Accuracy of telephone-administered dietary recalls in a group of free-living elderly subjects

The accuracy of telephone-administered dietary recalls was determined for a group of 159 elderly subjects. Accuracy was assessed by comparing a telephoned 24-hour recall that included a meal consumed at a congregate meal site on the day prior to the telephone interview with data on observed intake for that meal obtained by trained dietitians. For 12 of the 15 nutrients evaluated, the bias in the mean recalled intake data was less than 10 percent. After appropriate adjustment for the imprecision in the data on observed intake, there was evidence of an attenuated regression slope relating the individual recalls to the true intakes for only five of the 15 nutrients evaluated.Two statistically significant models for predicting the accuracy of the dietary recall data on the basis of health, dietary, social and demographic characteristics were developed. Although this latter work was of a hypothesis-generating nature, the clustering of variables related to heightened awareness of what is eaten in one of the models suggests that further evaluation of these types of predictive variables is appropriate