Infections acquired during extracorporeal membrane oxygenation in neonates, children, and adults^*

Abstract

Objective: 

To determine current rates, risk factors, and causal organisms related to infections acquired during extracorporeal membrane oxygenation (ECMO).

Design: 

A descriptive and retrospective case-control study.

Setting: 

ECMO centers belonging to the Extracorporeal Life Support Organization.

Patients: 

The Extracorporeal Life Support Organization Registry was queried for data related to all ECMO cases from 1998 through 2008. All culture-proven infections obtained from any site during ECMO support and not believed preexisting were included. Infection rates were analyzed by age category (i.e., neonatal, pediatric, adult), indication for ECMO (i.e., respiratory, cardiac, cardiopulmonary resuscitation), mode of ECMO (e.g., venovenous), and duration of ECMO support. Infected and noninfected ECMO patients were compared.

Interventions: 

None.

Measurements and Main Results: 

A total of 2,418 infections were reported during 20,741 (11.7%) ECMO cases for a rate of 15.4 per 1,000 ECMO days. Rates were highest in the adult vs. the pediatric and neonatal populations (30.6 vs. 20.8 vs. 10.1 infections per 1,000 ECMO days, respectively) and in those necessitating extracorporeal cardiopulmonary resuscitation (24.7 infections per 1,000 ECMO days). In each age category, venoarterial ECMO was the mode of support associated with the highest rate of infection. Prevalence of infection increased with duration of ECMO support from 6.1% of those requiring bypass for ≤7 days to 30.3% of those requiring ECMO for >14 days (p < .001). Coagulase-negative staphylococci (15.9%) were the most common organisms cultured followed by species of Candida (12.7%), and Pseudomonas (10.5%). Those with an infection acquired during ECMO support were significantly older, had a longer duration of ECMO, a longer duration of post-ECMO ventilatory support, and a higher prevalence of death than those without.

Conclusions: 

Infections acquired during ECMO are common and can have significant associated consequences. Knowledge of high-risk patients and common causal organisms may improve strategies for treatment and prevention, but further work to develop strategies and guidelines for prevention of these infections is urgently needed.

Since the first report of its use >30 yrs ago (¹), extracorporeal membrane oxygenation (ECMO) has evolved into an established method of support for reversible cardiac and pulmonary failure in neonates, children, and adults. Despite its efficacy in many clinical scenarios, ECMO carries significant associated risks, including the acquisition of infections during bypass (^2–6). In January 2010, the Extracorporeal Life Support Organization (ELSO) reported that 6.1%, 18.7%, and 20.5% of all neonatal, pediatric, and adult respiratory ECMO cases, respectively, were complicated by a culture-proven infection (⁷). These infections can have significant consequences to the ECMO patient population, including a prolonged need for bypass support (^{2, 3, 6}), an increased risk of mechanical and patient-related complications (²), a delay in post-ECMO lung recovery (²), and an increased risk of death before hospital discharge (^{2, 3}).

The ELSO Task Force on Infections and ECMO was established in 2009 to address this problem. The initial step in this process was to query the ELSO registry in an effort to determine current rates, risk factors, and causal organisms related to infections acquired during ECMO in the neonatal, pediatric, and adult populations. This investigation reports the results of those inquiries.

MATERIALS AND METHODS

Patient Population, Data Collection, and Definitions

ELSO is a voluntary international alliance of 110–130 active ECMO centers with a centralized registry for the purpose of exchanging data and ideas among the ECMO community (^{7, 8}). Descriptive and case-control analyses of a retrospective cohort were performed, utilizing data from this registry. Information was obtained on all ECMO cases reported from January 1, 1998 through December 31, 2008 and included the age, weight, sex, and race of each patient, the indication for ECMO support, the mode of ECMO support, the presence or absence of an infection acquired during ECMO, and, if present, the causal organism, the duration of ECMO support and of post-ECMO ventilatory support, and survival to hospital discharge.

An infection acquired as the patient received ECMO was defined as a culture-proven infection from any site (e.g., blood, urine) during ECMO support and not believed preexisting. To further differentiate infections acquired during bypass from those preexisting, any case reported to the ELSO registry as having both a pre-ECMO infection and an on-ECMO infection with the same organism was excluded. Only one acquired infection per ECMO course could be reported to the registry. Culture site-specific data were not available.

