Long-Term Survival
Median follow up was 2.28 years (IQR 0.04 to 4.50 years). One- and five-year unadjusted actuarial survival was 56.1% and 43.0% for the entire cohort, respectively (Figure 1A ). Figure 1Bdisplays unadjusted survival, stratified by mechanism of MCS. In this analysis, patients were categorized by the highest level of support used (ECMO > Impella > IABP). At one and five years, actuarial survival was highest in patients bridged with an Impella device and lowest in patients bridged with ECMO.
Cox proportional hazards modeling was performed to identify predictors of mortality in patients bridged with MCS from CBP. In a univariable analysis, postoperative IABP insertion (HR 1.45, 95% CI 1.00 to 2.10, P=0.05), and either intraoperative (HR 1.59, 95% CI 1.12 to 2.26, P<0.001) or postoperative (HR 2.52, 95% CI 1.82 to 3.49, P<0.001) ECMO insertion were associated with increased hazards for mortality. Usage of an Impella device, either intraoperatively (HR 0.54, 95% CI 0.23 to 1.32, P=0.18) or postoperatively (HR 0.90, 95% CI 0.37 to 2.17, P=0.81) was not found to be associated with mortality. There were too few patients with durable ventricular assist device insertion to model.
When adjusted for other significant baseline risk factors, postoperative ECMO cannulation was associated with a five-fold increased hazards for mortality in the final multivariable model (HR 5.12, 95% CI 2.04 to 12.85, P<0.001) (Table 5 ). Intraoperative ECMO cannulation did not reach statistical significance for mortality hazard (HR 2.47, 95% CI 0.96 to 6.33, P=0.06). Other factors associated with increased hazards for mortality include increasing age (per year, HR 1.04, 95% CI 1.01 to 1.07, P=0.01), presence of peripheral vascular disease (HR 3.55, 95% CI 1.93 to 6.52, P<0.001), and emergent operative status (HR 5.90, 95% CI 1.89 to 18.44, P<0.001). After risk adjustment, bridging with either Impella or IABP were not found to be associated with mortality, and were removed from the final model.