Vitamin D with calcium reduces mortality: patient level pooled analysis of 70,528 patients from eight major vitamin D trials

Abstract

Introduction: Vitamin D may affect multiple health outcomes. If so, an effect on mortality is to be expected. Using pooled data from randomized controlled trials, we performed individual patient data (IPD) and trial level meta-analyses to assess mortality among participants randomized to either vitamin D alone or vitamin D with calcium.

Subjects and methods: Through a systematic literature search, we identified 24 randomized controlled trials reporting data on mortality in which vitamin D was given either alone or with calcium. From a total of 13 trials with more than 1000 participants each, eight trials were included in our IPD analysis. Using a stratified Cox regression model, we calculated risk of death during 3 yr of treatment in an intention-to-treat analysis. Also, we performed a trial level meta-analysis including data from all studies.

Results: The IPD analysis yielded data on 70,528 randomized participants (86.8% females) with a median age of 70 (interquartile range, 62-77) yr. Vitamin D with or without calcium reduced mortality by 7% [hazard ratio, 0.93; 95% confidence interval (CI), 0.88-0.99]. However, vitamin D alone did not affect mortality, but risk of death was reduced if vitamin D was given with calcium (hazard ratio, 0.91; 95% CI, 0.84-0.98). The number needed to treat with vitamin D plus calcium for 3 yr to prevent one death was 151. Trial level meta-analysis (24 trials with 88,097 participants) showed similar results, i.e. mortality was reduced with vitamin D plus calcium (odds ratio, 0.94; 95% CI, 0.88-0.99), but not with vitamin D alone (odds ratio, 0.98; 95% CI, 0.91-1.06).

Conclusion: Vitamin D with calcium reduces mortality in the elderly, whereas available data do not support an effect of vitamin D alone.

Fig. 1.

IPD analysis on survival in participants randomized to vitamin D with or without calcium supplements compared with placebo/no supplements. Data represent adjusted HR (95% CI).

Fig. 2.

IPD analysis on survival in participants randomized to vitamin D alone (left) or vitamin D with calcium (right) vs. placebo/no supplements. For the RECORD study, participants treated with calcium without concomitant vitamin D (n = 1311) were excluded, whereas the placebo group (n = 1332) was included in both analyses. Data represent adjusted HR (95% CI). *, P < 0.05.

Fig. 3.

IPD analysis on dose-effect relationships of vitamin D treatment on risk of death. Note that none of the included studies used a daily dose between 10 and 20 μg. All studies and stratified by whether vitamin D was provided alone (Vit D alone) or in combination with calcium (Vit D + calcium). Data represent adjusted HR (95% CI). The Smith study using an average daily dose of 20 μg vitamin D₂ was regrouped into the low-dose (10 μg/d) group.

Fig. 4.

Sensitivity analysis: influence of removing individual studies from the IPD analysis stratified by whether vitamin D (Vit D) was provided alone or in combination with calcium (Vit D + calcium). Data represent adjusted HR (95% CI). w/o, Without.

Fig. 5.

Trial level meta-analyses (random effect model) of studies included in the IPD analysis (A) and meta-analysis including additional large- and medium-sized studies (B). For studies included in the IPD, all deaths occurring during trials were accounted for as detailed in Table 1. ¹⁾No between-study heterogeneity was evident for studies on vitamin D with calcium (P = 0.46; I² = 0%), vitamin D without calcium (P = 0.86; I² = 0%) or the overall effect (P = 0.69; I² = 0%). Influence analyses showed no statistically significant effect of treatment if the WHI study (19) (OR, 0.94; 95% CI, 0.86–1.03) or the Larsen et al. (13) study (OR, 0.94; 95% CI, 0.86–1.02) was removed from the analysis. ²⁾No between-study heterogeneity was evident for studies on vitamin D with calcium (P = 0.37; I² = 8%), vitamin D without calcium (P = 0.30; I² = 16%), or the overall effect (P = 0.30; I² = 13%). Influence analyses showed no effects of removal of studies one by one, except for removal of the Porthouse et al. (14) study, leaving a statistically significant (P = 0.02) risk estimate for the remaining seven studies (OR, 0.92; 95% CI, 0.87–0.99). Effect size changed slightly by removal of the cluster randomized studies by Larsen et al. (13) (OR, 0.95; 95% CI, 0.88–1.04) and by Law et al. (16) (OR, 0.96; 95% CI, 0.89–1.03).