Accuracy of platelet counting haematology analysers in severe thrombocytopenia and potential impact on platelet transfusion

Abstract

SummaryAlthough haematology analysers provide reliable full blood counts, they are known to be inaccurate at enumerating platelets in severe thrombocytopenia. If the thresholds for platelet transfusion, currently set at 10 · 10 9 /l, are to be further reduced, it is vital that the limitations of current analysers are fully understood. The aim of this large multicentre study was to determine the accuracy of haematology analysers in current routine practice for platelet counts below 20 · 10 9 /l. Platelet counts estimated by analysers using optical, impedance and immunological methods were compared with the International Reference Method for platelet counting. The results demonstrated variation in platelet counting between different analysers and even the same type of analyser at different sites. Optical methods for platelet counting on the XE 2100, Advia 120, Cell-Dyn 4000 and H3* were not superior to impedance methods on the XE 2100, LH750 and Pentra analysers. All analysers except one overestimated the platelet count, which would result in under transfusion of platelets. This study highlights the inaccuracies of haematology analysers in platelet counting in severe thrombocytopenia. It re-emphasizes the need for external quality control to improve analyser calibration for samples with low platelet counts, and suggests that the optimal thresholds for prophylactic platelet transfusions should be re-evaluated.

Comparison of platelet counts by CellDyn Sapphire (Abbot Diagnostics), LH750 (Beckman Coulter), ReaPanThrombo immunoplatelet method (ReaMetrix), and the international flow reference method, in thrombocytopenic blood samples

“…This may not be representative of other causes of thrombocytopenia especially those which are immune mediated like immune thrombocytopenic purpura (ITP). This is a significant factor as the superiority of optical and impedance methods over each other has often been debated (5). Improvements of optical platelet counting over impedance methods were described using various other CBC analyzers (16,17).…”

“…However, these smaller studies did not use thrombocytopenic samples from patients with hematological malignancies; the samples were specifically selected with low counts from patients with autoimmune thrombocytopenia and myelodysplasia, which are disorders where impedance platelet counting is known to be problematic because of the presence of large platelets. Recent studies have now concluded that the current methods of optical platelet counting may not be superior to impedance counts for all patient populations (5,18). This study shows similar results with the counts obtained from LH750 impedance (with curve fit correction) based counter were more precise and accurate when compared with CDS-O counts.…”

“…Various studies have evaluated the degree of inaccurate transfusion at different transfusion thresholds (5,19,20). Hong et al (20) evaluated the accuracy of three automated hematology analyzers [Sysmex XE-2100 (both optical and impedance mode), Bayer Advia 120, and Beckman Coulter LH-750] when compared with the immunoplatelet reference method for platelet measurement and also looked at the degree of under transfusion and over transfusion.…”

“…A large multicenter study to compare the inaccuracy of platelet counts of current analyzers in severe thrombocytopenia (compared with a reference flow cytometric method) showed that the most analyzers overestimated the count, which would result in under transfusion of platelets at any set threshold (5). This study highlighted the inaccuracies of hematology analyzers in platelet counting and reemphasizes the need for low-platelet count controls and calibrators to improve analyzer calibration.…”

Automated method and not automated: the evaluation of hematological parameters of rats

“…Similar findings were reported by Messias et al (2009). Nevertheless, the automatic analyzers can overestimate the platelet count, what may be related to poor sensitivity at levels below the normal range (Segal et al, 2005).…”