Regression models for analyzing radiological visual grading studies – an empirical comparison

BMC Medical Imaging

Table 3 Estimated parameters, goodness-of-fit statistics and estimated dose reduction for GQscore

Model	Coefficient						Goodness-of-fit		Dose Reduction
Model	logCTDI		id2		id4		AIC	Pseudo R²	id2	id4
	Est.	P-value	Est.	P-value	Est.	P-value	AIC	Pseudo R²	id2	id4
regress^a	1.424	<0.001	0.158	<0.001	0.175	<0.001	-	0.3560	10.53 %	11.57 %
regress^a	(1.217, 1.630)		(0.085, 0.232)		(0.102, 0.248)		-	0.3560	(6.46 %, 14.59 %)	(7.58 %, 15.56 %)
ologit^a	9.626	<0.001	1.011	<0.001	1.133	<0.001	1060.25	0.3573	9.97 %	11.10 %
ologit^a	(8.020, 11.232)		(0.547, 1.476)		(0.665, 1.600)		1060.25	0.3573	(6.13 %, 13.17 %)	(7.32 %, 14.88 %)
gologit2^a	8.627	<0.001	1.092	<0.001	1.484	<0.001	1113.99	0.3816	11.89 %	15.80 %
=2	(6.519, 10.735)		(0.558, 1.625)		(0.924, 2.044)				(6.74 %, 17.03 %)	(10.48 %, 21.12 %)
gologit2^a =3	9.652	<0.001	0.964	0.073	0.387	0.507			9.50 %	3.93 %
gologit2^a =3	(6.915, 12.388)		(−0.091, 2.019)		(−0.756, 1.529)				(1.29 %, 17.72 %)	(−6.76 %, 14.61 %)
slogit^a	18.523	<0.001	1.867	<0.001	2.148	<0.001	1061.13	0.3594	9.59 %	10.95 %
slogit^a	(15.277,21.769)		(0.997, 2.738)		(1.275, 3.022)		1061.13	0.3594	(5.80 %, 13.38 %)	(7.23 %, 14.67 %)
mixed^b	1.424	<0.001	0.158	<0.001	0.175	<0.001	1119.97	0.1857	10.53 %	11.57 %
mixed^b	(1.218, 1.630)		(0.085, 0.231)		(0.102, 0.248)		1119.97	0.1857	(6.47 %, 14.59 %)	(7.58 %, 15.55 %)
meologit^b	9.179	<0.001	0.967	<0.001	1.085	<0.001	1123.66	0.1853	10.00 %	11.15 %
meologit^b	(7.031, 11.328)		(0.468, 1.467)		(0.578, 1.592)		1123.66	0.1853	(6.03 %, 13.98 %)	(7.25 %, 15.05 %)

ISSN: 1471-2342