Reporting hemoglobin A1c: do the units matter?

Hemoglobin A1c (Hb A1c) 2 is the analyte most commonly measured in clinical laboratories for patients with diabetes mellitus. Physicians use Hb A1c to monitor long-term glycemic control, adjust therapy, and predict complications of diabetes. Recently, Hb A1c was added as a criterion for the diagnosis of diabetes. Unlike most analytes, many patients with diabetes know their Hb A1c value and are familiar with the target value they need to attain for good glycemic control. Glycation of proteins was recognized over 100 years ago. Hb A1c was first isolated (by chromatography) in 1958 (1 ) and was identified 8 years later as a Schiff base between hemoglobin and a ketone or aldehyde (2 ). Early commercial assays for Hb A1c were limited by the absence of standardization. For example, a comparison of 7 methods in 1992 revealed very large differences ( 200%) in Hb A1c values among methods (3 ). Standardization programs were developed soon after 1993, when the Diabetes Control and Complications Trial (DCCT) documented the clinical value of Hb A1c in patients with type 1 diabetes (4 ). The publication 5 years later of results for the UK Prospective Diabetes Study (UKPDS), which revealed that decreasing Hb A1c in patients with type 2 diabetes reduced complications (5 ), further emphasized the essential contribution of Hb A1c monitoring to patient care. Importantly, the Hb A1c results obtained in the UKPDS were aligned with those in the DCCT. Therefore, appropriate patient management requires that an individual’s Hb A1c value be traceable to DCCT/UKPDS values. The NGSP, previously known as the National Glycohemoglobin Standardization Program, was formed to achieve this objective. The NGSP laboratory network assists manufacturers of Hb A1c assays in calibrating their methods so that patient values correlate with those of the DCCT and the UKPDS (6 ). Analogous to the clinical-outcomes trials, Hb A1c is reported as a percentage of the total hemoglobin. The NGSP, which has markedly improved Hb A1c standardization (6 ), has been adopted by numerous countries. A few years after the establishment of the NGSP, the IFCC developed a higher-order reference method for Hb A1c. A comparison of the NGSP and IFCC networks revealed a linear relationship, but IFCC values were approximately 1.5% to 2% lower (7 ). An influential decision was made to report IFCC results in SI units [millimoles of Hb A1c per mole of Hb A0 (8 )], and several countries, predominantly in Europe and the Antipodes, have adopted SI units for Hb A1c (9 ). One of the motivations for the use of SI units was that the large difference in results (e.g., an Hb A1c value of 6.5% is equivalent to 48 mmol/mol) would eliminate confusion regarding the reporting schemes, but no evidence to support this premise has previously been published. The study reported by Kilpatrick et al. (10 ) in this issue of Clinical Chemistry has addressed this topic by evaluating the effect on glycemic control of changing Hb A1c reporting from DCCT/NGSP units (a percentage) to SI units (millimoles per mole). Data were derived from all the patients on the diabetes register in Hull and East Yorkshire, UK. In the UK, Hb A1c was reported exclusively as a percentage until June 1, 2009, when “dual reporting” of both DCCT/NGSP units and the new units was instituted. Beginning October 1, 2011, Hb A1c results were reported only in SI units. Kilpatrick et al. (10 ) compared Hb A1c values during the last year of dual reporting (October 1, 2010, to September 30, 2011) to those obtained during the first year after DCCT/NGSP units were eliminated (October 1, 2011, to September 30, 2012). The authors have drawn 2 general conclusions from their data analysis: (a) There was no difference in the overall glycemic control of the entire population before and after the units were changed, and (b) the change in Hb A1c in the patients who had poor glycemic control (defined in their study as an Hb A1c value 8%) was not altered by the units used for reporting. This concise—and timely—report, which provides valuable information, exhibits several positive attributes. The population was large, with 44 721 Hb A1c measurements (21 880 in the first year and 22 841 in the second year) obtained from 13 197 patients, all of whom have diabetes. All samples were analyzed in a single laboratory using a single method. This strategy 1 Department of Laboratory Medicine, National Institutes of Health, Bethesda, MD. * Address correspondence to the author at: Department of Laboratory Medicine, Bldg. 10, Rm. 2C306, 10 Center Dr., Bethesda, MD 20892. Fax 301-402-1885; e-mail [email protected]. Received July 10, 2013; accepted July 15, 2013. Previously published online at DOI: 10.1373/clinchem.2013.211227 2 Nonstandard abbreviations: Hb A1c, hemoglobin A1c; DCCT, Diabetes Control and Complications Trial; UKPDS, UK Prospective Diabetes Study. Clinical Chemistry 59:1