Little contribution of conventional factors in an algorithm to predicting death risk in Turkish adults

Abstract

Objective: Determinants of risk of death are highly relevant for the management strategy of individuals. We aimed to determine an algorithm for predicting risk of death in Turkish adults who have a high prevalence of metabolic syndrome (MetS). Methods: Nine-year risk of death was estimated in 3348 middle-aged adults, followed over 8.81 +/- 4.2 years. Cox proportional hazard regression was used to predict risk of death. Discrimination was assessed using C-statistics. Results: Death occurred in 565 subjects. In multivariable analysis, high-density lipoprotein (HDL) and non-HDL cholesterol levels were not predictive in either sex; in women, current smoking was also not predictive. Age, presence of diabetes, systolic blood pressure >= 160 mmHg and low physical activity were predictors in both sexes, beyond smoking status in men. Exclusion of coronary disease at baseline did not change risk estimates materially. Using an algorithm of the stated 7 variables showed an 11- to 18-fold spread in the absolute risk of dying among individuals in the highest than in the lowest of 4 risk score categories. C-statistics of the model using age alone was 0.790 inmen, 0.808 in women (p < 0.001 each), while the incorporation of 6 conventional risk factors contributed to C-index was >0.020 in males and 0.009 in females. Conclusions: In a middle-aged population with prevalent MetS, serum lipoproteins and, in women, smoking status, were not relevant for the risk of death. The contribution of conventional risk factors beyond age to estimating risk of death was modest among Turkish men, and little in women in whom autoimmune activation is operative.