A contemporary risk model for predicting 30-day mortality following percutaneous coronary intervention in England and...

Area of Science:

• Interventional Cardiology
• Health Outcomes Research
• Clinical Risk Prediction

Background:

• Existing UK percutaneous coronary intervention (PCI) risk models are outdated, reflecting an earlier era of interventional cardiology.
• A contemporary cohort is needed to develop a more accurate risk prediction tool for current PCI procedures.

Purpose of the Study:

• To develop and validate a new risk prediction model for 30-day mortality following percutaneous coronary intervention (PCI).
• To ensure the model provides good discrimination and calibration across a wide spectrum of patient risk.
• To create a tool applicable to modern interventional cardiology practices.

Main Methods:

• A large training dataset of 336,433 PCI cases (2007-2011) from England and Wales was used.
• 30-day mortality was determined via record linkage; variables were selected based on clinical consensus and data quality.
• Temporal validation was performed using 2012 PCI procedure data.

Main Results:

• The strongest predictors of 30-day mortality included cardiogenic shock, dialysis, and the urgency of the PCI procedure.
• The model achieved an area under the receiver operator characteristic curve of 0.85 (training) and 0.86 (validation).
• Calibration plots demonstrated a good model fit, maintained during validation.

Conclusions:

• A contemporary, validated risk model for PCI has been developed, suitable for diverse clinical scenarios from elective to emergency procedures.
• The model exhibits high discrimination and excellent calibration across the risk spectrum.
• This risk-adjusted model can enhance the examination of key outcomes in PCI audits.