Background: Persons living with human immunodeficiency virus (HIV) have greater risks for myocardial infarction (MI) than uninfected persons. These elevated MI risks, which are associated with both traditional and HIV-specific risk factors, may limit the applicability of cardiovascular risk prediction models for HIV-infected persons.
Methods: We evaluated the ACC/AHA atherosclerotic cardiovascular disease (ASCVD) Risk Estimator’s Pooled Cohort Equations (PCEs) and a new model incorporating HIV-specific variables using the Centers for AIDS Research (CFAR) Network of Integrated Clinical Systems (CNICS) multi-center U.S.-based cohort of HIV-infected adults. CNICS has rigorous central adjudication of type I and II myocardial infarction (MI), which may be distributed differently in HIV-infected versus uninfected populations. Because adjudicated stroke data were not available, ASCVD risks predicted by the PCEs were scaled to determine predicted MI rates using race-sex-specific proportions. MI rates predicted by the PCEs and the new model were compared with observed MI rates in CNICS. Harrell’s C-statistic and Hosmer-Lemeshow chi-square were used to determine discrimination and calibration of these models.
Results: Of 11,901 HIV-infected persons in CNICS with baseline data required for the PCEs, 180 experienced MIs over a mean follow-up of 4.3 years. The PCEs were poorly calibrated for white men and black women (Hosmer-Lemeshow chi-square = 25.3 and 42.9, respectively) and had inadequate discrimination for black men (Harrell’s C statistic = 0.65). A modified, data-derived model had excellent discrimination (Harrell’s C-statistic for men = 0.81; for women = 0.83) and was well calibrated (Figure).
Conclusions: A data-derived ASCVD risk estimator that incorporates HIV-specific coefficients had excellent discrimination and calibration in a large multi-center U.S. HIV cohort. Future studies are needed to validate this model in other HIV cohorts and evaluate the contributions of type I versus II MI to ASCVD risk.