The use of extracorporeal membrane oxygenation can be rationalised by the assumption that non-zero survival after refractory cardiorespiratory failure represents improved outcome. Survivors may have cognitive and or functional morbidities, require complex ongoing care, and as a consequence consume considerable healthcare resources.T he use of extracorporeal membrane oxygenation (ECMO) can be rationalised by the assumption that non-zero survival after refractory cardiorespiratory failure represents improved outcome. Survivors may have cognitive and or functional morbidities, require complex ongoing care, and as a consequence consume considerable healthcare resources. 1 We performed a retrospective evaluation of the survival, morbidity, and hospital based costs of children receiving ECMO between January 1997 and May 2002. The indication for ECMO, survival, neurological outcome as assessed by the pediatric cerebral and overall performance category scores, 2 and resource use were abstracted from medical records. Costs were based on in-house estimates. Local billing fees were applied where in-house estimates were not available. Five year hospital based costs in CAD$2003 were calculated beginning at ECMO initiation. A 5% discount rate was applied in arrears.We excluded resources consumed before the institution of ECMO, costs to families, and blood products (centrally funded). For logistic reasons, the cost of drugs was not included. Years with missing data were imputed based on resource use from survivors. Costs were converted into J (2003).Eighty of 83 children who received ECMO during the study period were evaluated; 47 (59%) survived 24 hours after decannulation, 31 (39%) survived to hospital discharge, and 28 (35%) were alive one year after hospital discharge. The five year costs were J5 117 551 in total and J165 082 per child surviving to hospital discharge (table 1); 85% of costs occurred in the first year, and 53% were due to hospital stay (mean 37.5, maximum 275 days).Morbidity was assessed in 30 (97%) survivors at hospital discharge. Three (10%) had no functional impairment, and nine (30%) were developmentally normal. Nineteen (68%) children were assessed at 12 months; function was normal in four (21%) and mildly or moderately impaired in 15 (79%) (table 2).These findings highlight the importance of identifying the children who are most likely to benefit from ECMO. In our institution, as in others, the limited evidence to support the use of ECMO for specific indications has necessitated ECMO use on a case by case basis. Reported patterns of ECMO use are changing, 3 and the indications for ECMO are broadening and now include sepsis and cardiac arrest. 4 We found, retrospectively and in a single centre cohort, that fewer than two of five children who received ECMO survived to hospital discharge. Improved outcome prediction before initiation of ECMO has great potential to facilitate improvements in the outcome of ECMO, improve end of life decision making, and will permit more efficient use of healthcare resource...