Physics Maths Engineering
Bias estimation for Sigma metric calculation
Şerif Ercan
Peer Reviewed
Abstract
I read with great interest the study of Keleş on the evaluation of analytical performances of clinical chemistry assays using the Six Sigma methodology (1). The author has computed Sigma metrics according to their laboratory performance as well as the manufacturer’s data in the reagent package inserts. For Sigma metric calculation according to laboratory performance, the author has estimated the precision using the internal quality control data from three months, and bias by the external quality assessment (EQA) data from twelve months. Keleş has stated that the contribution of bias values to the Six Sigma budget was less than the precision. This finding has been explained by the long-term bias evaluation. In addition, I would like to note a point for readers and the author about bias estimation.
Key Questions
What is the Sigma metric in clinical laboratories?
The Sigma metric is a statistical measure used in clinical laboratories to assess the performance of analytical processes. It combines bias (systematic error) and precision (random error) to evaluate the capability of a process to meet specified quality requirements.
Why is bias estimation important in Sigma metric calculation?
Bias estimation is crucial because it quantifies the systematic deviation of test results from the true value. Accurate bias estimation ensures reliable Sigma metrics, which are essential for maintaining the quality and accuracy of laboratory tests.
What are the arithmetic mean and quadratic mean methods for bias estimation?
The arithmetic mean method calculates the average of individual bias values, which can be misleading if biases have different signs (positive or negative). The quadratic mean (root mean square) method squares each bias value, averages them, and then takes the square root, providing a more accurate representation of overall bias by accounting for the magnitude of all deviations.
Why is the quadratic mean preferred over the arithmetic mean for bias estimation?
The quadratic mean is preferred because it eliminates the issue of positive and negative biases canceling each other out, which can occur with the arithmetic mean. By considering the magnitude of all bias values, regardless of their direction, the quadratic mean provides a more accurate and reliable estimate of the true bias in the analytical process.
