Introduction: Plasma hemoglobin (Hb) is measured for assessment of in vivo and in vitro hemolysis. The objective of the present investigation was to conduct a method comparison of five quantitative and one semi-quantitative Hb and H-index (hemolysis index) assays to evaluate their performance measuring plasma Hb in clinical specimens.
Methods: One hundred and fourteen clinical specimens previously tested for plasma Hb using a laboratory-developed spectrophotometric assay were also tested for Hb using a HemoCue Plasma/Low Hb assay (azide methemoglobin), a laboratory-modified Pointe Scientific Hb assay (cyanmethemoglobin), tested for H-index measurements using a Roche cobas c501, an Abbott Architect c8000, and a semi-quantitative (binned) H-index measurement on a Beckman AU5800. The reference result was defined as the median Hb score (median of all Hb or H-index results).
Results: The laboratory-developed spectrophotometric Hb assay and Roche H-index methods mostly closely matched the median Hb score across all data, as well as for lower range median Hb score results ≤2.0 g/L. Two-way frequency table analysis using an Hb (or H-index) cutoff of 0.5 g/L (or 0.5 H-index units) was then performed to compare methods to the median Hb score cutoff. The Beckman method had the highest accuracy at this cutoff, the Roche and Abbott methods had the highest positive predictive value (PPV), and the Beckman, HemoCue, and Pointe methods had the highest negative predictive value (NPV).
Conclusions: Plasma Hb and H-index results vary by method. Laboratories should evaluate the performance characteristics of their respective assays when considering adoption of spectrophotometric or chemical methods for plasma Hb assessment.
Keywords: accuracy; hemoglobin; hemolysis; plasma; spectrophotometry.
© 2021 John Wiley & Sons Ltd.