EDITOR, - David R McCance and colleagues state that the two hour plasma glucose, glycated haemoglobin, and fasting plasma glucose concentrations are equivalent predictors of the development of retinopathy and nephropathy.1 They therefore consider all three glycaemic variables to be equivalent for diagnosing diabetes. This conclusion should be viewed with caution.

With regard to the sensitivity and specificity of glycaemic variables used, their roles in predicting microangiopathy and diagnosing diabetes need to be clearly distinguished. On the basis of data presented in tables I and II in the authors' paper (table), the oral glucose tolerance test has the highest sensitivity in diagnosing diabetes.

It would be important to know whether all three glycaemic variables in an individual subject had identical centile positions in the distributions of the variables used. This requirement will be fulfilled if there is a positive linear correlation between the two hour glucose and glycated haemoglobin concentrations in an individual. No evidence can be found about this in McCance and colleagues' paper. This correlation is more likely in patients with poor metabolic control than in those with strict control. If the centile positions of the three glycaemic variables for an individual subject differ this means that the prediction of microangiopathy varies as a function of the glycaemic variables used.

The Diabetes Control and Complications Trial Research Group has documented that better glycaemic control can delay the onset and slow the progression of microangiopathy in patients with insulin dependent diabetes mellitus.2 Better control implies early diagnosis and treatment of diabetes. Since the oral glucose tolerance test has the highest sensitivity in detecting patients with impaired glucose homoeostasis it will continue to be the preferred method for diagnosing diabetes.