Back to Annual Meeting Program

Background: National pediatric ECMO cost and resource utilization are not well characterized. Hospital environments (HE) and their influence on ECMO outcomes are also unknown. The wisdom of extending regionalization initiatives in the provision of ECMO support requires evidence that such mandates would improve outcomes while reducing cost. We sought to define the prevalence of different HE’s where pediatric ECMO occurs, and the contemporary cost of pediatric ECMO.
Methods: The 2009 Healthcare Cost and Utilization Project (H-CUP) Kids’ Inpatient Database (KID) was searched for all discharges where ECMO support was provided. AHRQ HCUP cost-to-charge ratio files estimated hospital-specific costs from total hospital charges. Hospitals were segregated into terciles (high volume > 30 cases; medium volume 15-30 cases; low volume < 15 cases) based on total annual ECMO volume, and then were classified as small, medium, or large depending on total bed size as delineated in the KID. Four HE’s were then created which combined hospital bed size and annual ECMO volume, including: small, low-volume; small, high volume; large, low volume ; large, high volume. ECMO cases were classified as cardiac or non-cardiac based upon indication. Cardiac cases were mapped to Risk Adjustment for Congenital Heart Surgery (RACHS) categories. Weighted linear and logistic regression determined multivariable factors associated with prolonged hospital length of stay (LOS), increased hospital cost, and increased prevalence of in-hospital death.
Results: We identified 1670 (national estimate 2541 ± 220) pediatric ECMO cases from 7.3 million national pediatric discharges in 4,121 hospitals. Median age was less than 1 year (range birth - 20 yrs). Nearly all ECMO cases occurred at urban teaching hospitals (2221 ± 207; 99%), with only 1% occurring at either urban nonteaching or rural hospitals. Based on the four HE’s: ECMO cases occurred at 94±35 (4%) small, low-volume centers, 696±80 (27%) large, low-volume centers, and 170 ± 120 (7%) large, high volume centers. No small, high-volume centers were identified. Median hospital LOS was 31 days (range 0 - 332 days). Mean estimated hospital cost was (229,340 ±148,403 USD). Estimated hospital costs were significantly higher in small, low-volume ECMO centers (298,545 ± 140,690 USD) compared to others (227,475 ± 148,214 USD), P=0.007. Hospital LOS tended to be longer at small, low-volume hospitals as well (47.5 days) compared to others (44.7 days), though statistical significance was not reached. In-hospital death was higher at small, low-volume hospitals (51%) compared to others (44%). Multivariable analysis demonstrated that high volume ECMO centers (Odds ratio 0.26 [95% C.I. 0.07-0.93] P-value 0.03) and lower RACHS categories (Odds ratio 0.65 [95% C.I. 0.44 - 0.96] P=0.04) reduced the risk of in-hospital death.
Conclusions: Pediatric ECMO currently is performed in a range of HE’s, including small, low volume centers. The contemporary national cost of pediatric ECMO is substantial. Regionalization of care, in which the majority of pediatric ECMO support for complex cardiac surgery patients is concentrated in large, high volume centers, should reduce resource utilization and improve in-hospital survival.