Validation of current mortality risk prediction models in patients with cirrhosis undergoing cardiac surgery

Background: Liver cirrhosis is a known risk factor for mortality after cardiac surgery. In this study, we investigated the correlation between existing prediction models and actual mortality rates after cardiac surgery. Methods: Between July 2014 and December 2022, 1257 patients with a history of chronic liver disease underwent cardiac surgery, of whom 220 had a diagnosis of cirrhosis. Preoperative comorbidities, intraoperative details, and postoperative outcomes were analyzed. The Mayo Clinic and VOCAL-Penn scores were applied to the 220 patients and compared to observed mortality. Results: Among the 220 patients with cirrhosis, the preoperative median MELD score was 10. According to the Mayo Clinic score, the mean predicted mortality rate was 20.9% at 30 days and 29.6% at 90 days. In contrast, according to the VOCAL-Penn score, the mean predicted mortality rate was 9.0% at 30 days and 14.0% at 90 days, whereas the observed postoperative mortality of the 220 patients was 4.1% at 30 days and 12.6% at 90 days. VOCAL-Penn showed good calibration but still overestimated mortality by 3% at 30 days and by 1.3% at 90 days. The Mayo Clinic score demonstrated weak calibration, overestimating mortality by 15% at 30 days and by 17% at 90 days. Conclusions: In this single-center analysis, the Mayo Clinic and VOCAL-Penn scores overestimated the postoperative mortality in patients with cirrhosis undergoing cardiac surgery, although VOCAL-Penn provided a more accurate risk prediction. Ultimately, this heterogeneous group of patients often presents with a wide range of comorbidities, which complicate risk stratification; however, patients should not be denied surgery solely on the basis of current risk prediction models.