Season of sampling and the interpretation of serum homocysteine results: a within-person analysis of 268,877 health-screening examinations

Abstract Objectives Serum total homocysteine (tHcy) is interpreted against fixed reference intervals and a decision limit. Whether the month of blood collection is a material source of variation, and whether it reflects the seasonal cycle of 25-hydroxyvitamin D [25(OH)D], is unresolved because earlier studies were cross-sectional or small. Methods We analysed 268,877 health-screening examinations from 69,131 adults between January 2014 and December 2025 at a single Korean centre. Because no participant was examined twice in a calendar year, within-person seasonal contrasts span years; all models therefore combined person with calendar-year fixed effects. Seasonality was quantified by cosinor regression with cluster-robust standard errors. Results The crude amplitude of tHcy (0.473 μmol/L) fell to 0.255 μmol/L under person fixed effects, so about half of the apparent seasonality reflected who attended. After the secular trend was also removed the within-person amplitude was 0.301 μmol/L (95 % CI 0.281–0.324), peaking in early November, and was unchanged by time-varying covariates. The 25(OH)D acrophase was in August, 2.6 months away, and adjustment for 25(OH)D did not attenuate the tHcy amplitude (retention 108.0 %); the rhythm persisted where within-person 25(OH)D barely varied. Spring examinations carried lower within-person odds of exceeding 15 μmol/L (odds ratio 0.699, 95 % CI 0.654–0.748). Conclusions Month of sampling is a small but reproducible systematic source of within-person variation in tHcy that is not explained by 25(OH)D. It is modest relative to within-subject biological variation, but sufficient to alter classification near the decision limit and to bias reference intervals derived from seasonally unbalanced data.