Data Analysis

Once collected, data were stratified by age, indication for support, mode of ECMO support, and duration of therapy. Age category was based on age at the onset of ECMO and divided into neonatal (0–30 days of age), pediatric (>30 days and <18 yrs of age), and adult (≥18 yrs of age). The indication for ECMO support was categorized as respiratory, cardiac, or extracorporeal cardiopulmonary resuscitation.

The mode of ECMO support was classified as venoarterial (VA), venovenous, venovenous via a double lumen cannula, VA plus an additional venous cannula, venovenous via a double lumen cannula plus an additional venous cannula, venovenous converted to VA, VA converted to venovenous, and other. Given discrepancies in the use of certain modes of ECMO across patient populations (i.e., the predominant use of venovenous via a double lumen cannula ECMO in neonates), these data were stratified by age category for the purpose of analysis. The duration of ECMO support was separated into three categories for analysis: 0–7 days; 8–14 days; and >14 days. The specific timing of each infection was not available from the registry.

Data obtained from the ELSO registry were imported into a separate database and analyzed, using SPSS v. 15.0 (SPSS Inc., Chicago, IL). If an individual required ECMO on more than one occasion, only the first course was included for analysis. Infections acquired on ECMO were represented as both prevalence and rates. The prevalence was calculated as a percentage based on the number of infections divided by the total number of individuals within each population being studied. Rates were calculated and presented as the number of infections per 1,000 ECMO days for each population being studied.

Bivariate analyses were performed comparing infected and noninfected ECMO patients with respect to certain demographic characteristics, risk factors, and outcomes of interest. Continuous data were summarized, using the mean and sd from the mean, and dichotomous data were summarized as frequencies and percents. Continuous data were analyzed by the Student's t test. Dichotomous data were compared using a chi-square analysis, and the results were presented as odds ratios with 95% confidence intervals. Trends were compared, using the chi-square analysis of trends. A two-sided α of 0.05 was used to establish statistical significance.

The collection and analysis of these data were approved by the University of Michigan Medical School's Institutional Review Board for Human Subject Research.

RESULTS

Prevalence and Rates of Infections in Patients on ECMO

A total of 2,418 infections were reported in 20,741 ECMO patients from 1998 through 2008, representing a prevalence of 11.7% and a rate of 15.4 per 1000 ECMO days. The overall rate of infections remained relatively constant throughout the 11-yr study period (Fig. 1). Adult ECMO patients, irrespective of the indication for ECMO support, had the highest prevalence and rates of infection, whereas neonates had the lowest prevalence and rates of infection (Fig. 1 and Table 1). When neonates, children, and adults were analyzed separately, extracorporeal cardiopulmonary resuscitation (Table 1) and VA ECMO (Table 2) were the indications for and mode of ECMO support, respectively, associated with the highest rate of acquired infection in each age category.

Figure 1.: Trends in the number of culture-proven infections acquired during extracorporeal membrane oxygenation depicted per 1,000 extracorporeal membrane oxygenation days from 1998 through 2008. Rates for the entire study population are displayed as are rates by specific age category.

Table 1: Prevalence and rates of culture-proven infections acquired during extracorporeal membrane oxygenation by age category and indication for support

Table 2: Prevalence and rates of culture-proven infections acquired during extracorporeal membrane oxygenation by mode of support

The overall prevalence of infection was highest in patients who required ECMO for >14 days (30.3%) as compared with those requiring ECMO for 8–14 days (15.7%) and for ≤7 days (6.1%) (p < .001) (Fig. 2). This trend was observed in all age categories but was most evident in the adult population, where 51.6% of those who required ECMO support for >14 days were reported to have acquired an infection during bypass support as compared with 12.8% of those requiring ECMO for ≤7 days (p < .001) (Fig. 2).

Figure 2.: Prevalence of infections by duration of extracorporeal membrane oxygenation support depicted for the entire population and by age category.

