Desirable performance standards for HbA(1c) analysis - precision, accuracy and standardisation: consensus statement of the Australasian Association of Clinical Biochemists (AACB), the Australian Diabetes Society (ADS), the Royal College of Pathologists of Australasia (RCPA), Endocrine Society of Australia (ESA), and the Australian Diabetes Educators Association (ADEA)

Abstract

Background: HbA(1c) (glycohaemoglobin) is universally used in the ongoing monitoring of all patients with diabetes. There are many % HbA(1c) target control rating recommendations by national, regional and international expert bodies for diabetes patients and these are variable around the world. General patient target control ratings are currently most often recommended as either <6.5% or <7.0% HbA(1c), with <6.0% HbA(1c) stated for individual patients where clinically possible. This necessitates very precise HbA(1c) assays and the same patient values, irrespective of HbA(1c) method or area of the world.

Methods: HbA(1c) targets recommended by major expert groups and published HbA(1c) assay precision (coefficient of variation, %CV) levels have been detailed. These have been compared with published biological variation levels and with calculated HbA(1c) error ranges at various HbA(1c) levels and %CV levels. In addition, these have been compared with the analytical precision necessary to differentiate between the upper limit of the normal range for HbA(1c) and targets recommended by expert groups for diabetes control.

Results: Intralaboratory analytical CVs of <2% are necessary and are achievable on automated HPLC analysers, and are supported on grounds of both clinical need and biological variation, as well as the need to differentiate the national, regional and international target recommendations from the upper limit of the normal range (<6.0% HbA(1c) level).

Conclusions: Routine methods with tight long-term imprecision with CVs of <2% are recommended. International HbA(1c) targets essentially require that all HbA(1c) methods be precise, and have minimal standardisation bias and minimal methodological interferences in individual patients.