Introduction: Sepsis can be challenging to identify in patients who are on extracorporeal membrane oxygenation (ECMO) or continuous renal replacement therapy (CRRT) due to temperature regulation, frequent fluctuations in blood pressure, and immune dysregulation. Many centers in the United States perform daily surveillance cultures for these patients, though this may be of low diagnostic yield and may increase unnecessary antibiotic exposures. This study aimed to 1) determine the positivity rate of surveillance blood and urine cultures for true infections, 2) compare antimicrobial exposure for patients who received surveillance cultures vs those who did not, and 3) evaluate whether other clinical criteria (leukopenia, leukocytosis, c-reactive protein, procalcitonin) can be used to help in the diagnosis of bacterial infections. Methods: This was a retrospective, longitudinal, single-center cohort study of patients less than 18 years of age admitted to the neonatal, cardiac, or pediatric intensive care units who required ECMO and/or CRRT for at least 1 day over a 4-year period (2018-2022). Summary statistics, Wilcoxon rank-sum tests, and generalized linear mixed models were used to analyze the data. Results: Overall, 390 patients were included in the cohort. The total positivity rate of surveillance blood cultures was 1.3% vs 11.9% for cultures driven by clinical change (p< 0.001). For urine cultures it was 4.2% vs 9.4%, respectively. Patients without surveillance cultures sent received a median of 4 days of antibiotics (interquartile range (IQR): 3.00, 6.50) vs patients with surveillance cultures sent who received a median of 7 days of antibiotics (IQR: 4.0, 15.0, p< 0.001). The total percentage of days on antibiotics for patients on ECMO and/or CRRT was higher in those with 1+ surveillance culture sent versus those without (p=0.01). Multivariable analysis showed no significant associations between lab values and positive cultures, after adjusting for age, CRRT use, and oncologic illness. Conclusions: Surveillance cultures in children on ECMO and/or CRRT are very low yield compared to those prompted by clinical change. Surveillance cultures were associated with a greater number of days of antibiotics and a greater percentage of days of antibiotics while on extracorporeal support.
Dagher et al. (Sun,) studied this question.