Causal Organisms

Coagulase-negative staphylococci were the most common organisms cultured (15.9%) in the entire ECMO patient population, followed by species of Candida (12.7%), Pseudomonas aeruginosa (10.5%), Staphylococcus aureus (9.4%), Enterobacter spp. (5.7%), Klebsiella spp. (4%), Enterococcus spp. (4%), and Escherichia coli (3.9%) (Fig. 3). In the neonatal population, coagulase-negative staphylococci were most common, whereas Candida was most common in the pediatric and adult populations (Fig. 3). No major differences in causal organisms were observed by indication for, or mode or duration of, ECMO support.

Figure 3.: The most common organisms responsible for infections in the extracorporeal membrane oxygenation population depicted for all subjects and by specific age category. CONS, coagulase-negative staphylococci; spp., species; Pseudomonas aer., Pseudomonas aeruginosa; Staph. aureus, Staphylococcus aureus.

A Comparison of Infected Versus Noninfected ECMO Patients

Those patients with a culture-proven infection acquired on ECMO were significantly older, had a longer duration of ECMO support and post-ECMO ventilatory support, and had a significantly higher risk of death before discharge than those without (Table 3). When the cohort was stratified by age, those with an infection required a significantly longer duration of ECMO support in all age categories (neonatal: 302.6 hrs vs. 172.2 hrs; pediatric: 322.2 hrs vs. 161.7 hrs; adult: 284.3 hrs vs. 132.4 hrs; p < .001 for all three comparisons), a significantly longer duration of post-ECMO ventilatory support in the neonatal (87.9 hrs vs. 59.6 hrs; p < .001) and the adult cohorts (97.5 hrs vs. 68.8 hrs; p = .003), and had a significantly higher odds of death in the neonatal (odds ratio, 2.80; 95% confidence interval, 2.44, 3.21) and pediatric (odds ratio, 1.24; 95% confidence interval, 1.08, 1.42) ECMO patient cohorts.

Table 3: A bivariate comparison of infected and noninfected extracorporeal membrane oxygenation patients

DISCUSSION

We observed a high rate of culture-proven infections in the ECMO patient population over our 11-yr study period. This is likely due to several factors related to both the patient and the medical therapy. The ECMO population, in either acute cardiac and/or respiratory failure or arrest, tends to be among the most critically ill of all inpatient groups and are often immunosuppressed. ECMO patients require invasive procedures before and during bypass, are frequently exposed to broad-spectrum antibiotics, and require the prolonged use of invasive support devices, such as central lines, urinary catheters, and endotracheal tubes. Certain factors inherent to ECMO support may also contribute to the high rate of acquired infections in this population. Frequent entry into the ECMO circuit is often needed for laboratory assessments and for the administration of intravascular infusions. This can increase the likelihood that organisms which colonize access sites will enter the bloodstream. Furthermore, alterations in the immune response often accompany the utilization of bypass. In particular, a reduction in the number and normal function of circulating neutrophils and lymphocytes is frequently observed, which may further increase the susceptibility to infection (^9–12). Although the entire ECMO patient population is therefore vulnerable, certain subsets of the population are most at risk. We identified adults, those requiring extracorporeal cardiopulmonary resuscitation, those on VA ECMO (when stratified by age category), and those requiring ECMO support for >14 days, as those most susceptible to infection. This knowledge may be helpful when assessing risk and considering screening, treatment, and prevention strategies for individual ECMO patients.

Organism-specific data from the ELSO registry revealed that coagulase-negative staphylococci, Candida spp., P. aeruginosa, S. aureus, and Enterobacter spp. were responsible for the majority of positive cultures obtained from patients on ECMO support. Species of Candida were determined to be the most frequent cause of acquired infections in pediatric and adult ECMO patients and the second most common in neonates. This information may be useful in selecting an appropriate antimicrobial regimen (i.e., the addition of an antifungal) when an infection in a patient on ECMO is suspected. Also, many of the causal organisms identified from our investigation were determined to be mostly commensal species of bacteria and fungi with a predilection for support apparatus, such as central venous catheters, endotracheal tubes, and urinary catheters. In a survey of 463 hospitals, the National Healthcare Safety Network reported that these same organisms were also responsible for the majority of non-ECMO device and procedure-related healthcare-associated infections (¹³). Strategies similar to those utilized to prevent central line-associated bloodstream infections (^14–16) and ventilator-associated pneumonia (^{17, 18}) are therefore likely applicable to the ECMO population and, if adopted, may serve to reduce the risk of infections and their associated complications.

