High Risk Coronary Artery Bypass: Impact of Heart Failure Symptoms Versus Left Ventricular Function
Robert B Hawkins1, J Hunter Mehaffey1, Sarah A Schubert1, Clifford Fonner2, Jeffrey B Rich2, Alan M Speir3, Mohammed Quader4, Gorav Ailawadi1, Irving L Kron1
1University of Virginia, Charlottesville, VA;2Virginia Cardiac Services Quality Initiative, Falls Church, VA;3Inova Heart and Vascular Institute, Falls Church, VA;4Virginia Commonwealth University, Richmond, VA
Objectives: The evidence comparing patients with heart failure and preserved versus reduced ejection fraction (EF) is inconclusive regarding risk for coronary artery bypass grafting (CABG). Studies have previously defined ventricular dysfunction primarily as an EF < 50%. We hypothesize that a clinically derived EF threshold can be utilized to better characterize the interaction between heart failure symptoms and EF.
Methods: A total of 41,744 patients underwent isolated CABG within a regional consortium from 2005-2016 with both EF and heart failure status available. Hierarchic logistic regression was used to identify the lowest threshold for sufficiently preserved EF, over which EF was not a significant predictor of mortality. Patients were classified by heart failure status and reduced versus preserved EF based on this new EF threshold. Hierarchic logistic regression models identified independent predictors of mortality while adjusting for operative risk, clustering at the hospital level, and changing practices over time.
Results: The threshold identified for preserved EF was 35%, over which EF was not a significant predictor of mortality (OR 0.99, p=0.250). Of the cohort with preserved EF, 32,523 (77.9%) had no recent heart failure (NoHFpEF) while 3,336 (8.0%) had recent heart failure (HFpEF). Of patients with a reduced EF, 3,131 (7.5%) had no recent heart failure (NoHFrEF) while 2,754 (6.6%) had recent heart failure (HFrEF). The unadjusted mortality rate for each group was 1.4% (NoHFpEF), 4.1% (HFpEF), 3.5% (NoHFrEF), and 6.1% (HFrEF); the risk adjusted odds of mortality are provided in Table 1. The table also demonstrates that current risk models underestimate higher risk heart failure patients. After risk adjustment, the EF value in patients with a preserved EF adds no predictive clinical value (OR=0.99, p=0.226), while heart failure symptoms independently confer a 1.65 fold increase in odds of mortality (p<0.0001). For patients with a reduced EF, ejection fraction is independently associated with mortality (OR=0.97, p=0.014), and heart failure symptoms trend towards significance (OR=1.29, p=0.093).
Conclusions: Risk assessment suggests the current EF threshold of 50% is too high for surgical patients, and instead a sufficiently preserved ejection fraction is greater than 35%. The elevated risk of mortality associated with heart failure symptoms despite preserved ejection fraction represents a possible area of improvement with preoperative management. This analysis provides an updated understanding of clinically useful thresholds for the expanding heart failure surgical population.
Table 1. Risk-adjusted odds of operative mortality compared to no heart failure with preserved ejection fraction.
|STS Operative Mortality||OR||CI||p value||O:E|
|No heart failure with preserved EF (baseline)||1.0||0.94|
|Heart Failure with preserved EF||1.71||(1.37-2.14)||<0.0001||1.14|
|No Heart Failure with reduced EF||1.80||(1.41-2.29)||<0.0001||1.11|
|Heart Failure with reduced EF||2.35||(1.88-2.94)||<0.0001||1.13|
Back to 2017 Program