Clinical usefulness of estriol assay for predicting "light-for-dates" infants

A method was developed for determining clinical usefulness of a test for predicting "light-for-dates" infants before birth. The method was applied in a prospective study of assay of maternal estriol at 35-36 weeks of gestation as a means of identifying such infants. With the tenth centile for estriol values as the cutoff, low estriol values identified less than a third of all light-for-dates infants, and less than a third of pregnancies with low estriol values were associated with births of light-for-dates infants. Our reanalysis of data from three other published studies, using the same method of assessment, gave similar results. When laboratory tests are expected to provide clinically relevant information, we urge that data should be appropriately collected, analyzed, and reported, so clinical usefulness can be evaluated.