Variation in nutrient intake of infants and its implications for collecting reliable dietary intake data

Sources of variance in dietary intake data for infants in the fourth to sixth months of life were examined. Twenty-eight mothers of infants in the Special Supplemental Food Program for Women, Infants, and Children gave 24-hr recall information on their infants' diets on 4 random days over approximately 1 month. ANOVA found no day of the week or sequence effects on nutrient variation. Variance was partitioned into between- and within-subject (residual) components. Interindividual variation exceeded intraindividual variation for all nutrients except vitamin C. Variance ratios ranged from 0.3 to 1.3. Calculations of 95% confidence limits for group and individual mean intakes indicated that, whereas a single recall would estimate the group mean within 20% for most nutrients, five recalls would be necessary for the same accuracy in individual means. Magnitude and patterning of variance components differ substantially from those found in studies of adults, suggesting that infant data collection strategies should be tailored to the unique nature of the infant diet.