Validation of the 2016 USPSTF recommendations for primary cardiovascular prevention in a large contemporary cohort

Abstract

Aims The aim of this study was to evaluate the performance of the US Preventive Services Task Force (USPSTF) cholesterol recommendations in a contemporary non-US cohort. Methods and results This is a historical cohort analysis of electronic records from Israel's largest health provider. All patients in the Tel Aviv district eligible for primary cardiovascular prevention were followed between January 2005 and December 2015. Risk was estimated by the pooled cohort equations. Statin eligibility was determined by USPSTF and American College of Cardiology and American Heart Association (ACC/AHA) recommendations. Atherosclerotic cardiovascular disease events were retrieved from electronic registration. The mean ± standard deviation age of the 10,889 (98,258 person-years) participants was 60.3 ± 9.4 years, and 69.1% were women. Outcome events were recorded for 1351 patients (12.4%). Treatment recommendations were discordant in 901 patients (8.3%) whose treatment was indicated only by the ACC/AHA guidelines, implying a 26% reduction in newly eligible patients for statin treatment had the USPSTF recommendations been implemented. Among the statin-naive patients, the pooled cohort equations underestimated the risk, with a predicted-to-observed event ratio of 0.88. The recommended treatment thresholds provided excellent calibration, with ratios of 1.0 for USPSTF and 0.98 for ACC/AHA-eligible patients. Both models showed similar discrimination (Harrel's C = 0.63 (0.62–0.65) for USPSTF vs. 0.64 (0.63–0.66) for ACC/AHA, P = 0.26). The USPSTF recommendations were less sensitive and more specific for the detection of outcome events than the ACC/AHA recommendations (61% vs. 75% and 68% vs. 55%, respectively). The net reclassification index was −0.01. Conclusions Calibration, discrimination and net reclassifications were very similar for USPSTF and ACC/AHA recommendations. Applying the USPSTF recommendations could reduce over-treatment.