Accuracy of conversion formulae for estimation of glycohaemoglobin

To analyse the accuracy of the conversion formulae for estimation of glycohaemoglobin (GHb) measured by different methods, we analysed 210 samples for HbA1c using HPLC. Fifty of these specimens were analysed by micro-column chromatography (MC), 43 by electrophoresis (EP), 50 by IMX system (Abbott Laboratories), 38 by Primus HPLC and 29 by Diamat HPLC. Regression analyses were performed and the equations were used to estimate HbA1c values (HbA1c calc) for the five methods. The 95% limits of agreement between HPLC and the converted results were -1.77 to 1.71%, -1.54 to 1.54%, -0.92 to 0.88%, -0.46 to 0.56%, and -0.39 to 0.41% for MC, EP, IMX, Primus and Diamat equations, respectively. The mean relative errors were 3.4 (-28.2 to 35%), 1.3 (-22.9 to 25.5%), 0.4 (-14.6 to 15.0%), 0.51 (-6.55 to 7.57%), -0.20 (-5.8 to 5.4%), for MC, EP, IMX, Primus and Diamat, respectively. These results show that conversion formulae based on methods that do not measure HbA1c (MC, EP and IMX) are inaccurate and can mask a clinically relevant variation of HbA1c. However, GHb results obtained by HPLC methods could be interchangeably converted with an absolute variation of less than 1%. Converted HbA1c results from non-standardized methods should be interpreted with caution.