A REGRESSION METHOD INCLUDING CHRONOLOGICAL AND BONE-AGE FOR PREDICTING FINAL HEIGHT IN TURNERS-SYNDROME, WITH A COMPARISON OF EXISTING METHODS

A total of 235 measurement points of 57 Dutch women with Turner's synd rome (TS), including women with spontaneous menarche and oestrogen tre atment, served to develop a new Turner-specific final height (FH) pred iction method (PTS). Analogous to the Tanner and Whitehouse mark 2 met hod (TW) for normal children, smoothed regression coefficients are tab ulated for PTS for height (H), chronological age (CA) and bone age (BA ), both TW RUS and Greulich and Pyle (GP). Comparison between all meth ods on 40 measurement points of 21 Danish TS women showed small mean p rediction errors (predicted minus observed FH) and corresponding stand ard deviation (ESD) of both PTSRUS and PTSGP, in particular at the ''y ounger'' ages. Comparison between existing methods on the Dutch data i ndicated a tendency to overpredict FH. Before the CA of 9 years the me an prediction errors of the Bayley and Pinneau and TW methods were mar kedly higher compared with the other methods. Overall, the simplest me thods-projected height (PAH) and its modification (mPAH)-were remarkab ly good at most ages. Although the validity of PTSRUS and PTSGP remain s to be tested below the age of 6 years, both gave small mean predicti on errors and a high accuracy. FH prediction in TS is important in the consideration of growth-promoting therapy or in the evaluation of its effects.