Greater confidence in quality control compliance with moving averages

The concept of a ‘moving average’ was first introduced in the financial sector to carry out technical analysis of stock prices, housing markets, gross domestic product or unemployment levels. It is now attracting increased attention from the UK’s pathology services after changes to the inspection regime. Laboratory managers want to incorporate patient data analysis to provide greater confidence in the test results being reported between quality control (QC) runs.

Running a sample of known concentration and using this to check equipment calibration remains the basis of good QC. However, it is no longer sufficient to rely on periodic reagent checks. These are usually run only three or four times every 24 hours, as to run them more frequently would be too expensive and time-consuming. Quality control is usually performed after maintenance and calibration when the system is at peak performance. In addition, ‘matrix effects’ or interference from components in the QC sample can have a considerable effect on the consistency of material QC figures.
 Using patient data analysis is an important complement to commercial QC. ISO 15189 requires a laboratory to “evaluate the results from patient samples that were examined after the last successful quality control event”. Indeed, it goes further, requiring that “statistical and non-statistical techniques for process control should be used wherever possible to continuously monitor examination system performance”. With software such as REMISOL Advance, a moving average (MA) calculation can run in the background, bringing consistent, uninterrupted monitoring for minimal additional cost.

The more advanced exponential analysis gives greater weight to more recent results and this type of MA calculation is seen as offering the greater potential to UK laboratories (Fig 1). While individual results may vary greatly, the mean for any given parameter should remain roughly constant. Using MA continuous analysis, any variations or ‘drifts’ can be indicators of problems with samples, analysers, reagents or other factors.