PathToScale: The PERSIST Trial

The study will test a package of implementation strategies to support continuation on pre-exposure prophylaxis (PrEP), with a particular focus on long-acting injectable cabotegravir (CAB LA). The study will prioritize impacts on PrEP continuation at 5-months and will secondarily consider alternate continuation endpoints, acceptability, feasibility, appropriateness and fidelity of implementation strategies among CAB LA users and providers. The study will use an efficient hybrid type 3 pragmatic cluster randomization. Clusters are defined as delivery sites, including primary health centers (PHCs) and private clinics serving the broader population and drop-in centers/community-based organizations providing CAB LA to prioritized populations including female sex workers, clients, men who have sex with men and transgender persons.

Clusters will be randomized to receive the standard of care (SoC) per Ministry of Health guidelines for PrEP continuation vs. an enhanced bundle of clinical support and provider-focused implementation strategies to promote sustained PrEP use. Variable-constrained randomization will be applied to balance allocation by facility type (drop-in centre for female sex workers, drop-in centre for men who have sex with men, primary health clinic, hospital, private), and overall clinic/PrEP volume.

PrEP users will be passively enrolled into the cluster randomized trial and a subset actively enrolled for participation in the targeted research activities. De-identified programmatic data will be utilized from OCR technology called ScanForm, which was developed by QED PrEP to identify the number and demographic characteristics (gender, age, indication for PrEP) of oral and CAB LA PrEP users initiating PrEP modalities by site, switching from oral to CAB LA PrEP (and vice versa), and continuation rates.

Outcomes of PrEP persistence will be followed for at least 5 months. Follow-up will begin at the time of injectable or oral PrEP initiation. With a 6-month passive enrollment period across sites and at least 5 months of follow-up of all clients initiating oral or injectable PrEP, this will result in 5-11 months of follow-up data per facility.

In addition to the utilization of programmatic data documenting PrEP continuation overall, and by PrEP modality (oral vs. CAB LA), a number of discrete research activities will be undertaken to evaluate the secondary objectives of the cluster randomized trial. These include: (1) serial cross-sectional surveys with PrEP clients; (2) In-depth qualitative interviews with PrEP clients; (3) Fidelity assessments of implementation strategies.

During a quantitative analysis, the primary endpoint (continuation of CAB LA at 5 months) will be compared as an intention-to-treat analysis across clusters of clinics implementing the standard of care continuation support for CAB LA vs. clinics implementing the enhanced continuation support package. The primary analysis will be an unpaired t-test to compare cluster-level proportions between the two study arms for each of the primary endpoints. We will further apply log-binomial regression to evaluate the difference in the study arms adjusting for stratification variables and individual-level characteristics that may be associated with the outcome. Similar analytical approaches will be followed for secondary endpoints.

For qualitative analysis, clients across priority populations and delivery channels will be identified for in-depth interviews (IDIs). We will enroll 100 clients to complete an IDI. With a sample size of around n=100, we promote feasibility and representation. Notably as this is not an ethnographic study, we will not seek to reach saturation in any of the groups, but rather to ensure that a diverse set of perspectives and experiences are represented which will allow for greater understanding of consistencies and variations in CAB LA implementation across populations and delivery channels.

Across the IDIs, we will audio record transcripts and use rapid qualitative analysis methods to efficiently translate research findings. Rapid analysis will include the write-up of detailed notes within 24 hours of the interview by the note- taker, and generation of an analytic memo utilizing structured templates within 2-business days of the interview by the moderator, listening to the audio recording where necessary for clarification; the note taker and moderator will then jointly review the memo and reach consensus. Matrices will be completed for each interview and memos detailing emerging themes created, allowing for within and across participant comparisons. A thematic analysis utilizing the memos and field notes will be applied deductively using CFIR domains and valence rating, with inductive themes added; audio recordings will be transcribed for further cross-reference and quotes. This approach, common in implementation science, will allow for more rapid identification and application of findings.