Extracorporeal membrane oxygenation (ECMO) has been used in the care of critically ill neonates since the 1970s. Some neonatal ECMO indications include inadequate tissue oxygen delivery, severe hypoxic respiratory failure, persistence of the fetal circulation, and severe pulmonary hypertension with right and/or left ventricular dysfunction. Because ECMO remains resource-intensive and medically complex, there is ongoing interest in reporting trends and outcomes related to use across multiple centers.

We constructed a multicenter cohort of neonatal ECMO cases using hospital discharge data from the New York State Department of Health (SPARCS). Neonatal ECMO cases between January 1, 2018, and December 31, 2024, were isolated using relevant procedure and diagnosis codes. Key patient demographic, situation, and outcome variables were extracted. Incidence of neonatal ECMO per 100,000 inpatient discharges and survival to discharge were calculated. Comparisons across years and subgroups (race/ethnicity, birth weight quartiles, hospital case-volume quartiles) were performed using two-tailed Z-tests with significance defined as P < 0.05.

A total of 203 neonatal ECMO cases between 2018 and 2024 met inclusion criteria: 49.8% (n = 101) were male, and 29.9% (n = 61) were White. The overall incidence of neonatal ECMO was 1.33 cases per 100,000 discharges, and overall survival to discharge for the cohort was 58.6% (n = 119). Significantly lower incidence was seen in 2022 and 2023 (P < 0.05). No single year was associated with significant differences in survival to discharge relative to other years (P > 0.05). Similarly, no significant difference was seen in survival to discharge across patient race/ethnicity or hospital ECMO volume (P > 0.05). Neonates in the lowest quartile for birth weight (≤2700 g) were associated with significantly lower survival to discharge (P = 0.0009).

Our analysis of a large multicenter patient database from New York State reveals broad trends surrounding the use of ECMO in neonates. Further study is needed to identify if the trends observed are replicated in other settings.