BACKGROUND: In subjects with type 1 diabetes, persisting elevations of fetal hemoglobin (HbF) have been demonstrated. This study evaluated whether HbF levels typically seen in type 1 diabetes (up to 3%) interfere with glycohemoglobin determinations using a common immunologic method (DCA 2000). METHODS: HbA(1c) was measured by high-performance liquid chromatography (HPLC) using a Diamat analyzer in 90 type 1 diabetics with parallel determinations of HbF. Results were compared with HbA(1c) concentrations obtained using DCA 2000. RESULTS: Reproducibility was good for both methods with coefficients of variation <5% and correlation between the two methods was good (r(2)=0.939, p<0.0001). Mean difference between the two methods was small (0.007%). Limits of agreement varied between -0.92% and +0.93% (95% confidence interval [95% CI]) and constant bias (intercept: 0.73 95% CI 0.28-1.18) as well as a proportional bias (slope: 0.92 95% CI 0.87-0.97) were detected. At low concentrations of HbF, the DCA 2000 immunologic method tended to underestimate and at higher concentrations tended to overestimate HbA(1c) when compared with Diamat. Stepwise linear regression with HbA(1c) (DCA 2000) as dependent variable included HbA(1c) (Diamat) and HbF in the model (r(2)=0.946, p<0.0001), explaining 94.6% of the variability of HbA(1c) (DCA 2000). Partial correlation coefficient between HbA(1c) (DCA 2000) and HbF corrected for HbA(1c) (Diamat) was 0.337 (p=0.0012). CONCLUSIONS: DCA 2000 allowed measurements of HbA(1c) rapidly and with precision adequate for clinical purposes. However, agreement with Diamat results was comparatively weak with both constant as well as proportional biases. The 95% limits of agreement between Diamat and DCA 2000 fell within a range that significantly limited traceability between these two methods; therefore, the two methods should not be used interchangeably. Small but persistent elevations of HbF concentrations were identified as a significant cofactor, which may be relevant for limited traceability between the two methods.
BACKGROUND: In subjects with type 1 diabetes, persisting elevations of fetal hemoglobin (HbF) have been demonstrated. This study evaluated whether HbF levels typically seen in type 1 diabetes (up to 3%) interfere with glycohemoglobin determinations using a common immunologic method (DCA 2000). METHODS: HbA(1c) was measured by high-performance liquid chromatography (HPLC) using a Diamat analyzer in 90 type 1 diabetics with parallel determinations of HbF. Results were compared with HbA(1c) concentrations obtained using DCA 2000. RESULTS: Reproducibility was good for both methods with coefficients of variation <5% and correlation between the two methods was good (r(2)=0.939, p<0.0001). Mean difference between the two methods was small (0.007%). Limits of agreement varied between -0.92% and +0.93% (95% confidence interval [95% CI]) and constant bias (intercept: 0.73 95% CI 0.28-1.18) as well as a proportional bias (slope: 0.92 95% CI 0.87-0.97) were detected. At low concentrations of HbF, the DCA 2000 immunologic method tended to underestimate and at higher concentrations tended to overestimate HbA(1c) when compared with Diamat. Stepwise linear regression with HbA(1c) (DCA 2000) as dependent variable included HbA(1c) (Diamat) and HbF in the model (r(2)=0.946, p<0.0001), explaining 94.6% of the variability of HbA(1c) (DCA 2000). Partial correlation coefficient between HbA(1c) (DCA 2000) and HbF corrected for HbA(1c) (Diamat) was 0.337 (p=0.0012). CONCLUSIONS:DCA 2000 allowed measurements of HbA(1c) rapidly and with precision adequate for clinical purposes. However, agreement with Diamat results was comparatively weak with both constant as well as proportional biases. The 95% limits of agreement between Diamat and DCA 2000 fell within a range that significantly limited traceability between these two methods; therefore, the two methods should not be used interchangeably. Small but persistent elevations of HbF concentrations were identified as a significant cofactor, which may be relevant for limited traceability between the two methods.