Multiple imputation for handling missing outcome data when estimating the relative risk

Table 1 Results for X ₁ and X ₂ Binary, λ = 2, RR(X ₁, X ₂) = 2 and β ₁ = β ₂ = log(3)

Simulation scenario	Method	β ₁					β ₂
Simulation scenario	Method	Bias	Avg SE	Emp SE	Coverage	MSE	Bias	Avg SE	Emp SE	Coverage	MSE
Coordinated, prevalence =0.10	MVNI	−0.08	0.30	0.28	0.951	0.08	−0.28	0.35	0.28	0.896	0.15
	MVNI + deletion	0.06	0.30	0.31	0.955	0.10	−0.09	0.39	0.35	0.956	0.13
	FCS	0.02	0.30	0.31	0.955	0.10	0.00	0.40	0.40	0.962	0.16
	FCS + deletion	0.02	0.30	0.31	0.948	0.09	0.01	0.40	0.40	0.962	0.16
	CCA	0.01	0.34	0.34	0.953	0.12	0.03	0.40	0.40	0.964	0.16
Coordinated, prevalence =0.30	MVNI	0.03	0.16	0.15	0.952	0.02	−0.32	0.17	0.16	0.547	0.13
	MVNI + deletion	0.05	0.16	0.16	0.948	0.03	−0.15	0.20	0.19	0.872	0.06
	FCS	0.03	0.16	0.16	0.951	0.03	−0.11	0.20	0.21	0.893	0.05
	FCS + deletion	0.02	0.16	0.16	0.955	0.02	−0.06	0.21	0.21	0.932	0.05
	CCA	0.01	0.17	0.17	0.953	0.03	0.01	0.21	0.22	0.949	0.05
Opposite, prevalence =0.10	MVNI	−0.08	0.29	0.28	0.949	0.08	−0.26	0.34	0.26	0.908	0.13
	MVNI + deletion	0.05	0.30	0.30	0.955	0.09	−0.07	0.37	0.33	0.964	0.11
	FCS	0.01	0.30	0.31	0.952	0.10	0.03	0.39	0.39	0.963	0.16
	FCS + deletion	0.01	0.30	0.31	0.950	0.09	0.05	0.39	0.40	0.964	0.16
	CCA	0.03	0.39	0.41	0.956	0.17	0.03	0.39	0.39	0.965	0.15
Opposite, prevalence =0.30	MVNI	0.00	0.00	0.15	0.961	0.02	−0.20	0.18	0.16	0.805	0.06
	MVNI + deletion	0.03	0.16	0.15	0.961	0.02	−0.02	0.20	0.19	0.952	0.03
	FCS	0.00	0.16	0.16	0.951	0.02	0.01	0.20	0.20	0.948	0.04
	FCS + deletion	−0.02	0.16	0.15	0.949	0.02	0.07	0.21	0.21	0.947	0.05
	CCA	0.01	0.20	0.20	0.952	0.04	0.02	0.20	0.20	0.952	0.04

Abbreviations: MVNI multivariate normal imputation, FCS fully conditional specification, CCA complete case analysis, Avg SE average standard error, Emp SE empirical standard error, MSE mean square error

ISSN: 1471-2288