To the Editor: Bloodstream infections are frequent in healthcare settings and cause significant mortality and morbidity. 1,2 Most of these infections are caused by intravenous catheters, particularly central venous catheters (CVCs). Over 250 000 catheter-related bloodstream infections occur annually in the United States, 1 and over 3000 in Australia.2 Many CVCs are retained when no longer essential. For example, a recent one-day audit in a US teaching hospital found that 15% of CVCs (11/74) were “unjustified” most of these had been inserted in the intensive care unit but retained unecessarily after discharge from the unit.3

The risk of bloodstream infection is much higher with CVCs than with peripheral venous catheters (4.0 versus 0.2 per 1000 line-days).2,4 Such simple facts are often overlooked or inadequately emphasised in preventive programs, and CVCs may be retained for convenience.

Our intensive care unit maintained an informal clinical practice of routinely removing CVCs when patients were discharged from the unit. However, an audit found that many CVCs were retained, often inappropriately, thus exposing patients to needless increased risk.5 A formal intervention policy aimed at improving CVC removal was implemented. This included a month of staff education, culminating in introduction of a formal written policy in March 2003. CVCs were to be removed when no longer clinically required or at discharge from the intensive care unit, unless the patient met predetermined retention criteria (ie, administration of vasoactive or venotoxic drugs [eg, dopamine or vancomycin] or parenteral nutrition solutions; poor peripheral venous access [after two attempts] with ongoing need for intravenous therapy; or transfer to another intensive care or coronary care unit).

We undertook a prospective observational study of all patients with CVCs in the intensive care unit of our hospital in the period March to August 2003. Patients were grouped according to whether the CVC was removed per policy before or at discharge from the intensive care unit; whether it was retained per policy at discharge from the unit; or whether it was retained in breach of policy. All patients were followed up for 7 days after CVC removal. Those who died within this time were excluded from the analysis.

We studied a total of 305 CVCs in 272 patients (Box). We observed:

• high compliance with the written policy (91%),

• significantly lower CVC in-situ times when policy was followed (5.1 v 8.1 days),

• low CVC reinsertion rates (7%),

• no difference in incidence of bloodstream infections between the groups.

This study demonstrates that a formal policy directed at early CVC removal is effective in lowering CVC in-situ times without incurring clinical cost to the patients (eg, excessive CVC reinsertion rates). Policy breaches were infrequent (8% of all CVCs), but, when they occurred, CVC retention appeared unnecessary, and CVC in-situ times were significantly prolonged. The risk of sepsis with CVCs may be substantially lowered by policy-driven removal of CVCs, without compromising patient care.