Munger · JAMA 2006 · Prospective nested case-control study · n=771

Serum 25-hydroxyvitamin D levels and risk of multiple sclerosis.

Cited 1899 times in the scientific literature.

Level 4 - case-series / case-control

Prospective nested case-control study

PubMed 17179460 · doi:10.1001/jama.296.23.2832 · record verified 2026-08-28

What was done

A prospective nested case-control study was conducted using the Department of Defense Serum Repository, drawn from over 7 million US military personnel. Multiple sclerosis cases diagnosed between 1992 and 2004 were confirmed via medical record review (n = 257) and matched 1:2 to controls (n = 514) by age, sex, race/ethnicity, and blood collection dates. Serum 25-hydroxyvitamin D levels were measured and averaged across two or more samples collected before the onset of initial multiple sclerosis symptoms. Odds ratios for multiple sclerosis were estimated across continuous and quintile levels of 25-hydroxyvitamin D stratified by racial/ethnic group.

What was found

Among white participants (148 cases, 296 controls), multiple sclerosis risk decreased significantly with increasing 25-hydroxyvitamin D (odds ratio [OR] 0.59 per 50-nmol/L increase; 95% CI, 0.36–0.97). By quintile compared with the lowest (<63.3 nmol/L), ORs were 0.57, 0.57, 0.74, and 0.38 (P for trend = .02), with only the highest quintile (>99.1 nmol/L) reaching statistical significance (OR 0.38; 95% CI, 0.19–0.75; P = .006). The inverse association was strongest for 25-hydroxyvitamin D measured before age 20. Among Black and Hispanic participants (109 cases, 218 controls), who had lower average vitamin D levels, no significant association was found.

Why it matters

This study provides prospective evidence that higher circulating vitamin D levels in young adulthood are associated with reduced multiple sclerosis risk in white populations.

Limits

The observational design cannot establish causality. No significant protective effect was detected in Black and Hispanic subgroups, which had lower baseline levels and smaller sample sizes. Unmeasured confounders, such as sun exposure behaviors, genetic risk factors, or diet, were not detailed in the abstract, and the military cohort may not generalize to other demographic settings.

Cited by