Table 3.

Relative mean deviation in percent and range of imputed estimates to true incremental cost, effects and net monetary benefit of the item and the aggregate imputation (simulation study based on n = 289 patients)

	Item imputation					Aggregate imputation
	Relative bias from true incremental cost		Relative bias from true incremental effects		Sampling coverage probability	Relative bias from true incremental cost		Relative bias from true incremental effects		Sampling coverage probability
	Mean (%)	Rangea (%)	Mean (%)	Rangea (%)	Mean	Mean (%)	Rangea (%)	Mean (%)	Rangea (%)	Mean
Missing completely at random (MCAR)
10%	− 0.4	− 12−11	− 0.8	− 15−13	0.826	− 0.2	− 39−32	0.7	− 37−39	0.817
20%	− 0.7	− 18−15	− 1.6	− 21−17	0.824	− 0.6	− 52−39	− 0.7	− 47−47	0.800
40%	− 0.4	− 29−23	− 4.6	− 31−21	0.819	− 0.2	− 69−67	− 1.6	− 67–66	0.779
Missing at random (MAR)
10%	− 0.6	− 13−11	− 2.1	− 13−11	0.828	− 1.2	− 41−39	0.8	− 32−35	0.818
20%	− 0.7	− 21−19	− 2.1	− 19−13	0.826	− 2.4	− 45−42	3.8	− 39−44	0.812
40%	− 1.4	− 29−25	− 1.9	− 25−24	0.824	− 1.9	− 76−67	10.1	− 56−83	0.775
Missing not at random (MNAR)
10%	− 0.7	− 15−16	–	–	0.812	− 1.5	− 49−44	–	–	0.798
20%	− 0.8	− 37−25	–	–	0.782	− 1.3	− 55−55	–	–	0.740
40%	− 1.3	− 38−3	–	–	0.712	− 2.2	− 74−67	–	–	0.604

^aThe 5th and 95th percentiles were used to demonstrate the range of the relative bias to true incremental cost and QALYs