LAUSR

does not provide any essential macro or micro-nutrients, that caffeine intake has been often associated with harmful health effects and coffee drinking is often considered as a component of unhealthy behaviours, including smoking and sedentary lifestyle, we strongly believe that abstinence should be the value of reference in such analyses. We definitely agree that continuous variables compared to categorical ones allow better degrees of freedom in many, but not all types of analyses. Especially for coffee consumption and CVD risk, it has been reported that they do not have a linear relationship; thus, the classification in common drinking groups is the optimal way to test this research hypothesis. In addition, and from a public health perspective, providing real intake-based analyses, it makes the conclusions more comprehensible in daily life. Moreover, in this paper, consistently with all the ATTCA study papers, a composite CVD endpoint due to the relatively low 10-year incidence observed of specific CVD manifestations was generated (i.e., acute myocardial infarction, stroke, unstable angina, heart failure) [5]. This should not be seen as a limitation, but as strength of the study, especially from a methodological point of view, where the number of events is a key player in establishing robust effect estimates through statistical analyses. Finally, the fact that only baseline measurements were used in the analysis, as risk factors or potential mediators between the exposure and the 10-year CVD incidence, is a very common approach in prospective studies. It would be great if in epidemiologic studies we had continuous monitoring of participants’ characteristics, but this was not the case. Nevertheless, this has already been acknowledged as a limitation in the discussion section of the paper [1]. Conclusively, coffee consumption seems to play complex and multiple roles in CVD risk that may partially explain the contradictory findings reported by previous Dear Editor,