Results of a systematic review and meta-analysis published in JAMA Cardiology showcardiovascular disease (CVD) risk prediction models tend to underpredict CVD risk in patients with HIV infection. These findings reinforce recommendations from the American College of Cardiology and American Heart Association to consider HIV infection as a risk factor for CVD.
Researchers at the University of Maryland School of Medicine and Brown University searched publication databases through January 2021 for studies about CVD risk calculation and CVD outcomes in the setting of HIV. A total of 11 publications from 9 epidemiologic studies were included in this analysis.
The study population comprised 75,304 individuals, of whom the mean age was 42 years, 79% were men, 59% were White, 33% were Black, and 8% were Hispanic. Most (89%) of the study population were receiving antiretroviral therapy. The mean CD4+ count was 449 (range, 397-501) cells/µL, and the mean nadir CD4+ count was 207 (range, 95-220) cells/µL.
The researchers evaluated 6 general CVD risk prediction tools: Copenhagen risk score, Cuore Project, Framingham Risk Score (FRS), Framingham-Registre Gironí del COR, Pooled Cohort Equation (PCE), and Systemic Coronary Risk Evaluation. Four HIV-specific tools were also evaluated, including the 2010 Data Collection on Adverse Effects of Anti-HIV Drugs (D:A:D) risk prediction model, the 2016 D:A:D model, HIV Myocardial Infarction score, and Veterans Aging Cohort Study Index.
The majority of the models had moderate power to discriminate between higher and lower CVD risk (area under the receiver operating characteristic curve range, 0.7-0.8). The 3 best CVD risk prediction models were the 2016 D:A:D model, FRS, and the PCE 10-year model.
Despite moderate performance, the researchers reported CVD risk was underpredicted by 20% to 50% in most models, with the exception of the FRS and PCE 10-year models. In general, 5-year predictions of CVD risk were more prone to underprediction.
There were insufficient data to perform subgroup analyses for model prediction, though some evidence suggested the FRS and PCE models performed better among Black individuals compared with White individuals.
Limitations of this analysis include insufficient data to perform comprehensive comparative analyses.
According to the researchers, “As established in current guidelines, clinicians should consider HIV as a risk-enhancing factor when assessing CVD in people living with HIV using these risk calculators.”
Soares C, Kwok M, Boucher K-A, et al. Performance of cardiovascular risk prediction models among people living with HIV: a systematic review and meta-analysis. JAMA Cardiol. Published online December 28, 2022. doi:10.1001/jamacardio.2022.4873