Predictive accuracy of the ‘Framingham’s general CVD algorithm’ in a Middle Eastern population: Tehran Lipid and Glucose Study

Abstract

To examine the performance of 'the Framingham's general CVD algorithm' in estimating 5-year risk of cardiovascular disease (CVD) and coronary heart disease (CHD) in an adult Middle Eastern community. In a population-based prospective cohort, the Tehran Lipid and Glucose Study (TLGS), CVD risk estimates were calculated for 3838 individuals (2183 women) free of CVD at baseline. Over 8.6 years, there were 119 CVD (106 CHD) events among women, and 164 (137) among men aged ≥ 30 years. Standard risk factors [age, systolic blood pressure (SBP), antihypertensive treatment, total and high-density lipoprotein cholesterol, smoking and diabetes mellitus], measured at baseline, were significantly related to the incidence of CVD. The only exception, lack of significance for coefficients of SBP among women, was attributable to collinearity between age and SBP. For most risk factors, hazard ratios (HRs) for CVD were similar for the TLGS and the Framingham cohorts' participants. Few exceptions were lower HR for SBP among TLGS' women and lower HR for age among TLGS' men. The Framingham equations quite correctly discriminated participants [C statistic: CVD, 0.818 (women) and 0.774 (men); CHD: 0.822 (women) and 0.751 (men)] with good calibration (all calibration χ(2) values < 15, ps > 0.1). The Framingham's CVD algorithm was effective at ranking individuals and could be used to quantify risk and to guide preventive care in Iranian adults. For regions without established cohort, recalibration using risk factors and CVD (CHD) rates may be an effective method to develop CVD (CHD) risk prediction algorithms for local practice.