A total of 21 RCTs (11 in meta-analysis 1 and 12 in meta-analysis 2), providing 23 comparisons, and including 1260 participants in total (at follow-up).
Meta-analysis 1.
Included nine comparisons of vitamin D plus calcium versus no therapy/placebo (n=4) or calcium alone/calcium plus placebo (n=5). Two studies used in the sensitivity analyses compared vitamin D alone with either no therapy or calcium alone. One of the studies was a paediatric study and six studies were defined as prevention studies. No statistical heterogeneity was reported between any of the comparisons (P=0.14) and there was no evidence of publication bias (P=0.32). Pooled effect sizes for LS BMD: 1. Vitamin D plus calcium versus no therapy or calcium alone (n=9) 0.60; 95% CI: 0.34, 0.85, P<0.0001 (equivalent to a 3.2% difference in the % change in BMD between treatment and control, in favour of vitamin D plus calcium treatment)
2. Including two studies using vitamin D alone (n=11) 0.57; 95% CI: 0.36, 0.78, P<0.0001. 3. Effect size was not affected by removing the paediatric study (n=8).
4. Only peer-reviewed published articles (n=7) 0.59 (95% CI: 0.32, 0.85, P<0.0001).
5. Findings remained similar when studies were excluded that were not double-blinded (n=not stated) or that did not provide intention-to-treat results (n=not stated).
6. Vitamin D plus calcium versus no therapy/placebo (n=4) 0.66; 95% CI: 0.35, 0.97, P<0.0001.
7. Vitamin D plus calcium versus calcium alone or with placebo (n=5) 0.57; 95% CI: 0.13, 0.96, P=0.009.
8. Activated metabolites/analogs versus placebo/calcium alone (n=4)0.43; 95% CI: 0.04, 0.82, P=0.03.
9. Non-active vitamin D versus placebo/calcium alone (n=5) 0.74; 95% CI: 0.42, 1.06, P<0.0001.
10. Prevention studies only (n=5) 0.62; 95% CI:0.34, 0.89, P=0.0001.
11. Non-prevention studies only (n=4) 0.53; 95% CI: -0.08, 1.13, p=0.09.
Pooled effect sizes for fracture incidence:
12. Fracture reduction (n=3) -0.89; 95% CI: -1.90, 0.12, P=0.08 (equivalent to OR=0.41; 95% CI: 0.15, 1.13).
Meta-analysis 2.
Included 12 comparisons, six were with bisphosphonates, four with calcitonin and two with fluoride (none were with hormone replacement therapies). In one of the studies comparing fluoride, vitamin D was provided to patients in both arms if they had low serum levels of 25-hydroxyvitamin D. There was statistical heterogeneity among the bisphosphonate studies (P=0.003), but not among the calcitonin (P=0.12) studies. There was no evidence of publication bias for either bisphosphonate (P=0.60) or calcitonin (P=0.12) studies. There were insufficient fluoride studies (n=2) to make a comment about heterogeneity or publication bias.
Pooled effect sizes for LS BMD:
1. Bisphosphonates versus vitamin D (n=6) 0.57; 95% CI: 0.09, 1.05, P=0.02 (equivalent to a 2.0% difference in the % change in BMD in favour of bisphosphonate therapy).
2. Calcitonin versus vitamin D (n=4) 0.03; 95% CI: -0.39, 0.45, P=0.90.
3. Fluoride versus vitamin D (n=2) 0.66; 95% CI: 0.11, 1.22, P=0.02).
4. Excluding the study in which all participants could received vitamin D 0.88; 95% CI: -0.28, 2.04.