Vitamin D and Calcium for the Prevention of Fracture

Key Points Question What is the available evidence for the efficacy of vitamin D with or without calcium supplementation for reducing the risk of fracture? Findings This systematic review and meta-analysis of randomized clinical trials of vitamin D alone (11 randomized clinical trials with 34 243 participants) showed no significant association with risk of any fracture or of hip fracture. In contrast, daily supplementation with both vitamin D and calcium (6 randomized clinical trials with 49 282 participants) was associated with a 16% reduced risk of hip fracture. Meaning In this study, neither intermittent nor daily dosing with standard doses of vitamin D alone was associated with reduced risk of fracture, but daily treatment with both vitamin D and calcium was a more promising strategy.


Data Extraction
Studies that classified blood 25(OH)D concentrations into only two categories were excluded as at least three exposure categories are needed to estimate a dose response relationship. 1,2 Eligibility evaluation and data extraction were carried out by two independent reviewers (PY and RC) and any discrepancies were adjudicated by discussion with a third reviewer (JA). Data extracted from all the identified studies included: author, publication year, country, study design, participants' characteristics

Statistical Analysis
The analyses assumed that there was a linear relationship between the natural logarithm of RR and 25(OH)D concentrations. For the studies that reported risk estimates separately by race (white, black, Hispanic, Asian and Native American) 3 or gender, 4 we combined these estimates using a fixed-effects model and subsequently used pooled estimates for each meta-analysis.
Heterogeneity was assessed using the I² statistic (I² >50% was considered significant heterogeneity). Contour enhanced funnel plots were constructed to assess publication bias. Subgroup analyses by relevant study characteristics were performed in order to identify potential sources of heterogeneity including: study design, age, geographic region, length of follow-up, and baseline blood 25(OH)D concentration. The continuous variables were dichotomised above and below the median values.

Data Extraction
Two researchers (PY and RC) independently extracted the relevant data from each trial, including: author, publication year, country, participant characteristics (total number, age, sex, residential status and previous history of fractures/falls), dosing regimen for vitamin D or calcium, type of control, compliance, trial duration, incident fracture types, and blood 25(OH)D concentrations at baseline and year of trial (if appropriate). Disagreements were resolved by discussion with a third reviewer (JA).
For factorial or multi-arm randomised trials, relevant data were extracted only for the effects of vitamin D, or vitamin D co-administered with calcium versus placebo. 5

Risk of Bias Assessment
Trials were assessed for possible bias using the Cochrane Collaboration risk of bias tool for randomised trials. 6 The tool included the following domains: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting, and other sources of bias. Each domain was rated as low risk, unclear risk or high risk of bias, and overall risk of bias was classified as low if all domains were at low risk of bias, or high if at least one domain was at high risk of bias, or as unclear if at least one domain was at unclear risk of bias and no domain was at high risk of bias.   The categories for risk of bias for each domain are "low risk", "moderate risk", "serious risk", "critical risk" of bias and "no information". We classified the overall risk of bias as low if all domains were at low risk of bias, as moderate if all domains were at low/moderate risk of bias, as serious if at least one domain were at serious risk of bias but not at critical risk of bias in any domain.We classified risk of bias as critical if at least one domain was at critical risk of bias. No information was defined if there is a lack of information in one or more key domains of bias.