ALTERNATIVE APPROACHES TO NUTRITIONAL ASSESSMENT FOR STUDIES OF DIET AND DISEASE: AN ANTHROPOLOGICAL PERSPECTIVE (VALIDITY, ARIZONA, ELDERLY)

Abstract

This study profiles the usual dietary habits of independent-living elderly from alternative methodological perspectives. The primary objective was to validate a comprehensive dietary questionnaire developed for use in epidemiology against the results of household refuse analysis, an independent, continuous measure of dietary behavior. Members of 44 one-and two-person households residing in a retirement community in southern Arizona completed a semi-quantitative food frequency questionnaire. Subsequently, all refuse discarded by participating households during the following six weeks was collected and recorded by researchers. During the final week, respondents completed a recall questionnaire asking them to report their dietary habits during the study period. Analysis compared the results of survey and material measures of monthly household consumption at the levels of total intake, food group, and food item. Primary indices of agreement or association between measures included tests of mean difference, correlation coefficients, and percentage of subjects misclassified in tertile comparisons. Agreement between the results of measures was very good, with significant exceptions. Survey and refuse estimates of mean monthly quantity of total intake differed by less than one percent. Significant differences between mean estimates of consumption produced by each measure were found for three of 10 food group categories, and 19 of 73 food items compared. The correlation coefficient for comparisons between survey and refuse estimates of total intake was .72, and positive associations were also evident for the large majority of other items examined. Fifty-seven percent of subjects were classified into equivalent tertiles by both survey and refuse estimates of total intake. Percentages classified into equivalent tertiles in food group comparisons ranged from 48 to 70 percent. Potential effects of sources of error in refuse were also examined. Significant differences between results of the measures clustered among food items commonly associated with health risk or benefit, or items which serve as accessory elements in meals. These results suggest that, although brief food frequency questionnaires can provide valid estimates of usual diet for the majority of food categories, social desirability response effects may significantly affect reported consumption of some items.