Dear Sir: In a recent issue of the Journal, Wannamethee et al (1) described the results of an interesting cross-sectional study on the association between anthropometric measures and metabolic abnormalities in 2924 men aged 60 -79 y. They showed that body mass index and waist circumference had similar associations with cardiovascular disease risk factors, whereas the waist-to-hip ratio showed weaker correlations. We think that this conclusion could be misleading for several reasons. First, the

more »

... uthors' definition of the metabolic syndrome is flawed; in their study, waist circumference was not included, and hypertension [blood pressure ͧ140 (systolic)/90 (diastolic) mm Hg] was considered as one of the components of the metabolic syndrome phenotype, whereas the Adult Treatment Panel III definition of the metabolic syndrome includes elevated blood pressure [ͧ130 (systolic)/85 (diastolic) mm Hg], not hypertension (2). Therefore, the prevalence rate is not the true prevalence of the metabolic syndrome in this age category. This is why the prevalence of the metabolic syndrome was 14% by their definition, which is lower than that shown in elderly groups (3, 4) . Second, the authors did not state that they had excluded subjects who were taking antihypertensive or lipid-lowering drugs. This could confound the findings. Third, the authors did not control for the effect of hip circumference in their analyses, but studies have shown that larger hip circumferences independently contribute to a reduced risk of metabolic abnormalities in adult and elderly men (5). Fourth, the odds ratio estimated from logistic regression models is a valid estimator of the rate ratio only when the outcome variable has a low prevalence in the sample (generally defined as Ȃ10% or less); this is not the case for the metabolic abnormalities in this study, and, as the outcome condition becomes common, the odds ratio highly overestimates the rate ratio (6, 7). Fifth, a comparison of odds ratios is not a suitable method for judgments about the predictive ability of screening tools; a comparison of sensitivity, specificity, and accuracy between screening tools is recommended to reach this objective (8). Sixth, it is possible that the relation between each anthropometric measure and cardiovascular disease risk factors is mediated by another measure. Because the investigators did not control for the simultaneous effects of anthropometric measures, it is not clear which anthropometric measure has a higher correlation coefficient with metabolic risks. Seventh, the use of cutoffs for defining metabolic disorders implies a loss of information; therefore, the authors should have also assessed the relations between anthropometric variables and continuous metabolic variables by using a multiple linear regression. The recognition of adiposity measures associated with metabolic abnormalities is extremely important, because the prevention of these risk factors is of public health importance for the prevention of noncommunicable diseases. However, careful epidemiologic and statistical methods should be adopted to avoid any incorrect conclusions. Prevalence of metabolic syndrome in an urban population: Tehran Lipid and Glucose Study. Diabetes Res Clin Pract 2003;61:29 -37. 5. Snijder MB, Zimmet PZ, Visser M, Dekker JM, Seidell JC, Shaw JE. Independent and opposite associations of waist and hip circumferences with diabetes, hypertension and dyslipidemia: the AusDiab Study. Int J Obes Relat Metab Disord 2004;28:402-9. 6. Zhang J, Yu KF. What's the relative risk? A method of correcting the odds ratio in cohort studies of common outcomes. JAMA 1998;280: 1690 -1. 7. Lee J, Chia KS. Estimation of prevalence rate ratios for cross-sectional data: an example in occupational epidemiology. Br Dear Sir: Esmaillzadeh et al (1) recently published a study in Iranian men, which were much younger than those in our study, that claims that the waist-to-hip ratio (WHR) is a better marker of metabolic abnormalities than are body mass index (BMI) or waist circumference (WC). In their letter, they make a number of criticisms on the methods used in our article. Here, we respond to their main concerns. Prevalence of metabolic syndrome If our aim had been to estimate the prevalence of the metabolic syndrome in the population, then our definition of WC would not yield the correct prevalence. However, this article was not about prevalence per se. A major purpose of our article was to investigate the association between the different adiposity measures and metabolic abnormalities and to compare the predictive abilities of the different adiposity measures in the detection of metabolic abnormalities. Therefore, it would have been inappropriate to include WC as part of the metabolic syndrome definition. This would inevitably have led to a correlation between metabolic abnormalities and WC that would have been greater than those between metabolic abnormalities and the other indexes that we investigated. We stated this clearly in the Methods section but retained the term "metabolic syndrome," which is comparable with the definition used in a recent study that compared the association between WC and BMI and the metabolic syndrome (2). Hypertension and lipid-lowering drugs The results were not dependent on the definition of hypertension used; repeating the analyses with different definitions of hypertension did not affect the results. In our article, we stated LETTERS TO THE EDITOR