September 1987

Rapid Diagnosis of Intravascular Catheter-Related Sepsis

Author Affiliations

From the Bacteriology Department, Westmead (Australia) Hospital. Dr Collignon is now with the Infectious Diseases Unit, Royal Canberra and Woden Valley Hospitals, Canberra, Australia.

Arch Intern Med. 1987;147(9):1609-1612. doi:10.1001/archinte.1987.00370090085015

• The use of Gram-stained "impression smears" of the external surface of intravascular catheters for rapid detection of catheter-associated infection was studied. Gram's stain results of 322 catheters were correlated with clinical episodes of systemic sepsis and semiquantitative cultures of the catheters. Organisms were seen on Gram's stain of 82 catheters, 37 of which were positive on semiquantitative cultures (≥15 colonies per plate). Catheter-related bacteremia occurred on three occasions. All three catheters showed numerous organisms on Gram's stain, although one was negative on semiquantitative culture. All five catheters, in place during bacteremic episodes that were unrelated to catheter infection, were negative on Gram's stain. If the presence of any organisms on Gram's stain was taken as a positive test result, the sensitivity of Gram's stain in predicting the result of semiquantitative culture was 83%, the specificity was 81%, and the predictive value of a positive and negative culture was 44% and 96%, respectively. Slides took two to five minutes to examine microscopically. Gram-stained impression smears of intravenous catheters can be made by a simple, inexpensive, and rapid technique that is accurate in diagnosing catheter-related infection. However, in this study in which a relatively low prevalence of catheter-related bacteremia occurred, the positive predictive value of the Gram's stain result in the diagnosis of catheter-related bacteremia, in contrast to catheter colonization, was low. Only in a patient group with a high prevalence of catheter-related bacteremia would the test be likely to have a high positive predictive value. Thus, selectivity should be exercised in the application of this method.

(Arch Intern Med 1987;147:1609-1612)