The study clusters comprise 22 of the 30 administrative wards in the county. The 30 wards were categorized into 5 strata based on community type (ie, periurban, rural fishing, or rural nonfishing) and whether the proportion of girls on PrEP is above or below the median, using MoH programmatic data. (Although the geography and PrEP use criteria define 6 strata, only 1 ward in the county fell within the urban, low PrEP use stratum. Because it was not possible to create a pair of intervention and control wards in this stratum, it was excluded.) The 22 study wards were purposively selected to ensure an even number of wards within each stratum and a balanced distribution of wards across strata. Randomization was completed by RTI statisticians once the 22 study wards are selected and placed into strata. Within each stratum, wards were randomly allocated to the intervention or control arms in a 1:1 ratio. After randomization, we created pairs of intervention and control wards within each stratum and determined the order in which they would be enrolled in the study based on the location of each ward, to minimize the risk of contamination while ensuring the feasibility of implementation. Subsequently, one intervention and control site pair were randomly selected from each stratum for inclusion in the qualitative component. Allocation was revealed to the investigators before community mapping (see the Recruitment and Screening section) and to the participants after their enrollment into the study.

To be eligible, potential participants must be females aged 15 to 24, currently in a sexual relationship with a male partner for at least 1 month, and vulnerable to HIV per a modified version of the DREAMS eligibility screening tool. To reflect the population most likely to participate in this intervention in a programmatic setting, they must be either taking PrEP or interested in taking PrEP (ie, she thinks that she would benefit from PrEP but is not currently taking it). In addition, they must be residents of the applicable study ward; willing and able to attend support clubs; willing and able to provide adequate contact information; fluent in English, Dholuo, or Kiswahili; and able and willing to provide informed consent (or assent and parental consent for nonmature minor participants aged 15 to 17). Potential participants are excluded if they are living with HIV (by self-report); planning any long-term travel or relocation in the next 12 months; or have any condition that the site PI or designee determines would preclude participation.

Individuals who do not meet the criteria for participation in this trial (ie, screen failure) because of criteria that are likely to change over time may be rescreened. Examples include not being resident of the ward or not being interested in taking PrEP.

Recruitment occurs simultaneously in each pair of wards before moving on to the next pair and focuses on the catchment areas of 2 to 4 health facilities in each ward. Before recruitment, community mapping is conducted to identify key stakeholders, referral resources, locations where adolescent girls and young women gather, and locations where men gather. Recruitment takes place at venues identified through community mapping, including health facilities, schools, churches, DREAMS Safe Spaces, youth groups, and community markets. To ensure that the study findings are generalizable to adolescent girls and young women across the 15- to 24-year age range, we aim to enroll at least 25% of the study participants from the 15- to 19-year age group who have lower rates of PrEP use but have similar HIV incidence to the 20- to 24-year age group [47,48]. Attendees interested in the study are invited to meet with the research assistant after the recruitment meeting to discuss questions, obtain additional information, and schedule an appointment for informed consent, eligibility screening, and, if eligible, for enrollment.

Once a participant is enrolled in the study, the study team makes every reasonable effort to retain her to minimize bias associated with loss to follow-up. To ensure high retention, IRDO tracks retention rates and follows standard study operating procedures detailing strategies, such as clear explanation of the visit schedule at all visits, collection of detailed locator information with multiple means to contact participants, appropriate and timely visit reminders, offering weekend hours for study visits, and conducting immediate follow-up on missed visits through phone, home or other off-site visits.

Data collection visits occur simultaneously in each pair of wards at enrollment, and months 3, 6, and 12 (Table 3). At enrollment, interviewer-administered questionnaires are used to assess sociodemographics, relationship characteristics, IPV, behavioral HIV risk factors, and history of PrEP use. Measures are repeated in follow-up questionnaires to document changes in relevant characteristics, attitudes, and behaviors.

DBS samples are collected via fingerprick at months 6 and 12 from all participants who report initiation or continuation of PrEP since the last visit. For participants who report discontinuation of PrEP, reasons for discontinuation are recorded, but DBSs are not collected. Due to storage and transport challenges, DBS samples are also not collected for participants who have relocated outside the county and have follow-up visits conducted by phone or at a remote location. DBS samples are stored in freezers at the IRDO laboratory and then shipped to the laboratory at the University of North Carolina Center for AIDS Research Clinical Pharmacology and Analytical Chemistry Core for quantification of drug concentrations by liquid chromatography-tandem mass spectrometry technology [60].

