Time series modeling for quality control in clinical chemistry

Area of Science:

• Clinical Chemistry
• Statistical Process Control
• Time-Series Analysis

Background:

• Autocorrelation in clinical chemistry quality-control (Q/C) measurements violates the statistical independence assumption of traditional Levey-Jennings control charts.
• This violation degrades the performance and reliability of standard statistical quality control methods in laboratory settings.

Purpose of the Study:

• To propose a novel statistical quality control approach that addresses the issue of autocorrelation in Q/C data.
• To enhance the accuracy and robustness of quality control procedures in clinical chemistry laboratories.

Main Methods:

• Replaced the single raw Q/C data chart with two new charts: a common cause chart and a special cause chart.
• The common cause chart utilizes a Box-Jenkins ARIMA time-series model to capture and represent persistent nonrandomness.
• The special cause chart analyzes the residuals from the ARIMA model, which are statistically independent, meeting standard control chart criteria.

Main Results:

• The proposed two-chart method effectively removes the autocorrelation issue, fulfilling the independence assumption.
• Residuals from the ARIMA model are free of persisting nonrandomness, enabling the use of standard Levey-Jennings plotting and control rules.
• A comparative analysis demonstrated improved performance of the proposed approach over current practices.

Conclusions:

• The novel two-chart statistical quality control strategy offers a robust solution for managing autocorrelation in clinical chemistry Q/C data.
• This method enhances the reliability of quality control by ensuring statistical independence of observations, leading to better process monitoring.
• Implementation of this approach can significantly improve the overall performance and diagnostic accuracy of clinical laboratory testing.