Building a Novel Risk Prediction Model for Use in Pre-operative Cardiac Surgery Patients

Background: Frailty is a major predictor of adverse outcomes following cardiac surgery, including postoperative complications, mortality, functional decline, and increased healthcare utilization. However, frailty measurement remains inconsistent, and existing tools may not adequately capture the multidimensional vulnerabilities of patients undergoing non-emergent cardiac surgery. Purpose: This study aimed to develop and evaluate a cardiac-specific frailty measure for adults undergoing non-emergent cardiac surgery and to examine its association with mortality, health-related quality of life, functional outcomes, and health services utilization. Methods: This secondary analysis used prospectively collected cohort data from the Alberta CVICU Frailty Database, which included 529 adults aged 50 years and older undergoing elective or urgent cardiac surgery. The original objective was to derive a traditional cumulative deficit Frailty Index using established construction criteria. Because too few variables met formal inclusion requirements, a standard Frailty Index was not pursued. Instead, an adapted multidimensional frailty measure, the CVICU Risk Prediction Model (C-RPM), was developed using expert-informed variable selection and deficit accumulation principles. Associations between C-RPM and outcomes were assessed using adjusted logistic regression, Cox proportional hazards modelling, and regression analyses for quality of life, mobility, and length of stay. Model discrimination was evaluated using receiver operating characteristic analysis and bootstrapping. Results: Higher C-RPM scores were independently associated with increased in-hospital, 6-month, 1-year, and 5-year mortality. The C-RPM also predicted worse health-related quality of life, poorer functional mobility, and longer CVICU and hospital length of stay. Discrimination for mortality outcomes was good to excellent, with area under the curve values ranging from 0.84 to 0.89. Compared with simpler frailty measures, the C-RPM provided a more comprehensive assessment of vulnerability, while Timed Up and Go testing, hand grip strength, and the Clinical Frailty Scale offered useful complementary information. Conclusions: The C-RPM represents a pragmatic, theory-informed approach to frailty measurement in cardiac surgery. Multidimensional frailty assessment may improve perioperative risk stratification, support targeted intervention, and enhance patient-centred care planning.