Abstract
Consistency of intake over time for calcium, phosphorus, protien and energy, as estimated from diet diaries, was studied prospectively in two groups of women. One group provided short-term data (n = 165), and the other, data spanning intervals of up to 20 y (n = 164). Although group average intakes changed little, there was a great deal of individual variation. Even at intervals as short as 6 m, current intake estimates accounted for only about 40% of the variance of intakes at en earlier time. Over longer intervals, the predictive value of one estimate in respect to a second deteriorated even further. Thus, across 5 y, less than 25% of the variance of all four nutrient intakes could be accounted for by current intake. The 95% confidence interval for estimate of earlier intake based upon a later measurement was more than ±700 mg for calcium, the least consistent nutrient. The other three nutrients showed nearly identical interstudy consistency, with uncertainties relative to their respective means being about half as large as for calcium. These data show that accurate longitudinal estimates of intake require continuous, prospective monitoring and that current intake is not a good estimator of past intake for most nutrients, especially for calcium.