Verification of point-of-care analysers for C-reactive protein, lipid studies and glycated haemoglobin.

ABSTRACT

Due to increased convenience and faster test results, interest in point-of-care testing (PoCT) has grown significantly. Though PoCT may improve the speed and convenience of testing, the devices need to be fit for their intended purpose. Our aim was to verify the performance of Roche cobas b 101 and Abbott Afinion 2 for C-reactive protein (CRP), lipid studies and glycated haemoglobin (HbA1c), and Siemens Atellica DCA for HbA1c. For all PoCT analysers and measurands, accuracy was assessed by method comparison with central laboratory analysers. Passing-Bablok linear regression was performed, and Bland-Altman plots were generated. The proportion of samples within the Royal College of Pathologists of Australasia Quality Assurance Programs Analytical Performance Specifications (RCPAQAP APS) was assessed. Within-run and between-day imprecision was assessed and compared with manufacturer claims and biological variation or clinical guidelines for desirable imprecision. For CRP, both evaluated PoCT analysers had all samples within the RCPAQAP APS and had optimal imprecision according to biological variation. For lipid studies, the Roche cobas b 101 had most samples within the RCPAQAP APS, with two of 22 cholesterol, one of 22 high-density-lipoprotein-cholesterol (HDL-C) and zero of 22 triglyceride comparisons outside the RCPAQAP APS. The Abbott Afinion 2 had a positive bias with all three measured parameters, although the effect was more limited in the calculated parameters cholesterolHDL-C ratio, non-HDL-C and low-density-lipoprotein-cholesterol (LDL-C). For HbA1c, all analysers had acceptable imprecision for monitoring with coefficient of variation (CV) <3% and minimal bias at the treatment target (HbA1c 53 mmol/mol or 7.0%). However, significant biases were apparent at higher or lower HbA1c for all analysers. All evaluated analysers were fit for purpose for CRP and for serial monitoring of HbA1c, although bias in some analysers was present at extremes of HbA1c. For lipid studies, the Roche cobas b 101 had fewer results outside the RCPAQAP allowable limits, and better precision. The Abbott Afinion 2 had a positive bias on both the cholesterol and HDL-C, but there is limited clinical impact when calculating cholesterolHDL-C, LDL-C and non-HDL-C.