Background: Catheter-related bloodstream infections (CRBIs) are frequent complications of the use of long term central venous catheters (CVCs). Comparative quantitative culture of blood obtained via the CVC and a peripheral vein (PV) is a well-accepted method of diagnosing CRBI; however, an alternative definition for use when a PV culture is not available is desirable.
Methods: A computerized search of patient records identified all positive blood culture results from the St. Jude Children's Research Hospital Microbiology Laboratory between January 1996 and May 2001. Demographic data, catheter information and culture results were abstracted. Sensitivity, specificity, positive predictive value (PPV), and likelihood ratio were calculated for 2 alternative definitions of CRBI.
Results: Review of the medical records revealed 136 episodes of bacteremia that were evaluable for alternative definition 1 and 241 episodes that were evaluable for alternative definition 2. In patients with a double lumen CVC, CRBI can be diagnosed by a > or = 5-fold difference in colony-forming units/mL between the 2 lumens (alternative definition 1) with sensitivity, specificity, PPV and likelihood ratio of 61.8, 93.3, 92.2 and 9.22, respectively. In patients with a single or double lumen CVC, CRBI can be diagnosed when the CVC culture yields > or = 100 colony-forming units/mL (alternative definition 2) with sensitivity, specificity, PPV and likelihood ratio of 75.5, 69.1, 79.3, and 2.44, respectively.
Conclusions: Our study suggests that comparison of colony counts from 2 lumens of a double lumen catheter is acceptable for diagnosis of CRBI when a PV culture is not available. Further validation is needed before discontinuing the recommendation to obtain a PV culture.