Risks and benefits of glycoprotein IIb/IIIa antagonists in acute coronary syndrome

BACKGROUND: Recommendations for the use of glycoprotein (GP) IIb/IIIa antagonists should consider not only their effectiveness but also their links with other clinical findings (both favorable and adverse), the risk/benefit ratio in different subgroups of patients, and the existence of other therapeutic alternatives; additionally, the estimates underlying the recommendation should be explicit. OBJECTIVE: To establish explicit evidence-based recommendations regarding the use of GP IIb/IIIa antagonists in medically treated patients with acute coronary syndrome without ST-segment elevation. DATA SOURCES: MEDLINE-based search (1980 to November 1999) of randomized controlled trials for effectiveness data and nonsystematic review of published data regarding utilities of relevant status and clinical events. STUDY SELECTION: We included clinical trials in which the patients were randomly assigned either to an experimental group treated with intravenous GP IIb/IIIa antagonists or to a control group. We excluded studies in which the intention to perform a percutaneous procedure was a criterion for inclusion. DATA EXTRACTION: The effectiveness of the treatment was defined as the incidence of death or a nonfatal infarct at 30 days. The risks of the treatment were estimated using the incidence of moderate to severe hemorrhage. DATA SYNTHESIS: Meta-analysis of randomized controlled trials in patients with acute coronary syndrome and calculation of the threshold number needed to treat (t-NNT). RESULTS: Compared with conventional treatment, GP IIb/IIIa antagonists reduce the risk of death or nonfatal infarct at 30 days by approximately 11.7% (number needed to treat [NNT] = 65 for the basal risk of patients included in the studies; 95% CI 40 to 203). However, the use of GP IIb/IIIa antagonists increases the risk of moderate to severe hemorrhage by 32%. In an analysis biased in favor of the use of GP IIb/IIIa antagonists, these risks imply a t-NNT of approximately 150, which overlaps with the confidence interval of the basal NNT. The limits of a 95% confidence interval of the NNT are only lower than the t-NNT in patients with a high risk of death/infarct (at least 5%) and low risk of hemorrhage (less than the weighted basal risk in the trials analyzed). CONCLUSIONS: At present, there is no conclusive evidence that the expected benefits outweigh the risks in the average patient included in the available trials. The benefit is probably greater than the risks in patients with