Ability of Short-Term Risk Mortality Scores to Predict Long-Term Survival in Patients Undergoing Cardiac Surgery - The 65th Annual Conference of the Israel Heart Society & The Israel Society of Cardiothoracic Surgery, 24-25 April 2018

Eyal Ben-David Medical School, St George's Hospital, London, UK Orit Blumenfeld Centre For Disease Control, Israeli Ministry of Health, Ramat Gan, Israel Ayelet Shapira-Daniels Cardiothoracic Surgery, Hadassah Hebrew University Medical Centre, Jerusalem, Israel Arturo Carranza Hernandez Cardiothoracic Surgery, Hadassah Hebrew University Medical Centre, Jerusalem, Israel Amit Korach Cardiothoracic Surgery, Hadassah Hebrew University Medical Centre, Jerusalem, Israel Ehud Rudis Cardiothoracic Surgery, Hadassah Hebrew University Medical Centre, Jerusalem, Israel Uzi Izhar Cardiothoracic Surgery, Hadassah Hebrew University Medical Centre, Jerusalem, Israel Oz M. Shapira Cardiothoracic Surgery, Hadassah Hebrew University Medical Centre, Jerusalem, Israel

Objective

The Society of Thoracic Surgeons Predicted Risk of Mortality score (STS PROM), EuroSCOREI (ES1) and EuroSCOREII (ES2) reliably predict 30-day mortality after cardiac surgery. Data on the ability of these short-term indices to accurately predict long-term survival are sparse. We sought to assess the ability of STS PROM, ES1 and ES2 to predict long-term survival.

Methods

Demographic, clinical and procedural data on 1279 consecutive patients undergoing any of 5 cardiac operations with a calculable STS PROM were prospectively collected and used to calculate STS PROM, ES1 and ES2. Patients were stratified into 5 risk-score categories based on their scores (1: 0-0.99%, 2:1.0-1.99%, 3: 2.0-2.99%, 4: 3.0-4.99%, 5: >5%). Long-term survival data were obtained from the National Death Registry. Kaplan-Meier (KM) survival curves were plotted for each risk-score category and compared using Log-rank test. Model discrimination was assessed using the Area Under the Receiver Operating Curve (AUC). We used Cox regression analysis to identify independent predictors of long-term survival.

Results

Mean follow up was 62±28 months (range 1-107 months). Mean survival of the entire cohort was 95±1 months. Long-term survival was inversely related to the STS-PROM, ES1 and ES2 (Panels A, B and C). STS PROM and ES2 predicted long-term survival more reliably than ES1. Eight-year survival for the 5 STS risk-score categories was 96±1%, 81±3%, 78±4%, 70±5%, 57±4%, respectively (p<0.0001). STS PROM model discrimination (0.76±0.02) exceeded that of the ES1 (0.69±0.02) and ES2 (0.72±0.02) (Panel D). Cox regression identified age, incidence of surgery, diabetes mellitus, dialysis and STS PROM to be independent predictors of long-term mortality.

Conclusion

The STS PROM is a more reliable predictor of long-term survival in patients undergoing cardiac surgery compared to the ES1 and ES2. Mortality scores should be used in the discussion of long-term outcomes with patients and in the analysis of procedural cost-effectiveness.

Figure 1 - Kaplan Meier Curves and AUC for STS PROM, ES1 and ES2