Abstract Background/Introduction People with HIV (PWH) are at elevated cardiovascular risk, but existing calculators have poor calibration for this population. The American Heart Association developed new prediction equations to replace the pooled cohort equations (PCE), but these have not been validated among PWH. Purpose To evaluate the discrimination and calibration of the Predicting Risk of cardiovascular disease EVENTs (PREVENT) equations compared to the PCE in a large multicenter cohort study of PWH with adjudicated cardiovascular events. Methods Within the Center for AIDS Research Network of Integrated Clinical Systems (CNICS) Cohort study, we included individuals aged 40-75 with available data to calculate their cardiovascular risk from 2001 to 2021. We calculated predicted 10-year atherosclerotic cardiovascular disease (ASCVD) risk at baseline using both the pooled cohort equation (PCE) and the PREVENT 10-year ASCVD base equation. Individuals were followed until first myocardial infarction, stroke (cerebrovascular event), death, or end of follow-up. Myocardial infarctions and strokes were adjudicated according to standard protocols. To assess discrimination, we calculated Harrell’s C-index and to assess calibration we used the Greenwood-Nam-D’Agostino goodness-of-fit tests. As the cause of death was not adjudicated, only myocardial infarction and stroke were included as events with individuals censored at time of death; therefore, the PCE and PREVENT would be expected to predict more events than are observed since they also include cardiovascular-specific mortality. Results We included 13,135 individuals from 5 sites across the United States. The mean age at enrollment was 44 years and 18% were female (Table). The mean predicted 10-year ASCVD risk was 5.8% by PCE and 2.9% by PREVENT. Over a mean 5.7 (SD 3.5) years of follow-up, there were 628 individuals who had a myocardial infarction or stroke, with 502 myocardial infarctions and 131 strokes (5 concurrent events). Discrimination was improved with PREVENT compared to the PCE, with Harrell’s C-indexes of 0.722 (95% CI 0.701, 0.741) and 0.708 (95% CI 0.687, 0.729), respectively (p=0.008). Both equations underpredicted risk, with 766 events predicted by PCE and 385 events by PREVENT compared to 1037 calculated observed events (p0.001 for both). The calibration slopes were 0.932 for PCE and 1.998 for PREVENT, respectively (Figure). Conclusions Among this cohort of PWH in the United States, the PREVENT 10-year ASCVD base equation had better discrimination than the PCE but was poorly calibrated and severely underestimated composite risk for myocardial infarction and stroke, with observed risks more than double predicted risks.Table.Baseline Cohort Characteristics Figure.Calibration plots
Durstenfeld et al. (Sat,) studied this question.