The calcium scare—what would Austin Bradford Hill have thought?

The meta-analysis was found to be subject to several limitations including non-adherence to the clinical protocol, multiple endpoint testing and failure to correctly adjust for endpoint ascertainment. The main risk factors for myocardial infarction were not available for 65% of the participants, and none of the trials had cardiovascular disease as its primary endpoint. There were more overweight participants, more subjects on thyroxine and more men on calcium than on placebo. In particular, over 65% of all the heart attacks were self-reported. When the evidence was considered in the light of Austin Bradford Hill's six main criteria for disease causation, it was found not to be biologically plausible or strong or to reflect a dose–response relationship or to be consistent or to reflect the relationship between the trends in calcium supplementation and heart attacks in the community or to have been confirmed by experiment. The addition of a more recent trial on 1,460 women over 5 years reduced the relative risk to 1.23 (P = 0.0695).