Abstract
The currently used final height prediction methods are based on an estimate of the growth potential, which can be read from the Bayley-Pinneau tables once bone age (BA) and chronological age (CA) are known. Using a spline-fit of recent Dutch height for age percentiles and of the corresponding standard deviations, at any age a prediction of final height can be made. Such predictions are associated with a certain error. Knowledge of the error distribution allows the calculation of (age-dependent) statistical weights for both CA- and BA-based height predictions. Our procedure involves the application of a weighted linear regression procedure on all height predictions available of an individual (CA and BA-based). The best estimate is calculated at the age corresponding to the last data-point. In principle this method can be applied to all reasonably homogeneous groups, without use of the Bayley-Pinneau tables. In a population of healthy tall-for-age girls (N=60; mean age at time of prediction 12 yrs), for whom final height was known, predictions were obtained with a mean error of 0.2 cm; 1 SD=2.12 cm.
Article PDF
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Degenhart, H., Bak, A., Drop, S. et al. Height prediction by a linear regression procedure. Pediatr Res 18, 1218 (1984). https://doi.org/10.1203/00006450-198411000-00105
Issue Date:
DOI: https://doi.org/10.1203/00006450-198411000-00105