Table 2.

Risk of bias assessment of key trials.

	TRITON-TIMI 38		TRILOGY ACS		PLATO
	Authors judgement	Support for judgement	Authors judgement	Support for judgement	Authors judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk of bias	No information provided on randomization protocol aside from stratification based on presenting symptoms	Unclear risk of bias	No information provided on randomization protocol aside from stratification by age, country, and prior clopidogrel treatment status	Low risk of bias	Patients randomised using computer software (treating physician could not influence) operating on a randomisation schedule managed by a group not involved in the trial
Allocation concealment (selection bias)	Low risk of bias	Double-dummy trial method used	Low risk of bias	Double-dummy trial method used	Low risk of bias	Double-dummy trial method used
Blinding of participants and personnel (performance bias)	Unclear risk of bias	Trial is “double blinded" but does not refer explicitly who is blinded	Unclear risk of bias	Trial is “double blinded" but does not refer explicitly who is blinded	Unclear risk of bias	Trial is “double blinded" but does not refer explicitly who is blinded
Blinding of outcome assessment (detection bias)	Unclear risk of bias	Trial is “double blinded" but does not refer explicitly who is blinded	Unclear risk of bias	Trial is “double blinded" but does not refer explicitly who is blinded	Unclear risk of bias	Trial is “double blinded" but does not refer explicitly who is blinded
Incomplete outcome data (attrition bias)	Low risk of bias	Intention to treat analysis, only 0.1% of patients lost to followup, raw data for safety and efficacy endpoints provided for overall cohort	Low risk of bias	A similar proportion of the prasugrel and clopidogrel groups (76% versus 78%, resp.) continued to receive the study drug throughout the followup. A small proportion (120 patients from 4 sites) were excluded from analysis due to breaking protocol, this was done prior to unblinding, and therefore the effects were unknown to the authors.	Low risk of bias	Analysis carried out on entire prespecified stratum. Similar proportions of study drug (14.6) and control drug (14.0) treated patients dropped out of the study. Intention to treat analysis
Selective reporting (reporting bias)	Low risk of bias	Study protocol available and all primary, secondary and safety outcomes are reported	Low risk of bias	Study protocol available and all primary, secondary, and safety outcomes are reported	Low risk of bias	Study protocol available and all primary, secondary, and safety outcomes are reported
Other sources of bias	Low risk of bias	The study appears to be free of other sources of bias	Low risk of bias	The study appears to be free of other sources of bias	Low risk of bias	The study appears to be free of other sources of bias