Abstract
INTRODUCTIONGlycaemic control associates with better outcomes for hospitalised patients. Whether GLP-1 receptor agonists (GLP-1 RA) are suitable and effective drugs for inpatients is unclear. METHODSA retrospective, single centre, observational study using data from the electronic health record. Patients admitted using GLP-1 RA as outpatients, from 2016 to 2019, were identified. Outcomes were compared to those admitted using twice-daily (BD) mixed insulin. Capillary glucose, medication use, creatinine, and demographic data were collected. As drugs may be discontinued/not administered in hospital, days when GLP-1 RA was administered were 'GLP-1 RA active' and, for insulin, 'insulin active'. The primary comparison was rate of hypoglycaemia (<4 mmol/L) and severe hypoglycaemia (<3 mmol/L). A logistic regression model examined variables for hypoglycaemia. RESULTSGLP-1 RA comprised n = 262 admissions and BD insulin n = 166. The 'insulin active' cohort (n = 957 patient days) had higher risk of hypoglycaemia than 'GLP-1 RA active' (n = 806 days); occurring on 14.7% of days; 95% confidence interval [CI] 12.6-17.1 versus 9.9% days; 95% CI 8.0-12.2; p = 0.002, and severe hypoglycaemia 4.0% of days (95% CI 2.8-5.4) versus 2.0% (95% CI 1.1%-3.2%; p = 0.005). Daily glucose (mean ± standard deviation) was 10.8 ± 5.2 mmol/L in insulin active versus 9.6 ± 4.7 mmol/L in GLP-1 RA active; p < 0.001. Insulin use, age, and acute admissions predicted hypoglycaemia. The odds ratio for hypoglycaemia was 2.15 times greater (95% CI, 1.14-4.08; p = 0.019) with insulin than with GLP-1 RA. CONCLUSIONSGLP-1 RA provided better glycaemic control than BD mixed insulin and should be continued during hospitalisation unless there is a clear indication for cessation.