The number of replicates, and pooling versus individual measurements for analytical imprecision calculations: Does it matter?

Abstract

BACKGROUND: Recommendations from the American Society of Veterinary Clinical Pathology (ASVCP) are to calculate the between-run coefficient of variation (CV) based on measuring one replicate per day on quality control materials (QCMs) or pooled patient samples over a minimum of 20 days. However, this recommendation is not always followed by researchers. OBJECTIVES: We aimed to determine if a reduction in the number of replicates using QCM or individual or pooled samples would provide CV results similar to those obtained based on ASVCP recommendations. METHODS: CVs were calculated for three measurands, namely urea, creatinine, and C-reactive protein based on the analytic results of the following groups: (a) QCM measured once daily for 20 days (considered as the reference for comparison), b) QCM measured once daily for 5 days, (c) five different canine serum samples measured once daily for 5 days, and (d) a pooled canine serum measured once daily for 5 days. CVs were calculated for two different measurand concentrations. RESULTS: Compared with the reference method, significantly different CVs were obtained with all methods except for when the QCM was measured once daily for 5 days. The use of the five different individual samples also provided significantly different CVs compared with the use of a pooled sample. CONCLUSIONS: The results indicate that different protocols for determining between-run imprecision calculations can give different results compared with the reference procedure and that this should be taken into consideration when evaluating the total error associated with a test.