Development and relative validity of a food frequency questionnaire for the estimation of intake of retinol and beta-carotene.

The aim of this study was to develop and validate a semiquantitative food frequency questionnaire to classify individuals according to their intakes of retinol and beta-carotene. Food items for the questionnaire were selected both on the basis of their contribution to total population intake of retinol and beta-carotene and on the proportion of between-person variation explained, which was as calculated from data of two study populations in the Netherlands. Thus, 15 products containing retinol and 15 products containing beta-carotene were selected. These contributed over 90% to the total intake and explained 99% of the variation of retinol and beta-carotene, respectively. The questionnaire was validated against a dietary history in a population of 82 women (aged 30-49 years). The time elapsed between the two interviews was (on average) 25 days. Spearman rank-order correlation coefficients comparing the questionnaire with the dietary history were 0.54, 0.59, and 0.64 for retinol, beta-carotene, and total vitamin A, respectively. The proportion of exact agreement in the two extreme categories of vitamin A intake, based on quintiles, was 56%. The corresponding gross misclassification (from 1 extreme category into the opposite) was 3%. These data indicate that a very short questionnaire can classify subjects into categories according to their vitamin A intake.