Patterns of variability in the nutrient intake of nutritionally vulnerable pregnant women.

Intra/inter-individual variance ratios for energy, protein, fat and retinol-equivalent intake during the third trimester of pregnancy were calculated. Data were collected on 743 pregnant women from rural Indonesia between 1982 and 1984. Food intake of these women was directly weighed for three consecutive days per month during months 6-9 of pregnancy, yielding 7139 observations for analysis. Ratios were calculated separately for a single month of pregnancy and for the last trimester of pregnancy. The intra/inter-individual variance ratio for kilocalorie and protein intake was less than 1.0 for a single month; it was greater than 1.0 for the trimester. This suggests that women varied their intake of kilocalories and protein more over the course of their pregnancy than they did during one month of their pregnancy. Both the monthly and trimester variance ratios for fat and retinol-equivalent intake were greater than 1.0. Our findings suggest that usual mean intake during the third trimester of pregnancy can be measured with relative precision with two replicates per individual. However, more measurements are needed for regression analyses: measurement error greatly attenuated the coefficients, even though data were collected by the direct weighing method. Thus the number of measurements per individual needed in a study of nutrition and pregnancy depends on the nutrient, the period of 'usual intake' that is of interest, as well as the analytical strategy.