Glucose Control in Critical Care Oncology

Abstract

The optimal strategy for glucose control in critically ill patients remains controversial and may vary depending on their diagnostic groups. This retrospective study addresses this question in a cohort of critically ill oncology patients. Serial blood glucose levels were measured in a specialist oncology intensive care unit (ICU) in England between January 2009 and May 2010. Intravenous sliding scale insulin was started when blood glucose levels were greater than 8.3 mmol/L, aiming for a target glucose level below 10 mmol/L. There were 565 patients admitted to the ICU, of whom 181 (32%) were medical and 384 (68%) surgical. The mean blood glucose was 7.8 mmol/L. Mortality rates relative to mean glucose levels followed a U-shaped curve. There was a statistically significant increase in mortality among patients who spent longer periods with blood glucose levels less than 6 mmol/L, and those whose glucose was greater than 10 mmol/L (20% vs 7.3%; p <0.001). Fluctuations in blood glucose levels measured by the coefficient of variation against mean glucose levels demonstrated a significant increase in mortality in patients with higher variability (p<0.01). Higher coefficients of variation were also associated with an increased length of ICU stay and increased readmission rates to ICU. In this study, hypoglycaemia, hyperglycaemia and increased variability in glucose levels were each independently associated with worse outcomes in critically ill patients.