The mixed methods process evaluation uses implementation data; quantitative adolescent girls and young women participant questionnaires; exit surveys with men who attend intervention activities; qualitative in-depth interviews (IDIs) with adolescent girls and young women participants, their male partners, and intervention providers; and a project diary. Our process evaluation framework (Figure 3 [61]) highlights the key evaluation components, relationships between them, and key measures and data sources [62-65].

Eligibility criteria for adolescent girls and young women participants are described in the Eligibility Criteria section. Participants in exit surveys or male partner IDIs must be males aged ≥15 years. In addition, the exit survey participants must have attended a study Buddy Days or community sensitization event and have been present for at least half of the event, while IDI participants must be a sexual partner of the adolescent girls and young women clinical trial participant and that adolescent girls and young women participant must have provided permission for the study staff to contact the male partner. Intervention providers must be aged ≥18 years and have been involved in delivering Tu’Washindi intervention activities. In addition, all participants must be fluent in English, Dholuo, or Kiswahili; willing and able to provide informed consent; and not have any condition that the site PI or designee determines would preclude participation.

Implementation data includes staff hiring and training records; minutes from youth advisory board, stakeholder advisory board, and technical advisory committee meetings; and study activation checklists to track activities, materials, and systems required for successful implementation, such as community introductions, final intervention manuals, and vetted lists of referral agencies. Providers maintain activity logs and attendance records for each meeting or event, and supervisors record minutes from feedback and coaching meetings. Quality and fidelity of intervention activities is documented on observation forms as described above.

Participant questionnaires are administered to adolescent girls and young women participants at each visit to assess mechanisms of change (for all participants) and perceived intervention quality (intervention arm) or potential contamination (control arm). The hypothesized mechanisms of change include PrEP self-efficacy and outcome expectations, PrEP knowledge, healthy communication, disclosure of PrEP use to male partner and partner reaction, community knowledge of PrEP, PrEP stigma, and social assets.

Exit surveys are brief, anonymous, quantitative, interviewer-administered questionnaires that gather information on men’s views of PrEP, their willingness to support their partner’s PrEP use, and feedback on intervention activities.

IDIs are conducted in a private location and facilitated by trained female interviewers following a semistructured guide as follows:

A program diary is maintained by the study coordinator to record external events (eg, elections, police activity, festivities, and health campaigns) and internal events (eg, increased funding or supply shortages) that may affect intervention delivery or study outcomes. Before intervention implementation in each ward, the study coordinator identifies a set of sources, including study staff, health facility personnel, community members or others who are well-positioned to provide this contextual information, and contacts them at regular intervals to record any updates.

The primary outcomes of the study will be the proportion of study participants with effective PrEP use, defined in Table 4, evaluated immediately postintervention, at month 6, and at month 12 to gauge the persistence of intervention effect. TFVdp levels will be dichotomized by Clinical Pharmacology and Analytical Chemistry Core laboratory specialists to delineate between consistent (≥4 doses per week) versus inconsistent (<4 doses per week) adherence [68]. The aim 1 hypothesis is that the proportion of participants with effective PrEP use at (a) month 6 and (b) month 12 will be higher in the intervention versus control arm.

The secondary outcomes of the study will be the proportion of participants reporting any IPV and severe IPV (Table 4) between enrollment and month 6, and between month 6 and month 12, and the difference in IPV intensity scores at months 6 and 12. The aim 2 hypothesis is that the proportion of participants reporting any IPV and severe IPV and IPV severity scores will all be lower at both time points in the intervention arm versus control arm.

In exploratory analyses, we will first explore effect modification by age (15-19 vs 20-24) and DREAMS participation (current, former, or never) for the primary and secondary outcomes. Second, we will break down the primary outcome into steps along the continuum (PrEP uptake, execution, and persistence) and IPV outcomes by type—physical, sexual, emotional, and economic—to understand where the intervention had its strongest effects. Third, we will compare the proportion of participants per arm with emtricitabine triphosphate (FTCtp) levels indicating consistent PrEP adherence over the past 2 weeks. Comparing TFVdp and FTCtp levels will yield insights on adherence patterns over both a long and short window of time before collection. We will also analyze the alignment of adherence measures with HIV risk perception and behavior to better understand adherence patterns.

