Prediction of adult height

Abstract

Different methods of height prediction were compared in healthy subjects of known adult height, 30 males and 30 females. Predictions were made yearly at ages 8-17 years. Epiphyseal bone age (BA) was determined with Greulich-Pyle Atlas (G-P) and TW 2 RUS. The methods assuming constant SDS for age (CD) or for BA (IPH), and the Walker method were inaccurate with mean absolute errors up to 10 cm, and systematic overprediction except for IPH with TW 2 RUS. The methods of Bayley and Pinneau (BP), Roche et al. (RWT) and Tanner et al. (T) had about equal accuracy with mean absolute errors of 2-4 cm. With G-P BA adjusted to the growth of Finnish children, the IPH was as accurate as BP, RWT and T. RWT, and BP in girls, tended to overpredict. The year-to-year variation of prediction was greatest with BP, IPH, and chronological age based T. In a group of 8 boys with delayed growth and maturation, IPH with the adjusted G-P BA proved to be the most accurate method.