Determining growth faltering with a tracking score

Abstract

A tracking score for determining growth faltering in children was developed and tested. A graphical method was developed for classifying by visual inspection whether or not a given child had faltered in growth. This method was used to classify all children in a sample of Guatemalan children as to faltering in both weight and length between 9 and 24 months of age; 80 of 345 children had faltered in weight, and 44 of 336 children had faltered in length. The accuracies for determining this faltering of seven versions of the tracking score and that of three other commonly used scores were assessed and compared using ROC curves, the areas under the curves, and two other ways. The other scores tested were the increment score, the residual score, and a growth‐curve parameter score. The accuracies of the scores were overall lower for weight than for length. For both weight and length, the growth‐curve parameter score had the lowest accuracy, and the tracking score that used an intermediate number of measurements (four in the interval of interest and three in the previous interval) had the highest accuracy. The tracking score has a number of features that make it attractive as an indicator of faltering. It is easy to calculate and interpret, preserves magnitude and direction, incorporates several growth measurements into a single indicator of faltering, gives equal weight to the information in all the measurements used, allows for missing measurements, and can be generalized to include other aspects of growth.