of the study’s primary findings. Additionally, in both countries, we have oriented policy-makers and implementing partners to study goals and progress. Results will be disseminated in peer-reviewed journals and at national and international meetings and conferences. Dissemination of final results is expected in 2018.