Background: The effects of vitamin D on fractures, falls, and bone mineral density are uncertain, particularly for high vitamin D doses. We aimed to determine the effect of vitamin D supplementation on fractures, falls, and bone density.
Methods: In this systematic review, random-effects meta-analysis, and trial sequential analysis, we used findings from literature searches in previously published meta-analyses. We updated these findings by searching PubMed, Embase, and Cochrane Central on Sept 14, 2017, and Feb 26, 2018, using the search term "vitamin D" and additional keywords, without any language restrictions. We assessed randomised controlled trials of adults (>18 years) that compared vitamin D with untreated controls, placebo, or lower-dose vitamin D supplements. Trials with multiple interventions (eg, co-administered calcium and vitamin D) were eligible if the study groups differed only by use of vitamin D. We excluded trials of hydroxylated vitamin D analogues. Eligible studies included outcome data for total or hip fractures, falls, or bone mineral density measured at the lumbar spine, total hip, femoral neck, total body, or forearm. We extracted data about participant characteristics, study design, interventions, outcomes, funding sources, and conflicts of interest. The co-primary endpoints were participants with at least one fracture, at least one hip fracture, or at least one fall; we compared data for fractures and falls using relative risks with an intention-to-treat analysis using all available data. The secondary endpoints were the percentage change in bone mineral density from baseline at lumbar spine, total hip, femoral neck, total body, and forearm.
Findings: We identified 81 randomised controlled trials (n=53 537 participants) that reported fracture (n=42), falls (n=37), or bone mineral density (n=41). In pooled analyses, vitamin D had no effect on total fracture (36 trials; n=44 790, relative risk 1·00, 95% CI 0·93-1·07), hip fracture (20 trials; n=36 655, 1·11, 0·97-1·26), or falls (37 trials; n=34 144, 0·97, 0·93-1·02). Results were similar in randomised controlled trials of high-dose versus low-dose vitamin D and in subgroup analyses of randomised controlled trials using doses greater than 800 IU per day. In pooled analyses, there were no clinically relevant between-group differences in bone mineral density at any site (range -0·16% to 0·76% over 1-5 years). For total fracture and falls, the effect estimate lay within the futility boundary for relative risks of 15%, 10%, 7·5%, and 5% (total fracture only), suggesting that vitamin D supplementation does not reduce fractures or falls by these amounts. For hip fracture, at a 15% relative risk, the effect estimate lay between the futility boundary and the inferior boundary, meaning there is reliable evidence that vitamin D supplementation does not reduce hip fractures by this amount, but uncertainty remains as to whether it might increase hip fractures. The effect estimate lay within the futility boundary at thresholds of 0·5% for total hip, forearm, and total body bone mineral density, and 1·0% for lumbar spine and femoral neck, providing reliable evidence that vitamin D does not alter these outcomes by these amounts.
Interpretation: Our findings suggest that vitamin D supplementation does not prevent fractures or falls, or have clinically meaningful effects on bone mineral density. There were no differences between the effects of higher and lower doses of vitamin D. There is little justification to use vitamin D supplements to maintain or improve musculoskeletal health. This conclusion should be reflected in clinical guidelines.
Funding: Health Research Council of New Zealand.
Copyright © 2018 Elsevier Ltd. All rights reserved.