Utilization of extracorporeal membrane oxygenation (ECMO) for adult critically ill patients is increasing, but data in obstetric cohorts are scant. This study analyzed ECMO utilization and maternal outcomes in obstetric patients in the United States.
Data were abstracted from the 1999–2014 National Inpatient Sample (NIS), a 20% US national representative sample. ECMO hospitalizations (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] code 39.65) in patients ≥15 years of age were categorized into obstetric ECMO and nonobstetric ECMO. Obstetric patients included 4 categories: (1) loss or termination of pregnancy, (2) delivery (term or preterm), (3) postdelivery hospitalization, and (4) pregnancy without an obstetrical outcome. Possible underlying causes for obstetric ECMO were identified by analysis of ICD-9-CM codes in individual records. In-hospital death was abstracted from the NIS, and ECMO complications were identified using ICD-9-CM algorithms. Statistical significance in time-effect was assessed using weighted regression models.
During the 16-year study period, 20,454 adult ECMO cases were identified, of which 331 occurred in obstetric patients (1.6%; 95% confidence interval [CI], 1.4–1.8). Obstetric ECMO utilization rate was 4.7 per million obstetric discharges (95% CI, 4.2–5.2). The top 3 possible indications were sepsis (22.1%), cardiomyopathy (16.6%), and aspiration pneumonia (9.7%). Obstetric ECMO utilization rate increased significantly during the study period from 1.1 per million obstetric discharges in 1999–2002 (95% CI, 0.6–1.7) to 11.2 in 2011–2014 (95% CI, 9.6–12.9), corresponding to a 144.7% increase per 4-year period (95% CI, 115.3–178.1). Compared with nonobstetric ECMO, obstetric ECMO was associated with decreased in-hospital all-cause mortality (adjusted odds ratio [aOR] 0.78; 95% CI, 0.66–0.93). In-hospital all-cause mortality for obstetric ECMO decreased from 73.7% in 1999–2002 (95% CI, 48.8–90.8) to 31.9% in 2011–2014 (95% CI, 25.2–39.1), corresponding to a 26.1% decrease per 4-year period (95% CI, 10.1–39.3). Compared with nonobstetric ECMO, obstetric ECMO was associated with significantly increased risk of both venous thromboembolism without associated pulmonary embolism (aOR 1.83; 95% CI, 1.06–3.15) and of nontraumatic hemoperitoneum (aOR 4.32; 95% CI, 2.41–7.74).
During the study period, obstetric ECMO utilization has increased significantly and maternal prognosis improved.