A prospective study of response error in food history questionnaires: implications for foodborne outbreak investigation.

To explore the problem of response error in food history data, a prospective study examined the validity of food questionnaire data obtained five days after the study meal. Unobtrusive observation of 64 persons selecting two different foods at a buffet-style luncheon were compared with subsequent histories of food consumption. The predictive value of a positive response was 0.73 for one food and 0.82 for the second food. The response error measures obtained were then applied to data from a published foodborne outbreak to illustrate the impact of predictive value positive and predictive value negative levels on the significance of a food-illness association. Public health workers engaged in food questionnaire administration and analysis must consider response error and should explore methods of reducing this problem through attention to both interviewer-respondent interaction and questionnaire design.