Aim To triangulate three data sources and report the characteristics and long-term outcomes of patients with diabetes requiring emergency department (ED) care for hypoglycaemia.

Method Three data sources were used—ambulance electronic records, hospital episode statistics and patient administration system. Hypoglycaemia (capillary blood glucose <4.0 mmol/L)-related attendances to a single hospital's ED between 1 April 2012 and 31 March 2013 were studied.

Results Using the three sources, there were 165 hypoglycaemia-related attendances in 132 patients with diabetes [type 1–59 episodes in 43 patients, type 2–106 episodes in 89 patients (therapy—54 (51%) insulin, 35 (33%) sulfonylurea, 11 (10%) both, 6 (6%) others)]. At best only 65% of episodes would have been identified were a single data source used. Patients with type 2 vs type 1 diabetes were older (median age 79 vs 61 years, p<0.0001), had more comorbidities (median Charlson comorbidity index (CCI) 4 vs 3, p=0.002) but no difference in HbA1c (median 7.8% vs 8.4%, p=0.065). Compared with insulin-treated type 2 patients with diabetes, sulfonylurea-treated patients (33%) were older (median age 82 vs 76 years, p=0.007), had worse renal function (median estimated glomerular filtration rate 38 vs 56 mL/min/1.73 m², p=0.019) and lower HbA1c (median 6.7% vs 8.4%, p<0.0001). At least 17 (10%) hypoglycaemic episodes resulted in additional serious harm. The 30-day, 90-day and 1-year all-cause mortality were 10.6% (14), 16.7% (22) and 28% (37), respectively. Age, CCI and hospitalisation were risk factors for long-term mortality.

Conclusions Dependence on a single data source would have at best identified only 65% of episodes. One-third of episodes were sulfonylurea related in patients with type 2 diabetes, and one-fourth of all patients with diabetes who required ED care for hypoglycaemia died the following year.