Objectives: To evaluate whether cardiac and noncardiac variables may be used to predict survival in children treated with extracorporeal membrane oxygenation (ECMO) after cardiopulmonary bypass and to determine when to discontinue ECMO support.
Design: Retrospective review.
Setting: Neonatal and pediatric intensive care units of Kosair Children's Hospital.
Patients: Fifty-nine children treated with ECMO after cardiopulmonary bypass from 1987 through 1996.
Interventions: None.
Measurements and main results: Medical, nursing, operative, and perfusion records for each patient were reviewed. The primary outcome measure was survival to hospital discharge. Cardiac and noncardiac variables were recorded at serial times. Nineteen of 59 patients (32%) survived. No cardiac variable was a clinically useful predictor of survival or marker for when to discontinue ECMO. Among the noncardiac variables, progressive multiple organ system dysfunction and development of a nosocomial infection were significantly associated with nonsurvival. No patient with a positive blood culture (n = 3) within the first 24 hrs of ECMO survived, and 21 of 24 children with a positive culture from any site during ECMO died (p = .007). Despite their higher mortality, children with positive cultures were supported with ECMO significantly longer than those with negative cultures (275+/-168 vs. 135+/-108 hrs, respectively; p = .0004). For all patients, the longest duration of ECMO that resulted in survival was 256 hrs. For children with a positive culture, the longest duration of support that resulted in survival was 200 hrs.
Conclusions: Support with ECMO beyond 256 hrs was not associated with survival. Progressive multiple organ system dysfunction and nosocomial infections have a negative impact on survival. Serious consideration should be given to discontinuing ECMO support whenever there is a progressive increase in the number of abnormally functioning organ systems, a nosocomial infection occurs, or native cardiac function has not improved significantly by 250 hrs of ECMO support.