HIV-ASSIST Uses Algorithms in Appropriate ART-Regimen Selection

The US Department of Health and Human Services (HHS) provides guidelines on the complex process of selecting the most appropriate antiretroviral therapy (ART) for an individual patient living with HIV. There are a multitude of drugs and combinations available, and the number of options can be daunting for trainees and non-HIV specialists.

The investigators of a new study, published in Clinical Infectious Diseases, developed and tested an interactive decision-support program. They found that trainees in the study cohort using HIV-ASSIST were significantly more likely to choose appropriate ART regimens compared with those using guidelines alone.

HIV-ASSIST uses a multiple-criteria decision analysis framework to compare roughly 2000 possible ART combinations consisting of 2 to 4 drugs. The tool then generates a ranked list of ART recommendations individuated to the particular patient.

“Specifically, HIV-ASSIST algorithmically generates a ‘weighted score’ for every potential ART regimen by quantitatively evaluating individualized covariates (eg, resistance mutations, concurrent medications, comorbidities, HIV viral load, treatment history, etc.) to generate a composite utility score that attempts to maximize likelihood of viral suppression while optimizing tolerability (eg, minimizing drug interactions, pill burden, etc.,” the study authors explained.

The team conducted a randomized study of 118 medical students and internal medicine residents at John Hopkins University in early 2019. Participants with access to DHHS guidelines alone were compared with those accessing DHHS guidelines alongside HIV-ASSIST.

The investigators created 10 case-scenarios involving people with HIV, ranging from case-vignettes with ART-naïve patients to ART experienced patients. Scenarios included patients presenting with and without ongoing viremia, comorbidities, resistance, and concurrent medications.

Responses from 17 experienced HIV clinicians at hospitals around the country were used to define the list of possible appropriate ART selections. Selections were also considered appropriate in ART-naïve scenarios if they were consistent with HHS guidelines for initial treatment regimens.

All medical student participants had completed a curriculum on infectious diseases and microbiology which included lectures on ART pharmacology. The internal medicine resident participants had completed clinical rotations on an inpatient HIV unit. The participants were randomized 1:1 to either arm of the study. All participants were shown a 7-minute video on navigating HHS guidelines.

Out of the 118 participants, 62 used HIV-ASSIST and 56 were randomized to the HHS-alone group. In both study groups, a majority of participants reported that they had been involved in the care of fewer than 25 patients living with HIV thus far.

Ultimately, trainees randomized to HIV-ASSIST were more likely to select appropriate ART regimens. The median percentage of appropriate selections for HHS-alone was 40%, but for the HIV-ASSIST group the median percentage was 90%.

Despite better results, the HIV-ASSIST group also took less time to arrive at their selections. The median time in the HHS-alone arm was 36 minutes, compared with 21 minutes in the HIV-assist arm.

For ART-naïve patients, 89% of participants agreed or strongly agreed that HHS guidelines were helpful in guiding ART selections. Comparatively, 98% of HIV-ASSIST participants agreed or strongly agreed that HIV-ASSIST was useful in those cases.

For cases concerning ART-experienced patients, only 53% of participants considered HHS guidelines useful compared to 85% of HIV-ASSIST participants finding the tool useful.

Less than 4% of participants disagreed that HIV-ASSIST was useful for making ART selections in patients with comorbidities, resistance, or comedications.

In comparison, 27% of HHS-only participants disagreed with the statement that existing guidelines are useful in making ART selections for patients with comorbidities, 35% disagreed for patients with comedications, and 46% for patients with resistance mutations.

“Our results show that an interactive approach to guideline delivery through HIV-ASSIST was more effective at allowing trainees to apply HIV treatment principles and recommendations and may fill a needed niche for trainees and providers with less experience in HIV medicine,” study authors concluded.

With a growing shortage of HIV-care providers, a tool to educate students and assist those who are not specialists has a lot of potential.

While experienced infectious disease clinicians can balance guidelines with their own knowledge and familiarity with the latest research, HIV-ASSIST appears to make understanding what guidelines actually recommend far easier.