ABSTRACT
BACKGROUND: A Spanish multicentre evaluation of the third generation of Roche Diagnostics immunoturbidimetric inhibition method (TINIA) is presented for quantification of haemoglobin A1c (HbA1c) in whole blood. METHODS: The analytical performance of the TINIA test was evaluated and blood sample results were compared with two other widely used analysers, Bio-Rad Variant II and Adams Arkray HA-8160, based on HPLC. RESULTS: Within- and between-batch imprecision (coefficients of variation (CVs)) for HbA1c levels of 5, 6, 7 and 8% were 0.77, 1.23, 1.35 and 1.26% and 2.38, 1.51, 1.76 and 2.16%, respectively. For low (5.4% A1c) and high (10.1% A1c) quality control samples, the within and between-batch %CV were: 1.26; 1.43 and 2; 1.71 respectively. The test met the expected performance in most aspects, except for linearity, that is under the reported range, and HbF interferences, detected for levels over 7.5%. There was a good concordance between the results of TINIA and Variant-IIt in the whole range and with HA-8160 only up to levels of 9%. Between-batch imprecision suggests more frequent calibrations than reported by the provider to maintain variability within the limits established by clinical practice guidelines. CONCLUSIONS: The assay meets the necessary quality standards for routine use, as long as we keep the analytical variability within narrow limits. The results may be interchangeable with the tested HPLC systems, but HbF interference is not detected and it happens at lower levels than reported.