Descriptive and bivariate statistics will be used to compare participant characteristics in the intervention and control arms and those who completed the study versus those lost to follow-up. Outcomes will be modeled using an intent-to-treat approach with mixed effect logistic and linear regression models. Model predictors will include treatment assignments, randomization strata, and a random effect for ward (cluster) within stratum that is assumed to be independent and normally distributed with mean 0. If the cluster randomization does not adequately balance baseline characteristics across arms, we will conduct exploratory analyses controlling for those characteristics. We will analyze the data from each of the 2 time points separately and the primary outcomes analyses will use a type I error rate of 0.025 at each time point.

In any situation with missing data, we will perform appropriate secondary analyses adjusting for variables that may be related to the missingness mechanism. If missing data rates are higher than anticipated (>10%), we will include covariates that are related to missingness in regression models. We will also perform sensitivity analyses to assess the potential impact of the missing data.

Implementation inputs, processes, and outputs will be analyzed descriptively, and we will draw comparisons across intervention wards. Intervention effects on each mechanism of change will be modeled individually following the approach described for the cRCT. We will regress each mechanism on the exposure (intervention arm assignment), controlling for baseline levels of the mechanism. If the primary analyses show no effect of the intervention on PrEP or IPV outcomes, the results will inform our interpretation of whether the intervention failed to affect the intended mechanism or whether the change in mechanism did not lead to the hypothesized outcome. If the primary analyses suggest a positive intervention effect on PrEP or IPV outcomes, we will conduct causal mediation analysis [72,73]. Coverage, uptake, and exit survey data on perceived impact will also be evaluated as moderators of intervention effectiveness at the site level in exploratory analyses of mechanisms of change.

The coding process will involve a core group of about 2 to 5 analysts from IRDO and RTI who will develop one or more codebooks and establish qualitative coding procedures. Data will be coded using a qualitative software package, such as Dedoose (SocioCultural Research Consultants, LLC). A set of preliminary codes will be developed based on key themes or topics identified a priori and the findings highlighted in the debriefing reports for each of the 3 datasets (eg, adolescent girls and young women, male partners, and providers). Each group member will apply the initial set of thematic codes to a common transcript, discuss their coding experiences, and agree on expanding and modifying code names and definitions when necessary. We will continue with an iterative process of revising the codebooks as we continue to read and code the data. Once finalized, the codebooks will be used for a final recoding of all transcripts.

A preselected number of transcripts will be double-coded by 2 coders to establish intracoder and intercoder reliability. Following this process, the coding team will discuss discrepancies, which will ultimately be resolved through consensus. This process will continue until the intercoder reliability is at least 80%. Thereafter the remaining texts will primarily be coded by 1 analyst only, with one or more rounds of consistency checks and regular discussions among the coding team to ensure that coding remains standardized and reliable.

Comprehensive listings of all coded quotations for every code or code “family” will be generated in the qualitative analysis software. Coding memos will be used to summarize and explore the relationship between the constructs of the evaluation framework (eg, the relationship between implementation processes and hypothesized mechanisms of change, or between mechanisms of change and outcomes). Analysts will also review and analyze coded serial IDI content within each participant’s dataset to look specifically for patterns over time. The findings and interpretations of the data will be critically discussed until there is group consensus on the dominant themes and meanings contained in the data [74].

For each construct of the evaluation framework, findings from the qualitative and quantitative analyses will be integrated to build a more comprehensive picture of whether the target intervention inputs, processes, and outputs were achieved, whether these resulted in the hypothesized changes in mechanisms of action, whether these changes resulted in the intended outcomes, and how external contextual factors influenced implementation, thereby both evaluating the success of study implementation and testing the theory of change.

The study team continuously reviews recruitment and follow-up data to ensure compliance with the protocol and ethical standards. The IRDO study coordinator regularly monitors protocol compliance at study sites. The IRDO internal research monitor conducts quarterly spot checks of 100% of informed consent forms and 10% of data files. Any issues are documented in quality control logs and reviewed with the study staff. Protocol deviations or violations are reported to the study leadership and institutional review board (IRB) within 5 business days.

The trial is monitored by a data and safety monitoring board (DSMB) composed of 4 members: a statistician, a medical physician, a researcher with content expertise in IPV, and a researcher with expertise in PrEP use among adolescent girls and young women. All DSMB members are independent from any professional or financial conflict of interest with the research project, study investigators, and study institutions. The DSMB established a charter that outlines their responsibilities and meets approximately semiannually to review data quality and integrity, protocol adherence, participant safety, study conduct, and progress. On the basis of their review, they make recommendations on study continuation and modifications.

