Glycosylated haemoglobin (HbA1c) and cortisol levels on admission to intensive care as predictors of outcome

Abstract

Objective: To evaluate the predictive value of glycosylated haemoglobin and cortisol on admission, in critical care patients. Design: Prospective, observational, single centre study. Setting: 14 bedded Intensive care unit of a tertiary-level university hospital. Patients: 124 consecutive emergency medical and surgical patients. Methods: Data collected on admission included patient demographics, medical history, medication, diagnosis, type of nutrition, TISS28 score, serum blood glucose, Glycosylated haemoglobin (HbA1c), cortisol, mean arterial blood pressure, and the use of inotropes in the first 24hrs. Daily baseline tests included complete blood count, urea and electrolytes, creatinine, twice weekly liver function tests. The primary outcome measure was intensive care unit mortality. Secondary outcome measures were ITU stay, days of ventilation, peak urea, peak creatinine, lowest platelet count, peak bilirubin, lowest Pa/FiO2, and the number of transfusions. Measurements and results: 124 patients (mean age 56.2 years SD 23.2) were included. Regression analysis was used to identify any potential predictors of outcome: HbA1c levels on admission were not found to be significantly associated with mortality (p=0.51), or any other secondary endpoints listed above. However, subgroup analysis revealed a predictive role of HbA1c with regards to length of ITU stay (p= 0.01) and number of days of ventilation (p=0.007) in those patients with a history of diabetes. Glucose level on admission emerged as an independent marker of mortality (p=0.009). Conclusions: This study suggests that HbA1c may not be a predictor of outcome in the general ITU population but may be of predictive value in diabetic ITU patients. On the other hand, blood glucose levels on admission emerged as a predictor of mortality, whilst no association was found between HbA1c and cortisol levels on admission.