Understanding Patient-Based Real-Time Quality Control Using Simulation Modeling.

Area of Science:

• Clinical Chemistry
• Laboratory Medicine
• Quality Control

Background:

• Patient-based real-time quality control (PBRTQC) offers an alternative to traditional quality control methods.
• PBRTQC requires optimization of parameters like algorithm, truncation, block size, and control limits for individual laboratories.

Purpose of the Study:

• To assess the effectiveness of different PBRTQC methods in early bias detection using computer simulations.
• To provide recommendations for optimizing PBRTQC parameters based on simulation results.

Main Methods:

• Computer simulations were performed using 460,000 historical patient measurements for 10 analytes.
• Biases were introduced to simulate erroneous measurements, and various PBRTQC algorithms were evaluated for their detection capabilities.
• A web application was developed to facilitate the estimation of PBRTQC performance.

Main Results:

• "Percentiles of daily extremes" proved effective for control limit calculation, though sensitive to changes in measurement distribution.
• Box-Cox transformation enhanced error detection, and Winsorization outperformed simple outlier removal.
• Parameter choices significantly impacted performance; for example, block size affected medians more than Harrell-Davis 50 percentile estimators (HD50s).

Conclusions:

• Computer simulations are valuable for optimizing PBRTQC parameters.
• Certain parameters demonstrate superior performance and can be considered as default settings for PBRTQC implementation.