Calibration Factors for STS Risk Model Predictions: Why, How and When They Are Used.

Area of Science:

• Cardiology
• Health Services Research
• Data Science

Background:

• The Society of Thoracic Surgeons (STS) Adult Cardiac Surgery Database is a leading global registry for adult cardiac surgery outcomes.
• Accurate risk prediction is crucial for evaluating surgical performance and patient safety.

Purpose of the Study:

• To explain the methodology behind updating STS risk models.
• To clarify the rationale and process of quarterly calibration of observed-to-expected (O/E) ratios.
• To differentiate between calibrated and uncalibrated O/E ratios and guide their appropriate use.

Main Methods:

• Description of the STS risk model update process.
• Explanation of the quarterly calibration technique to maintain O/E ratios at 1.
• Comparative analysis of calibrated versus uncalibrated O/E ratios.

Main Results:

• STS risk models are periodically updated to reflect current clinical practices and outcomes.
• Quarterly calibration ensures that the aggregate predicted adverse events match observed events annually (O/E = 1).
• Understanding the differences between calibrated and uncalibrated O/E ratios is essential for accurate interpretation.

Conclusions:

• The STS registry employs a dynamic approach to risk modeling and outcome assessment.
• Regular calibration of O/E ratios maintains the predictive accuracy and relevance of the STS database.
• This tutorial provides essential guidance for interpreting and utilizing STS cardiac surgery outcome data effectively.