Mortality control charts for comparing performance of surgical units: validation study using hospital mortality data

Area of Science:

• Biostatistics
• Surgical Quality Improvement
• Health Services Research

Background:

• Evaluating surgical unit performance is crucial for patient safety and quality improvement.
• Traditional methods often fail to account for variations in patient case mix and surgical volume.
• Objective performance metrics are needed to identify areas for improvement in surgical care.

Purpose of the Study:

• To develop and validate a statistical methodology for assessing surgical unit performance.
• The method aims to adjust for case volume and case mix in evaluating operative mortality.
• To create a reliable tool for monitoring surgical performance and identifying outliers.

Main Methods:

• A validation study utilized routinely collected in-hospital mortality data from two UK databases (ASCOT and RISC).
• A two-level hierarchical logistic regression model was employed to adjust operative mortality for patient case mix.
• Risk-adjusted mortality was visualized on control charts, with performance outliers identified using 90%, 95%, and 99% confidence intervals.

Main Results:

• The study included 1042 patients undergoing gastro-oesophageal cancer surgery across 29 hospitals.
• Mean in-hospital mortality was 12%, with unit case volumes ranging from 1 to 55 cases annually.
• Risk-adjusted analysis identified fewer units outside control limits compared to crude figures, indicating the method's sensitivity to case mix.

Conclusions:

• Mortality control charts offer an accurate, risk-adjusted approach to identifying surgical units with significantly divergent operative mortality.
• This graphical method serves as an effective early warning system for substandard surgical performance.
• The methodology holds potential for adaptation across various surgical specialties to monitor performance.