The effect of controlling for kilocalories on classification to quintiles of nutrient intake: One- to seven-day food records as estimators of usual intake

The effect of adjustment for kilocalories on accuracy of classification to quintiles of intake was studied. Categorization of 106 women to quintiles by one-, three-, and seven-day food records was compared to categorization by usual intake estimated from 37 to 72 days of records per subject. Subjects in the study were volunteers, aged 35-65, from the Madison, WI area. During three years (1979-1982), food records were collected on a structured, pre-coded form. Two techniques of adjusting intake of calcium, vitamin A and vitamin C for kilocalorie intake were investigated: nutrient density ratios of intake to kilocalorie intake and kilocalorie-adjusted residuals from the regression of nutrient intake on energy intake. Correlations with energy intake for calcium, vitamin A and vitamin C were r=0.65, 0.48 and 0.05, respectively. For both kilocalorie-controlled measurements, misclassification was approximately equal to misclassification without controlling for kilocalories; controlling for kilocalories did not reduce misclassification due to intraindividual variability.