Plasma 25-hydroxyvitamin D concentrations, vitamin D deficiency and mortality in community-dwelling Japanese adults

Blood 25-hydroxyvitamin D (25(OH)D) concentrations vary considerably by season and sex. The present study aimed to determine associations between vitamin D deficiency and mortality in Japanese adults and identify risk thresholds according to 25(OH)D concentrations. This was a cohort study with an 11-year follow-up. Participants were 8285 community-dwelling Japanese adults aged 40-74 years. Plasma 25(OH)D concentrations were measured by chemiluminescent immunoassay at baseline and divided into quintiles for each of the subgroups stratified by season and sex (denoted as season- and sex-stratified quintiles). The main outcome was all-cause mortality. Hazard ratios (HR) were calculated using a Cox proportional hazards model. Mean age and 25(OH)D concentration were 59·9 years (sd = 9·1) and 50·1 nmol/l (sd = 18·1), respectively. Lower season- and sex-stratified quintiles were associated with higher hazards of all-cause mortality (P_{for trend} = 0·0015), with the first quintile (median = 28·2 nmol/l) having a higher HR (HR = 1·46, 95 % CI, 1·13, 1·88) than the highest quintile (reference). When crude quintiles were used, the overall association was similar (P_{for trend} = 0·0027), with the first (median = 28·0 nmol/l) and second (median = 39·7 nmol/l) quintiles having higher HR (HR = 1·40, 95 % CI, 1·06, 1·85 and 1·38, 95 % CI, 1·07, 1·77, respectively) than the reference. The risk threshold difference was estimated to be approximately 10 nmol/l. In conclusion, low blood 25(OH)D concentrations are associated with high mortality risk. Crude blood 25(OH)D concentration may modulate the estimated risk threshold for vitamin D deficiency associated with mortality.