Table 1

Summary of the main findings for each imputation method considered

Method	Imputed mean disease activity		Imputed remission rate		Effect of random missingness	Sensitivity to quantity of data around follow-up
Method	Baseline	Follow-up	Baseline	Follow-up	Effect of random missingness	Sensitivity to quantity of data around follow-up
LOCF	Unusable	Over-estimated	Unusable	Under-estimated	None	Yes
LFE	Unusable	Under-estimated	Unusable	Over-estimated	None	Yes
NAO	Strong under-estimation	Close to unbiased	Strong over-estimation	Close to unbiased	None	Small
LE	Strong under-estimation	Close to unbiased	Strong over-estimation	Slightly biassed	None	Small
PE	Strong under-estimation	Slightly biassed	Strong over-estimation	Close to unbiased	None	Yes
LME3	Strong under-estimation	Close to unbiased	Close to unbiased	Strong under-estimation	None	No
MICE	Strong under-estimation	Close to unbiased	Close to unbiased	Strong under-estimation	Reduction of bias of remission	No

LE, linear extrapolation; LFE, linear forward extrapolation; LME3, linear mixed effects cubic regression; LOCF, last observation carried forward; MICE, multiple imputation by chained equation; NAO, nearest available observation; PE, polynomial extrapolation.