Skip to content
Licensed Unlicensed Requires Authentication Published by De Gruyter June 24, 2009

Multicenter evaluation of the hemolysis index in automated clinical chemistry systems

Giuseppe Lippi, Gian Luca Salvagno, Norbert Blanckaert, Davide Giavarina, Sol Green, Steve Kitchen, Vladimir Palicka, Anne J. Vassault and Mario Plebani


Background: In vitro hemolysis, the prevailing cause of preanalytical error in routine laboratory diagnostics, might influence the reliability of several tests, affect the quality of the total testing process and jeopardize patient safety. Although laboratory instrumentation is now routinely equipped with systems capable of automatically testing and eventually correcting for hemolysis interference, to our knowledge there are no reports that have compared the efficiency of different analytical platforms for identifying and classifying specimens with hemolysis.

Methods: Serum from a healthy volunteer was spiked with varying amounts of hemolyzed blood from the same volunteer, providing a serum free hemoglobin concentration ranging from 0.0 g/L to 2.0 g/L as measured by the reference cyanmethemoglobin assay. The spiked serum samples were shipped to seven separate laboratories and the hemolysis index (HI) was tested in triplicate on the following analytical platforms: Roche Modular System P (n=4) and Integra 400 Plus (n=1), Siemens Dimension RxL (n=3), ADVIA 2400 (n=1) and ADVIA 1800 (n=1), Olympus AU 680 (n=1) and Coulter DXC 800 (n=1).

Results: Satisfactory agreement of HI results was observed among the various analytical platforms, despite a trend toward overestimation by the ADVIA 2400 and 1800. After normalizing results according to the instrument-specific alert value, discrepancies were considerably reduced. All instruments except for the Dimension RxL gave values normalized to the instrument-specific alert value, <1.0 for the sample with 0.048 g/L free hemoglobin, and >1.0 for the sample with 0.075 g/L free hemoglobin. The results of the four Modular System P tests were also highly reproducible among the different facilities. When evaluating instruments that provided quantitative HI results, the mean intra-assay coefficient of variation (CV) calculated for the triplicate determinations was always between 0.1% and 2.7%.

Conclusions: The results of this multicenter evaluation confirm that efficiency of different analytical platforms to correctly identify and classify unsuitable samples is satisfactory. However, more effort should be placed on the standardization of reporting HI. All the instruments that we tested provide either quantitative or qualitative results that are essentially comparable, but which should always be compared with the instrument-specific alert values to harmonize their efficiency.

Clin Chem Lab Med 2009;47:934–9.

Corresponding author: Prof. Giuseppe Lippi, MD, Sezione di Chimica e Microscopia Clinica, Dipartimento di Scienze Morfologico-Biomediche, Università degli Studi di Verona, Ospedale Policlinico G.B. Rossi, Piazzale Scuro, 10, 37134 Verona, Italy Phone: +39-045-8124308, Fax: +39-045-8201889, ;

Received: 2009-3-24
Accepted: 2009-5-6
Published Online: 2009-06-24
Published in Print: 2009-08-01

©2009 by Walter de Gruyter Berlin New York

Scroll Up Arrow