Back

Vitamin D Deficiency, Supplementation, and Risk of Mortality and Chronic Disease: Evidence from Matched Cohorts in Israel and the US

Israel, A.; Weizman, A.; Israel, S.; Ashkenazi, S.; Ruppin, E.; Magen, E.; Merzon, E.; Vinker, S.

2025-06-04 public and global health
10.1101/2025.05.29.25328548 medRxiv
Show abstract

Vitamin D deficiency is common worldwide and has been linked to excess morbidity and mortality, yet its causal role remains debated. We analyzed electronic health-record data from two large healthcare networks: Leumit Health Services (LHS) in Israel and the US-based TriNetX Research Network. We examined more than 1.67 million serum 25-hydroxyvitamin D [25(OH)D] measurements from over 468,500 adults in LHS (2009-2020). In longitudinal matched-cohort analyses, we compared 13,352 severely deficient individuals (<10 ng/mL) to 12,352 with sufficient levels (>30 ng/mL) in LHS and validated findings in 223,000 matched pairs in TriNetX. Severe deficiency was associated with increased risks of all-cause mortality, diabetes, diabetic retinopathy, myocardial infarction, cerebrovascular accident, dementia, dialysis, and foot/toe amputation, while skin malignancy showed an inverse association consistent with lower UV exposure. To test modifiability, we fitted time-dependent Cox models in LHS incorporating monthly pharmacy-dispensed vitamin D supplementation. Supplementation was independently associated with dose-dependent risk reductions for mortality and most complications, but did not affect skin-malignancy risk, indicating supplementation effect is unlikely to be confounded by sun-exposure or other health-behavior factors. These findings suggest severe vitamin D deficiency is a modifiable, clinically important risk factor, providing a strong rationale for evaluating targeted supplementation strategies in deficient populations.

Matching journals

The top 9 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.