Objective: To identify and propose the optimal waist circumference cut-off points (WCp) for the diagnosis of central obesity (CO) in a Brazilian population, so as to compare these cut-off points with those recommended by the ATPIII (WC-ATPIII), and to estimate the difference in prevalence rates of metabolic syndrome (MS) using the two criteria.
Methods: Cross-sectional study conducted in a population subgroup of 1439 adults in the city of Salvador, Brazil. ROC curves of waist circumference (WC) were plotted to identify diabetes mellitus (DM) and obesity. ROC curve sensitivity and specificity values >60% and the closest to each other were used to define WCp. The prevalence of MS was estimated using WCp and WC-ATPIII.
Results: Eight hundred and twenty nine women comprised 57.7% of the sample. The WCp selected were 84 cm for women and 88 cm for men. These cut-off points detected DM with a 68.7% and 70% sensitivity, and a 66.2% and 68.3% specificity, respectively. For obesity, sensitivity and specificity were 79.8% and 77.6% in women and 64.3% and 71.6% in men, respectively. Using WC-ATPIII, 88 cm (for women) and 102 cm (for men), the sensitivity was 53.3% and 26.5%, respectively, to diagnose DM. For obesity, sensitivity was 66.5% (for women) and 28.6% (for men). The prevalence of MS using WCp was 23.7%, 95%CI (21.6-25.9), whereas using WC-ATPIII it was 19.0%, 95%CI (17.1-20.9), 1.2 times higher using WCp.
Conclusion: WC-ATPIII were inappropriate and underestimated the prevalence of MS in the population studied, particularly among men. We suggest that the WC cut-off points > 84 cm for women and > 88 cm for men should be tested in other Brazilian populations.