ABSTRACT
OBJECTIVE: To assess the predictive validity of the meats, eggs, dairy, fried foods, fat in baked goods, convenience foods, fats added at the table, and snacks (MEDFICTS) questionnaire, a rapid dietary fat screening instrument, when used with African-American women. DESIGN: A case series design was utilized to assess the validity of MEDFICTS compared to the Arizona Food Frequency Questionnaire. SUBJECTS/SETTING: Data for this study were collected from 184 healthy premenopausal African-American women who completed both the MEDFICTS and the Arizona Food Frequency Questionnaire during screening for eligibility in a nutrition intervention study. STATISTICAL ANALYSES PERFORMED: Analyses of sensitivity, specificity, predictive values, and receiver operating characteristic analysis were used to examine the predictive validity of MEDFICTS. Covariates of correctly and incorrectly identified groups were examined with contingency table analysis and t tests. RESULTS: MEDFICTS was a statistically significant predictor of dietary fat consumption, but underestimated fat consumption of >or=30%. MEDFICTS' sensitivity to detect those consuming >or=30% fat was 57.3%, whereas its specificity (detection of those consuming <30% dietary fat) was 66.0%. Positive and negative predictive values were 80.6% and 38.5%, respectively. A reduction in the threshold score resulted in increased sensitivity but decreased specificity. Women identified by MEDFICTS as consuming <30% dietary fat but who were actually consuming 30% or more reported greater consumption of fats from foods classified as mixed foods. CONCLUSIONS: Study findings suggest MEDFICTS underestimates fat consumption of >or=30%. The inclusion of a category to assess mixed foods could improve the sensitivity and specificity of MEDFICTS for predicting dietary fat consumption. These results demonstrate the need for population-based validation of dietary screening instruments.