We appreciate the comments raised by Giugliano, Ceriello, and Esposito (1) on our study (2). When we wrote that participants in the highest score of the Mediterranean food pattern had lower levels of risk factors, we meant in terms of comparisons between extreme values (score = 0 vs. score = 9), not grouped categories (as shown in Table 1 of our study). Average levels of factors (all of them considerably below the cut-off point for the metabolic syndrome) across grouped categories of the Mediterranean food pattern were shown simply to describe our sample of fairly healthy participants.
Our end point was the incidence of the metabolic syndrome. There is no inconsistency in the finding of small differences in mean risk factor levels, while the multivariate odds ratio for the metabolic syndrome showed a strong inverse association. According to the International Diabetes Federation (3), for a person to be classified as having the metabolic syndrome, they must have central obesity plus any two of four additional factors (triglycerides ≥150 mg/dl, HDL cholesterol <40 mg/dl in male or <50 mg/dl in female subjects, raised blood pressure [systolic ≥130 or diastolic ≥85 mmHg], and fasting plasma glucose ≥100 mg/dl). Therefore, central obesity is a sine qua non criterion and has substantially higher weight than the additional factors. Moreover, the results of a multivariate-adjusted logistic model (with metabolic syndrome as the dichotomous outcome) are not directly equivalent to simple comparisons between means of risk factors. Besides uncontrolled confounding, comparisons of means can be influenced by skewed data, extreme values, or upward trends within the normal range without clinical significance.
The speculation that our participants who were initially free of the metabolic syndrome were in fact very close to the cut-off values of the International Diabetes Federation definition is not supported by our data. Our study was designed as an epidemiological cohort including initially healthy participants, not as a trial for patients with the metabolic syndrome. Therefore, we quoted only studies within the field of prevention, not trials for the treatment of patients with established metabolic syndrome at baseline. Indeed, we agree that experimental studies (trials) can better support a causal relationship than observational epidemiological studies. However, taken together, our study, the Italian trial (4), and a more recent Spanish trial (5) provide good and specific evidence to support that the Mediterranean food pattern is inversely associated with the risk of the metabolic syndrome.
