Mark Bolland, Andrew Grey and Ian Reid, the authors of the article, comment:
Our discussion of the evidence for fracture efficacy of calcium and/or vitamin D included the DIPART meta-analysis.5 The claim of very significant antifracture efficacy of co-administered calcium and vitamin D in this meta-analysis is not supported by even superficial scrutiny. There was an 8% relative risk reduction in total fractures with calcium and vitamin D, with a number needed to treat of 213 to prevent one fracture over three years. For hip fractures, the relative risk reduction was 16% and the number needed to treat was 255 to prevent one hip fracture over three years. However, the hip fracture results were heavily dependent on one cluster randomised controlled trial,6 the results of which are problematic to interpret. When this trial was excluded the relative risk reduction was only 3%.5 Thus, the DIPART meta-analysis does not provide compelling evidence for the antifracture efficacy of calcium and vitamin D.
The Women’s Health Initiative study permitted widespread use of non-protocol vitamin D and calcium7 which obscured both adverse cardiovascular risks and potential benefits on cancer incidence.8 The Women’s Health Initiative investigators have now repeated our analyses on the complete dataset and have produced very similar results to ours.9 Given this, we do not think it is credible to claim that the original analysis provides reassurance about cardiovascular risks for patients.
Observational studies are hypothesis-generating, not hypothesis-testing. There are numerous examples of discrepant results between observational studies and randomised clinical trials, when positive benefits of drugs observed in observational studies are not observed in clinical trials. Examples include hormone replacement treatment and cardiovascular risk, vitamin D and various outcomes, and folic acid and antioxidants and cardiovascular disease and cancer. It is therefore unwise to emphasise the results of observational studies when there is a large database of randomised controlled trials that shows clear, consistent evidence of modest increases in myocardial infarction and stroke from calcium supplement use.
However, we acknowledge the correspondents’ point that the recent very large National Institutes of Health-sponsored observational study from the USA10 as well as similar large observational studies from Europe 11-13report increases in cardiovascular effects in association with calcium use.
Finally, our conclusion aligns with the recent recommendation of the US Preventive Services Task Force, whose members are free from both commercial and academic conflicts of interest, that vitamin D and calcium should not be administered for primary prevention of fractures in non-institutionalised postmenopausal women.14