450 - Effect of bilirubin assays on treatment decision in neonatal hyperbilirubinemia: past and present using Dutch external quality assessment results

Poster Number: 450
Publication Number: 450.1246

Background: Accurate measurement of bilirubin presents a significant challenge, especially in the context of managing hyperbilirubinemia in newborns. Although prevention of both over- and under-treatment requires similar results, different methods for neonatal bilirubin quantifications repeatedly show to have method- specific imprecision and bias. To this end, an external quality assessment (EQA) program is in place in the Netherlands.

Objective: To evaluate the accuracy of bilirubin methods and its evolution over a 10 year timeframe, focusing on bilirubin concentrations relevant for clinical neonatal hyperbilirubinemia.

Design/Methods: In retrospect, we analyzed prospectively collected Dutch external quality assurance (EQA) data from 2013 until 2023. Pooled adult blood serum material, spiked with unconjugated bilirubin (ranging from 100-500 µmol/L) was sent to participating laboratories across the Netherlands. Bias was deduced at the level of 200, 300 and 350 µmol/L by the regression between the sample results and the golden standard Doumas method, whereas imprecision was calculated using the residuals from regression. Statistical significance of bias was established with Kruskal-wallis and pairwise Wilcoxon-ranked tests.

Results: We analyzed 24.370 data points in 88 participating laboratories across the Netherlands, using 8 different laboratory methods. Notably, bias of laboratories differs across years and between platforms, and the degree of bias (i.e. the between-laboratory variability) also depends on the platform used. Over the years, most platforms had significant changes in performance at 200, 300 and 350 µmol /L (Kruskal-wallis test, except Siemens Advia and Atellica). From 2015 onwards, at all concentration levels, platforms differed significantly from another (Kruskal-wallis). Furthermore, we noted large within- and between laboratory variability that dependents on platforms.

Conclusion(s): Despite the well-known method-specific imprecision and bias, EQA results continue to show bias in bilirubin test outcomes over time, influencing clinical treatment decisions and underlining the stagnation in progress in bilirubin measurements. The unique 10 year timeframe allowed us to visualize improvement and deterioration of the method-specific analytical performance in which lessons can be learned. Close collaboration between laboratory specialists, clinicians, and manufacturers is required to move towards optimizing bilirubin measurements. To support future assessment, the total allowable error and clinically relevant limits of bilirubin quantification should be defined in international guidelines.