The results of HbA1c measurement and its comparison with reference method values in an EQA programme

Abstract

This article describes the effect of a change in reference method for HbA1c measurement on results in our EQA programme. Important indicators of analytical quality, particularly reproducibility, bias and uncertainty, were assessed. Basic analytical quality indicators for the measurement of HbA1c in an EQA survey during 2000–2008 were evaluated. Three EQA surveys per year were performed. Usually there were more than 200 participating laboratories in each survey. Reproducibility, systematic differences, bias, and methodology changes before and after IFCC reference method implementation were determined. In the EQA surveys, target values traceable to the IFCC reference method are used. Estimation of the combined uncertainty of the result was a voluntary part of the surveys and was reported by 50% of participating laboratories. Reproducibility of all participants’ measurements was close to the target value of CV = 5.0% (coefficient of variation). The group of participants using HPLC methodology had reproducibility lower than 5.0% in each survey, whilst the immunochemistry group usually had CV values more than 6.0%. The differences in individual results from target values ranged in the interval ±12% for HPLC methods but more (−8 to +30%) for immunochemistry methods. During the monitored time interval the number of participants using HPLC methods increased up to 70% whilst the number of participants using immunochemistry methods decreased approximately to 30%. The combined uncertainty (uc) reported by participants varies between 4.0 and 4.5% for HbA1c values between 30 and 90 mmol/mol. No significant differences between precision and systematic errors (bias) after the implementation of IFCC method of measurement for HbA1c to routine processes for HPLC were observed. Immunochemical methods do not fulfil requirements for analytical reproducibility.