Covariate balance-related propensity score weighting in estimating overall hazard ratio with distributed survival data

Table 2 Comparisons of proposed weight, global weight and local weight to estimate the overall hazard ratio in the simulations, with heterogeneous design and treatment assignment generated with \(X_{1}\)~\(X_{4}\) and \(X_{1}^{2}\), \(X_{1}X_{4}\)

			\(\boldsymbol K\boldsymbol=\mathbf{5}\)			\(\boldsymbol K\boldsymbol=\mathbf{10}\)			\(\boldsymbol K\boldsymbol=\mathbf{20}\)
		Global weight	Local weight	Proposed weight	Global weight	Local weight	Proposed weight	Global weight	Local weight	Proposed weight
\({n}_{k}=500\)	Bias	0.045	0.041	0.033	0.026	0.023	0.015	0.025	0.019	0.009
	RMSE	0.129	0.125	0.110	0.082	0.097	0.065	0.069	0.070	0.049
	r-RMSE	1.176	1.140	1.000 (Ref)	1.255	1.484	1.000 (Ref)	1.400	1.421	1.000 (Ref)
\({n}_{k}=1000\)	Bias	0.039	0.024	0.022	0.027	0.018	0.012	0.031	0.020	0.012
	RMSE	0.096	0.092	0.076	0.083	0.079	0.067	0.105	0.084	0.072
	r-RMSE	1.269	1.216	1.000 (Ref)	1.247	1.187	1.000 (Ref)	1.458	1.166	1.000 (Ref)
\({n}_{k}=2000\)	Bias	0.031	0.014	0.011	0.037	0.021	0.017	0.021	0.005	0.002
	RMSE	0.093	0.094	0.077	0.196	0.175	0.161	0.042	0.049	0.032
	r-RMSE	1.210	1.223	1.000 (Ref)	1.215	1.084	1.000 (Ref)	1.320	1.540	1.000 (Ref)

Bias Absolute bias, RMSE Root mean squared error, r-RMSE Ratio of RMSE of global weight or local weight against proposed weight

ISSN: 1471-2288