Aims/hypothesis: The main disadvantage of intensive treatment in the Diabetes Control and Complications Trial (DCCT) was an increased risk of hypoglycaemia that was not explained by the difference in HbA(1c) values alone. This study re-analysed DCCT data to establish whether mean blood glucose (MBG) and/or glucose variability add to the predictive value of HbA(1c) for hypoglycaemia risk in type 1 diabetes.
Methods: The times to first and subsequent severe hypoglycaemic events were compared with MBG, HbA(1c) and within-day SD of blood glucose using Cox regression after adjusting for other known risk factors for hypoglycaemia.
Results: On its own, the incidence of time to first hypoglycaemic event increased 1.05-fold for each 1 mmol/l decrease in MBG and 1.07-fold for every 1 mmol/l increase in glucose SD. MBG and SD of blood glucose also both added to the ability of HbA(1c) to predict repeated hypoglycaemic events: after adjusting for HbA(1c), a 1 mmol/l increase in SD was associated with a 1.09-fold increased risk of a first event, increasing to a 1.12-fold risk of a fifth event. A 1 mmol/l fall in MBG added a constant 1.02-1.03-fold risk of repeated events. Daytime events were predicted more accurately than nocturnal episodes.
Conclusions/interpretation: This study has established that HbA(1c), MBG and glucose variability measurements each have an independent role in determining an individual's risk of hypoglycaemia in type 1 diabetes. All three aspects of glycaemic assessment should thus be considered in patients in whom hypoglycaemia is a real or potential problem.