Use of HbA1c in the diagnosis of diabetes

BMJ 2012; 345 doi: http://dx.doi.org/10.1136/bmj.e7293 (Published 01 November 2012) Cite this as: BMJ 2012;345:e7293

This article has a correction. Please see:

  1. Andrew Farmer, professor of general practice
  1. 1NIHR School for Primary Care Research, Department of Primary Care Health Sciences, University of Oxford, Oxford OX2 6GG, UK
  1. andrew.farmer{at}phc.ox.ac.uk

Be aware of clinical circumstances in which results may mislead

Before 2010, guidelines recommended the measurement of blood glucose for diagnosing diabetes. Improved standardisation in the measurement of glycated haemoglobin (HbA1c) and wider availability of the assay led to a recommendation in 2011 by the World Health Organization that HbA1c should be used instead.1 This recommendation was included within a recent UK public health guidance for the identification of people at high risk of diabetes.2 In addition, a UK expert advisory body convened by the Department of Health has—after an extended period of consultation—now stated how the WHO recommendations should be implemented.3

The expert advisory group recommends that an HbA1c cut-off point of 48 mmol/mol or more (6.5%) be used for the diagnosis of diabetes. Unless the clinical diagnosis is clear, a second confirmatory measurement is needed as soon as possible. If either the initial or follow-up measurement is less than 48 mmol/mol, the diagnosis of diabetes should not be made.3 The group recommended that patients with an HbA1c of 42-47 mmol/mol should be considered at high risk of diabetes and provided with intensive lifestyle advice and retested annually. Those with an HbA1c of less than 42 mmol/mol may still have a high risk of diabetes and should be treated according to clinical indication, with retesting at least every three years.2 The group also recommended that glucose testing should not be carried out alongside or after HbA1c measurement for confirmation of the diagnosis unless the patient has a condition in which HbA1c cannot be measured accurately.

A major benefit of the new diagnostic test is that it avoids the need for fasting. Furthermore, the biological variability of HbA1c, which reflects glucose exposure over the 120 day average lifespan of an erythrocyte, is lower than that of glucose. HbA1c is also more stable than glucose in samples transported to the laboratory and is a good predictor of future complications of diabetes, such as cardiovascular disease.4 Assays for HbA1c are more variable than glucose assays, however, and achieving an acceptable level of precision requires that analysis is carried out in an accredited laboratory that takes part in an external quality assurance scheme. The measurement of HbA1c is also more expensive, and it may not be possible in settings with limited resources.

A major concern in recommending the measurement of HbA1c as a diagnostic test for diabetes is that it might be used inappropriately (box). HbA1c should not be used to diagnose diabetes when type 1 diabetes is suspected, in people who are acutely ill, in children or young adults, or in women who may have gestational diabetes. In these conditions glucose values can change quickly, and the HbA1c value might not accurately reflect glycaemic exposure. In acutely ill patients, management should still be guided by the measurement of finger prick capillary glucose and confirmation of glucose concentrations with laboratory assayed samples. Asymptomatic children and adolescents should continue to be tested for diabetes with glucose based measures rather than HbA1c assays.5

Situations in which glycated haemoglobin (HbA1c) cannot be used for diagnosis3

  • All children and young people.

  • Women who are currently pregnant or have been pregnant in the past two months

  • People with suspected type 1 diabetes, regardless of age

  • People whose diabetic symptoms are of short duration

  • Patients at high risk of diabetes who are acutely ill (HbA1c ≥48 mmol/mol confirms pre-existing diabetes, but a value <48 mmol/mol does not exclude it, and such patients must be retested once the acute episode has resolved)

  • Patients taking drugs that may cause a rapid rise in glucose, such as corticosteroids or antipsychotic drugs (≤2 months). HbA1c can be used in patients taking these drugs longer term (>2 months) who are not clinically unwell

  • Patients with acute pancreatic damage or who have undergone pancreatic surgery

  • Patients with renal failure

  • Patients with HIV infection

HbA1c values are also affected by certain haemoglobin variants and haemolytic anaemia, which, along with other conditions, affect erythrocyte survival. Severe iron deficiency anaemia should be treated before measuring HbA1c. Other conditions may interfere with the measurement of HbA1c. Most assays are based on immunochemistry or high performance liquid chromatography, and their use in specific circumstances needs to be guided by local laboratories. The prevalence of individual types of haemoglobinopathy, many of which can be identified on testing, and the assay being used must be taken into account.6 Nevertheless, HbA1c can be used to diagnose diabetes in most people.

The HbA1c cut-off point for the diagnosis of diabetes has been determined from observational studies on the association between intervals of glycaemic measures and the prevalence of diabetes specific retinopathy.7 It is uncertain how much the new diagnostic strategy will increase or decrease the prevalence of diabetes compared with using the oral glucose tolerance test. It is also unclear whether the new strategy will identify the same individuals as having diabetes as were identified using the old method. HbA1c and glucose concentrations may be less strongly associated across the range of glycaemia, particularly in the sub-diabetic range, than was previously thought.8 In the absence of a gold standard test for diagnosing diabetes, the recommendations made in the recent guidelines for active management of people whose HbA1c values are below the new diagnostic cut-off point are welcome. The recommendation for close follow-up of those who fall below the diagnostic threshold, but may be at risk, addresses concerns that the use of HbA1c may “miss” some people with diabetes.

Undiagnosed diabetes is a serious problem.9 In England, one in four people with diabetes is estimated to be undiagnosed.10 Annual costs of diabetes to the NHS are estimated at £23.7bn (€29.5bn; $38.1bn) and projected to rise to £39.8bn by 2035.11 The WHO recommendations, now endorsed by the expert advisory group, aim to improve the detection of diabetes by making the process of testing easier. The recent report provides practical guidance about the use of the new tests and detailed advice on management, along with an educational programme to ensure widespread awareness of the new advice. This should help to avoid the many adverse consequences that might arise from inappropriate use of the HbA1c test for the diagnosis of diabetes.


Cite this as: BMJ 2012;345:e7293


  • Competing interests: The author has completed the ICMJE uniform disclosure form at www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author) and declares: no support from any organisation for the submitted work; no financial relationships with any organisations that might have an interest in the submitted work in the previous three years; no other relationships or activities that could appear to have influenced the submitted work.

  • Provenance and peer review: Commissioned; not externally peer reviewed.