Interesting data. A few thoughts -
Using a dose of 2000IU per day, the very large (~25,000 participants) VITAL study that was discussed in the OP’s article found a small (HR .99) but not statistically significant decrease in all-cause mortality. There was a statistically significant reduction in cancer mortality when they did analyses that excluded cancer deaths in the first or second year of the trial, as that would likely be too early for any benefit to show up and those cancers may have already been established or “simmering” before the trial started.
Based on that, you may be thinking “well, it must have been the large bolus dosing of 60k IU per month that was problematic.” Another very large trial in Australia called D-Health used that same dose and found a small increase in all-cause mortality (HR 1.04), though it was nothing close to the 1.29 HR in the study that tweet is discussing, and it did not reach the level of statistical significance. This study did find an increase in cancer mortality (HR 1.15), but that also did not reach statistical significance. Even though neither was statistically significant, this does point to the possibility that at a given monthly dose of Vitamin D, it’s better to split it up into daily doses rather than taking it all at once.
Although weekly or monthly doses have been found just as effective as daily doses at correcting Vitamin D deficiencies, a larger dose is metabolized quite differently, which could produce different biological effects. There could also be differences based on one’s ethnicity and genetic background, as this trial that found such a large increase in death rates drew its subjects mostly from Indian and Asian populations.
My takeaway from all the Vitamin D research I’ve looked at over the years - the benefits and risks of supplementing likely vary depending on your baseline characteristics (BMI, baseline vitamin D levels, existing illnesses, maybe even your ethnicity) and what outcome you’re actually looking at (bone health, auto-immune disorders, depression, all-cause mortality, cancer occurrence/CVD events and cancer/CVD mortality, etc). This could also be another example of a biomarker that shows big impacts when you look at one’s “natural baseline” level, but changing that level with medications or supplements might have much less significant effects. The classic one that I’m thinking of is HDL cholesterol, where there is evidence that having higher baseline levels lowers CVD risk, but when they studied medications designed to raised HDL, they found either no benefits, or an increase in risk that was so significant that they had to stop the trial.
Selected articles of interest:
