| Abstract: | We prospectively studied 45 central venous catheters to determine whether Gram''s staining and culture of skin swabs from the entry site could be used to predict catheter-related infection. Data were collected from insertion site swabs, intracutaneous and intravascular catheter segments, and blood cultures. Surveillance site cultures at the time of dressing changes showed that bacterial growth, once established, persisted until removal of the catheter but that the time of onset of infection was not predictable. Gram''s staining alone and Gram''s staining combined with culture were tested for their ability to predict catheter colonization and catheter-related infection. Bacteria seen with Gram''s staining invariably denoted catheter colonization. When bacteria were not seen with Gram''s staining, positive results of culture did not change the pretest probabilities of colonization or infection; however, negative results of culture reduced the probability of colonization to low levels (likelihood ratio less than 0.06). We conclude that Gram''s staining and culture of skin swabs from the entry site provide, without line removal, a simple, inexpensive and practical test for the diagnosis of catheter-related infection. |
