Prognostic cut-off values of caffeine and cardiovascular events in a cohort of unselected men and women from general population

Background Several studies focused on the effect of caffeine on the cardiovascular (CV) system and on CV outcome, with fairly contrasting results. Dietary caffeine is usually taken into consideration as a continuous variable, but researcher would prefer to use cut-off values more useful in clinical setting. Purpose The present study is aimed at defining in epidemiological setting and in population-based context the cut-off values of caffeine, if any, having a CV prognostic values, to confirm them in men and women of general population, and to establish whether caffeine of food origin contained in multiple sources used in everyday life. Methods The study cohort was represented by 1,668 unselected men and women aged 59.6±17.1 years living in an area in North-East Italy and sharing homogeneous lifestyle, randomly taken from general population. Multivariate dichotomic Cox regression models having fatal and morbid CV event as dependent variables, adjusted for age, sex, basal heart rate, body mass index, LDLC, individualized FEV1, β-blocking therapy, ethanol intake, diabetes mellitus, arterial hypertension, smoking, smoking, blood pressure were preliminarily used to search for an association between caffeine as a continuous variable and incident CV event. Prognostic cut-off values, identified by means of receiver operating curves (ROC) and able to discriminate between subjects doomed to develop the CV event, were then used as independent predictors to divide people into those cut-off in further multivariate Cox models adjusted for the confounders listed above. Results Daily intake of caffeine over a specific cut-off determined via the ROC curve analysis are associated to lower incidence of CV events in a 20-year follow-up period. This cut-off corresponds, for each events, to an interval between two groups of quintiles of caffeine intake. The cerebrovascular events and those due to heart failure and arrhythmias were reduced also in multivariate models adjusted for confounders. On the contrary, intake of caffeine over the cut-off was only apparently associated to lower incidence of coronary disease, but this unadjusted crude association was wales and not confirmed after adjustment. Conclusions In men but not in women, higher voluntary daily consumption of caffeine from any source reduces significantly and to a considerable extent, in a long follow-up, the incidence of cerebrovascular events, heart failure and arrhythmias. To have preventive effects on events the consumption of caffeine must be higher than a cut-off that corresponds to 280, 230 and 280 mg/day respectively, cut-off values always falling between the third and fourth quintile of caffeine consumption. No prognostic cut-off can be identified in women and in any sex in the case of coronary events. Specific studies will be mandatory to understand the reason of this sex-specific difference.