The cardiovascular safety of calcium supplements has been revisited with a further meta-analysis, (1) which concludes that calcium supplementation does not increase coronary heart disease in women, without providing data for men. Their conclusion is at odds with that of our meta-analyses, which reported that calcium increased the risk of myocardial infarction (MI) and possibly stroke in men and women together. (2,3) There are important differences between approaches to the meta-analyses. In the current article and previously, the authors suggest that including men and self-reported events may have explained the increased risk of MI from calcium. However, in sensitivity analyses, we reported that sex did not interact with the effect of calcium supplements on cardiovascular risk (4) and that the risk of MI from calcium is increased after men or self-reported events are excluded. (4,5) The major effect of restricting the analyses to the subgroup of women is a reduction in the power of the analyses, and, for trials that included men and women, potentially introducing bias through loss of the balancing effects of randomization on possible confounders. These, and other major limitations of subgroup analyses, have been well described. (6) Additionally, even though they state the question is whether calcium has an impact on MI, the authors chose to use a composite endpoint of a much broader cluster of cardiovascular diagnoses of different clinical significance for their primary endpoint, which is not recommended practice and does not provide a better measure of safety when there is an increased risk of a clearly defined single endpoint. (7) The composite endpoint included angina pectoris and chronic coronary heart disease, although there is no evidence that these conditions are influenced by calcium supplementation, and they lack a standardized objective definition, which precludes reliable ascertainment of these events. Including such conditions in the composite endpoint might obscure the elevated risk of MI from calcium. Surprisingly, the authors also did not seek data for their primary endpoint from our previous trial, which would have increased the size of their data set of calcium monotherapy trials by about 40%. (8) In addition, a group that previously supplied individual patient data for our meta-analysis did not agree to participate in the Lewis meta-analysis.In their analyses of calcium monotherapy and MI, the authors identified only one trial (with four MIs) published after our metaanalysis, (2) In their analyses of calcium plus vitamin D, the authors included two trials not in our meta-analysis, (3) one trial with one MI, and the second a trial by Larsen and colleagues. (9) The latter was an open-label study of multifactorial interventions, one of which included calcium and vitamin D, administered to people living in different regions of the city. Although described as a cluster randomized trial, there was only one cluster per treatment group, meaning it is better described as quasi-randomized by location of...