There was a time when making a diagnosis of diabetes was relatively easy. If the fasting plasma glucose was above 7mmol/L or the random glucose (two-hour value on oral glucose tolerance test) was above 11.1mmol/L on one occasion in a symptomatic patient or two occasions in an asymptomatic patient, you had your diagnosis.
However, HbA1c has now landed in the surgery as a diagnostic test. In principle, this brings advantages - it requires no fasting or oral glucose testing and is much more easily interpreted, with a single diagnostic cut-off (48mmol/mol).
But its arrival has also created uncertainty. Is it applicable in all situations? No. Is there still a role for fasting/random glucose testing? Yes. Will each test pick up the diagnosis with equal reliability? Maybe not. How do I know when I should do an HbA1c and when I should do a plasma glucose? We'll come to that.
In most situations, a one-off HbA1c >48mmol/mol (6.5% in 'old money') is sufficient to make a diagnosis of diabetes mellitus. Obviously, it does not tell you what type of diabetes the patient has, but statistically, it will be type 2 in 90% of cases.
For the exceptions, you use clinical acumen to recognise the young, slim, ketotic type 1 patient, the pregnant gestational diabetic, the emaciated malabsorbing chronic pancreatitis patient and so on.
Beneath this simplicity, there is a layer of important caveats.
An osmotically symptomatic, hyperglycaemic, ketotic, likely type 1 diabetes candidate may have developed hyperglycaemia so rapidly, their HbA1c has not had time to cross the diagnostic threshold.
The HbA1c timescale is related to slowly progressing, non-enzymatic covalent glycosylation of the Hb molecule, the slow turnover time of Hb and the relatively long lifespan of red blood cells.
A normal HbA1c does not rule out the diagnosis of diabetes in the presence of symptoms and hyperglycaemia. This is the first instance where there is still a role for plasma glucose testing, and in this sort of presentation, HbA1c would be the wrong test.
Gestational diabetes (GDM) is a distinct entity and is the next example of where HbA1c is the wrong test.
There is, separately, much debate about how one should diagnose GDM, with the American Diabetes Association, NICE, WHO and the International Association of the Diabetes and Pregnancy Study Groups all joining in the discussion.
Local antenatal clinics nearly always have their own policy on the preferred test for GDM, but a combination of lack of sensitivity and limitations around the duration/speed of onset of hyperglycaemia mean HbA1c is not the right test for diagnosing GDM.
Haemoglobinopathy is another example of where HbA1c testing for the diagnosis of diabetes has its limitations. Sometimes the patient will have an undiagnosed haemoglobinopathy and there is little that can be done other than to treat with caution a non-diagnostic HbA1c where your index of clinical suspicion is high.
However, in cases where you are aware of a pre-existing diagnosis of haemoglobinopathy, it may be prudent to fall back on fasting glucose, random glucose or even a formal oral glucose tolerance test.
Iron deficiency represents a particular challenge to the use of HbA1c as a diagnostic test in primary care.
The DH advisory committee on the use of HbA1c recommends that it is not used in the presence of iron deficiency. The prevalence of iron deficiency in the older primary care population is relatively high, which begs the question: what level of iron deficiency is significant?
The answer is that this is not clearly delineated, so at present, the policy of perfection has to be to avoid using HbA1c as a diagnostic test in the presence of any known degree of iron deficiency.
In addition to these three circumstances, there is a well-recognised list of other situations where HbA1c should not be used for diagnosis and plasma glucose based methodologies should be employed. These are:
- Children and young people.
- Patients who have suspected diabetes of less than two months' duration.
- Patients at high risk of diabetes who are acutely unwell.
- Those taking medications that can cause rapidly worsening glycaemia (for example, steroids).
- After acute pancreatic damage (pancreatitis, pancreatic surgery).
- Other factors influencing Hb dynamics that may invalidate the test, such as recent transfusion, renal failure and haematological illness.
The other matter that arises with a switch from plasma glucose based diagnosis to HbA1c based diagnosis is what to do with those who previously would have been diagnosed with impaired fasting glucose or impaired glucose tolerance.
The short answer is that such 'pre-diabetes' still exists by HbA1c criteria, but is now characterised by HbA1c values of 42-48mmol/mol.
HbA1c has replaced plasma glucose as the first-line test for the diagnosis of diabetes. However, there are circumstances in which it should not be used and these are largely to do with a rapidly rising glucose level that has not yet translated into increased Hb glycosylation (for example, type 1 diabetes) or circumstances that alter Hb turnover.
Plasma glucose is still a perfectly acceptable test and is the first-line test where HbA1c is not recommended. The diagnostic cut-off to be used for HbA1c is 48mmol/mol and for plasma glucose is 7mmol/L fasting and 11.1mmol/L random or two- hour post glucose load in a 75g oral glucose tolerance test.
- Dr Turner is a consultant diabetologist at the Norfolk and Norwich University Hospitals NHS Foundation Trust. He is the author of www.diabetesbible.com, which offers doctors and nurses free practical guidance on diabetes diagnosis and management.