860 An assessment of the agreement between laboratory and POCT blood gas analysis measurements of haematological indices on the NICU: a service evaluation

Abstract

<h3>Background</h3> Accurate biochemical monitoring is an essential component of neonatal care and fluid management. Point of care testing (POCT) offers a rapid, low blood volume alternative to laboratory testing, reducing the risk of iatrogenic anaemia in neonates and potentially improving patients’ prognoses. If the blood gas analyser (BGA) can act as a reasonably accurate proxy for laboratory analyses, this will enable more rapid biochemical analyses, accelerating clinical decision making. <h3>Objectives</h3> To assess the agreement of haematological indices (sodium, potassium, chloride, and bilirubin concentrations, and haemoglobin levels) when measured by BGA and laboratory testing. <h3>Methods</h3> This service evaluation took place in the neonatal intensive care unit (NICU) at Liverpool Women’s Hospital (LWH). The BGA used on the NICU at LWH is the Siemens Rapidlab 1265 and the Abbott Alinity analysers are used at Alder Hey Children’s Hospital Laboratory. Temporally paired sets of results from POCT blood gas analysis and laboratory testing were identified from the historical NICU Badger database to allow comparison of agreement. The blood gas sample taken within the closest proximity to the biochemistry or haematology sample was used, with samples taken within two hours of each other regarded as adequately paired. Statistical analyses were undertaken using the SPSS software version 26.0 and within Excel 2010. The data were processed using the method outlined by Bland and Altman, for the calculation of Bland Altman plots where repeated measures have occurred. Grubb’s test for outliers was performed on the data to identify and remove obviously aberrant values. <h3>Results</h3> Paired samples for analyses were available from 99 patients over a two-month period. Following the application of Grubb’s outlier test and considering the matched pairs for which not all data were available, there were 195 paired samples for sodium analyses, 186 paired samples for potassium, 185 paired samples for chloride, 135 paired samples for bilirubin and 124 paired samples for haemoglobin. The mean difference for sodium measurements between the laboratory and blood gas measurements was +1.72 mmol/L (95% confidence interval (95%CI) −4.95 mmol/L to 8.40 mmol/L); for potassium measurements +0.29 mmol/L (95%CI −0.88 mmol/L to 1.46 mmol/L); for chloride measurements +1.93mmol/L (95%CI −3.74mmol/L to 7.60 mmol/L) for bilirubin measurements +5.39 μmol/L (95%CI −42.57μmol/L to 31.80 μmol/L) and for haemoglobin measurements −1.89 g/L (95%CI −15.89 g/L to 12.12 g/L). <h3>Conclusions</h3> Assessment of some of these measurements (namely bilirubin concentration and haemoglobin levels) by POCT may be acceptable for intermittent monitoring of haematological parameters in neonates, providing that the extremes of the estimated true value (indicated by the range of the 95%CI) would not mandate a different treatment course. The relatively wide 95%CI for some of the electrolyte levels (namely sodium and potassium concentrations), which exceeds the normal range of those values, limits the value of these measurements as independent measures, without laboratory corroboration, although trends may still be inferred. It may be necessary to explain these caveats to clinical staff interpreting the results and provide further education regarding reference ranges.