BACKGROUND: The purpose of this study was to evaluate the performance and agreement among HbA(1c) values measured using selected analyzers certified by the National Glycohemoglobin Standardization Program (NGSP) and standardized by the International Federation of Clinical Chemistry and Laboratory Medicine (IFCC). METHODS: HbA(1c) determined using D-10 (Bio-Rad, USA), Variant II Turbo (Turbo; Bio-Rad, USA), Cobas Integra 800 (Integra; Roche, Switzerland) and Afinion AS100 (Afinion; Axis-Shield, Norway) were compared with each other. Precision and method comparisons with Deming regression were evaluated according to CLSI recommendations. We also compared the HbA(1c) values obtained with each analyzer using either IFCC or NGSP methods by correlation analysis and kappa statistics. RESULTS: The repeatability and method/device precisions of D-10 and Afinion were acceptable. The correlation coefficients of HbA(1c) were 0.986 for D-10 vs. Afinion, 0.997 for D-10 vs. Turbo, 0.988 for D-10 vs. Integra, and 0.991 for Integra vs. Afinion. The average biases of HbA(1c) Afinion (IFCC) and HbA(1c) Integra (IFCC) against HbA(1c) D-10 (NGSP) were -1.90% and -1.79%, respectively. Kappa agreement statistics for the three diabetic control group HbA(1c) values of "less than 6.5%," "6.5%-7.5%," and "greater than 7.5%" for D-10 vs. Turbo, D-10 vs. Integra, and D-10 vs. Afinion were 0.872, 0.836, and 0.833, respectively. CONCLUSIONS: The strong correlations and good clinical agreements of HbA(1c) between each analyzer expressed in terms of either NGSP or IFCC-derived NGSP indicate that these analyzers can be used interchangeably.
BACKGROUND: The purpose of this study was to evaluate the performance and agreement among HbA(1c) values measured using selected analyzers certified by the National Glycohemoglobin Standardization Program (NGSP) and standardized by the International Federation of Clinical Chemistry and Laboratory Medicine (IFCC). METHODS: HbA(1c) determined using D-10 (Bio-Rad, USA), Variant II Turbo (Turbo; Bio-Rad, USA), Cobas Integra 800 (Integra; Roche, Switzerland) and Afinion AS100 (Afinion; Axis-Shield, Norway) were compared with each other. Precision and method comparisons with Deming regression were evaluated according to CLSI recommendations. We also compared the HbA(1c) values obtained with each analyzer using either IFCC or NGSP methods by correlation analysis and kappa statistics. RESULTS: The repeatability and method/device precisions of D-10 and Afinion were acceptable. The correlation coefficients of HbA(1c) were 0.986 for D-10 vs. Afinion, 0.997 for D-10 vs. Turbo, 0.988 for D-10 vs. Integra, and 0.991 for Integra vs. Afinion. The average biases of HbA(1c) Afinion (IFCC) and HbA(1c) Integra (IFCC) against HbA(1c) D-10 (NGSP) were -1.90% and -1.79%, respectively. Kappa agreement statistics for the three diabetic control group HbA(1c) values of "less than 6.5%," "6.5%-7.5%," and "greater than 7.5%" for D-10 vs. Turbo, D-10 vs. Integra, and D-10 vs. Afinion were 0.872, 0.836, and 0.833, respectively. CONCLUSIONS: The strong correlations and good clinical agreements of HbA(1c) between each analyzer expressed in terms of either NGSP or IFCC-derived NGSP indicate that these analyzers can be used interchangeably.