Diagnosis of catheter-related bloodstream infections among pediatric oncology patients lacking a peripheral culture, using differential time to detection

Background: Current methods for in situ diagnosis of catheter-related bloodstream infections require concurrent collection of central venous catheter (CVC) and peripheral vein (PV) blood cultures. Both the pain and inconvenience of PV cultures are undesirable. Methods: A prospective study was conducted (August 2002 to March 2004) to assess the accuracy of diagnosing catheter-related bloodstream infections based on the difference in time to detection of blood cultures drawn concurrently from 2 lumens of a multilumen CVC. This difference in time to detection between 2 lumens was compared with results of the standard criterion with paired CVC and PV blood cultures. Results: Twenty-one infectious episodes were categorized as catheter-related bloodstream infections and 38 as non-catheter-related bloodstream infections. With a cutoff in difference in time to detection between 2 lumens of 180 minutes, the sensitivity of this test to diagnose a catheter-related bloodstream infection was 61% (95% confidence interval, 39-80%) and the specificity was 94% (95% confidence interval, 82-99%). In 4 of 7 episodes with false-negative results, the colony counts in cultures from both lumens were > 400 colony-forming units/mL (maximal value reported), indicating the limitation of this method when both lumens of the catheter are colonized. With the pretest probability of catheter-related bloodstream infections ranging from 28% to 54%, the positive predictive value of a difference in time to detection between 2 lumens of >= 180 minutes for diagnosis of catheter-related bloodstream infections ranged from 81% to 93% and the negative predictive value ranged from 67% to 86%. Conclusion: Within the context of its limitations, this novel method provides an alternative for diagnosing catheter-related bloodstream infections among patients with a CVC, without PV cultures.