PROJECT SUMMARY Louisiana exemplifies the disparity between HIV pre-exposure prophylaxis (PrEP) need and uptake in the South, ranking 4th among US states in HIV incidence in 2018 while ranking 46th in PrEP uptake the following year. To date, few solutions have emerged to address barriers to optimal PrEP utilization in Louisiana and the South overall. Our team has previously demonstrated proof-of-concept of the utility of electronic health record (EHR)-based machine learning (ML) algorithms for identifying incident HIV cases (surrogate for PrEP candidates) within healthcare systems, outperforming current Centers for Disease Control and Prevention (CDC) PrEP indication guidelines. This promising methodology has never been implemented in a Southern healthcare system, and the best approach for incorporating health system-based EHR risk prediction models into community HIV prevention efforts is unclear. The proposed project seeks to evaluate two novel approaches to expanding EHR-based model implementation beyond their originating health systems and into the communities they serve: 1) an asynchronous strategy involving study team and local community-based personnel notifying community members at risk of HIV infection using a monthly report generated by the EHR risk model 2) a real-time strategy using best practice advisories to alert ED and UC providers of persons flagged as increased risk for HIV by the model during acute care encounters. We will test these strategies within two healthcare systems in Southeastern Louisiana: LCMC Health in New Orleans and Our Lady of the Lake Health in Baton Rouge. To capture a high HIV risk population, the study will focus on persons in the health system who exclusively engage the health system through emergency department (ED) and urgent care (UC) encounters. The project’s specific aims are to: 1) Derive and validate an EHR-based HIV risk prediction model utilizing clinical data from ED and UC encounters in two Southeastern Louisiana health systems. 2) Develop stakeholder-informed implementation strategies for extending the reach of the EHR-based prediction model beyond the health system. 3) Evaluate feasibility and acceptability of two community-facing implementation approaches to EHR HIV risk prediction model deployment. Aim 1 will adapt our EHR-based risk prediction model into the local HIV epidemiologic context. Aim 2 will obtain key stakeholder input to guide the development of culturally-responsive strategies for risk status notification of at-risk individuals identified by the model. Aim 3 will feature a pilot implementation trial to assess the two implementation strategies: To execute these objectives, we have assembled a multidisciplinary team of experts in HIV health services research, HIV prevention epidemiology, health informatics and implementation science. This team will partner with key community-based organizations (Camp ACE of the St. John 5 Missionary Baptist Church in New Orleans and Metro Health of Baton Rouge), to leverage the power and reach of health system EHR towards empowering community members with the data they need to make informed decisions about using PrEP.

PROJECT NARRATIVE HIV pre-exposure prophylaxis (PrEP) uptake in Louisiana is among the lowest in the country, and few solutions have been developed to address the unique barriers to optimal PrEP uptake in the state. This project focuses on the use of machine learning algorithms embedded within the electronic health record (EHR) of large health systems in Southeastern Louisiana to identify persons at increased risk for HIV infection. The project will also evaluate approaches to best utilize the output from these algorithms to inform effective health-system based HIV risk status notification strategies on a population level, with input from community stakeholders.