Finally, a UK consensus on the use of HbA1c to diagnose diabetes



Cover image: Computer model of a glycated haemoglobin molecule. The alpha and beta subunits of the haemoglobin are blue and pink, and the iron-containing haem groups are grey. Glycated haemoglobin is formed when glucose in the blood binds to a haemoglobin molecule. Credit: Ramon Andrade 3dciencia/Science Photo Library.

Diabet. Med. 29, 1349 (2012)

Wikipedia defines consensus decision making as a group decision-making process that seeks consent, not necessarily agreement, of participants and the resolution of objections. This is a very apt definition to describe the UK consensus statement on the use of HbA1c to diagnose diabetes. Following an International Expert Committee Report in 2009, the American Diabetes Association in 2010 adopted HbA1c for the diagnosis of diabetes and a range of HbA1c to identify those at risk of progression to disease. In 2011 the World Health Organization recommended that diabetes could be diagnosed with an HbA1c ≥ 48 mmol/mol (≥ 6.5%), but failed to make a recommendation for those at risk. These events prompted the UK Department of Health Advisory on Diabetes to publish a brief summary of a consensus statement in 2011 in Practical Diabetes, but the full statement, published today (page 1350), took another year to ‘achieve the resolution of objections’. Nonetheless, and despite the delays, the committee, comprising all major stakeholders in the UK, should be congratulated on finally providing guidance to clinicians about the use of HbA1c in the diagnosis of diabetes. They have recommended that HbA1c of greater or equal to 48 mmol/mol (6.5%), rather than the fasting glucose or a glucose tolerance test, should be used to diagnose diabetes, unless the person is acutely symptomatic, there is a suspicion of Type 1 diabetes, pregnant, there are contra-indications to the use of HbA1c or a number of other conditions as listed in the consensus document. Furthermore, they have recommended that an HbA1c between 42 and 47 mmol/mol (6.0–6.4%) should be considered at high risk of diabetes. The use of HbA1c has a number of advantages over glucose measurement, which importantly include that the person does not need to be fasting for diagnosis.

Why has the issue of using HbA1c as a diagnostic test caused so much controversy and required 3 years before the UK finally published its consensus document? There is a misconception amongst some that the oral glucose tolerance test (OGTT) is the gold standard to diagnose diabetes and therefore any replacement test should identify the same group of patients; they do not and this is illustrated by the article from Bhansali and colleagues (page 1385 of this issue), as well as many other similar publications. What is often forgotten is that there is no precise cut-off point in the glucose distribution that defines diabetes. Furthermore, the same person can change diagnostic criteria if the oral glucose tolerance test is repeated; hardly the performance of a gold standard. Currently, we use a glucose measure (whether fasting glucose, oral glucose tolerance test or HbA1c) to predict the occurrence of macrovascular or microvascular complications of diabetes; as this is not bimodal, but a skewed continuous distribution, by definition no such method will ever be a gold standard. There is, however, a genuine fear that HbA1c might be used inappropriately under certain clinical circumstances; this particularly applies to the diagnosis of Type 1 diabetes or someone who is acutely unwell. Therefore, the tables and flow diagrams in the consensus statement published today in Diabetic Medicine require wide dissemination to put the use of HbA1c as a diagnostic marker of diabetes into the appropriate context (page 1350).

The adoption of HbA1c as a diagnostic marker cannot be currently worldwide, as some emergent countries, and in certain settings, the use of HbA1c may not be affordable. It is also important to stress that, if HbA1c is used as a diagnostic marker, the measurement needs to be performed in a properly accredited laboratory, operating to defined standards.

Finally, are there circumstances in which both HbA1c and glucose should ever be used together? In the majority of instances, the answer is no, as they do not necessarily identify the same groups. There may be exceptions to this rule. In this issue of Diabetic Medicine, Liu and colleagues present an interesting paper (page 1456) suggesting that the combination of HbA1c and glucose can be used to identify a group of high-risk patients at high risk of mortality after acute myocardial infarction; this will clearly require further research.