AGREEMENT IN ALCOHOL-CONSUMPTION LEVELS AS MEASURED BY 2 DIFFERENT QUESTIONNAIRES

Objective: We evaluated the agreement in reported alcohol consumption between a standard food frequency questionnaire (FFQ) and a risk facto r questionnaire (RFQ) developed for a genetic epidemiologic study of b reast cancer. Method: The FFQ measured intake of alcoholic beverages b y asking which of nine levels were consumed during a single time perio d. In contrast, the RFQ used open-ended questions to measure intake of alcoholic beverages during numerous time periods. Subjects (N = 765) completed both questionnaires at home. Results: Mean daily alcohol con sumption levels were consistently higher in the FFQ than in the RFQ; f or example, the mean alcohol consumption for all subjects was 7.0 g/da y in the FFQ versus 5.3 g/day in the RFQ. Moreover, the RFQ overestima ted the number of nondrinkers relative to the FFQ. Nonetheless, the Sp earman correlation coefficients between daily alcohol consumption leve ls as measured by the two questionnaires were relatively high: total a lcohol, r = 0.72; beer, r = 0.69; wine, r = 0.69; and distilled spirit s, r = 0.54. Conclusions: The reasonable agreement between these quest ionnaires supports the validity of historical alcohol consumption leve ls measured by the RFQ.