Diagnosis of ECMO Cannula Infection: Prospective Comparison of Sampling Methods

BACKGROUND ECMO is a life-support technique widely used in refractory respiratory and/or circulatory failure. It is associated with infectious complications, notably cannula infections, whose incidence is estimated at about 10%. These infections are associated with an increased risk of septic shock and contribute to longer ICU stay and higher mortality. There is to date neither a standardized definition nor consensus diagnostic criteria, particularly regarding the nature and diagnostic performance of the microbiological samples to be collected. Superficial swabbing of the insertion site is simple and minimally invasive but frequently yields contaminated results or results reflecting colonizing flora. More invasive methods (needle aspiration, culture of the cannula tip at decannulation) may be more specific but are less commonly performed and have not been comparatively validated. A recent international Delphi consensus identified cannula aspiration and cannula swab as the two preferred microbiological methods. The lack of an established diagnostic algorithm leads to heterogeneous management, with potential consequences for antibiotic stewardship and patient outcomes.

DESIGN Single-center, prospective, non-interventional study conducted in the medical ICU of Pitié-Salpêtrière Hospital (AP-HP, Paris, France), which manages about 300 ECMO patients per year. All adult patients on veno-venous or veno-arterial ECMO are eligible. The patient (or trusted person, relative or legal representative) is informed and the patient is included unless an objection is expressed. Inclusion period: 24 months. Participation: duration of ICU stay (maximum 2 months). Total study duration: 26 months.

SAMPLING

• Swab (performed for the research): a sterile single-use swab is applied to the cannula insertion orifice.
• Needle aspiration (standard of care): 1 mL of sterile saline is injected along the cannula through a syringe and a plastic catheter, re-aspirated and transferred to a sterile container.
• Cannula tip (standard of care): at decannulation, the distal 5 cm of the cannula is cut with a sterile blade and placed in a sterile container.

On the day the clinician in charge suspects a cannulation site infection, a swab and a needle aspiration are performed at each cannulation site. On the day of decannulation, a swab and a needle aspiration are performed at each cannulation site and the tip of each cannula is sent for culture.

MICROBIOLOGY Samples are cultured in the microbiology laboratory using the Brun-Buisson technique: suspension in 1 mL of sterile saline and vortexing for about 1 minute, then 100 microliters are plated on fresh blood agar, chocolate agar and Sabouraud agar, incubated for 48 hours to 10 days under aerobic and anaerobic conditions and read daily. In case of positive culture, identification is performed by mass spectrometry (MALDI-TOF). Each sample is recorded as culture positive or culture negative.

REFERENCE STANDARD The reference diagnosis of ECMO cannula infection is the clinical diagnosis established by an adjudication committee of three clinician experts, ruling retrospectively and blinded to the results of the index tests, on the basis of the available clinical, biological, microbiological and outcome data.

DATA Data are collected prospectively from the electronic medical record into a REDCap electronic case report form hosted by AP-HP. At each sampling: local appearance of the insertion site (photograph), general and hemodynamic status, biological inflammatory markers, nosocomial infection during the stay, antibiotic therapy within the previous 7 days, rectal colonization with a multidrug-resistant organism, positive blood culture within the previous 48 to 72 hours, duration of ECMO support.

STATISTICS The sample size was set to obtain at least 50 cases of cannula infection; with an expected incidence of 10%, 500 patients will be included. Sensitivity, specificity and predictive values of each sampling strategy are computed against the reference diagnosis. Paired comparisons use the McNemar test. Agreement between methods is assessed with the kappa coefficient. Exploratory logistic regression analyses will identify factors associated with microbiological positivity. Analyses are performed on complete cases; a failed sampling technique is considered negative.