Abstract

AIMS: To determine the accuracy of a novel endoluminal brush method for the diagnosis of catheter related sepsis (CRS), which is performed in situ and hence does not require line sacrifice. METHODS: 230 central venous catheters in 216 patients were examined prospectively for evidence of CRS or colonisation using an endoluminal brush method in conjunction with peripheral blood cultures. The results were compared with those obtained using methods that require line sacrifice: extraluminal sampling (Maki roll) or endoluminal sampling (modified Cleri flush) of microorganisms. RESULTS: Only 16% of 128 patients suspected clinically of having line associated infection were confirmed as having CRS. In addition, 2 of 102 patients not suspected of having line associated infection had CRS. Line colonisation was apparent in approximately twice as many catheters using the Maki roll criteria (92%) compared with either the endoluminal brush (43%) or Cleri flush (43%). Furthermore, colonised catheters sampled using the Maki roll technique yielded mixed growth twice as often as when examined by endoluminal methods (17 and 8 cases, respectively). It was rare to detect either only endoluminal (4 of 22 episodes) or extraluminal (1 of 22 episodes) microorganisms in cases of CRS. In contrast, catheters defined as being colonised most frequently (59% of episodes) yielded only significant extraluminal growth. Only one case of CRS (5%) would have been "missed" if lines yielding a negative result from endoluminal brush sampling had been left in situ. Conversely, four episodes of CRS (18%) would not have been diagnosed by relying on extraluminal sampling alone. CONCLUSIONS: Diagnosis of CRS by the endoluminal brush method can be achieved without line sacrifice and is more sensitive (95%) and specific (84%) than extraluminal sampling of the catheter tip by the Maki roll technique (82% and 66%, respectively).