Abstract 616 Poster Session I, Saturday, 5/1 (poster 181)

ECMO is a technique of providing cardiopulmonary bypass to newborn infants with severe but potentially reversible pulmonary failure. While it has been in use for over 20 years there has been persistent discussion on whether enough evidence exists to justify its widespread use. Even with the recent UK collaborative study there is still comment on the ethics of this large study (Paed.Peri.Epid. July 1997). The purpose of this abstract is to review systematically and cumulatively the effects of ECMO on infant deaths in the four randomized trials. The four trials are varying methodological quality, most with small sample sizes, with one study published only in a preliminary form. Only infants entering the trials by randomization were included in the analysis. With some trial having no deaths in the ECMO group, to aid in the calculations, a correction factor of .5 was added to each cell in all trials. A logarithm of the RR method and a fixed effect model was used. Below is a table of the cumulative meta-analysis with 1st trial author, Relative Risk(RR), lower confidence interval(UCI), upper confidence interval(UCI), P association and P heterogeneity. The systematic review (complete results not in abstract) suggests that only with the completion of the UK study was a high level of statistical significance obtained in demonstrating a reduced risk of death in the ECMO group (Pooled common relative risk .48(95 % confidence interval: .36,.80, Chi square p = <.001 and Q Cochran test for heterogeneity = .47). The cumulative meta-analysis (presented in table above) suggests that statistical significance was obtained after the third study (Gross et al), although this trial has been only presented in a preliminary form and has a lower methodological quality score than the other trials.

Table 1 No caption available.
Full size table