Comparison of point-of-care device DiaSpect against the HemoCue and laboratory analyser in an ICU population

Abstract

Background: Haemoglobin levels guide many diagnostic and therapeutic interventions. They form part of the basic daily management of ICU patients. Laboratory results are frequently delayed. This can have important clinical consequences, including withholding potentially life-saving interventions. Point-of-care devices, if accurate, provide a solution to this problem. Inaccurate devices may provide misleading results leading to unnecessary and hazardous blood transfusions or under-estimation of disease severity. The aim was to compare the accuracy and reliability of the DiaSpect Hemoglobin T (DiaSpect. Medical GmbH, Sailauf, Germany) haemoglobinometer with the HemoCue (HemoCue AB, Angelholm Sweden) haemoglobinometer and the gold standard laboratory analyser for this trial (XT-200i Sysmex/Coulter LH750). Methods: Venous blood samples were collected via a central venous line from Intensive Care Unit (ICU) patients (n = 265) for determination of haemoglobin (Hb) concentration by DiaSpect as well as the HemoCue, and the automated haemoglobin analysers: XT-200i Sysmex or the Coulter LH750. Agreement between the methods was tested using Bland–Altman plots. A Hb variation of 0.5 g/dl was considered clinically significant. There were a significant number of wasted samples with the DiaSpect, as a total of 350 cuvettes were used but only 214 complete samples obtained. This wastage was attributed to errors in sampling, cuvette shortage, lack of experience with the device and lack of ‘user-friendliness’. Results: A total of 265 samples were obtained, of which 207 had complete data sets and were analysed further in the study. There were 58 incomplete samples, of which 51 were DiaSpect samples. Mean haemoglobin was 9.11 (standard deviation ± 1.74), 9.07 (standard deviation ± 1.96) and 9.02 (standard deviation ± 2.06) using the laboratory analyser, HemoCue and DiaSpect respectively. Laboratory and HemoCue: the mean difference between the two readings was -0.04 (95% limits of agreement -2.15 to 2.07 g/dl) with an average between 5.00 g/dl and 15.10 g/dl. Laboratory and DiaSpect: the mean difference between the readings was -0.09 (95% limits of agreement -2.64 to 2.46 g/dl) with an average between 5.10 g/dl and 14.9 g/dl. HemoCue and DiaSpect: the mean difference between the two haemoglobin levels was 0.04 (95% limits of agreement 2.73 to -2.64 g/dl). Both methods returned measurements within the pre-defined clinical acceptable limits of ± 0.5 g/dl. Conclusion: The haemoglobin concentration assessment by the HemoCue and DiaSpect showed an acceptable limit of agreement when compared against the laboratory analyser. There were a significant number of wasted samples when using the DiaSpect.