Predicting hospital mortality in cardiac care unit patients: external validation of the M-CARS

Abstract

Abstract Background Resource availability for cardiac patients particularly specialized intensive care unit is often limited in Asian developing countries. Risk score specifically developed for cardiac patients admitted to cardiac care unit (CCU) are scarce or not widely validated. The Mayo Cardiac Intensive Care Unit Admission Risk score (M-CARS) has shown excellent performance in predicting in-hospital mortality in US population. However, the M-CARS has not yet been validated in the Asian population where characteristics of cardiac patients may be substantially different. In addition, the value of the M-CARS in predicting intermediate-term mortality is unknown. Purpose We sought to validate the M-CARS for in-hospital and 1-year all-cause mortality in consecutive patients admitted to the cardiac care unit. Methods Consecutive patients admitted to the CCU of a tertiary care center between July 2015 and December 2019 were included into the present study. Patients who were transferred from the other in-hospital wards were excluded from the study. Missing data were handled with multiple imputation using multivariable imputation by chained equations. Seven variables including cardiac arrest, cardiogenic shock, respiratory failure, Braden skin score, BUN, anion gap and RDW were used to calculate the M-CARS in the imputed data. Patients were stratified into 3 groups according to the M-CARS (<2, 2–6, >6). In-hospital mortality was compared among groups with different M-CARS. Kaplan-Meier curves were used to demonstrate 1-year all-cause mortality among groups. Discrimination was assessed with C-statistic. Results Of 1709 patients included in the present study, 72% was male with mean age of 64.3 years. Prevalence of diabetes, chronic kidney disease, cardiogenic shock and respiratory failure prior to or within 1 hour of CCU admission were 22.4%, 13.6%, 8.1% and 1.8% respectively. Majority of patients (68.2%) had ACS as principal diagnosis. Ninety-three patients died during the index hospital admission (mortality rate of 5.44%). The in-hospital mortality rate in patients with M-CARS of <2, 2–6 and ≥6 was 1%, 8.7% and 21.4% respectively. The Kaplan-Meier estimate of 1-year mortality in the group with M-CARS ≥6 was 54.8% while it was 6.3% in those with M-CARS <2. C-statistic of M-CARS for in-hospital mortality was 0.865 (95% CI 0.825–0.899) whereas it was slightly lower for 1-year all-cause mortality (c-statistic 0.840, 95% CI 0.801–0.874). Calibration for in-hospital mortality was satisfactory in the group with M-CARS ≤6 whereas actual mortality was underestimated in M-CARS >6. Conclusions The M-CARS had excellent and clearly useful discrimination with satisfactory calibration in our external validation study. M-CARS identified high-risk patients in CCU (M-CARS ≥6) who had very high risk for mortality. M-CARS may be helpful for physicians in risk stratifying and allocating health-care resource for cardiac patients. Funding Acknowledgement Type of funding sources: None.