A Clinical Risk Scoring Tool to Predict Readmission After Cardiac Surgery: An Ontario Administrative and Clinical Population Database Study.

Background: Reducing readmission after cardiac surgery remains a quality improvement priority yet most readmission risk models examine only coronary artery bypass grafting (CABG). Our objective was to develop a predictive risk score for readmission after discharge in cardiac surgery.

Methods: All adults > 18 years undergoing isolated CABG, isolated/multiple valve, or combined CABG/valve surgery from 2008 to 2016 in Ontario were eligible. Risk factors for 30-day readmission after discharge were obtained through linkages of the CorHealth Ontario Cardiac Registry to other administrative health databases. Hazard ratios (HR) for risk factors were calculated using Cox proportional hazards regression with 95% confidence intervals (95% CI). We developed a clinical risk scoring tool weighted by beta coefficients from the final model. Discrimination and calibration was performed using c-statistics and comparing the predicted with observed probabilities across deciles of predicted risk.

Results: A total of 63,336 patients underwent CABG and/or valve surgery from 2008 to 2016. The 30-day readmission rate was 11.5% overall. Patients who were readmitted were older with higher incidences of cardiac comorbidities compared with nonreadmitted patients. Significant risk factors for readmission from the final model were prolonged length of stay (HR: 1.45; 95% CI: 1.57, 1.86; P < 0.0001), isolated valve surgery (HR: 1.35; 95% CI: 1.26, 1.44; P < 0.0001), in-hospital complications of sepsis (HR: 1.47; 95% CI: 1.05, 2.07; P = 0.024), and acute myocardial infarction (HR: 1.36; 95% CI: 1.09, 1.71; P = 0.007). A clinical risk scoring tool with 22 variables was derived that delineated patients into 1 of 5 risk quintiles. The c-statistic for the overall model was 0.63.

Conclusions: Readmission after cardiac surgery is common and moderately predictable in this contemporary cohort.