Objective: To investigate the role of diabetes as risk factor for ICU-acquired bloodstream infections (BSI).
Design: Prospective observational study.
Setting: A general eight-bed ICU of a tertiary hospital.
Patients: Three hundred and forty-three consecutive patients (63 diabetic and 280 nondiabetic) admitted in the ICU.
Methods: BSI episodes in the ICU were recorded and classified as primary, secondary, catheter-related and mixed according to strict criteria. In all patients, blood glucose was strictly controlled with a continuous insulin infusion within a range of 80-120 mg/dl.
Results: One-hundred and eighteen patients (34.4%) developed at least one BSI episode. Diabetic patients had an increased probability of developing at least one BSI episode compared with nondiabetic patients (hazard ratio = 1.66, 95% confidence interval 1.04-2.64, P = 0.034) in a Cox proportional hazards regression model adjusting for age, gender, admission category and APACHE II score at admission in the ICU and comorbidities.
Conclusions: Despite strict glycemic control, diabetic patients have a 1.7-fold probability of developing an ICU-acquired BSI compared to nondiabetic subjects.