. Author manuscript; available in PMC: 2014 Mar 14.

Published in final edited form as: Biometrics. 2013 Nov 13;70(1):144–152. doi: 10.1111/biom.12105

Table 3.

CCC results: Logistic model for hazard of developing end-stage liver disease as a function of HCV clearance. Naïvt refers to unweighted logistic regression. Variance estimates were obtained using a robust sandwich estimator for tht Naïve and IPTW methods and the efficient influence curve for the TMLE. Each method was performed on 50 imputed datasets and the inference combined.

Method	Est	SE	95% CI
γ₂Intercept
Naïve	−3.05	0.29	(−3.62,−2.49)
IPTW	−3.30	1.03	(−5.32,−1.27)
TMLE	−3.37	0.68	(−4.70,−2.04)
γ₁Coefficient of exposure status
Naïve	−0.12	0.37	(−0.85,0.62)
IPTW	−0.44	0.82	(−2.05,1.17)
TMLE	−0.35	0.46	(−1.26,0.55)
γ₂Coefficient of time
Naïve	0.10	0.07	(−0.05,0.24)
IPTW	0.22	0.22	(−0.21,0.66)
TMLE	0.22	0.15	(−0.08,0.52)