The Impact of Postrandomization Crossover of Therapy in Acute Coronary Syndromes Care
What is the implication of post-randomization crossover of care in patients with acute coronary syndrome (ACS)?
The authors assessed the outcome of patients enrolled in the SYNERGY trial who crossed over to another therapy after randomization, and compared them to those without crossover. The impact of crossover on the overall outcome of the trial was assessed using inverse probability weighting (IPW). The analysis included 9,613 (96.3%) patients who received at least 1 dose of randomized therapy.
Post-randomization crossover occurred in 740 (7.7%) patients (554 enoxaparin, 186 unfractionated heparin [UFH]). In unadjusted analysis, crossover patients had higher rates of 30-day death/myocardial infarction (MI) (18.9% vs. 14.0%), TIMI major bleeding (16.9% vs. 7.6%), GUSTO severe bleeding (4.5% vs. 2.3%), and transfusions (32.3% vs. 15.2%). Adjustment using the IPW technique suggested that crossover had minimal impact on study outcome, with the 30-day death or MI rate changing from 13.9% and 14.5% (enoxaparin vs. UFH) to 13.3% and 14.2%. IPW adjustment for TIMI bleeding changed the rates from 9.1% versus 7.6% to 8.2% versus 7.2%, respectively.
Patients with post-randomization crossover had worse outcome compared with those without crossover, but use of appropriate adjustment suggested that crossover had only minimal impact on overall treatment effect.
This study highlights the significant bias that limits any interpretation of outcome of a subgroup of patients who cross over from one therapy to another in a clinical trial. Data from such an analysis should be interpreted cautiously and should serve at best as hypothesis generating.
Clinical Topics: Acute Coronary Syndromes
Keywords: Myocardial Infarction, Acute Coronary Syndrome, Cross-Over Studies, Probability
< Back to Listings