@article{2989275,
    title = "Epidemiological evidence for the non-random clustering of the components of the metabolic syndrome: Multicentre study of the Mediterranean Group for the Study of Diabetes",
    author = "Thanopoulou, A. and Karamanos, B. and Angelico, F. and Assaad-Khalil, S. and Djordjevic, P. and Katsilambros, N. and Migdalis, I. and Mrabet, M. and Petkova, M. and Roussi, D. and Tenconi, M.-T. and Archimandritis, A.",
    journal = "European Journal of Clinical Nutrition",
    year = "2006",
    volume = "60",
    number = "12",
    pages = "1376-1383",
    publisher = "Nature Publishing Group",
    issn = "0954-3007, 1476-5640",
    doi = "10.1038/sj.ejcn.1602467",
    keywords = "glucose;  high density lipoprotein cholesterol;  low density lipoprotein cholesterol;  triacylglycerol, adult;  aged;  Algeria;  article;  blood pressure;  body height;  body mass;  body weight;  Bulgaria;  cluster analysis;  eating habit;  Egypt;  female;  geographic distribution;  Greece;  human;  Italy;  major clinical study;  male;  Mediterranean diet;  metabolic syndrome X;  normal value;  nutritional assessment;  prevalence;  questionnaire;  risk factor;  smoking habit;  waist circumference;  Yugoslavia",
    abstract = "Objective: To determine: (a) whether the components of metabolic syndrome (MetS) cluster more frequently than predicted by chance alone and (b) whether increased risk for MetS is associated also with values of each component below, but close to the cutoff points defining MetS. Research design and methods: Anthropometrical and biochemical measurements were performed and a dietary questionnaire was filled-in in 1833 randomly selected non-diabetic subjects, 916 men and 917 women, 20-74 years old, in nine centres in five Mediterranean countries. The prevalence of MetS and of possible combinations of its individual components was measured. The expected frequencies of the above combinations were calculated according to the mathematical formula of probabilities. Results: The overall prevalence of MetS was 27.2%, but varied greatly among countries, from 5.8% in Algeria to 37.3% in Greece. The observed prevalence of each combination diagnostic of MetS was higher than the expected by chance. Thus, the observed overall prevalence of MetS was also higher than the expected, 27.2 vs 24.0%, P=0.03. Furthermore, for each individual component (except high-density lipoprotein), as values in the normal range, approached the cutoff point, the risk of having MetS (i.e. clustering of the other components) increased significantly (odds ratio 2.2-4.6, P<0.001). Conclusions: The MetS is not related to the Mediterranean type of diet and its prevalence varies greatly among five Mediterranean countries. The clustering of the components defining the MetS is not due to chance and moreover even 'high normal' levels of each component confer increased risk for the syndrome."
}