Patient-based real-time quality control in medical laboratories: On the design and robustness of the moving average control chart with truncation limits


Murat U., TESTİK M. C., PINAR A.

Quality and Reliability Engineering International, 2024 (SCI-Expanded) identifier

  • Publication Type: Article / Article
  • Publication Date: 2024
  • Doi Number: 10.1002/qre.3502
  • Journal Name: Quality and Reliability Engineering International
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Academic Search Premier, Aerospace Database, Communication Abstracts, Compendex, INSPEC, Metadex, Civil Engineering Abstracts
  • Keywords: internal quality control, medical laboratories, moving average control chart, patient-based real-time quality control, robustness to non-normality, truncation limits
  • Hacettepe University Affiliated: Yes

Abstract

Quality is an indispensable requirement in medical laboratories, where controls are performed at different phases of the laboratory processes. In the following, we consider the internal quality control activities during the analytical phase of medical laboratories for verifying the test results prior to their release. To monitor the measurement process for errors in the test results, one approach that is considerably attracting the health professionals is patient-based real-time quality control. In this respect, real-time patient test results are monitored, commonly by implementing the Moving Average (MA) control chart with truncation limits. A usual practice is the use of truncation limits to exclude abnormal patient test results from the MA calculations. However, an open issue in the literature is how to determine the width of the truncation limits and the window size in the MA calculations. Furthermore, the distributions of the test results are often nonnormal and difficult to model by personnel having insufficient training in statistics. Consequently, robustness of the performance of the MA control chart to non-normal distributions of test results is another open topic in the literature. In this study, a detailed robustness analysis is performed by considering various distributions having different shapes. Performance of several MA control chart designs constructed by altering the truncation limits and window sizes are investigated. Based on our simulation results with selected distributions, we conclude that the truncation limits should not be used and a window size of 20 is recommended. Under such a setting, the MA control chart performs similar to its intended performance in terms of false alarms and in detecting systematic errors, hence being robust to non-normality of the test results.