LAUSR

Purpose: To investigate the relationship between dysglycemia and hospital mortality in patients with and without a preadmission diagnosis of insulin treated diabetes mellitus (ITDM). Materials and methods: An observational multicentre cohort study using the ANZICS‐APD database on adult patients admitted to ICU with sepsis between January 1st 2006 and December 31st 2015. Four domains of dysglycemia were investigated (highest, mean and lowest blood glucose levels and glycemic variability: the absolute difference between the highest and lowest level). The association between a preadmission diagnosis of ITDM and hospital mortality in each domain was analysed. Results: We studied 90,644 septic patients including 5127 patients with ITDM. We found that septic ICU patients with ITDM have lower adjusted hospital mortality with higher peak blood glucose levels in the first 24h while non‐ITDM patients have increased mortality (interaction p 0.012). We found that this significant difference was replicated when assessing glycemic variability (interaction p 0.048). Conclusions: Septic patients with a pre‐existing diagnosis of ITDM show a different relationship between hospital mortality and highest glucose levels and glycemic variability in the first 24h than those without ITDM. These findings provide a rationale for an ITDM‐specific approach to the management of dysglycemia. HIGHLIGHTSA pre‐admission diagnosis of insulin treated diabetes mellitus (ITDM) modified the relationship between dysglycemia and mortality in over 90,000 ICU patients with sepsisIn patients with a pre‐admission diagnosis of ITDM, a higher peak blood glucose in the first 24 hours was associated with lower mortality.In patients without a pre‐admission diagnosis of ITDM, hospital mortality increased as peak blood glucose in the first 24 hours increased by quintile.A diabetic‐specific approach to the management of blood glucose levels in the ICU may be justified