Impact of vitamin D supplementation on all-cause mortality: Randomized trials revisited.

Emulated trials weighted to match the 25(OH)D distributions of the original VITAL and D-Health trials reproduced null mortality results similar to those reported in the original trials.

• Hazard ratios for all-cause mortality in VITAL-emulated and D-Health-emulated trials were 0.97 (95% CI: 0.92–1.02) and 1.02 (95% CI: 0.97–1.07), respectively.
• These results closely mirrored the null findings reported in the original VITAL and D-Health randomized controlled trials.
• Study populations meeting trial inclusion criteria were n=237,502 (VITAL emulation) and n=185,809 (D-Health emulation) from the UK Biobank cohort.
• The emulation used Cox models to estimate expected effects over the mean trial follow-up times of 5.3 and 5.7 years.

When emulated trials were restricted to people with vitamin D insufficiency, major reductions in all-cause mortality were expected.

• Hazard ratios for all-cause mortality in the vitamin D insufficiency-restricted emulations were 0.85 (95% CI: 0.79–0.91) for the VITAL emulation and 0.81 (95% CI: 0.76–0.86) for the D-Health emulation.
• These analyses were conducted by restricting the UK Biobank study populations to participants with vitamin D insufficiency.
• The expected increases in serum 25(OH)D concentrations modeled were 30 nmol/L (VITAL) and 38 nmol/L (D-Health).

When emulated trials were restricted to people with vitamin D deficiency, even larger reductions in all-cause mortality were expected compared to the insufficiency-restricted analyses.

• Hazard ratios for all-cause mortality in the vitamin D deficiency-restricted emulations were 0.79 (95% CI: 0.72–0.87) for the VITAL emulation and 0.75 (95% CI: 0.69–0.81) for the D-Health emulation.
• These results suggest larger expected mortality benefits in more severely deficient populations compared to those with insufficiency.
• The analyses were performed using data from the UK Biobank cohort with Cox proportional hazards models.

The original VITAL and D-Health trials were conducted in mostly vitamin D-sufficient populations, which the authors identified as the primary reason for their null mortality findings.

• VITAL and D-Health are described as 'two large randomized controlled trials conducted in mostly vitamin D-sufficient populations.'
• The study used UK Biobank data to emulate both trials, weighting study populations to yield distributions of 25(OH)D as observed in the original trials.
• Vitamin D insufficiency and deficiency are described as 'common worldwide and linked to adverse health outcomes, including higher all-cause mortality.'

The emulation methodology using UK Biobank data successfully reproduced the null results of the original vitamin D trials when baseline vitamin D status distributions were matched to those of the original trials.

• Two alternative analytical approaches were used: weighting to the 25(OH)D distributions of the original trials, and restricting to vitamin D insufficient or deficient participants.
• The weighting approach yielded HRs of 0.97 and 1.02, consistent with null findings, while the restriction approach yielded HRs of 0.75–0.85, indicating meaningful mortality reductions.
• The UK Biobank cohort provided large sample sizes of 237,502 and 185,809 participants meeting respective trial inclusion criteria.
• Expected effects were estimated over mean follow-up times of 5.3 years (VITAL) and 5.7 years (D-Health).