Simplified model for end-stage liver disease score predicts mortality for tricuspid valve surgery.

ABSTRACT

OBJECTIVES: The model for end-stage liver disease score (MELD = 3.8*LN[total bilirubin] + 9.6*LN[creatinine] + 11.2*[PT-INR] + 6.4) predicts mortality for tricuspid valve surgery. However, the MELD is problematic in patients undergoing warfarin therapy, as warfarin affects the international normalized ratio (INR). This study aimed to determine whether a simplified MELD score that does not require the INR for calculation could predict mortality for patients undergoing tricuspid valve surgery. Simplified MELD score = 3.8*LN[total bilirubin] + 9.6*LN[creatinine] + 6.4. METHODS: A total of 172 patients (male 66, female 106; mean age, 63.8 ± 10.3 years) who underwent tricuspid replacement (n = 18) or repair (n = 154) from January 1991 to July 2011 at a single centre were included. Of them, 168 patients in whom the simplified MELD score could be calculated were retrospectively analysed. The relationship between in-hospital mortality and perioperative variables was assessed by univariate and multivariate analysis. RESULTS: The rate of in-hospital mortality was 6.4%. The mean admission simplified MELD score for the patients who died was significantly higher than for those surviving beyond discharge (11.3 ± 4.1 vs 5.8 ± 4.0; P = 0.001). By multivariate analysis, independent risk factors for in-hospital mortality included higher simplified MELD score (P = 0.001) and tricuspid valve replacement (P = 0.023). In-hospital mortality and morbidity increased along with increasing simplified MELD score. Scores <0, 0-6.9, 7-13.9 and >14 were associated with mortalities of 0, 2.0, 8.3 and 66.7%, respectively. The incidence of serious complications (multiple organ failure, P = 0.005; prolonged ventilation, P = 0.01; need for haemodialysis; P = 0.002) was also significantly higher in patients with simplified MELD score ≥ 7. CONCLUSIONS: The simplified MELD score predicts mortality in patients undergoing tricuspid valve surgery. This model requires only total bilirubin and creatinine and is therefore applicable in patients undergoing warfarin therapy.