Background: Individual randomized clinical trials have not shown conclusively that taking vitamin D can prevent type 2 diabetes.

Methods: We conducted a systematic review and meta-analysis of individual participant-level data to determine whether vitamin D supplementation lowers the risk of diabetes in people with prediabetes. Trials published through February 12, 2021, that compared oral vitamin D supplementation (alone) with placebo over a minimum 2-year follow up, and had new-onset diabetes as a primary outcome were included. Proportional hazards models were used to analyze individual participant data, with pre-specified intention-to-treat (ITT) and per-protocol analyses, without and after adjustment for baseline age, gender, BMI, race and hemoglobin A1c. Safety was also assessed.

Results: Three trials were eligible (testing D3 20,000 IU/week [Tromsø], D3 4,000 IU/day [D2d], and eldecalcitol 0.75 μg/day [DPVD] vs. matching placebos) and contributed data from 4190 randomized participants (mean age 60.4 years, 44.2% women). After a median 3.0 years of follow-up, new-onset diabetes occurred in 475/2097 participants assigned to vitamin D and 524/2093 assigned to placebo (8.42 and 9.50 events per 100 person-years, respectively). The unadjusted hazard ratio (95% CI) for vitamin D was 0.88 (0.77 to 0.99) in the ITT and 0.85 (0.75 to 0.97) in the per-protocol analyses. Adjusted hazard ratios were 0.85 (0.75 to 0.96) in the ITT and 0.83 (0.73 to 0.94) in the per-protocol analyses. There was no evidence of heterogeneity of treatment effect among the three trials. The incidence of hypercalcemia, hypercalciuria, or kidney stones did not differ between the vitamin D and placebo groups.

Conclusion: In adult patients with prediabetes, vitamin D supplementation is safe and effective for lowering the risk of diabetes by about 15%.


A. G. Pittas: None. T. Kawahara: None. R. Jorde: None. J. P. Nelson: None. E. Angellotti: None. E. Vickery: None. B. Dawson-hughes: None. T. Trikalinos: None. E. M. Balk: None.


American Diabetes Association (1-14-D2D-01 to A.G.P.); National Institutes of Health (U01DK098245)

Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at