Implementation guidance needed for PrEP risk-prediction tools

Abstract

In a recent Comment,1Ortblad KF Baeten JM Electronic health record tools to catalyse PrEP conversations.Lancet HIV. 2019; 6: e644-e645Scopus (3) Google Scholar Ortblad and Baeten suggested that risk-prediction tools based on electronic health algorithms2Krakower DS Gruber S Hsu K et al.Development and validation of an automated HIV prediction algorithm to identify candidates for pre-exposure prophylaxis: a modelling study.Lancet HIV. 2019; 6: e696-e704Summary Full Text Full Text PDF PubMed Scopus (24) Google Scholar, 3Marcus JL Hurley LB Krakower DS Alexeeff S Silverberg MJ Volk JE Use of electronic health record data and machine learning to identify candidates for HIV pre-exposure prophylaxis: a modelling study.Lancet HIV. 2019; 6: e688-e695Google Scholar could catalyse patient-provider conversations about pre-exposure prophylaxis (PrEP). They asserted such tools “might normalise conversations around PrEP…and thus amplify PrEP use, reducing disparities in PrEP access…”. Whether this vision is realised depends on how such tools are implemented in practice; accompanying guidance is needed to avoid exacerbation of existing restrictions and disparities in PrEP access. Like questionnaire-based risk prediction, electronic-algorithm-based approaches rely on pre-established criteria to identify PrEP candidates. Because risk prediction is imperfect, use of these tools as screening mechanisms to decide whether to discuss PrEP with, or to offer PrEP to, clients inevitably excludes some people at risk. Moreover, though potentially mitigating some biases,1Ortblad KF Baeten JM Electronic health record tools to catalyse PrEP conversations.Lancet HIV. 2019; 6: e644-e645Scopus (3) Google Scholar, 2Krakower DS Gruber S Hsu K et al.Development and validation of an automated HIV prediction algorithm to identify candidates for pre-exposure prophylaxis: a modelling study.Lancet HIV. 2019; 6: e696-e704Summary Full Text Full Text PDF PubMed Scopus (24) Google Scholar, 3Marcus JL Hurley LB Krakower DS Alexeeff S Silverberg MJ Volk JE Use of electronic health record data and machine learning to identify candidates for HIV pre-exposure prophylaxis: a modelling study.Lancet HIV. 2019; 6: e688-e695Google Scholar automated risk prediction informed by people's self-disclosure of sensitive behaviour, self-identification as sexual minorities, or recorded medical history might bias selection against groups less trusting of providers, more inclined to maintain privacy around their sexuality, or less likely to be continuously engaged in a given health-care system. Therefore, providers should not use risk-prediction tools to determine PrEP candidacy, but rather to support people in evaluating their own candidacy. PrEP should be discussed with all clients as part of routine sexual health care.4Calabrese SK Krakower DS Mayer KH Integrating HIV pre-exposure prophylaxis (PrEP) into routine preventive healthcare to avoid exacerbating disparities.Am J Public Health. 2017; 107: 1883-1889Crossref PubMed Scopus (90) Google Scholar Shared decision-making resources that incorporate risk estimation and are conducive to routine discussions have shown early promise.5Krakower D, Powell V, Maloney K, Wong J, Wilson I, Mayer K. Impact of a personalized clinical decision aid on informed decision-making about HIV pre-exposure prophylaxis among men who have sex with men. 13th International Conference on HIV Treatment and Prevention Adherence; Miami, FL, USA; June 8–10, 2018 (abstr 106).Google Scholar Risk-prediction tools that could perpetuate selective PrEP education and access require accompanying implementation guidance to specify their intended use and limitations, to optimise their impact, and to avoid unintended consequences. I declare compensation from the DC Department of Health for input on provider training related to PrEP. Electronic health record tools to catalyse PrEP conversationsPre-exposure prophylaxis (PrEP) prevents HIV acquisition, and its scale-up reduces HIV infections at the population level.1–3 However, PrEP's impact currently falls short of its potential. In the USA, the Centers for Disease Control and Prevention (CDC) estimates that 1·1 million individuals are potentially eligible for PrEP but less than 10% of them are prescribed it.4 Globally, the gap is even greater. And, sadly, at the national and international level, PrEP use breaks along lines of established racial, economic, gender, and other disparities. Full-Text PDF Use of electronic health record data and machine learning to identify candidates for HIV pre-exposure prophylaxis: a modelling studyPrediction models using EHR data can identify patients at high risk of HIV acquisition who could benefit from PrEP. Future studies should optimise EHR-based HIV risk prediction tools and evaluate their effect on prescription of PrEP. Full-Text PDF Development and validation of an automated HIV prediction algorithm to identify candidates for pre-exposure prophylaxis: a modelling studyAutomated algorithms can efficiently identify patients at increased risk for HIV acquisition. Integrating these models into EHRs to alert providers about patients who might benefit from PrEP could improve prescribing and prevent new HIV infections. Full-Text PDF