Biometrical analysis of individual growth curves

Longitudinal data for height (length) between birth and 2 years of age were examined for 690 Dutch Registry twin pairs. A two-stage analysis was performed, where individual growth curves were first fit to available data for each subject using a linear multiple regression procedure and estimated individual growth curve parameters were then subjected to multivariate biometrical analysis. Quadratic polynomial curves were found to adequately represent observed growth patterns for the majority of cases (median R2 = .98). A specific scalar model of sex limitation best characterized individual variation in growth curve parameters. That is, there was significantly greater genetic variation for boys than for girls in both the predicted length and rate of growth at 1 year of age and the amount of deceleration in individual growth curves across age.