A Simple Scoring System for Predicting In-Hospital Mortality after Heart Valve Surgery in A Developing Country

Abstract

Background: Various scoring systems predict mortality after cardiac surgery, but not many were designed for specific valvular surgery. Developing countries have different characteristics of patients and conditions in cardiac centers compare to developed countries. We aimed to develop a simple scoring system for predicting in-hospital mortality after valve surgery and further validate the scoring system. Methods: For developing the scoring system, the data was taken from the medical record of patient underwent valve surgery in 2012 - 2014, and for the validation study, it was from 2015 to 2016. The scoring system was developed using logistic regression models, then validated using calibration and discrimination analysis. Result: For developing a scoring system, we recruited 1040 patients in the study. The in-hospital mortality rate was 68 (6.5%). Eight variables were incorporated, including; functional class, hypertension, previous open-heart surgery, impaired renal function, right ventricular dysfunction, emergent operation, coronary artery bypass surgery, and tricuspid valve surgery. The mortality risk score has Hosmer Lemeshow (H-L) test p-value = 0.212; AUC = 0.813 (CI 95% = 0.758–0.867); and cut-off point of 5, predicting 14% risk of death (sensitivity 72.1%, specificity 75.3%). In the validation study, 789 subjects were recruited. The observed and predicted mortality were 8.6% and 11.9% respectively, with H-L test p-value = 0.169 and AUC 0.761 (95% CI; 0.702-0.821) Conclusion: We have developed a simple scoring system for predicting in-hospital mortality after valve surgery. The mortality risk score was well-calibrated with a moderate discrimination value in the validation study.

Keywords

• Heart valves, multiple; Cardiac, Statistics, risk analysis/modeling; Cardiac, Surgery, complications; Cardiac, Mitral valve replacement; Cardiac, Mitral valve repair; Cardiac, Aortic valve, replacement