Kidney Week

Abstract: FR-PO420

External Validation of Models to Predict Risk of Major Cardiovascular Events and Death for People with Kidney Failure Having Noncardiac Surgery

Session Information

• Hemodialysis Epidemiology and Outcomes
  October 25, 2024 | Location: Exhibit Hall, Convention Center
  Abstract Time: 10:00 AM - 12:00 PM

Category: Dialysis

• 801 Dialysis: Hemodialysis and Frequent Dialysis

Authors

• Pabla, Gurpreet S., University of Manitoba Faculty of Health Sciences, Winnipeg, Manitoba, Canada

Gurpreet S. Pabla, MS, MSc

• Harrison, Tyrone, University of Calgary Cumming School of Medicine, Calgary, Alberta, Canada

Tyrone Harrison, MD, PhD

• Tangri, Navdeep, University of Manitoba Max Rady College of Medicine, Winnipeg, Manitoba, Canada

Navdeep Tangri, MD, PhD

• Whitlock, Reid, Seven Oaks General Hospital, Winnipeg, Manitoba, Canada

Reid Whitlock, MSc

• Ferguson, Thomas W., Seven Oaks General Hospital, Winnipeg, Manitoba, Canada

Thomas W. Ferguson, MS, MSc

Background

Patients with kidney failure undergoing non-cardiac surgery have a significantly higher risk of adverse cardiovascular events and mortality compared to people with normal kidney function. Existing tools for perioperative risk stratification are not valid for patients with kidney failure. Recently, three risk prediction models were developed from a population-based cohort of people with kidney failure in Alberta, Canada. We externally validated the established Alberta models for major cardiovascular events and mortality in patients with kidney failure within 30 days of non-cardiac surgery in Manitoba, Canada.

Methods

Data was sourced from the Manitoba Centre for Health Policy. The cohort included adults (≥ 18 years) with pre-existing kidney failure (estimated glomerular filtration rate < 15 mL/min/1.73m² or on maintenance dialysis) undergoing non-cardiac surgery procedures between April 1, 2007, and December 31, 2019. The primary outcome of this study was a composite of acute myocardial infarction, cardiac arrest, ventricular arrhythmia, and all-cause mortality. The models performance was evaluated using C-statistics, Brier scores, and calibration on Manitoba data using two approaches:
1. Model deployment: Used coefficients from the Alberta models to predict outcomes on Manitoba data.
2. Model refitting: Re-estimated model coefficients using logistic regression on Manitoba data while maintaining the same variables as the Alberta models.

Results

We identified 12,082 surgeries and 569 outcomes (5%). All three models performed well with both approaches, with C-statistics ranging from 0.82 for model 1 to 0.87 for model 3 in the first approach. The calibration slopes for models 1, 2, and 3 were 1.3, 1.4, and 1.2, respectively. Once refit, discrimination remained strong with C-statistics ranging from 0.83 (model 1) to 0.86 (model 3). Calibration slopes were 1 across all the models. Brier scores were consistently low at 0.04 for all the models in both approaches.

Conclusion

Our external validation study confirms the original Alberta models' robustness in a geographically distinct Canadian population. Future research should test the impact of these models in clinical care.

Funding

• Government Support – Non-U.S.