This study adheres to the World Health Organization Ethical and Safety Recommendations for Intervention Research on Violence Against Women [75]. All data collection staff receive training on IPV assessment, recognizing warning signs of immediate danger, discussing sensitive topics with participants, managing distressed participants, legal reporting requirements, and referral protocols. We aim to ensure that all communication with participants protects against unintentional disclosure of PrEP use, study participation, or reported IPV and does not place participants at additional risk of harm. To increase confidentiality, the study is presented to the community and to the participants as a study on HIV prevention methods in young women, and the focus on IPV is not advertised.

The MoH clinics conduct clinical safety monitoring for PrEP according to Kenyan guidelines. This study focuses on monitoring for adverse events that could be associated with the intervention, including IPV, social harms, serious adverse events, and unanticipated problems. All instances of these events are recorded on case report forms and followed up until resolution or 30 days after the participant exits from the study. These adverse events are reviewed monthly by the protocol team and reported to the IRB and DSMB according to their established protocols.

All staff have been trained to respond to any reports on IPV and social harms using the World Health Organization’s Listen, Inquire, Validate, Enhance Safety, and Support (LIVES) approach through referrals [76]. Staff follow up by phone or during subsequent visits to check on referral uptake and provide additional assistance if needed. Participants who report being in immediate danger or feeling unsafe returning home are offered an immediate escorted referral to vetted rescue or shelter services, and study staff follow up with the participant weekly until their situation is resolved or stabilized, or until 30 days after their exit from the study. If staff have concerns about the safety of a participant’s enrollment, they alert the site PI or designee to make a final determination about eligibility.

The study protocol and associated documents have been reviewed and approved by the Maseno University Ethical Review Committee (MUERC; MSU/DRPI/MUERC/00991/21). The RTI International Committee for Human Subjects Research established a reliance on MUERC for the study under a signed IRB authorization agreement. No amendment to the protocol or informed consent forms is implemented without prior ethics committee approval. Protocol violations are reported in writing to MUERC in accordance with their policy.

IRDO study staff obtain informed consent or assent from all participants taking part in the cRCT or IDIs. Study staff explain the study purpose and procedures, risks and benefits to the eligible participants, and compensation for their participation. They emphasize that their participation is voluntary and that there is no penalty if the individual decides not to participate. Participants aged ≥18 years provide written consent in English, Kiswahili, or Dholuo. The IRB has granted a waiver of parental consent for adolescent girls and young women participants who meet the criteria for “mature minors” outlined in MoH guidelines for Conducting Adolescent HIV Sexual and Reproductive Health Research in Kenya [77], and for male partners aged 15 to 17 years to take part in IDIs, because they constitute a minimal-risk research activity. All minor participants may choose to involve their parent in the consent process. For minor participants not meeting the above criteria, or who involve their parents, we obtain parental consent and child assent. The IRB has also granted a waiver of written consent for the brief and anonymous exit interviews focused on program acceptability.

All study procedures occur in private settings. Each participant receives a unique study identification number, and personally identifiable information does not appear on study data. All electronic data and study records are stored on secure, password-protected devices and uploaded to a secure RTI project server. Access to the system is restricted to trained study staff. Hard-copy data, consent and contact information forms with personally identifiable information are kept in separate double-locked files at the site and not transmitted to RTI.

To minimize confidentiality risks during the follow-up visits, we use contact methods agreed to by the participants, including aliases for study staff. To protect confidentiality during group activities, participants are asked to follow ground rules and pledge not to disclose others’ personal details outside the group. At each session’s start and end, participants are reminded to respect confidentiality and that there is a risk that others may disclose what they say.

Participants’ study information will not be released without their written permission, except as required by law or as necessary for review, monitoring, and auditing by the following:

Participants can withdraw from the study at any time for any reason. The PI, site PI, or designee may also withdraw participants for safety or noncompliance reasons, with the DSMB notified of such terminations. Efforts are made to complete a final evaluation for participants who withdraw. Staff document withdrawal reasons in study records. Participants who withdraw may rejoin and resume intervention or follow-up visits up to their original exit date, with PI and DSMB consultation.

The study may be suspended or terminated at any time by the NIH, RTI, site IRBs, or other authorities due to funding issues or substantial concerns, such as DSMB recommendations. If this occurs, the PI will be notified in writing with the reasons, and will inform participants, the IRB, and the sponsor. The study may resume if safety, protocol compliance, and data quality issues are resolved and approved by the relevant parties.