On the use of propensity scores in case of rare exposure

Table 1 Results for the reference configuration

Method	y	n	Bias	V ratio	RMSE	1-coverage	% match
PSW-ATE	10	364	0.056	0.914	0.406	0.091
	20	728	0.028	0.982	0.271	0.065
	30	1092	0.018	1.009	0.216	0.057
PSW-ATT	10	364	0.026	0.983	0.321	0.060
	20	728	0.013	1.019	0.222	0.047
	30	1092	0.008	1.031	0.180	0.046
PS-matching	10	364	0.051	0.925	0.473	0.062	99.0
	20	728	0.026	0.964	0.316	0.056	99.5
	30	1092	0.017	0.990	0.250	0.053	99.7

Bias, variability ratio, RMSE, and 1-coverage according to analytical method, number of events in the exposed group y, and mean analyzed sample size n, for one scenario (p=5 %, σ _U,B=σ _U,C=0.3, σ _B,C=0, \(\exp (\delta _{B}) = \exp (\delta _{C}) = 1.5\), H R=1, 2 confounding factors, censoring rate r _c=50 %). The mean percentage of matched exposed subjects is reported for the PS-matching method

ISSN: 1471-2288