Estimation and development of 10-and 20-year cardiovascularmortality risk models in an industrial male workers database

摘要

We examined the performance of the Framingham Heart Study (FHS) and the European Systematic Coronary Risk Evaluation (SCORE) models for cardiovascular disease (CVD) mortality prediction in Israeli industrial workers, and developed and validated new risk prediction models for CVD mortality incidence in the same population. Our database was a longitudinal Israeli industrial cohort (CORDIS cohort) of 4809 adultmales followed-up for 22 years. Performance of the FHS and the SCORE prediction models was analyzed by insertion of the CORDIS cohort measurements to each model separately. The standard prognostic variables and results obtained from the new refined Cox regression analyses were used to construct two new 10- and 20-year CVD mortality risk scoring systems: a modified FHS model (FHS/Cox) and an omnibus model with Cox regression (Omnibus/Cox). The SCORE model of high-risk and low-risk charts yielded 10-year mortality mean risks of 1.12% and 0.64%, respectively, for male subjects aged > 30 years. The new FHS/Cox and Omnibus/Cox models generated a mean predictive 10-year risk of 1.12% and 1.50%, respectively. The mean 20-year risk predicted by the new FHS/Cox and the Omnibus/Cox models was 2.66% and 3.75%, respectively. Internal validation of both models demonstrated a high and stable area under the receiver operating characteristic curve > 0.85. No significant differences were found between the two models. In conclusion, the CVD mortality risk prediction scoring systems tailored for the Israeli workers population demonstrated good performance. Additional studies to externally validate these algorithms will indicate which of these quantitative risk estimation platforms should be used in specific settings. (C) 2017 Elsevier Inc. All rights reserved.