To estimate the frequency and morbidity of insulin-induced hypoglycaemia, a retrospective survey was undertaken of the frequency of severe hypoglycaemia in 600 randomly selected patients with insulin-treated diabetes who were attending a large diabetic outpatient clinic in a teaching hospital. The resulting morbidity (hypoglycaemia-related injuries, convulsions, and road traffic accidents) was ascertained in 302 patients. One hundred and seventy-five (29.2 %) of the 600 patients reported a total of 964 episodes of severe hypoglycaemia in the preceding year, giving an overall frequency for the group of 1.60 episodes patient-1year-1. The frequency of severe hypoglycaemia which was documented in 544 Type 1 (ketosis prone) diabetic patients was double that observed in a subgroup of 56 Type 2 diabetic patients who were being treated with insulin (1.70 vs 0.73 episodes patient-1year-1). In the subset of 302 patients, those who had experienced severe hypoglycaemia had greater morbidity associated with an estimated rate of injury of 0.04 injuries person-1year-1. Twenty (6.6 %) patients reported a total of 37 convulsions associated with hypoglycaemia, 5 of which had occurred in the preceding year (0.02 convulsions person-1year-1). Five patients reported road traffic accidents in the preceding year which had been caused by hypoglycaemia. The only reliable predictors of severe hypoglycaemia were a history of previous severe hypoglycaemia (p < 0.001), a history of hypoglycaemia-related injury (p < 0.001 ) or convulsion (p < 0.001), and the duration of insulin therapy (p < 0.001). Those patients with a history of severe hypoglycaemia had been treated with insulin for longer (17.4 vs 14.3 years, p < 0.02) and tended to have a lower mean glycated haemoglobin concentration. This study confirmed that severe hypoglycaemia is common and demonstrated that it is associated with significant morbidity in a conventionally treated, insulin-requiring, diabetic population. The multifactorial aetiology of severe hypoglycaemia confounds attempts to predict the individual patient 'at risk'.
