Differences in reported food frequency by season of questionnaire administration: the 1987 National Health Interview Survey.

We assessed seasonal reporting bias in a 59-item food frequency questionnaire (FFQ) administered throughout 1 year using data from the 1987 National Health Interview Survey (N = 20,143 adults). Few meaningful differences were found in the proportion of individuals reporting rarely or never consuming a food by season of questionnaire administration. Seasonal reporting bias is evident in FFQs, however, and appears to be due to reporting most recent consumption. Using gender-specific median servings per week, an analysis using logistic regression showed that the estimated proportion of individuals reporting food intake at greater than the yearly median differed between any two seasons by at least 5% of the population for 22 foods. We compared gender-specific quintiles of selected nutrients/food groups for the whole year and each season; these showed that quintile assignment never varied by more than one adjacent quintile. The most frequent shift in quintile assignment, involving as many as 18.5% of women in the summer, occurred for citrus fruits. The intake biases are small and do not greatly affect population estimates if the FFQ is administered in all seasons, but they may somewhat affect classification of individuals into quantiles for some foods/nutrients.