Comparison of IFCC-calibrated HbA<inf>1c</inf>from laboratory and point of care testing systems
Manley SE., Hikin LJ., Round RA., Manning PW., Luzio SD., Dunseath GJ., Nightingale PG., Stratton IM., Cramb R., Sikaris KA., Gough SCL., Webber J.
ï¿½ 2014 Elsevier Ireland Ltd. Objective: WHO, IDF and ADA recommend HbA 1c ≥6.5% (48mmol/mol) for diagnosis of diabetes with pre-diabetes 6.0% (42mmol/mol) [WHO] or 5.7% (39mmol/mol) [ADA] to 6.4% (47mmol/mol). We have compared HbA 1c from several methods for research relating glycaemic markers. Research design and methods: HbA 1c was measured in EDTA blood from 128 patients with diabetes on IE HPLC analysers (Bio-Rad Variant II NU, Menarini HA8160 and Tosoh G8), point of care systems, POCT, (A1cNow+ disposable cartridges and DCA 2000ï¿½+ analyser), affinity chromatography (Primus Ultra2) and the IFCC secondary reference method (Menarini HA8160 calibrated using IFCC SRM protocol). Results: Median (IQ range) on IFCC SRM was 7.5% (6.8-8.4) (58(51-68) mmol/mol) HbA 1c with minimum 5.3%(34mmol/mol)/maximum 11.9%(107mmol/mol). There were positive offsets between IFCC SRM and Bio-Rad Variant II NU, mean difference (1SD), +0.33%(0.17) (+3.6(1.9) mmol/mol), r 2 =0.984, p < 0.001 and Tosoh G8, +0.22%(0.20) (2.4(2.2) mmol/mol), r 2 =0.976, p < 0.001 with a very small negative difference -0.04%(0.11) (-0.4(1.2) mmol/mol), r 2 =0.992, p < 0.001 for Menarini HA8160. POCT methods were less precise with negative offsets for DCA 2000ï¿½+ analyser -0.13%(0.28) (-1.4(3.1) mmol/mol), r 2 =0.955, p < 0.001 and A1cNow+ cartridges -0.70%(0.67) (-7.7(7.3) mmol/mol), r 2 =0.699, p < 0.001 (n=113). Positive biases for Tosoh and Bio-Rad (compared with IFCC SRM) have been eliminated by subsequent revision of calibration. Conclusions: Small differences observed between IFCC-calibrated and NGSP certified methods across a wide HbA 1c range were confirmed by quality control and external quality assurance. As these offsets affect estimates of diabetes prevalence, the analyser (and calibrator) employed should be considered when evaluating diagnostic data.