Skip navigation
Link to Back issues listing | Back Issue Listing with content Index | Subject Index

Catheters, infections, and dressings

Systematic review

Infection in hospital, especially in people more seriously ill, is a major problem. Central venous catheters frequently cause bloodstream infection, with up to half a million annually in the USA. Between one in four and one in five bloodstream infections results in death. Epidural catheters also have problems, with deep epidural infection and permanent neurological consequences following epidural abscess. Any intervention that can reduce catheter infections could be of value in helping to minimise serious hospital-acquired infections.

Chlorhexidine gluconate is widely used as a surgical scrub and skin disinfectant, and is now available as a dressing that releases chlorhexidine onto the underlying skin surface over a 10 day period when placed over a catheter exit site. A new meta-analysis indicates that it is likely to be of benefit [1].

Systematic review

Three major electronic databases were searched to the end of 2005 for randomised trials comparing chlorhexidine-impregnated dressings with placebo or povidine-iodine dressings, as well as reference lists. Outcomes were the proportion of patients with exit-site or catheter colonisation with bacteria, and systemic infections like bloodstream or central nervous system infection related to a vascular or epidural catheter.


Eight randomised trials reported on 2,558 patients. Two trials (112 patients) reported on epidurals, and six (2,446) on vascular catheters. None of the trials were blind. All trials used what was essentially a placebo dressing as a comparator, except one trial that used twice-weekly povidine-iodine dressing compared with weekly chlorhexidine dressing. The duration of catheter use varied; usually it was less than a week, but was 17 days on average in one neonatal intensive care study and 67 days on average for patients with tunnelled intravascular catheters for chemotherapy.

The results for bacterial colonisation of exit site or catheter are shown in Figure 1. Overall, the colonisation rate was 27% with control and 14% with chlorhexidine-impregnated dressing. The relative risk was 0.5 (0.45 to 0.62), with a number needed to treat to prevent one colonisation of 8 (6 to 10).

Figure 1: Results of individual trials for exit site or catheter colonisation. Dark symbols indicate epidural catheters, half-tone symbol represents povidine control

The results for bloodstream or central nervous system infections are shown in Figure 2. Overall, the rate was 3.8% with control and 2.3% with chlorhexidine-impregnated dressing. The relative risk was 0.6 (0.37 to 0.92), with a number needed to treat to prevent one colonisation of 64 (34 to 500). In the four trials which estimated bloodstream infections for vascular catheters only compared with placebo dressing (light symbols in Figure 2), the infection rates were 4.0% and 1.8%, the relative risk 0.4 (0.24 to 0.79), and the number needed to prevent one infection was 44 (26 to 148).

Figure 2: Results of individual trials for bloodstream or central nervous system infections. Dark symbols indicate epidural catheters, half-tone symbol represents povidine control


Clearly this is not a straightforward analysis. It is complicated by considerable clinical heterogeneity, particularly in the types of catheter, patients and circumstances (epidural, vascular; paediatric and adult; surgery, intensive care, cancer treatment). No trial was blind, and not all of them clearly indicated that the results were intention to treat. Moreover, even in total the numbers are limited, with no new trials published since the review to bolster them.

Having said that, the results show a reasonable degree of consistency for a less harmful outcome (exit site or catheter colonisation) and more harmful (bloodstream infection or central nervous system infection) outcome. Catheter-related bloodstream infections are expensive to treat (getting on for £17,000), and have a high mortality. The cost of each chlorhexidine-impregnated dressing is about £2, so even using the upper confidence interval of the NNT to prevent one bloodstream infection (148) indicates a potential for significant cost saving. Spending £300 to save £17,000 and save lives would look good on any health economic analysis, and certainly makes chlorhexidine-impregnated dressing worthy of consideration.


  1. KM Ho, E Litton. Use of chlorhexidine-impregnated dressing to prevent vascular and epidural catheter colonization and infection: a meta-analysis. Journal of Antimicrobial Chemotherapy 2006 58: 281-287.

previous story