Vitamin D's cold-fighting effect shrinks in big trials

Vitamin D became one of the most widely taken supplements on the promise that it shores up defences against respiratory infection, a claim that gathered momentum through the pandemic. Trials run between 2006 and 2021 suggested supplementation could prevent acute respiratory infections and severe COVID-19, but larger trials published in 2022 failed to confirm any benefit in COVID patients. That contradiction is not just a vitamin story; it raises the question of whether small trials systematically flatter treatments. This review set out to test whether trial size itself predicted the reported benefit.

The authors ran a systematic search for randomised trials of vitamin D supplementation and admission to intensive care in COVID patients, identifying nine trials that ranged from 50 to 548 patients. For acute respiratory infections, they extracted data from an existing 2021 meta-analysis, covering 37 trials ranging from 25 to 16,000 patients. Within each set, trials were split into those below and above the median size, and random-effects models produced summary odds ratios with 95% confidence intervals. They then tested for effect modification by trial size and assessed publication bias using the Louis-Furuya-Kanaruori index and the trim-and-fill method.

For acute respiratory infection prevention, the summary odds ratio across all 37 trials was 0.92 (95% CI 0.86-0.99) — but this split into 0.69 (0.57-0.83) in the smaller trials and 0.98 (0.94-1.03) in trials of 248 to 16,000 patients, with an interaction p of 0.0001. The bias index was -3.11, outside the no-bias range, and after trim and fill the pooled estimate moved to 0.96 (0.88-1.05). The intensive care analysis showed the same pattern: 0.61 overall, 0.34 in the smallest trials, 0.88 in trials of 106 to 548 patients, and 0.80 (0.40-1.61) after trim and fill.

This complicates rather than overturns the earlier meta-analytic literature: the pooled benefit was real in the numbers, but it was carried almost entirely by the smallest studies, which is the classic signature of publication bias. It aligns with the large 2022 trials that found no COVID benefit and gives a statistical explanation for why they disagreed with what came before. The authors' warning is aimed at systematic reviewers as much as readers. It leaves open whether specific groups — for example people who are genuinely deficient — might still benefit, since that question was not the focus of this size-based analysis.

This is a reason to treat blanket claims that vitamin D prevents colds and flu with real scepticism, because the effect largely evaporates in the biggest and best-powered trials. It is also a useful general lesson: when a supplement's evidence base rests on lots of small studies, the headline benefit is more likely to shrink as bigger trials arrive. None of this speaks to vitamin D's established roles beyond respiratory infection, nor to people with diagnosed deficiency, whose situation the analysis did not address. Correcting a deficiency and hoping to dodge a winter cold are different propositions.