Evaluating the 'all-comers' design: a comparison of participants in two 'all-comers' PCI trials with non-participants

We aimed to asses the generalizability of two 'all-comers' randomized clinical trials (AC-RCTs) in patients undergoing percutaneous coronary intervention (PCI).

Recently two large AC-RCT's comparing drug-eluting stents were performed in our institution (LEADERS and RESOLUTE-III). During the inclusion period of these trials 1242 consecutive PCI patients were treated of whom 579 (48%) were actually included. The most important reasons for non-participation were inability to provide informed consent (33.5%), refused to participate (19%), or patient met one of the other exclusion criteria (26.9%). Trial participants more frequently had stable angina (42.5 vs. 34.4%) and less frequently acute myocardial infarction as indication for PCI (31.4 vs. 42.4%) than non-participants. Hypertension (52.8 vs. 49.1%) and hypercholesterolaemia (56.3 vs. 49.1%) were seen more frequently in trial participants; heart failure was less common (2.1 vs. 4.4%). A significant difference in 30-day mortality was observed between AC-RCT participants and non-participants [0.7 vs. 4.5% events; adjusted hazard ratio (aHR) 0.18 and 95% confidence interval (CI) 0.06-0.52]. One-year mortality was also lower (3.1 vs. 6.9% events; aHR: 0.51 and 95% CI: 0.29-0.91, but 1-year mortality in 48 h survivors was similar (3.1 vs. 4.2% events; aHR: 0.74 and 95% CI: 0.41-1.34).

Applying the all-comers design did not result in inclusion of all consecutive patients, as only half of the target population was enrolled. It should be noted, however, that this design included more patients than observed in classical RCTs. AC-RCT participants and non-participants were different in terms of baseline characteristics and outcome.