In the Journals

EHRs can help identify potential PrEP candidates

Julia L. Marcus, PhD, MPH,
Julia L. Marcus

Researchers identified ways that clinicians may be able to use electronic health records to identify candidates for HIV prevention medication.

According to two studies published recently in The Lancet HIV, prediction models that use EHR data and automated algorithms that can be integrated into EHRs both effectively identified patients at high risk for HIV who could benefit from pre-exposure prophylaxis (PrEP).

“PrEP is more than 90% effective in preventing HIV infection, but of 1.1 million people in the United States who are eligible for PrEP, only 7% used it in 2016,” Julia L. Marcus, PhD, MPH, assistant professor of population medicine at Harvard Medical School and Harvard Pilgrim Health Care Institute, told Infectious Disease News.

“The federal government recently announced the Ending the HIV Epidemic initiative, which aims to reduce new HIV infections by 90% by 2030. Scaling up PrEP is a key pillar of that initiative. One barrier to PrEP prescribing is that providers have difficulty identifying patients who are at high risk of HIV acquisition ... Automated prediction tools that use EHR data to identify potential PrEP candidates could shift the paradigm for how PrEP is prescribed in the U.S., ultimately improving PrEP uptake and reducing new HIV infections.”

Model predicts 38% of cases

In one study, Marcus and colleagues assessed a population of more than 3.75 million HIV-uninfected adults at Kaiser Permanente Northern California who were not yet using PrEP and had at least 2 years of previous health plan enrollment with at least one outpatient visit from Jan. 1, 2007, to Dec. 31, 2017. According to the study, they used 81 EHR variables — all potential predictors of HIV risk — to predict new HIV diagnoses within 3 years in a subset of patients who entered the cohort from 2007 through 2014.

Marcus and colleagues reported that the model flagged 2.2% of the general patient population as potential PrEP candidates, and identified 38.6% of the 83 incident HIV cases, including nearly half of the 69 male HIV cases, but none of the 14 female cases.

Marcus noted that the full model outperformed simpler models that included only MSM status and STI positivity, equally predicting HIV risk for black and white patients, whereas the simpler models identified fewer black cases.

“Although risk prediction tools are imperfect and cannot replace the clinical judgment of skilled providers, our model may be substantially more efficient than efforts to identify PrEP candidates in current practice,” Marcus said. “Our study suggests that HIV prediction models could be embedded into EHRs as an automated screening tool to help providers identify the subset of their patients who are most likely to benefit from PrEP.”

Douglas S. Krakower, MD
Douglas S. Krakower

Algorithm could have prevented 40% of infections

The results were similar to findings from a second study in which Douglas S. Krakower, MD, from the division of infectious diseases at Boston’s Beth Israel Deaconess Medical Center, and colleagues developed an automated algorithm to identify individuals at increased risk for HIV infection using EHR data from more than 1.1 million patients between 2007 and 2015 at Atrius Health.

“We envisioned an automated algorithm as a tool to prompt clinicians about individual patients who are most likely to benefit from PrEP, instead of relying exclusively on each provider’s or patient’s own initiative to engage in discussions about HIV risk and PrEP,” Krakower told Infectious Disease News. “This could reduce missed opportunities to discuss and offer PrEP and improve providers’ efficiency in dedicating time to this area of preventive health care.”

The study compared 42 candidate machine learning algorithms for identifying individuals at high risk for HIV acquisition, using 180 predictors for risk that were drawn from EHR data. Krakower and colleagues validated the best performing algorithm prospectively at Atrius Health using 2016 data and externally at Fenway Health, an independent community health center in Boston, calculating HIV risk scores for HIV-uninfected patients not on PrEP, according to the study.

They prospectively identified 1.8% of 536,384 patients at Atrius Health in 2016 and 15.3% of 28,702 Fenway Health patients as potential PrEP candidates. The researchers deduced that nearly 40% of all HIV infections at Atrius Health in 2016 could potentially have been averted if clinicians had discussed and offered PrEP to the 2% of patients with the highest HIV risk scores.

“The [U.S. Preventive Services Task Force] has called for validated tools to help providers identify potential candidates for PrEP given the urgent need to improve PrEP implementation in the U.S. Our model holds promise to empower providers to identify potential candidates and prescribe PrEP more frequently, which could prevent new HIV infections.” – by Caitlyn Stulpin

Disclosures: Krakower reports conducting research supported by Gilead Sciences. Marcus reports receiving research grants support from Merck and consulting on a research grant from Gilead Sciences. Please see the studies for all other authors’ relevant financial disclosures.

