Table 2.  Predictive factors for adult height gain in patients with growth hormone deficiency: final model Variable* Regression coefficient (95% CI) P value No of patients in final model (r2) 1885 (0.58) Patients' characteristics:   Target height (SD score) 0.22 (0.19 to 0.25) <0.0001   Birth weight (SD score) 0.11 (0.08 to 0.13) <0.0001   Sex (male=1, female=0) 0.54 (0.62 to 0.47) <0.0001 Baseline variables:   Age (years) 0.31 (0.28 to 0.33) <0.0001   Pubertal (yes=1, no=0) 0.19 (0.26 to 0.11) <0.0001   Bone age delay (years) 0.18 (0.15 to 0.20) <0.0001   Weight (SD score) 0.19 (0.23 to 0.14) <0.0001   Log peak growth hormone concentration (µg/l) 0.08 (0.13 to 0.03) 0.002 Treatment variables:   Completion of treatment as scheduled (yes=1, no=0) 0.30 (0.37 to 0.22) <0.0001   Duration of treatment (years) 0.22 (0.19 to 0.25) <0.0001 SD=standard deviation. * The model includes baseline height SD score, time interval between baseline and adult height measurements, and the interaction between these two variables; therefore, the model similarly predicts the factors for adult height itself (in SD). The regression coefficient represents the change in SD score per unit change in predictor. The number of subjects corresponds to those with no missing value for any predictor variable.