Calcium intake and risk of fracture: systematic review

Journal Club

27 January 2016

Facilitated by Dr Ray Moynihan

1. Background

Recommendations in favour of calcium are common but are increasingly being questioned by those looking closely at the evidence. Given that CREBP is currently running focus groups on osteoporosis, this systematic review offered a good opportunity to think about the intersection of evidence and health recommendations. The session also threw up some interesting and unexpected discussion about conflicts of interest.

2.  Paper presented

Mark J Bolland, William Leung, Vicky Tai, Sonja Bastin, Greg D Gamble, Andrew Grey, Ian R Reid. Calcium intake and risk of fracture: systematic review. BMJ 2015;351:h4580

Design: Systematic review


P: People over 50 who are generally healthy, without any pathology, but who may be at risk of osteoporosis.

I: dietary calcium intake or calcium supplements

C: placebo (possibly plus other drugs or supplements)

O: risk of fracture (total, hip, vertebrae and forearm)

Study designs reviewed: RCTs or cohort studies.

Study Appraisal

 Was there a focused PICO ? – Yes

This article was appraised using the FAITH method:

Find: the authors conducted a search, initially only in English, then later, during an update, there was a focused search including non-English as well. Three databases are mentioned, but the group discussion/appraisal suggested two of them are very similar, and questions were raised about the reproducibility of such a long and detailed search strategy. It appears that one person (one of several) conducted initial screening – which may be acceptable when there is a broad/liberal inclusion at that point.

Appraise: For RCTs authors used the Cochrane Handbook, which is seen as acceptable. For the cohort studies, authors reported there was no critical appraisal. The CREBP discussion suggested there may be tools, albeit in early forms, which could be used to appraise the quality of cohort studies – as the main risk of bias comes from confounding.

Include: Authors included high risk of bias RCTs, but in analysis grouped studies according to their level of risk of bias.


Meta-analysis and pooling was only done for the RCTs (but not cohort studies which had too many different sorts of outcomes) which is acceptable.

Heterogeneity: There was heterogeneity, but there seems to be little reported in the text about heterogeneity, and there was some confusion around reading the reports of heterogeneity in the tables. There was discussion of visual inspection of funnel plots, and an Eggers regression, and a suggestion of some publication bias.

Summary of results

There are several elements to the results. In Summary, there were only 2 small RCTs of dietary calcium. There were 44 cohort studies of dietary calcium: most studies reported no association between calcium intake and fracture. Summarising 26 RCTs of calcium supplements, supplements reduced the risk of total fracture and vertebral fracture but not hip or forearm. (see paper for details) For RCTs at lowest risk of bias, there was no effect on risk fracture. Most of the 11 cohort studies reported no association between supplements and fracture risk.

Discussion/Journal Club commentary 

Overall, notwithstanding a few questions and criticisms, the CREBP group believed the systematic review was trustworthy, and raises important questions about current recommendations on calcium intake and common beliefs about the benefits of calcium intake to reduce fracture risk.

Discussion expanded to talk about conflicts of interest disclosure and the associated documents, available through the BMJ website – including Rapid Responses and the Peer Review documents.

One author of the Systematic Review disclosed relations with a number of companies which market osteoporosis medications, sparking discussion about the marketing of those drugs. The Rapid Responses and the Peer Review document provoked a discussion about whether the Systematic Review authors had in any way been too negative in their conclusions about calcium.