Results
75 patients were included for 712 ECMO days with an average duration of 9.5 days. 76.6% (n=58) of patients were male with an average age of 54.1 years. 48% (n=35) utilized a right ventricular assist device (RVAD) with in-line oxygenator which was included as a form of VV ECMO. 70.2% (n=45) of cannulations occurred in the operating room. Femoral-femoral cannulation was utilized in 38.7% (n=29), femoral-internal-jugular in 30.7% (n=23), central in 17.3% (n=13) and internal jugular only in 13.3% (n=10) of patients. 62.2% (n=46) were decannulated while 52.7% (n=39) were alive at discharge.
70.6% (n=53) of patients were on antimicrobials prior to cannulation while 56% (n=42) received peri-operative antimicrobials. Blood cultures, based on clinical suspicion, were obtained in 59.6% (n=49) of patients with seven positive cultures. Five were contaminants while two were E-BSI resulting in a E-BSI rate of 2.7% (3 per 1000 ECMO days).
Escherichia Coli, Enterobacter cloacae , and Streptococcus Pneumoniae were identified in the patients with E-BSI. Patient characteristics are summarized in Table 1. Of the contaminants, two wereStaphylococcus epidermidis while an unidentified Coagulase Negative staphylococcus, Bacillus not anthracis, andAbiotrophia species comprised the other positive cultures. These were determined to be contaminants due to the organisms and review of Infectious Diseases consultation notes.
Of the two E-BSI, one patient only had positive peripheral cultures. While the peripheral cultures may not have been directly related to his ECMO, we felt the need to include it given the organism (E. Coli)in the setting of femoral-femoral cannulation.
Our organism profile is similar to that reported in ECMO patients.1,2,4,6 Amongst VV-ECMO, common isolates include gram negative rods, candida species, andStaphylococcus/Streptococcus species.2,7,8Compared to central venous catheter associated infections, the ECMO population microbiological spectrum shifts toward gram-negatives and fungi rather than Staphylococcus species.7 This shift appears to be multifactorial including exposure to antimicrobials, complexity of illness, and complications from secondary infections.2,7
Inherent differences exist between VV and Venous-Arterial (VA) ECMO patients. VV-ECMO patients require support longer with more exposure to antimicrobials.2 Cannulation sites may also differ with VA-ECMO patients undergoing higher rates of central cannulation and open chests.7 Within our cohort, 17% of VV patients were cannulated centrally by way of RVAD and nearly 70% of cannulations occurred in the OR. These factors may influence the rate of E-BSIs and therefore we chose VV patients exclusively.
The current rate of E-BSI ranges from 9.7%-35%.2,5,7,8 Amongst VV patients only, an incidence of 13.1%-17% as been reported.2,7 Notably, Menaker et al. reported a BSI incidence of 5.7% amongst their VA-ECMO population. Per their institutional protocols, only VA-ECMO would receive routine antimicrobial prophylaxis.7
The use of antimicrobials in the pre-, peri-, or post-cannulation settings is not standardized and the routine use of prophylactic antimicrobials is currently discouraged by the Extracorporeal Life Support Organization.3,5 Despite this, many ECMO centers utilize a prophylactic antimicrobial regimen.3,5
We postulate our institutional use of antimicrobials in the perioperative setting may have decreased our overall rate. The majority of cannulations occurring within the OR likely also contributed to our low rate. Both E-BSI patients were cannulated in non-OR settings. Improving our rate of antimicrobial use in the perioperative setting may help reduce our overall institutional E-BSI incidence. Although E-BSI has not been linked with increased mortality, it is reasonable to assume that a correlation might emerge in a large cohort and this remains an open question.
The current study has several limitations including the retrospective nature, single institution, and an older data set. Duration of support has been suggested as a contributing risk factor for E-BSI, however, our cohort was not large enough to further study this. In the era of COVID-19, in which patients may have prolonged periods of VV-ECMO, an analysis including duration of support might be revealing. Our low E-BSI rate did not allow for further analysis into risk factors and effect on morbidity and mortality.