BACKGROUND: An approved IFCC reference method for measuring HbA1c, with a firm and reproducible correlation with NGSP values, is now available. We established (i) the degree of agreement of HbA1c results between the NGSP-certified and the IFCC-calibrated (converted to NGSP values) immunochemical method and (ii) the difference in the classification of control of diabetes mellitus (DM) in individual patients between the methods. METHODS: HbA1c was measured on the same hemolysate from each patient by both methods (n=92). Results were analyzed by the kappa statistic, linear regression, and McNemar's chi(2) test. RESULTS: Both methods achieved acceptable analytical performance. The kappa statistic measure of agreement was 0.65 and r(2)=0.937. Overall, 21 (22.8%) patients were classified differently in the extent of diabetes control (good, acceptable, or poor), with a significant difference between the 2 methods (p<0.0005). CONCLUSIONS: The IFCC converted to NGSP values were significantly different from the previously used NGSP-certified method. The differences between the 2 methods are of sufficient magnitude that HbA1c results from these methods are not interchangeable. The IFCC method being scientifically superior in concept and design should replace the NGSP method. New HbA1c targets for DM management need to be established for the IFCC method.
BACKGROUND: An approved IFCC reference method for measuring HbA1c, with a firm and reproducible correlation with NGSP values, is now available. We established (i) the degree of agreement of HbA1c results between the NGSP-certified and the IFCC-calibrated (converted to NGSP values) immunochemical method and (ii) the difference in the classification of control of diabetes mellitus (DM) in individual patients between the methods. METHODS: HbA1c was measured on the same hemolysate from each patient by both methods (n=92). Results were analyzed by the kappa statistic, linear regression, and McNemar's chi(2) test. RESULTS: Both methods achieved acceptable analytical performance. The kappa statistic measure of agreement was 0.65 and r(2)=0.937. Overall, 21 (22.8%) patients were classified differently in the extent of diabetes control (good, acceptable, or poor), with a significant difference between the 2 methods (p<0.0005). CONCLUSIONS: The IFCC converted to NGSP values were significantly different from the previously used NGSP-certified method. The differences between the 2 methods are of sufficient magnitude that HbA1c results from these methods are not interchangeable. The IFCC method being scientifically superior in concept and design should replace the NGSP method. New HbA1c targets for DM management need to be established for the IFCC method.