Performance evaluation of four scoring systems for mortality prediction in a contemporary cardiac intensive care unit

There are limited data addressing the scoring systems for clinical outcome prediction in patients admitted to a contemporary cardiac intensive care unit (CICU). We aimed to compare the Acute Physiology and Chronic Health Evaluation (APACHE), Sequential Organ Failure Assessment (SOFA), Mayo CICU Admission Risk Score (MCARS), and Clinical Frailty Scale (CFS) scores, in predicting 90-day mortality in an unselected CICU population.

This is a single-center prospective, cohort study that included all patients admitted to the CICU of a tertiary care center during two consecutive years. The performances of the risk scores in predicting mortality were compared with respect to discriminative ability, calibration, accuracy and net reclassification improvement (NRI). Multivariate Cox regression analysis was also performed.

A total of 1449 consecutive patients were included. The median (IQR) age was 73 (62-82) years, 970 (67 %) were males. The CICU, hospital and 90-day mortality rates were 11 %, 14 % and 17 %, respectively. Compared to APACHE II and SOFA scores, MCARS demonstrated superior calibration (Hosmer-Lemeshow test P = 0.661), accuracy (Brier score = 0.069) and overall performance (area under curve [AUC] = 0.905), as well as the NRI (0.76). It was also independently associated with 90-day mortality after adjustment for several factors (HR = 1.578, 95 % CI: 1.478-1.683, P < 0.001). CFS was also independently associated with 90-day mortality (HR = 1.153, 95 % CI: 1.123-1.185, P < 0.001). However, its goodness-of-fit, overall performance and reclassification improvement were inferior to MCARS. Additionally, MCARS was a reliable predictor for CICU mortality (AUC = 0.935).

MCARS outperforms standard risk scores for prediction of 90-day mortality in patients admitted to a contemporary CICU.