A non-linear beta-binomial regression model for mapping EORTC QLQ- C30 to the EQ-5D-3L in lung cancer patients: a comparison with existing approaches

Table 3 Testing of models using independent data

Model	Predicted mean (SE) [95% CI]	R²	RMSE	% of 95% CI containing the observed mean^‡
Beta (a)*	0.747 (0.0069)	0.75	0.132	77%
Beta (a)*	[0.733, 0.760]	0.75	0.132	77%
Beta (b)^†	0.622 (0.0057)	0.61	0.159	59%
Beta (b)^†	[0.608, 0.631]	0.61	0.159	59%
CLAD (a)*	0.671 (0.0091)	0.56	0.027	28%
CLAD (a)*	[0.652, 0.689]	0.56	0.027	28%
CLAD (b)^†	0.652 (0.0054)	0.47	0.154	19%
CLAD (b)^†	[0.639, 0.660]	0.47	0.154	19%
Linear Mixed (a)*	0.738 (0.0051)	0.63	0.019	45%
Linear Mixed (a)*	[0.728, 0.747]	0.63	0.019	45%
Linear Mixed (b)^†	0.642 (0.0059)	0.58	0.095	23%
Linear Mixed (b)^†	[0.630, 0.653]	0.58	0.095	23%
Quadratic (a)*	0.768 (0.0056)	0.63	0.018	37%
Quadratic (a)*	[0.757, 0.778]	0.63	0.018	37%
Quadratic (b)^†	0.636 (0.0039)	0.55	0.093	14%
Quadratic (b)^†	[0.628, 0.643]	0.55	0.093	14%
TOBIT (a)*	0.739 (0.014)	0.56	0.021	65%
TOBIT (a)*	[0.702, 0.757]	0.56	0.021	65%
TOBIT (b)^†	0.644 (0.0084)	0.59	0.112	24%
TOBIT (b)^†	[0.627, 0.660]	0.59	0.112	24%
Quantile (a)*	0.772 (0.0060)	0.62	0.190	21%
Quantile (a)*	[0.754, 0.778]	0.62	0.190	21%
Quantile (b)^†	0.661 (0.0084)	0.58	0.148	8%
Quantile (b)^†	[0.644, 0.677]	0.58	0.148	8%

ISSN: 1477-7525