Infections Acquired During Venoarterial Extracorporeal Membrane Oxygenation Postcardiac Surgery in Children: A Retrospective Observational Study

Introduction

Extracorporeal membrane oxygenation (ECMO) is increasingly being used in refractory cardiac and pulmonary dysfunction as a rescue modality.¹ It provides reliable and sustained circulatory support in children who have undergone cardiac surgery. The common indications for establishing venoarterial ECMO (VA-ECMO) support in children postcardiac surgery are failure to wean from cardiopulmonary bypass (CPB), postcardiotomy cardiogenic shock (PCCS), refractory pulmonary arterial hypertension, and as a bridge to recovery or transplant.², ³ PCCS is a significant concern in children, particularly neonates. Occasionally, PCCS is refractory to inotropic support and, in the absence of mechanical circulatory support, may prove fatal.⁴ For nearly five decades, the VA-ECMO has been the first-line mechanical circulatory support in patient's refractory to conventional therapies after cardiac surgery.⁵, ⁶ VA-ECMO allows blood to be drained from a central vein and returned to the arterial system. Therefore, VA-ECMO provides both respiratory and circulatory support. The survival rate of children on VA-ECMO support is 45%.⁷ Timely initiation of VA-ECMO in selected candidates is essential to achieve maximum survival benefits. VA-ECMO has evolved and undergone improvements over the decades, but it still carries significant complications. The most frequently encountered complications during VA-ECMO are bleeding, thrombosis, acute kidney injury, and infections.⁸, ⁹ Among those, infections acquired during VA-ECMO lead to high morbidity and mortality.¹⁰, ¹¹, ¹²

Hence, this study aimed to determine infection rates, causal microorganisms, and mortality risk factors in children developing infections on VA-ECMO support after cardiac surgery. The primary objective of our study was to determine infection rates, causal microorganisms, and mortality risk factors in children developing an infection during VA-ECMO. Our secondary objectives were to compare infection rates among survivor and nonsurvivor groups and to understand the pattern of change in infectious agents and their antibiotic resistance over the study period.

Methods

This retrospective, observational study was conducted between January 2016 and December 2020 in All India Institute of Medical Sciences, New Delhi, India. The study protocol was approved by the Institute Ethics Committee (IECPG-278/3/2021). The study included children <14 years of age who underwent elective or emergency cardiac surgery and required VA-ECMO. We excluded children who expired within 48 hours of initiation of VA-ECMO, and preoperative positive cultures were excluded from the study.

Results

In total, 133 children who underwent VA-ECMO between 2016 and 2020 were analyzed for this study. Among them, 27 were excluded (16 expired before 48 hours, 9 children were more than14 years old, and 2 had inadequate data). The remaining 106 children were included in the study (Fig. 1).

Close

Fig. 1

Algorithm for patient distribution of VA ECMO for this study.

Export to PPT

Discussion

A total of 49 infections were reported in 106 VA-ECMO patients from 2015 to 2020, representing a prevalence of 46.23% and an infection rate of 186.4 per 1,000 ECMO days. We also found a higher prevalence of approximately 26.42% and an acquired infection rate of 215.07/1,000 ECMO days in the nonsurvivor group. We also found a higher prevalence of approximately 26.42% and an acquired infection rate of 215.07/1,000 ECMO days in the nonsurvivor group. Austin et al reported fewer VA-ECMO infections as 7.2 to 18%, with an estimated incidence between 4.82 and 17.2 episodes per 1,000 ECMO days.¹⁹ Furthermore, Bizzarro et al and Kutleša et al, in their studies, also found similar ECMO infection rates of 20.9 and 35%, respectively.²⁰, ²¹

Our study showed that BSI and CAUTI episodes were 53.04 and 68.19 per 1,000 ECMO days and VAP rate 44.50 per 1,000 ventilator days, and all were higher in the nonsurvivor group than in the survivor group. Another study showed that 64% (142 out of 220) patients had developed hospital-acquired infections (mostly VAP and BSI) with VA-ECMO >48 hours duration. This study also found higher septic shock episodes with BSI followed by VAP and mediastinitis.⁸ Aubron et al proved that the adult population had a higher risk of nosocomial infection than the pediatric population during ECMO support (20.5 vs. 6.1%; p < 0.001).⁹ In our study, children aged between 1 and 12 months had a higher incidence of BSI and CAUTI than children less than 1 month of age. The incidence of VAP was higher in children less than 1 month of age compared with 1 to 12 months old.

The most frequently isolated organism was A. baumannii for both VAP and BSI in our study. The other organisms isolated were K. pneumoniae, E. coli, P. aeruginosa, and S. maltophilia, and B. cepacia. A recent study by Kim et al also reported gram-negative bacteria as common isolates from VAP and BSI in unison with our findings.²² In contrast to our study, the Extracorporeal Life Support Organization Registry reported coagulase-negative staphylococcus as the most frequently isolated organism from ECMO patients. Other infectious agents reported were Candida species, P. aeruginosa, Enterobacteriaceae, Staphylococcus aureus, and Enterococcus species.² The reason for different infection rates and microorganisms in our study may be distinct microbial flora and customized antimicrobial stewardship programs of our ICU. The incidence of polymicrobial growth (more than three organisms reported in samples) was VAP, 9 out of 62; drain fluid, 6 out of 11; BSI, 3 out of 18; and CAUTI, 2 out of 23. Schmidt et al also reported a similar result in their study.⁸ The occurrence of progressive antimicrobial resistance in gram-negative bacteria is a sign of concern.

