Factors affecting the benefit of glasses alone in treating childhood amblyopia: an analysis of PEDIG data

BMC Ophthalmology

Table 4 Results of stepwise, linear regression models for changes in VAAE, IOD-VA and stereopsis for the strabismic (group S) and combined-mechanism (C) amblyopes (n = 117 participants, 66 out of 183 excluded for incomplete data). T The importance (I) of a predictor is the residual sum of squares with the predictor removed from the model, normalized so that the importance values sum to 1. The percentage is the model’s accuracy, defined as the adjusted R². C: coefficient. Significant p-value (*, p < 0.05). Cells for model terms not contributing significantly to the model’s predictive power are empty. The factors missing from the table, do not contribute to the model. The numbers between parentheses indicate the order of importance in the model for each statistically significant variable

	ΔVA (23.1%)			ΔIOD-VA (22.7%)			ΔStereoacuity (sg/arc) (14.3%)
Model Term	C	P value	I	C	P value	I	C	P value	I
Intercept	-0.255	< 0.001*		-0.401	< 0.001*		-115.091	0.328
Baseline VA-AE	-0.229	0.003*	0.233 (2)	0.383	0.007*	0.127 (4)
Baseline IOD-VA				-0.586	< 0.001*	0.287 (1)	328.207	0.014*	0.201 (3)
Baseline Stereo	< 0.001	0.011*	0.050 (5)	< 0.001	0.004*	0.146 (3)	-0.315	0.005*	0.265 (2)
Mixed vs. Strabismic	0.096	0.004*	0.220 (3)	0.089	0.005*	0.143 (5)	103.338	0.076	0.103
Poor vs. Excellent Compliance	0.457	0.009*	0.254 (1)	0.447	0.006*	0.186 (2)
Fair vs. Excellent Compliance	0.202	0.101	0.254	0.203	0.072	0.186
Good vs. Excellent Compliance	0.056	0.325	0.254	0.022	0.683	0.186
Baseline spherical equivalent AE	-0.020	0.029*	0.123 (4)	-0.015	0.079	0.054	-28.800	0.061	0.115
Baseline NHT Size							9.534	0.002*	0.315 (1)
Age				0.029	0.070	0.057

ISSN: 1471-2415