In our cohort of 20,741 ECMO patients, those with a culture-proven infection experienced a longer duration of ECMO and post-ECMO ventilatory support and had a higher prevalence of death before hospital discharge as compared to those without an infection during bypass. These findings are consistent with those reported by Meyer et al (²) who examined 5,123 neonates placed on ECMO between 1987 and 1993 and determined that those with an infection from any site during ECMO had a higher rate of mechanical complications, required a longer duration of ECMO support and post-ECMO ventilatory support, and had a lower rate of survival than those without an infection. Similarly, O'Neill et al (⁶), in a cohort of 141 neonatal and nonneonatal patients, determined that those with an infection while on ECMO necessitated a longer duration of ECMO support. Douglass et al (³), in a cohort of 5,001 neonates, determined that an infection was associated with an increased duration of bypass and risk of death. These findings stress the need for efforts to curtail these infections and their associated consequences.

Our investigation and its results have limitations. There are gaps and potential inconsistencies in data reporting to the ELSO registry, which may have biased our results. The registry includes data from approximately 120 international centers. It is likely that there is variability among these centers with respect to infection control (e.g., prophylactic antimicrobial usage) and surveillance practices (e.g., daily vs. targeted culturing, selection of culture sites), which could complicate data reporting and interpretation. Also, there are no standardized criteria for the diagnosis of ECMO-related infections. Lack of culture site-specific data makes it difficult to determine whether or not infections are acquired from ECMO-specific support apparatus (e.g., the ECMO circuit, the cannula) or are merely the sum total of other healthcare-associated infections (e.g., central line-associated bloodstream infections, urinary tract infections, and ventilator-associated pneumonia). There are no specific criteria within the ELSO registry that assist in discerning true infection from contamination. This distinction is instead left to the discretion of the individual reporting the information and could therefore further bias the prevalence rates. Also, the registry only allows for one infection per ECMO course to be reported, which may result in underreporting of the problem. Another potential pitfall relates to difficulties in the ability to discern preexisting from on-ECMO infections, which may have resulted in an overestimation of the prevalence and rate of infection. We attempted to address this problem by eliminating any patient reported as having had both a pre-ECMO and on-ECMO infection with the same organism, but this approach assumes that adequate and reliable pre-ECMO cultures were obtained.

Gaps in data from the registry also made it difficult to adequately assess the relationship between infection and selected risk factors and outcomes of interest. For example, we observed an association between infection and increased duration of ECMO support. However, no data were available with respect to the timing of each infection during the ECMO course. We, therefore, could not adequately assess whether the infection was the cause or effect of prolonged ECMO support. Furthermore, lack of data on other potential confounding risk factors for prolonged ECMO support (or any other outcome of interest) made more accurate and valid assessment of the impact of an infection via multivariate analysis unreliable. Standard definitions for infections acquired during ECMO support, specifically those believed ECMO-associated, and data specific to the timing and site of each infection, if added to the ELSO registry, would thereby facilitate more accurate data reporting, interpretation, and application for future infection control initiatives.

CONCLUSIONS

Rates of culture-proven infections on ECMO are high and may result in additional morbidity and mortality to an already vulnerable patient population. The potential need for prolonged ECMO support and mechanical ventilation may also present an additional burden to the healthcare system. Although considerable efforts have been made to prevent healthcare-associated infections, such as central line-associated bloodstream infections (^14–16), ventilator-associated pneumonia (^{17, 18}), and surgical site infections (¹⁹) in high-risk populations, little attention has been focused on the ECMO population which, given its need for invasive support apparatus and procedures, is at high risk for any and all of these infections. Although this investigation has helped to identify certain high-risk ECMO patient populations and common causal organisms, it is clear that more concerted efforts are needed.

ACKNOWLEDGMENTS

We thank all members of the ELSO Task Force on Infections and ECMO for their dedication and support.