Huub H van Rossum1, Hans Kemperman2. 1. Department of Clinical Chemistry and Haematology, University Medical Center Utrecht, Utrecht, The Netherlands; Department of Clinical Chemistry, The Netherlands Cancer Institute - Antoni van Leeuwenhoek Hospital, Amsterdam, The Netherlands. Electronic address: h.v.rossum@nki.nl. 2. Department of Clinical Chemistry and Haematology, University Medical Center Utrecht, Utrecht, The Netherlands.
Abstract
BACKGROUND: A moving average (MA) optimization and validation method, based on a realistic simulation of MA bias detection using reported consecutive results, is demonstrated for creatinine. METHODS: We aimed for reproducible and fast bias detection between scheduled internal quality control measurements and a manageable number of false MA alarms. For this, multiple MA procedures were investigated for their bias detection properties by power function analysis and simulation of the number of results needed for MA bias detection. An optimal MA procedure was chosen based on the range of bias detectable within scheduled QC measures. RESULTS: Power function analysis showed that increasing batch sizes and more stringent truncation limits always improved MA bias detection probabilities per MA result. However, these variables could significantly delay MA bias detection over time. The selected optimal MA procedure resulted in a reproducible creatinine bias detection of a 20% bias in 23-145 creatinine results and a 40% bias in 22-60 creatinine results. CONCLUSIONS: Our method of simulating MA bias detection gives a more realistic estimate of the MA bias detection properties when compared to power function analysis and is therefore useful for implementation of MA as a continuous analytical quality control instrument.
BACKGROUND: A moving average (MA) optimization and validation method, based on a realistic simulation of MA bias detection using reported consecutive results, is demonstrated for creatinine. METHODS: We aimed for reproducible and fast bias detection between scheduled internal quality control measurements and a manageable number of false MA alarms. For this, multiple MA procedures were investigated for their bias detection properties by power function analysis and simulation of the number of results needed for MA bias detection. An optimal MA procedure was chosen based on the range of bias detectable within scheduled QC measures. RESULTS: Power function analysis showed that increasing batch sizes and more stringent truncation limits always improved MA bias detection probabilities per MA result. However, these variables could significantly delay MA bias detection over time. The selected optimal MA procedure resulted in a reproducible creatinine bias detection of a 20% bias in 23-145 creatinine results and a 40% bias in 22-60 creatinine results. CONCLUSIONS: Our method of simulating MA bias detection gives a more realistic estimate of the MA bias detection properties when compared to power function analysis and is therefore useful for implementation of MA as a continuous analytical quality control instrument.