Background: We analyzed the difference between off-pump and on-pump coronary bypass surgery towards early patient outcome.
Methods: All consecutive patients (1997 to 2005) receiving primary or redo coronary bypass surgery were included. Acute infarct patients were excluded since they would unfairly compromize the analysis. Off-pump surgery was performed in 2864 patients, 1583 patients were operated on-pump. Focus of the analysis was placed on early outcome (90-day mortality and freedom from stroke). Control of potential confounders was provided through a saturated propensity model and Cox proportional hazards modelling.
Results: The propensity score reached a c-statistic of 0.84 and included 14 variables (demographic, clinical, interventional and institutional). Univariate comparison revealed that off-pump patients had significantly lower 90-day mortality (2.6% vs 4.1%, p=0.009) and significant lower occurrence of postoperative stroke (0.8% vs 1.7%, p=0.008). After adjustment for propensity score, multivariate Cox analysis revealed a p-value of 0.1 for 90-day mortality. Unfortunately, the statistical power of this analysis only reached 0.77, making a non-significant p-value unreliable. Concerning freedom from postoperative stroke, multivariate analysis showed clear superiority of the off-pump group (p=0.001). When reviewing the most recent 1000 patients receiving coronary artery surgery in our center, off-pump surgery was performed in 98.8% of cases. While the predicted EuroSCORE risk for this specific group was 5.2%, the observed postoperative mortality was 0.4% at day 10, 0.7% at day 30 and 1.3% at day 90.
Conclusion: Off-pump coronary bypass surgery provides better early survival (in multivariate setting not yet significant) and less postoperative stroke. The multivariate analysis concerning survival reached a power of 0.77, which implicates a 23% chance for a type II error (beta error). Most investigators agree that beta error rates greater than 20% (study power less than 80%) are subject to unacceptably high risks of false negative results. With the observed differences in 90-day mortality, one can easily calculate that at least 2300 patients are needed in each group to reach the 80% power limit.