Candida species is a commonly isolated organism from patients during ECMO support.²³ Candida species were isolated in 95.23% (20 out of 21) of urine samples in our study. Sun et al reported that Candida species were isolated in two-third of the urine cultures, whereas E. coli was isolated in one-third of the samples.²⁴ Children requiring ECMO support are critically ill and undergo immunomodulation due to blood contact with a nonbiological foreign surface and the use of steroids. They also receive broad-spectrum antibiotics, multiple invasive devices, and frequent blood transfusions, all risk factors for Candida infection.

Central VA-ECMO is a complex, invasive procedure that requires open sternum, skin, and the prolonged use of invasive devices that makes children vulnerable to infections. A prolonged ECMO support leads to dysregulated immune response and an incessant inflammatory reaction. The reexploration surgeries for bleeding, multiple blood transfusions during ECMO, and hypoperfusion state aggravate this inflammatory cascade.²⁵, ²⁶ Immunomodulation occurring during ECMO serves as a risk factor for infection. A recent study done by Burket et al reported that the rate of BSI increased proportionately from 9.5 to 64.5 per 1,000 ECMO days with an increased duration of ECMO.²⁷ Another study reported that the only predictor for acquired nosocomial BSI during VA-ECMO was a duration of support >10 days.²⁸ Meyer et al concluded that sepsis in ECMO patients is an independent negative predictor of survival for the neonates.²⁹ A logical conclusion is that prolonged duration of ECMO leads to BSI and sepsis, especially in neonates, thus increasing mortality. In our study, the duration of ECMO therapy also significantly correlated with adverse outcomes (p = 0.0305).

In our cohort, male children predominate in both survivor and nonsurvivor groups. Younger children had more survival benefits in the survivor group (p = 0.008). The survivor's group had fewer preoperative admission hours than nonsurvivors, which was statistically significant (p < 0.001). A prolonged preoperative stay in nonsurviving children may indicate sicker children requiring medical optimization or intensive care therapy before surgery.

In our study, VA-ECMO was instituted for children with hemodynamic instability and poor ventricular function while weaning from CPB in the operating room. The mean AXC, CPB time, and VIS were significantly more in the nonsurvivor group than in the survivor group (p < 0.001). Because not all the children were operated on by the same surgeon, prolonged AXC and CPB time may reflect the complexity of the surgery and the surgeon's experience.

In total, 93 (87.73%) out of 106 children were weaned successfully from VA-ECMO support, of which 59 (55.66%) were subsequently discharged from ICU. However, the remaining 34 weaned from ECMO support died in the ICU. These patients' primary cause of death was sepsis and consecutive multiorgan failure. In 13 (12.26%) out of 106 children who required ECMO support, cardiac function did not improve, weaning was not possible, VA-ECMO support had to be withdrawn, and they died subsequently. The leading cause of death in patients with failed weaning was persistent low cardiac output (10 out of 13). Other causes of mortality were coagulopathy (8 out of 13), culture-proven infections (4 out of 13), and cerebral hemorrhagic stroke (2 out of 13). Rastan et al and Hsu et al reported a better weaning rate with VA-ECMO following PCCS patients ranging from 31 to 60%, despite high “in-hospital” mortality of 59 to 84%.³⁰, ³¹ In this study, we could not establish the percentage of children discharged from the hospital due to the study's retrospective nature. An earlier prospective study of our group showed a 45% survival at discharge and at 4 months follow-up.¹³

In our study, 21 (19.81%) children had AKI, in which 13(12.26%) received PD, the incidence of which was statistically significant between survivor and nonsurvivor groups (Table 6). The AKI could be due to systemic hypotension, systemic inflammatory response, nephrotoxic drugs, and microembolization to renal vasculature during ECMO. Lee et al in their study reported a higher incidence (55.6%) of AKI in VA-ECMO patients and associated it with increased mortality.³² Another study in children by Selewski et al showed continuous renal replacement therapy reduced fluid overload during ECMO in survivors compared with nonsurvivors (24.5 vs. 38%, p = 0.006).³³

Two children (1.88%) in our cohort had a cerebral bleed after 48 hours of ECMO initiation, and they were declared dead on ECMO. These two children had thrombocytopenia and required >100 mL/kg PRBC transfusion. Seven children in the survivor group and five in the nonsurvivor group developed arterial thrombus due to arterial lines during and after weaning VA-ECMO. A logistic regression analysis model identified weight <4 kg, CBP time >120 minutes, surgical exploration, PRBC transfusion >100 mL/kg, and culture-positive children as independent predictors of ECMO mortality.

They are certain limitations of the study. Ours is a retrospective study. We did not consider culture-negative sepsis and biochemical markers of infections. The number of children more than 1 year of age was less.

In conclusion, the prevalence of acquired infection during VA-ECMO was 46.23% in our study. The prevalence was highest in 1 to 12 months group followed by <1month group and >12 months group which were 27.36, 16.98, and 1.89%, respectively. The incidence rates of BSI, CAUTI, and VAP per 1,000 ECMO days were higher in nonsurvivor group compared with survivors. A. baumannii was the most common cultured gram-negative organism in VAP and BSI during VA-ECMO with 67.65% Acinetobacter samples resistant to carbapenems and CAUTI was predominately due to Candida species during VA-ECMO.