Objective: To assess four commercially available point-of-care HbA1c analytical systems with respect to (1) inaccuracy against the central laboratory HbA1c method and (2) imprecision against the HbA1c analytical goal of coefficient of variation < 3%.
Methods: Analytical inaccuracy was assessed by analysis of 110 patient samples on all five analytical platforms (Biorad Diastat, Drew DS5, Bayer DCA 2000, Nycomed Nycocard and Roche Tinaquant (used in central lab)). Analytical imprecision was assessed by analysis of two levels of patient sample four times daily for six days, as well as analysis of two levels of commercial control.
Results: Deming linear regression for agreement: Diastat=0.98 x Tinaquant + 0.36; DS5=1.23 x Tinaquant - 0.65; DCA2000=0.95 x Tinaquant + 0.63; Nycocard=0.94 x Tinaquant + 0.92. Analytical coefficients of variation (CVs) at Tinaquant HbA1c levels of 6.2-10.8% were: Tinaquant 0.8-1.1%, Diastat 1.6-6.6%, DCA2000 2.6-7.2%, DS5 5.1-11.7%, Nycocard 8.5-15.3%. Two HbE samples gave elevated HbA1c results with the DS5 method.
Conclusions: The Diastat and DCA2000 systems gave the best performance with acceptable imprecision and good agreement with both the central lab and each other. The DS5 was less precise with a significant positive bias compared to the other methods and interference from HbE, while the Nycocard system showed the poorest precision in the evaluation. The Diastat and DCA2000 systems appear to be satisfactory analytical alternatives to both central laboratory (Tinaquant) testing and each other.