ASSESSMENT OF THE RELIABILITY OF A DIET HISTORY QUESTIONNAIRE

The purpose of this study was to estimate the reliability of a self-ad ministered food frequency questionnaire developed at the National Canc er Institute (NCI) (Bethesda, MD). Food intake information was collect ed from 48 women with breast cancer and 50 women who had a familial br east cancer risk. These women were participants in a long-term investi gation of diet and breast cancer at the Ohio State University (Columbu s, OH). Pearson correlation coefficients were obtained for 29 nutrient s. The effect of time between questionnaires, age, weight, and caloric intake was also examined. The results showed moderate reliability coe fficients for most of the nutrients (r = 0.5-0.8). For the majority of nutrients, the women with breast cancer had higher coefficients than did the high-risk women. High-risk women with shorter elapsed time per iods between the first and second questionnaire had higher reliability coefficients as did women who consumed fewer calories. Age and weight (standardized by height) showed no effect on reliability. Recommendat ions for improving the NCI questionnaire in addition to using it as a standard method of measuring food intake are discussed.