Skip to main content

View full-text article in PMC

. 2012 Jan 1;14(1):31–36.

Table 1. Comparison between imputation models in terms of composition and performance.

Variable	Level	Median model		Regression model		E-M model		MICE model
Variable	Level	HR^a (95% CI^b)	Pvalue			HR^a (95% CI^b)	Pvalue	HR^a(95% CI^b)	Pvalue
Stage	1	1		1		1		1
	2	3.79 (1.96, 7.33)	<0.001	2.57 (1.39, 4.75)	0.003	3.84 (1.94, 7.22)	<0.001	3.13 (1.64, 5.97)	<0.001
	3	2.99 (1.24, 7.13)	0.014	2.17 (0.94, 4.99)	0.07	3.21 (1.35, 7.65)	0.01	2.53 (1.05, 6.12)	0.03
Grade	1	1		1		1		1
	2	1.69 (0.82, 3.49)	0.16	2.03 (1, 4.10)	0.05	1.56 (0.76, 3.19)	0.22	2.46 (1.15, 5.24)	0.02
	3	1.25 (0.56, 2.80)	0.59	1.51 (0.67, 3.37)	0.32	1.27 (0.56, 2.84)	0.57	1.52 (0.65, 3.60)	0.34
Age	<48	1		1		1		1
Age	≥48	1.80 (0.95, 3.45)	0.07	2.12 (1.41, 3.95)	0.02	1.72 (0.89, 3.32)	0.11	1.92 (1.01, 3.65)	0.04
Benign	No	1		1		1		1
Benign	Yes	2.26 (1.25, 4.11)	0.01	2.29 (1.27, 4.13)	0.01	2.13 (1.15, 3.94)	0.02	2.32 (1.24, 4.33)	0.01
Performance of models
Sensitivity		23%		36%		88%		88%
Specificity		59%		48%		56%		60%

^a HR: Hazard ratio

^b CI: Confidence interval, SD: Standard deviation