Table 4.

Statistical analysis for the cohort of 185. Predictors of Scoliosis: Logistic Regression – Univariate Models Results

	B	S.E.	Wald	Sig.	Exp(B)
Age Comm.	.039	.067	.335	.563	1.040
Duration	-.166	.099	2.814	.093	.847
Mg per Kg1	2.544	1.277	3.965	.046	12.728
Mg per Kg2	2.420	1.122	4.656	.031	11.245
Mg per Kg3	.716	2.085	.118	.731	2.046
Mg per Kg4	7.827	4.661	2.819	.093	2507.794
Last PCTL	-.015	.016	.863	.353	.985
Turner syndr.	-1.790	.471	14.430	<.001	.167
First PCTL_Age	-.046	.076	.364	.546	.955
First_PCTL	-.026	.024	1.255	.263	.974

B = Multiple regression co-efficient

S.E = Standard error

Wald (Wald Abraham. Selected papers in statistics and probability. Stanford University Press. New York, Toronto, London. 1955.)

Sig.= Significance

Exp (B) = Risk ratio

Each variable was used in turn to predict the occurrence of scoliosis in a logistic regression model. The size of the first and second doses of HGH (MG per KG1 and 2) and having Turner syndrome were predictive of having scoliosis. Variables for analysis where numbers exceeded 30 –

a) Age of commencement

b) Duration of HGH treatment

c) Dose of HGH in Mg/Kg body weight

d) First percentile height at commencement of HGH treatment (PCTL)

e) Last percentile height at cessation of HGH treatment (PCTL)

f) Presence of Turner syndrome