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

Table 1 Comparisons of proposed weight, global weight and local weight to estimate the overall hazard ratio in the simulations, with homogenous design and treatment assignment generated with \(X_1\)~\(X_4\) and \(X_{1}^{2}\), \(X_{1}X_{4}\)

			\(\boldsymbol K\boldsymbol=\mathbf5\)			\(\boldsymbol K\boldsymbol=\mathbf{10}\)			\(\boldsymbol K\boldsymbol=\mathbf{20}\)
		Global weight	Local weight	Proposed weight	Global weight	Local weight	Proposed weight	Global weight	Local weight	Proposed weight
\({n}_{k}=500\)	Bias	-0.018	-0.014	-0.017	-0.006	-0.012	-0.007	-0.002	-0.004	0.001
	RMSE	0.116	0.129	0.106	0.089	0.088	0.073	0.066	0.076	0.056
	r-RMSE	1.093	1.215	1.000 (Ref)	1.225	1.211	1.000 (Ref)	1.176	1.354	1.000 (Ref)
\({n}_{k}=1000\)	Bias	-0.006	-0.008	-0.006	-0.002	-0.003	-0.001	0.000	-0.004	0.001
	RMSE	0.089	0.100	0.077	0.066	0.068	0.058	0.122	0.093	0.087
	r-RMSE	1.160	1.303	1.000 (Ref)	1.140	1.175	1.000 (Ref)	1.401	1.068	1.000 (Ref)
\({n}_{k}=2000\)	Bias	-0.002	-0.003	-0.001	0.000	-0.004	-0.001	-0.002	-0.005	-0.002
	RMSE	0.066	0.065	0.061	0.122	0.082	0.077	0.050	0.045	0.040
	r-RMSE	1.087	1.071	1.000 (Ref)	1.578	1.060	1.000 (Ref)	1.249	1.124	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