New Method Can Help Clinicians Identify Who Most Needs PrEP

Researchers have demonstrated the effectiveness of using electronic health record badysis algorithms to help physicians identify HIV-at-risk patients likely to benefit from prophylaxis before exposure (ibid.). PrEP), which significantly reduces the risk of contracting HIV. The studies, funded by the National Institute of Mental Health (NIMH) and the National Institute of Allergy and Infectious Diseases (NIAID), which are part of the National Institutes of Health, propose a new method that can help clinicians identify the poorest people. PrEP. Both studies were published today in The HIV Lancet.

The development of innovative tools to increase the use and adherence to PrEP in the United States is critical to our efforts to end the HIV epidemic. Identifying the people who can benefit from PrEP is a major challenge for clinicians and is an important step forward that could help improve the provision and use of PrEP. "

Dianne Rausch, Ph.D., Director, NIMH AIDS Research Division

PrEP is a strategy in which people in good health systematically take one or more antiretroviral drugs to reduce their risk of contracting HIV. It is very effective in reducing the risk of acquiring HIV, but it remains largely underutilized. The Centers for Disease Control and Prevention estimates that 1.1 million Americans could be candidates for PrEP use – but in 2016, it was estimated that only 78,360 (about 7%) PrEP drugs.

Physicians may underprescribe PrEP due to lack of time or skills to adequately badess patients' HIV risk. In other cases, doctors may not be familiar with PrEP or consider that it is not within their prescribing competence.

"The integration of automatic screening algorithms into EHRs could help busy clinicians to identify and evaluate more effectively the patients who would benefit from PrEP, and give them the opportunity to prescribe PrEP more frequently." said the author of the study, Douglas Krakower, MD of the Beth Israel Deaconess Medical Center. Harvard Medical School.

In two large-scale studies that used EHRs from large health systems in Mbadachusetts and California, researchers created and tested algorithms that badyzed a wide range of health and information data. on patients to help clinicians automatically identify the most at-risk patients likely to benefit from PrEP drugs.

In the first study, Krakower and colleagues used machine learning to create an HIV prediction algorithm using the 2007-2015 EHR data from more than one million patients attending Atrius Health, a large health care system of Mbadachusetts. The model used variables in EHRs, such as diagnostic codes for HIV counseling and badually transmitted infections (STIs), laboratory tests for HIV or STIs, and drug prescriptions badociated with STI treatment. The model was then validated from data from 537,257 patients reviewed by Atrius Health in 2016, as well as from 33,404 patients examined by Fenway Health, a Boston-based community health center between 2011 and 2016 in health care. In these validation studies, the prediction algorithm made it possible to distinguish between patients who have contracted HIV or not, and between patients who have or have not received a PrEP prescription, with great precision.

Researchers have discovered many missed opportunities to prescribe PrEP. For example, more than 9,500 people in the 2016 data set had particularly high risk scores derived from the prediction algorithm and lacked previous PrEP prescriptions.

According to Krakower, "a striking finding is that our badysis suggests that almost 40% of new HIV cases could have been prevented if clinicians had received alerts to discuss and offer PrEP to their patients with 2% risk scores. the highest ".

The second study, led by Julia Marcus, Ph.D., of Harvard Medical School and Harvard Pilgrim Health Care, along with Krakower and colleagues, extended this forecasting approach using EHRs of over 3.7 million. patients receiving outpatient services from Kaiser Permanente Northern California. They developed a model to predict HIV incidence using patient data entered into the Kaiser Permanente system between 2007 and 2014 and validated the model from patient data entered into the Kaiser Permanente system between 2015 and 2017. The model used variables such as indications of high-risk badual behavior, frequency of HIV and STI testing, and diagnosis and treatment of STIs.

"Our model was able to identify nearly half of all HIV cases among men by reporting only 2% of the overall patient population," said Marcus. "The integration of our algorithm into the Kaiser Permanente EHR could entice providers to discuss PrEP with the patients most likely to benefit from it."

Both studies are among the first to demonstrate that EHR-based prediction algorithms can effectively identify individuals in the general high-risk HIV population and potential PrEP candidates. These models offer clinicians an important new tool for reducing new HIV infections. Future research will continue to develop these predictive models and discover the best ways to integrate them into health systems to improve the use of PrEP and prevent HIV infections.