Emergency attendance for acute hyper- and hypoglycaemia in the adult diabetic population of the metropolitan area of Milan: quantifying the phenomenon and studying its predictors.

BACKGROUND: We quantified, among diabetic adults, the frequency, costs, and factors associated with visits to the emergency department (EDs) and subsequent hospitalizations for acute hypoglycaemic and hyperglycaemic events. METHODS: We included adults with diabetes residing in the Milan Agency for Health Protection between 2015 and 2017. From healthcare databases, we identified demographic variables, comorbidities, type of treatment, insulin treatment duration, previous ED attendances for acute glycaemic events, and two indicators of glycaemic monitoring. Using a validated ICD-9-CM coding algorithm, we identified all ED attendances for acute glycaemic events from the ED database and calculated their incidence. We computed the direct costs from health databases and presented them as average annual mean costs for those having had at least an ED attendance. The analysis of the association between the number of ED attendances and potential determinants was performed using zero-inflated negative binomial regression models. These two-part models concomitantly estimate two sets of parameters: the odds-ratios (ORs) of having no attendances and the incidence rate ratios (IRRs) of attendance. RESULTS: The cohort included 168,285 subjects, 70% of subjects were older than 64 years, 56% were males, and 26% were treated with insulin. The incidence of acute glycaemic events for those attending the ED was 7.0 per 1000 patient-years, followed by hospitalization 26.0% of the time. The total annual direct cost for ED attendances due to acute glycaemic events was 174,000 €. Type of antidiabetic treatment had the strongest association with ED attendances for hypoglycaemia. Patients assuming insulin only had a lower probability of having no attendances (OR compared to those who assumed non-insulin antidiabetic drugs =0.01, 95% CI = 0.00-0.02). These patients also had the highest rate of hyperglycaemic episodes (IRR = 7.7, 95% CI = 5.1-11.7 for insulin only vs. non-insulin antidiabetic drugs). Subjects having had a previous episode of the same type leading to an ED visit had a higher rate of subsequent attendances (IRR for hypoglycaemia = 5.3, 95% CI = 3.9-7.3 and IRR for hyperglycaemia = 3.7, 95% CI = 1.3-10.2). CONCLUSION: Insulin treatment and having had a prior acute glycaemic event leading to an ED visit were major predictors of ED attendance for hyper and hypoglycaemia in a population of adults with diabetes.