An error in data analysis occurred in this paper by Bertrand et al.⁽Reference Bertrand, Giovannucci and Liu¹⁾, affecting the derivation of predicted 25-hydroxyvitamin D (25(OH)D) scores in ‘test samples’ used in validation analyses for the Nurses' Health Study (NHS).

On page 1891, in the second paragraph of the Results section, the reported NHS batch-adjusted Spearman correlation coefficient between predicted score and actual 25(OH)D level (0·23, 95 % CI: 0·16, 0·29) and the NHS batch-, season-, and age-adjusted correlation (0·23, 95 % CI: 0·16, 0·29) are incorrect. The correct age-adjusted correlation coefficient for NHS is 0·32 (95 % CI: 0·25, 0·38). The correct batch-, season-, and age-adjusted correlation for NHS is 0·33 (95 % CI: 0·26, 0·39). The reported difference in mean actual 25(OH)D level between extreme deciles of predicted 25(OH)D score for NHS (8·7 ng/ml (95 % CI: 5·4, 11·9)) is incorrect. The correct value is 9·7 ng/ml (95 % CI: 6·7, 12·7).

The following text in the last paragraph of the Results section on page 1893 should read: ‘Among these 443 women, the age- and season-adjusted Spearman correlation coefficient between average measured 25(OH)D based on two blood samples and long-term average predicted 25(OH)D over the same time period was 0·35. We corrected for within-person variation in plasma 25(OH)D to obtain a deattenuated correlation coefficient of 0·43’.