Design, validation and performance of aspartate aminotransferase- and lactate dehydrogenase-reporting algorithms for haemolysed specimens including correction within quality specifications.

Background: In vitro haemolysis is a major operational challenge for medical laboratories. A new experimental design was used to investigate under what conditions algorithms could be designed to report either quantitative or qualitative aspartate aminotransferase and lactate dehydrogenase results outside the manufacturer's haemolysis specifications. Quantitative corrections were required to meet prespecified quality specifications.

Methods: Twenty-five patient samples were used to design reporting algorithms and another 41 patient samples were used to validate the algorithms. Aspartate aminotransferase, lactate dehydrogenase and haemolysis index were determined using a Cobas 6000 analyser (Roche diagnostics, Mannheim, Germany). Correction factors were determined, and the accuracy of the correction was investigated. Reporting algorithms were designed based on (i) the manufacturer's cut-off for the haemolysis index, (ii) corrections within the total allowable error specification and (iii) qualitative reporting based on obtained results. The impact of the reporting algorithms was retrospectively determined by recalculating six months of aspartate aminotransferase and lactate dehydrogenase results.

Results: No correction for aspartate aminotransferase/lactate dehydrogenase was possible for results below the upper reference interval limit, while results equal to or greater than the upper reference interval limit could, up to mild haemolysis, be corrected within the total error criterion. All samples generated from the validated patient cohort fulfilled the set criteria. The algorithms allowed reporting 88.5% and 85.9% of otherwise unreported aspartate aminotransferase and lactate dehydrogenase results, respectively.

Conclusions: An approach is presented that allows to generate and validate reporting algorithms for aspartate aminotransferase and lactate dehydrogenase compatible with prespecified quality specifications. The designed algorithms resulted in a significant reduction of otherwise unreported aspartate aminotransferase and lactate dehydrogenase results.