To determine optimum anthropometric cutoffs for predicting the likelihood ratios of type 2 diabetes mellitus (DM) and hypertension (HT) in Mexicans. Data from a randomly selected, nationally representative health survey (2000) with 11,730 men [37.4 (+/- 12.9) years] and 26,647 women [37.3(+/- 12.9) years] were assessed for values of body mass index (BMI) and waist circumference (WC) for predicting DM or HT by receiver operating characteristic curve analyses. Likelihood ratios for DM and HT were calculated, and BMIs or WCs for public-health screening were developed. Subanalyses included regional data. Likelihood ratios of DM and HT increased from BMI values of 22 to 24 kg/m(2) in both sexes and with WC values of 75 to 80 cm in men and 70 to 80 cm in women. The best BMI cutoffs for predicting DM were 26.3 to 27.4 kg/m(2) in men and 27.7 to 28.9 kg/m(2) in women, with similar values for HT, i.e., 26.2 to 27.0 kg/m(2) and 27.7 to 28.5 kg/m(2), for men and women, respectively; WC cutoffs for DM were 93 to 98 cm in men and 94 to 99 cm in women, and cutoffs for HT were 92 to 96 cm and 93 to 96 cm for men and women, respectively. The WC cutoffs had higher sensitivity and specificity than those of BMI. The risk for DM and HT starts at lower levels of BMI and WC than those suggested by WHO. WC is a better discriminator than BMI measures for use in public health.
