Femoral lines might not be so bad after all for infection risk

...

Jan 4, 2013

"We've got to get that femoral line out of there!" The attending's face as he says it shadowed with a simmer of fear, a dash of anger. How could the moonlighter have been so incompetent or lazy as to choose the benighted femoral site for a central venous line when the internal jugular and subclavian were available? Why, it's virtually guaranteed that line will get infected, and soon! And what will the team coming on tomorrow think, if I don't change that line out today? That attending was me, and I was only trying to do what I was taught so forcefully during training, and what the Centers for Disease Control and Prevention (CDC) and Infectious Disease Societies of America (IDSA) still command: "avoidance of the femoral vein" being their strong 1A recommendation in 2011 for the prevention of central venous catheter associated bloodstream infections (CRBIs or CLABSIs). But medical lore does have a vulnerability: evidence. Paul Marik, Mark Flemmer, and Wendy Harrison examined 10 large studies that reported rates of central venous catheter-related bloodstream infections (CRBI, CLABSI) and found no clear evidence (only a soft trend) that the femoral site was any more prone to bloodstream infection than the internal jugular or subclavian sites. The best news: catheter-related infections at all sites seem to have dramatically declined in recent years thanks to widespread adoption of improved, cleaner techniques. Marik et al report their results in the August 2012 Critical Care Medicine.

What They Did

Authors identified 2 randomized trials including 1,006 non-tunneled central venous catheters, and 8 cohort studies including 16,370 catheters, in which infection rates were reported. Most catheters (~11,000) were in the internal jugular vein, with ~3,200 each in the subclavian vein or femoral vein. There were a total of 113,652 catheter-days for analysis. Catheters included small-bore central venous "triple lumen" catheters, as well as large-bore dialysis catheters.

What They Found

The average rate of catheter-related bloodstream infection was 2.5 per 1,000 catheter-days. (This translates to 1 infection for every 100 patients with a central line in place for 4 days). The randomized trials showed no difference in infection rates (although they only included 1,000 catheters, so might not have been powered to do so). There was no statistical difference in infection rates between the femoral site and the subclavian site (supposedly the cleanest, and the recommended site by some). The internal jugular site did in the initial analysis appear "cleaner" than the femoral, with a risk ratio of 1.90 for infection at the femoral site. Marik et al explain this away by pointing out that 2 studies that were statistical outliers accounted for nearly all this observed difference. Without those studies, the femoral's risk ratio was 1.35 (non-significant). More recent studies showed a lower rate of infection at the femoral site, with differences between sites virtually disappearing statistically in studies published in recent years. Most notably: the largest and most recent study, from an infection surveillance program in Wales (55,000 catheter-days) showed a risk of CRBI of only 0.6 per 1,000 catheter-days. Including only those with positive blood cultures reduced this to a mere 0.22 CRBIs per 1,000 catheter days; that's somewhere between one-fourth and one-tenth of the historically observed rates of catheter-related bloodstream infections. That large cohort did not show an increased risk for infection for lines placed at the femoral location. The suspected reason: increased adherence internationally with barrier precautions, use of chlorhexidine and ultrasound.

What It Means

Since no randomized trials showed an increased risk for CRBI at the femoral site, the authors rightly call out the IDSA and CDC to explain how their recommendation against the femoral site can be "1A" (which is accepted to mean, "based on strong evidence from randomized trials"). I've always been puzzled by the contrast between the hysteria among internists / intensivists to avoid the femoral site, and the calm of the nephrologist who blithely dialyzes a patient through a femoral catheter for weeks, unfazed by the intervening diarrheal illness that would seem to represent an infection risk. Marik et al close their article recommending we chuck the mislabeled "1A" recommendation in favor of a "pragmatic approach to site selection," emphasizing the following:

The chosen site should depend on the expertise and skill of the operator and the risks associated with placement.
In emergencies or in high-risk patients (like a demented, agitated patient) femoral placement may be best.
All catheters that are placed under non-sterile or emergency circumstances should be removed and resited within 2 days.
Ultrasound should be used for catheters placed in the internal jugular and femoral site to reduce the risk of complications from placement.
Avoid the subclavian site in patients with advanced renal failure to preserve the arm veins and subclavian vein for future fistula placement. Avoid the femoral veins in renal transplant patients.
One unequivocal downside of the femoral site is its interference with early mobilization, particularly in the case of patients with dialysis catheters.
There may be a higher risk of CRBI with femoral placement in massively obese patients and IJ or subclavian should be used instead, if possible.

Want to hear from Paul Marik himself, who compares physicians who blindly follow guidelines or medical teaching to "lemmings" who "will believe in the Tooth Fairy, or anything"? Check out Scott Weingart's fantastic podcast with Paul Marik on EMCrit.org:

EMCrit.org's Podcast With Paul Marik

(may not work in all browsers; go to EMCrit.org to retry)

Marik PE et al. The risk of catheter-related bloodstream infection with femoral venous catheters as compared to subclavian and internal jugular venous catheters: A systematic review of the literature and meta-analysis. Critical Care Medicine 2012; 40:2479-2485.