Fig. 3.
(a) Subgroup analysis of risk difference of relapse rates according to study characteristics (treatment strategy, follow-up period and relapse threshold). Discontinuation studies: Boonstra et al (2011) ,31 Chen et al (2010),30 Crow et al (1986),28 Kane et al (1982)27 and McCreadie et al (1989).29 Targeted discontinuation studies: Gaebel et al (2002, 2011).23,9 Follow-up ≤1 year studies: Chen et al (2010),30 Gaebel et al (2002),23 Kane et al (1982)27 and McCreadie et al (1989).29 Follow-up >1 year studies: Boonstra et al (2011),31 Crow et al (1986)28 and Gaebel et al (2002).23 Relapse studies with lower threshold: Boonstra et al (2011),3,1 Chen et al (2010)30 and Kane et al (1982).27 Relapse studies with higher threshold: Crow et al (1986),28 Gaebel et al (2002, 2011)23,9 and McCreadie et al (1989).29 (b) Subgroup analysis of risk difference of relapse rates according to study characteristics (sample size, blinding and risk of bias). Studies with sample size of <40 participants: Boonstra et al (2011),3,1Kane et al (1982)27 and McCreadie et al (1989).29 Studies with sample size of >40 participants: Chen et al (2010),30 Crow et al (1986)28 and Gaebel et al (2002, 2011).23,9 Blinded studies: Chen et al (2010),30 Crow et al (1986),28 Kane et al (1982)27 and McCreadie et al (1989).29 Open studies: Boonstra et al (2011)31 and Gaebel et al (2002, 2011).23,9 High risk of bias: Crow et al (1986),28 Kane et al (1982)27 and McCreadie et al (1989).29 Low risk of bias: Boonstra et al (2011),31 Chen et al (2010)30 and Gaebel et al (2002, 2011).23,9