Julia L. Marcus, PhD, MPH,
Julia L. Marcus

Researchers identified ways that clinicians may be able to use electronic health records to identify candidates for HIV prevention medication.

According to two studies published recently in The Lancet HIV, prediction models that use EHR data and automated algorithms that can be integrated into EHRs both effectively identified patients at high risk for HIV who could benefit from pre-exposure prophylaxis (PrEP).

“PrEP is more than 90% effective in preventing HIV infection, but of 1.1 million people in the United States who are eligible for PrEP, only 7% used it in 2016,” Julia L. Marcus, PhD, MPH, assistant professor of population medicine at Harvard Medical School and Harvard Pilgrim Health Care Institute, told Infectious Disease News.

“The federal government recently announced the Ending the HIV Epidemic initiative, which aims to reduce new HIV infections by 90% by 2030. Scaling up PrEP is a key pillar of that initiative. One barrier to PrEP prescribing is that providers have difficulty identifying patients who are at high risk of HIV acquisition ... Automated prediction tools that use EHR data to identify potential PrEP candidates could shift the paradigm for how PrEP is prescribed in the U.S., ultimately improving PrEP uptake and reducing new HIV infections.”

Model predicts 38% of cases

In one study, Marcus and colleagues assessed a population of more than 3.75 million HIV-uninfected adults at Kaiser Permanente Northern California who were not yet using PrEP and had at least 2 years of previous health plan enrollment with at least one outpatient visit from Jan. 1, 2007, to Dec. 31, 2017. According to the study, they used 81 EHR variables — all potential predictors of HIV risk — to predict new HIV diagnoses within 3 years in a subset of patients who entered the cohort from 2007 through 2014.

Marcus and colleagues reported that the model flagged 2.2% of the general patient population as potential PrEP candidates, and identified 38.6% of the 83 incident HIV cases, including nearly half of the 69 male HIV cases, but none of the 14 female cases.

Marcus noted that the full model outperformed simpler models that included only MSM status and STI positivity, equally predicting HIV risk for black and white patients, whereas the simpler models identified fewer black cases.

“Although risk prediction tools are imperfect and cannot replace the clinical judgment of skilled providers, our model may be substantially more efficient than efforts to identify PrEP candidates in current practice,” Marcus said. “Our study suggests that HIV prediction models could be embedded into EHRs as an automated screening tool to help providers identify the subset of their patients who are most likely to benefit from PrEP.”

PAGE BREAK
Douglas S. Krakower, MD
Douglas S. Krakower

Algorithm could have prevented 40% of infections

The results were similar to findings from a second study in which Douglas S. Krakower, MD, from the division of infectious diseases at Boston’s Beth Israel Deaconess Medical Center, and colleagues developed an automated algorithm to identify individuals at increased risk for HIV infection using EHR data from more than 1.1 million patients between 2007 and 2015 at Atrius Health.

“We envisioned an automated algorithm as a tool to prompt clinicians about individual patients who are most likely to benefit from PrEP, instead of relying exclusively on each provider’s or patient’s own initiative to engage in discussions about HIV risk and PrEP,” Krakower told Infectious Disease News. “This could reduce missed opportunities to discuss and offer PrEP and improve providers’ efficiency in dedicating time to this area of preventive health care.”

The study compared 42 candidate machine learning algorithms for identifying individuals at high risk for HIV acquisition, using 180 predictors for risk that were drawn from EHR data. Krakower and colleagues validated the best performing algorithm prospectively at Atrius Health using 2016 data and externally at Fenway Health, an independent community health center in Boston, calculating HIV risk scores for HIV-uninfected patients not on PrEP, according to the study.

They prospectively identified 1.8% of 536,384 patients at Atrius Health in 2016 and 15.3% of 28,702 Fenway Health patients as potential PrEP candidates. The researchers deduced that nearly 40% of all HIV infections at Atrius Health in 2016 could potentially have been averted if clinicians had discussed and offered PrEP to the 2% of patients with the highest HIV risk scores.

“The [U.S. Preventive Services Task Force] has called for validated tools to help providers identify potential candidates for PrEP given the urgent need to improve PrEP implementation in the U.S. Our model holds promise to empower providers to identify potential candidates and prescribe PrEP more frequently, which could prevent new HIV infections.” – by Caitlyn Stulpin

Disclosures: Krakower reports conducting research supported by Gilead Sciences. Marcus reports receiving research grants support from Merck and consulting on a research grant from Gilead Sciences. Please see the studies for all other authors’ relevant financial disclosures.