Changing the glucose cut-off values that define hypoglycaemia has a major effect on reported frequencies of hypoglycaemia

Aims/hypothesis The aim of this analysis was to quantify the relationship between the frequency of hypoglycaemia and various glucose cut-off points for the definition of hypoglycaemia, within a range of HbA(1c) strata. Methods Data from two trials examining insulin glargine dose titration in 12,837 type 2 diabetic participants starting insulin therapy were combined. Curves for hypoglycaemia frequency plotted against endpoint HbA(1c) level were constructed, using a range of glucose cut-off points for hypoglycaemia. Results During the 12-week study period, 3,912 patients recorded 21,592 hypoglycaemic episodes, comprising 242 severe, 8,871 symptomatic and 12,479 asymptomatic events, corresponding to hypoglycaemia event rates of 0.10, 3.8 and 5.3 events per patient year. Increasing the hypoglycaemia cutoff point from, for instance, <3.1 to <3.9 mmol/l more than doubled the percentage of affected patients, e. g. from 17.7 to 43.3% at HbA(1c) 7.0-7.2%. At higher hypoglycaemia cut-off points the proportion of patients having only asymptomatic hypoglycaemia increased, e. g. from 30.7% at <3.1 mmol/l to 61.7% of patients at a cut-off point of <3.9 mmol/l. In sensitivity analysis, 121 of 1,756 patients with at least one self-monitored blood glucose value <3.1 mmol/l experienced severe hypoglycaemia, compared with 149 of 3,912 patients with a self-monitored blood glucose level of <3.9 mmol/l. Thus, to identify 28 more patients with severe hypoglycaemia, the number of patients experiencing only non-severe hypoglycaemia more than doubled. Conclusions/interpretation The glucose cut-off point defining hypoglycaemia greatly affects the reported frequency of hypoglycaemia. When hypoglycaemia is to be defined by a predetermined glucose level, to have clinical relevance the cut-off should be set at a lower level than the threshold of 3.9 mmol/l proposed by the American Diabetes Association.