RESUMO
OBJECTIVE: To compare the accuracy of three techniques that do not require central venous catheter removal to diagnose catheter-related bloodstream infection. DESIGN: Prospective cohort study of central venous catheters from suspected cases of catheter-related bloodstream infection. SETTING: University teaching hospital. PATIENTS: One hundred and twenty-five central venous catheters from patients with suspected catheter-related bloodstream infection (a raised peripheral white blood cell count, temperature >37 degrees C, and/or local signs of infection at the catheter skin entry site) in intensive care and surgical patients in a large teaching hospital were assessed. INTERVENTIONS: None. MEASUREMENTS: Three techniques were compared: the differential time to positivity of central venous catheter vs. peripheral-blood cultures, quantitative culture of central venous catheter vs. peripheral blood, and the endoluminal brush with peripheral blood culture. MAIN RESULTS: Central venous catheters with a median dwell time of 11 days were examined. There were 36 episodes of catheter-related bloodstream infection, defined as a positive result from at least two of the three tests in the presence of a peripheral blood culture growing the same microorganism and without an identifiable alternative source of sepsis. The sensitivities of the endoluminal brush, quantitative culture, and differential time to positivity techniques were 100%, 89%, and 72%, respectively, with corresponding specificities of 89%, 97%, and 95%. Blood could be directly aspirated from only 231 of 312 (74%) lumens. In the 20 cases of catheter-related bloodstream infection associated with multiple-lumen central venous catheters, endoluminal brushing was positive for one, two, and three lumens in nine (45%), six (30%), and five (25%) cases, respectively. CONCLUSIONS: All three techniques had relatively high sensitivity. However, inability to obtain samples via central venous catheters is a major drawback of the differential time to positivity and quantitative blood culture approaches. Differential time to positivity is simple to perform and has high specificity and therefore could be used as a first line approach, with the endoluminal brush reserved for cases where blood cannot be obtained. All lumens of multiple-lumen central venous catheters must be sampled to ensure maximal sensitivity.