Predictive Analytics and Clinical Decision Support to Improve PrEP Prescribing in Community Health Centers (PrEDICT)

项目总结

PROJECT SUMMARY Rates of new HIV infections are disproportionately high, and uptake of preexposure prophylaxis (PrEP) low, in Black, Latino/a/x, and uninsured individuals in the US. Healthcare providers in community health centers (CHCs) could play a critical role in increasing PrEP prescribing to racial and ethnic minorities and other underserved populations. However, providers face barriers to PrEP prescribing, such as difficulty identifying candidates for PrEP; discomfort discussing sexual behavior; implicit biases about sexuality, race, and substance use; and lack of familiarity with PrEP care. We have found that providers are enthusiastic about the potential benefits of decision support tools to mitigate these barriers to PrEP provision, and that patients would find such tools acceptable if implemented sensitively. We previously showed that data from electronic health records (EHRs) can be used to identify patients at increased risk of HIV acquisition in two large, general practice healthcare systems. In our formative R34 research, we expanded on this approach in a safety-net setting, incorporating strategies to support not only identification of PrEP candidates but also PrEP discussions and prescribing. In a national network of CHCs serving 6.2 million patients in 46 states (OCHIN), we used machine learning with EHR data to identify patients at increased risk for incident HIV diagnosis (area under the curve 0.84). Using stakeholder-engaged qualitative methods, we then built an EHR-based decision support tool that uses our prediction model to prompt PrEP discussions with patients likely to benefit. The tool features a suite of resources to support initial PrEP prescribing, including suggested language for patient-centered discussions; information about PrEP indications, formulations, and dosing; laboratory order sets; diagnosis codes; and automated clinical notes. We piloted this tool at 3 CHCs, establishing feasibility and acceptability. We now propose Predictive Analytics and Clinical Decision Support to Improve PrEP Prescribing in Community Health Centers (PrEDICT) to evaluate the impact of our tool on PrEP provision in OCHIN CHCs. Our specific aims are to 1) expand and refine the decision support tool to facilitate PrEP follow-up care, and therefore patients’ persistence on PrEP; 2) quantify the impact of the decision support tool on PrEP initiation and persistence in a pragmatic stepped-wedge trial across 16 CHCs; and 3) identify patient populations with whom providers are less inclined to discuss PrEP when prompted to do so, and explore facilitators and barriers to equitable selection of patients for PrEP discussions. We will engage a diverse advisory group of patients from OCHIN CHCs in tool expansion, refinement, and implementation. This project is innovative in its use of predictive analytics and decision support to improve PrEP provision in safety-net settings. The research is significant because it has the potential to facilitate large increases in PrEP utilization using highly scalable tools. Our intervention addresses NIH priorities, aligns with the federal Ending the HIV Epidemic initiative, and could become a best practice for how CHCs and other healthcare systems support PrEP care delivery.