Stress hyperglycaemia and mortality in critical illness

None listed

Hyperglycaemia in hospitalised patients is independently associated with increased morbidity and mortality in a wide range of patient groups, including the critically ill. The association between hyperglycaemia and poor inpatient outcomes is strong in patients without diabetes. However, despite a greater elevation in plasma glucose concentration and poorer prognosis, hyperglycaemia is a weaker predictor of morbidity and mortality in patients with diabetes. A high plasma glucose concentration in a hospitalised patient can occur because of chronic poor diabetes control and be “normal” for that patient, represent a transient physiologic response to an inter­current illness (stress hyperglycaemia), or be a combination of the above. Stress hyperglycaemia arises because of the inflammatory and neuro­hormonal derangements that occur during a major illness. Therefore, stress hyperglycaemia develops in proportion to the severity of an inter­current illness, and an association between hyperglycaemia and adverse patient outcomes will be, at least in part, a reflection of this. However, stress hyperglycaemia may also directly contribute to adverse outcomes by inducing endothelial dysfunction and oxidative stress. The Endocrine Research Unit, Southern Adelaide Diabetes and Endocrine Services recently developed a metric for stress hyperglycaemia. Estimated average glucose concentration was calculated from glycosylated haemoglobin in 2290 patients acutely admitted to Flinders Medical Centre, Adelaide. Relative hyperglycaemia was defined by the stress hyperglycaemia ratio (SHR), calculated by dividing admission glucose by estimated average glucose. Thus, a patient with a SHR of 1.5 has an admission glucose concentration 50% higher than their average glucose over the prior 3 months. Using data gathered from a research primarily focused on validating the use of HbA1c for diagnosis of diabetes mellitus, a multivariable analysis including glucose, SHR and other potential predictors of poor outcome (e.g age, sex, renal function), the odds ratio for in­-hospital death or ICU admission per 0.1 SHR increment was 1.20 (p<0.001). In contrast, the odds ratio per one mmol/L glucose increment was 1.03 (p=0.31). In contrast to glucose, the association between diabetes and SHR was not affected by diabetic status. These data suggest that SHR is a better predictor of adverse outcomes than glucose and that SHR is associated with poor outcomes in patients with and without diabetes. In this study, we will evaluate whether SHR is more strongly associated with in-­hospital mortality in critically ill patients than absolute glucose. If true, this may create a new paradigm whereby patients in ICU are selected for glucose­lowering therapy based on relative, rather than absolute, hyperglycaemia.