Long COVID was 4X less common in those on 1,000 IU of vitamin D for 3+ years - observational

---####Chronic cholecalciferol supplementation and adequate vitamin D status are associated with reduced risk of Long COVID Journal of Endocrinological Investigation, March 23, 2026, https://doi.org/10.1007/s40618-026-02858-8 (open access)

Luigi di Filippo, Umberto Terenzi, Simona Bolamperti, Camilla Bechini, Simonetta Bossoni, Marco Campaniolo, Licia Gifuni, Mauro Doga, Andrea Giustina

PDF

Summary by Claude - August, 2026

A modest daily dose of vitamin D3 — taken for years before infection — was associated with much less Long COVID. This is a retrospective observational study of 132 Italian adults (median age 56) who caught COVID between March 2020 and May 2021, all previously seen at the endocrine clinic at San Raffaele, Milan. Fifty of them (38%) had been on cholecalciferol for at least 24 months beforehand, median 38 months.

Long COVID (NICE criteria: 2+ new symptoms still present 6 months after recovery) occurred in 30 of 132 (22.7%).

All supplemented vs. not: 7/50 (14%) vs. 23/82 (28%) — p = 0.086, not significant
Median dose ≥1000 IU/day: 4/38 (10.5%) vs. 26/94 (27.6%) — p = 0.039
1:1 matched cohort (age, sex, comorbidities, COVID severity), n=86: 5/43 (11.6%) vs. 14/43 (32.6%) — p = 0.036
Matched, ≥1000 IU/day, n=70: 3/35 (8.5%) vs. 12/35 (34.2%) — p = 0.018

Achieved levels tell the real story: supplemented patients had a median 25(OH)D of 28 ng/mL vs. 15 ng/mL in the unsupplemented; deficiency (under 20 ng) was 8.8% vs. 78%. Crucially, blood was drawn before infection (median 5.5 months prior), which defeats the usual objection that low vitamin D in COVID patients is just an acute-phase artifact. Pre-infection levels below roughly 18–23 ng/mL predicted Long COVID (AUROC 67–74%). Obesity was the only other independent risk factor.

What this does not show / limitations

Not a trial — no randomization, no causation. The strongest result rests on 3 events vs. 12; small shifts in a handful of patients would erase it. The dose "gradient" is not a dose-response: the ≥750 and ≥1000 IU subgroups are nested inside the same 50 supplemented patients, so the p-value crossing 0.05 partly reflects which 12 people were excluded. Long COVID was ascertained by unblinded retrospective telephone interview. Blood levels were available for only 96 of 132 patients. People who take a supplement faithfully for 3+ years differ from those who don't in diet, activity, sun, and income — the authors say residual confounding can't be excluded. Single center, pre-vaccine era, early variants; may not transfer to today. Data are held by the sponsor; di Filippo received an institutional grant from Abiogen and Giustina has consulted for them. Finally, this does not show 1,000 IU is the right dose — it shows that whatever got these patients to ~28 ng/mL and held them there tracked with less Long COVID.

Long COVID was 4X less common in those on 1,000 IU of vitamin D for 3+ years - observational

Related in VitaminDWiki