Supplementary Table 3.

Assessments of risk bias

Study, year	Randomization sequence generation	Allocation concealment	Blinding of participants, personnel	Blinding of outcomes assessment	Incomplete outcome data	Selective reporting	Other sources of bias
REAL/ZEST-LATE, 2010	Low risk: A preestablished, computer-generated randomization scheme	Unclear risk	High risk: Open-label study	Low risk	Low risk: 17 lost to follow-up (0.6%), but with an intention-to-treat analysis	Low risk	Low risk
RESET, 2012	Low risk: Web-based response system	Unclear risk	High risk: Open-label study	Low risk	Low risk: 31 lost to follow-up (1.5%), but with an intention-to-treat analysis	Low risk	Low risk
OPTIMIZE, 2013	Low risk: A dedicated web-based system and stratified by the presence of DM and institution	Unclear risk	High risk: Open-label study	Low risk	Low risk: 76 lost to follow-up (2.4%), but with an intention-to-treat analysis	Low risk	Low risk
DAPT, 2014	Low risk: A central interactive voice response system	Unclear risk	High risk: Open-label study	Low risk	Low risk: 571 lost to follow-up (5.7%), but with an intention-to-treat analysis	Low risk	Low risk
EXCELLENT, 2012	Low risk: A web-based online randomization system	Unclear risk	High risk: Open-label study	Low risk	Low risk: 15 lost to follow-up (1%), but with an intention-to-treat analysis	Low risk	Low risk
SECURITY, 2016	Low risk: By electronic case report, and balanced within center by blocks of 4	Unclear risk	High risk: Open-label study	Low risk	Low risk: 263 loss to follow-up (19%), but none excluded from the analysis	Low risk	Low risk

DAPT: Dual antiplatelet therapy; DM: Diabetes mellitus.