Adult height in boys and girls with untreated short stature and constitutional delay of growth and puberty: accuracy of five different methods of height prediction.

To determine how accurately several methods of height prediction estimate adult height, we compared height predictions calculated by the Bayley-Pinneau, Roche-Wainer-Thissen (RWT), target height, and Tanner-Whitehouse Mark I (TW-MI), and Mark II (TW-MII) methods with final adult height in 37 boys and 32 girls with short stature and constitutional delay of growth and puberty. They were first seen at a chronologic age (mean +/- SD) of 14.80 +/- 1.70 years (boys) and 12.87 +/- 2.56 years (girls). Adult height at 23.14 +/- 1.95 years and 21.05 +/- 2.02 years was 170.4 +/- 5.4 cm (boys) and 157.8 +/- 4.2 cm (girls), respectively, and thus within the lower range of normal. Height predictions were calculated for the total group and for patients with parents of normal (group 1) as well as short stature (group 2). For boys, the RWT method gave very accurate results, underestimating adult height by -0.6 cm for the total group. The prediction errors for the other methods were -7.3 cm (TW-MI), -4.2 cm (TW-MII), and +3.1 cm (Bayley-Pinneau method) or +1.7 cm (target height). For girls, no method was superior in estimating adult height. The mean prediction error was -0.8 cm, -2.1 cm, and -1.8 cm with the Bayley-Pinneau, TW-MI, and TW-MII methods, respectively. In contrast, adult height was overpredicted by +2.3 cm and +1.2 cm with the RWT and target height methods. We conclude that patients with short stature and constitutional delay of growth and puberty reach an adult height in the lower range of normal. Height prediction methods differ with respect to their accuracy and their tendency to overestimate or underestimate adult height.