Related Experiment Videos
Monitoring paired binary surgical outcomes using cumulative sum charts
S H Steiner1, R J Cook, V T Farewell
1Department of Statistics and Actuarial Sciences, University of Waterloo, Ontario, Canada. shsteine@setosa.uwaterloo.ca
Statistics in Medicine
|February 17, 1999
Summary
New bivariate cumulative sum (CUSUM) charts effectively monitor failure rates in correlated data. These charts, utilizing a Markov chain model, efficiently detect gradual changes in undesirable outcomes, enhancing medical research and quality control.
Area of Science:
- Statistics
- Quality Control
- Medical Research
Background:
- Correlated binary data are common in medical research, system reliability, and quality control.
- Monitoring failure rates in these scenarios requires robust statistical methods.
Purpose of the Study:
- To propose simultaneous bivariate cumulative sum (CUSUM) charts with secondary control limits for monitoring failure rates.
- To evaluate the run length properties and effectiveness of these charts for detecting changes in failure rates.
Main Methods:
- Utilized a Markov chain model to determine the run length properties of the proposed monitoring scheme.
- Developed bivariate CUSUM charts incorporating secondary control limits.
- Applied the procedure to bivariate outcome data from paediatric surgeries.
Main Results:
- The proposed CUSUM charts are easy to implement.
- Demonstrated high effectiveness in detecting small, gradual changes in the rate of undesirable outcomes.
- The methodology proved effective for bivariate outcome data.
Conclusions:
- The simultaneous bivariate CUSUM charts offer an effective and practical approach for monitoring failure rates in correlated binary data.
- This method is particularly adept at identifying subtle, progressive shifts in adverse event rates.
- The approach is adaptable to various data types, including multivariate normal, binomial, and Poisson responses.