Monitoring binary outcomes using risk-adjusted charts: a comparative study

Area of Science:

• Biostatistics
• Health Services Research
• Medical Quality Improvement

Background:

• Monitoring binary outcomes is crucial for healthcare performance evaluation.
• Cumulative sum (CUSUM) charts, including the risk-adjusted CUSUM (RA-CUSUM), are used for this purpose, particularly for 30-day mortality after cardiac surgery.
• The RA-CUSUM chart, while effective at early change detection, suffers from a type I error probability of 1 and a high rate of false alarms.

Purpose of the Study:

• To propose and evaluate alternative sequential curtailed and risk-adjusted charts.
• To control the type I error rate in monitoring 30-day mortality following cardiac surgery.
• To compare the performance of these new methods against the RA-CUSUM chart.

Main Methods:

• Development of novel sequential curtailed and risk-adjusted charts.
• Evaluation of methodologies based on average run lengths and type I error probabilities.
• Application of the methods to monitor the performance of seven cardiac surgeons.

Main Results:

• The proposed charts demonstrate control over the type I error rate, unlike the RA-CUSUM chart.
• Analysis of average run lengths and type I error probabilities highlights the merits of the new methodologies.
• The study illustrates the practical application of these charts using real-world surgical performance data.

Conclusions:

• The developed sequential charts offer a superior alternative to the RA-CUSUM chart when controlling false alarms is critical.
• These new methods provide a more reliable approach to monitoring healthcare performance, especially in high-stakes scenarios like cardiac surgery outcomes.
• The findings support the adoption of these improved statistical tools for enhancing patient safety and quality of care.