Matching methods to create paired survival data based on an exposure occurring over time: a simulation study with application to breast cancer

Table 3 Simulations results: time-dependent bias $b_{• Z^{(w)}}$ with ${RMSE}_{• Z^{(w)}}$ for the LWA _i model and the mean of the number of pairs inside each prognostic profile and according to the matching methods are presented with the percent of imperfect pairs for Method 2

Prognostic profiles Z^{( w)}	% of exposure	Method 1				Method 2
		$b_{• Z^{(w)}}$	${RMSE}_{• Z^{(w)}}$	Mean of pairs number	$b_{• Z^{(w)}}$	${RMSE}_{• Z^{(w)}}$	Mean of pairs number	% of imperfect pairs
(0,0,0)	0.82	-0.97	1.05	40.26	0.13	0.32	146.88	0.81
(0,0,1)	0.48	-0.46	0.59	85.46	0.04	0.33	109.05	0.44
(0,1,0)	0.67	-0.70	0.79	68.82	0.08	0.31	133.47	0.64
(0,1,1)	0.30	-0.22	0.46	70.87	-0.03	0.36	74.07	0.26
(1,0,0)	0.76	-0.83	0.91	54.48	0.10	0.31	141.70	0.73
(1,0,1)	0.39	-0.34	0.51	82.27	0.007	0.34	92.61	0.35
(1,1,0)	0.58	-0.57	0.67	79.94	0.05	0.31	123.44	0.54
(1,1,1)	0.23	-0.09	0.46	56.15	-0.06	0.39	56.88	0.20

ISSN: 1471-2288