Longitudinal evaluation of laboratory results and method precision in worldwide erythropoietin external quality assessments

Abstract

Introduction: This study presents a longitudinal analysis of external quality assessment (EQA) results for erythropoietin (EPO) determinations conducted between 2017 and 2022 with a continuously increasing number of participating laboratories. The aim of this work was to evaluate participant performance and methodological aspects. Methods: In each of the eleven EQA surveys, a blinded sample set of lyophilized human serum containing one sample with lower EPO concentrations (L) and one with higher EPO concentrations (H) was sent to the participating laboratories. Results: A total of 1,256 measurements were included. The median (interquartile range) fraction of participants not meeting the criteria of acceptance set at 20% around the robust mean of the respective survey was 9.5% (6.1%-10.7%) (sample L) and 9.1% (5.8%-11.8%) (sample H) but lacked a clear trend in the observed period. Some surveys exhibited unusually high interlaboratory variation, suggesting interfering components in the EQA samples. Different immunological methods and reagent manufacturers also showed variability in measurement outcomes to some extent. Conclusion: These findings highlight the need for continuous quality assessment in EPO measurements to ensure patient safety and identify areas for further research and investigation.

Keywords: erythropoietin; external quality assessment; immunoassay; method precision; proficiency testing.

FIGURE 1

General outcome/information of the INSTAND EPO EQA from 2017 to 2022. (A) Number of laboratories participating between 2017 and 2022 (green) and a corresponding trend line (blue) starting with one survey (S) in 2017 (2017-S1) and continuing with two runs per year (S1/S2) until 2022. (B) The percentage of measurements outside the criteria of acceptance (CoA; %) calculated for each survey for sample L (red) and sample H (turquoise). The CoA was defined as ± 20% around the robust mean for the individual surveys shown. (C) Mean absolute deviation (MAD)/median ratio (%) for sample L (red) and sample H (turquoise) for every survey.

FIGURE 2

Method-dependent analysis of EQA results for EPO levels from 2017 to 2022 (A) Distribution of the EPO measurement results (IU/L) for the individual methods CLIA (red), ELISA (green), LEIA (turquoise), and “other” (violet) in relation to the overall distribution of all measured values in the individual surveys (black) for sample L from 2017 to 2022. In this plot, whiskers span 1.5 times the IQR above and below the box, capturing the middle 50% of the data. The red, green, turquoise, violet and black dots mark outliers, which are defined as observations that exceed 1.5 times the IQR from either edge of the box. (B) The same consideration used for (A) but for sample H. (C) Percentage of the frequencies for the respective measurement methods of the total of all measurements per survey per sample.

FIGURE 3

Manufacturer-dependent analysis of EQA results for EPO levels from 2017 to 2022. (A) Distribution of the EPO measurement results (IU/L) for the individual reagent manufacturers BE (red), DG (green), IB (turquoise), and “other” (violet) in relation to the overall distribution of all measured values in the individual surveys (black) for sample L from 2017 to 2022. In this plot, whiskers span 1.5 times the IQR above and below the box, capturing the middle 50% of the data. The red, green, turquoise, violet and black dots mark outliers, which are defined as observations that exceed 1.5 times the IQR from either edge of the box. (B) The same consideration used in (A) but for sample H. (C) Percentage of the frequencies for the respective manufacturers of the total of all measurements per survey per sample.