Evaluation of a Differentiated Point-of-care Active Case Finding & Management Model to Eliminate Mother-to-Child Transmission of HIV in Malawi

Purpose: To conduct a hybrid effectiveness-implementation stepped-wedge trial to: 1. Estimate the effect of the PAC-Man model, compared to SOC, on the proportion of infants who receive timely EID testing. 2. Estimate the effect of the PAC-Man model, compared to SOC, on the proportion of pregnant and breastfeeding women living with HIV who receive guideline-adherent viral load testing. 3. Compare implementation outcomes between the PAC-Man model and SOC. Participants: The total study sample size is 2,426, including 2,304 mother-infant pair (MIP) survey participants, 48 high-risk pregnant/breastfeeding women living with HIV for in-depth interviews (from high-risk MIPs who participated in a survey); 10 PMTCT policy makers and Ministry of Health (MOH)/ President's Emergency Plan for AIDS Relief (PEPFAR) implementing partner senior managers for key informant interviews; and about 48-64 professional and lay health providers for focus group discussions and implementation actor surveys. To rigorously evaluate the PAC-Man model, investigators will use a hybrid (type 3) effectiveness-implementation incomplete stepped-wedge trial design. The hybrid incomplete stepped-wedge design is a rigorous quasi-experimental design that allows for incremental "rollout" of the PAC-Man model and serial measurement of both health outcomes (for infants and mothers in Objectives 1 and 2, respectively), as well as implementation outcomes (Objective 3). The incomplete stepped-wedge design is more efficient than a complete design, minimizes costs, and reduces participant burden, without sacrificing statistical precision. Using this design, sequential crossover from control (i.e., the SOC) to intervention (i.e., SOC plus the PAC-Man model) takes place at each 'step' using cluster randomization until all clusters (defined as a group of "sites," or health facilities plus the surrounding communities in its catchment area) receive the intervention. The 12 sites included in our study will be divided into 4 clusters of 3 sites each.