LAUSR

DOI: 10.5935/abc.20190023 Atherosclerosis is the main cause of morbidity and mortality in adults in Brazil and worldwide. Classical risk factors have shown its causal association from randomized experimental studies, such with cholesterol and hypertension. Other risk factors, including abdominal adiposity, show positive associations with atherosclerosis-related outcomes. The ELSA-Brazil observational study1 began collecting data in 2008 with clinical, epidemiological and laboratorial variables of 15,105 public servants aged 35 to 74 years. Several articles on these data have already been published and have brought relevant information about the association between risk factors and varied outcomes. In the present issue, Michaela Eickemberg et al.2 present data from a cross-sectional study that explores different measures of abdominal adiposity and its association with carotid intima-media thickness (C-IMT) measurement. Epidemiological studies seek to find plausible associations between risk factors and clinical outcomes or "surrogates" (here represented by C-IMT). Associations may or may not be causal. For an association between variables to indicate possible causality, it is necessary that some criteria, proposed by British statistician Austin Bradford Hill,3 be considered. They are: a) strength of association (magnitude of effect); b) consistency (or reproducibility); c) specificity (one disease, one variable); d) temporality (cause before effect); e) biological gradient (greater exposure, more disease); f) plausibility (known mechanism); g) coherence (between laboratory and clinical data); h) experiment (not always possible); i) analogy (comparison with similar situations). When applying these criteria to the study in question we have: a) the magnitude of effect of abdominal adiposity on C-IMT is a modest one (odds ratio around 1.4); b) there are other studies that prove this association; c) abdominal adiposity is not the only cause of atherosclerosis; d) it is probable that the adiposity precedes the intima-media thickening; e) we do not have definitive evidence of a biological gradient; f) there is biological plausibility for the assessed association; g) coherence between laboratory and clinical data is present; h) animal experiments have shown that a hyperlipidemic diet directly affects the arteries; i) in analogy with other risk factors, abdominal adiposity may indicate more arterial adiposity.