Methods: A search of the English-language literature from January 1994 until March 2006 was performed. Inclusion criteria for articles were: patients who underwent endovascular repair were compared to patients who underwent open surgery, each treatment group included at least 5 patients, information about patient's hemodynamic condition upon presentation to the hospital was reported, and (30-day) mortality was reported for each treatment group. Two reviewers independently extracted the data, and discrepancies were resolved by an arbiter. Random effects models and meta-regression analysis were used to calculate crude and adjusted odds ratios (endovascular versus open repair).

Results: Ten studies met the inclusion criteria in which the results of 478 procedures were reported (n=148 for endovascular repair, n=330 for open surgery). All studies were observational; no randomized controlled trials were found. The pooled 30-day mortality was 22% (95%-confidence interval 17-30%) for endovascular repair and 40% (95%-confidence interval 34-47%) for open surgery. The pooled systemic complication rate was 30% (95%-confidence interval 20-46%) for endovascular repair and 50% (95%-confidence interval 38-66%) for open surgery. The crude odds ratio for the 30-day mortality for endovascular repair compared to open surgery was 0.45 (95%-confidence interval 0.28–0.72). After adjustment for patients' hemodynamic condition, the odds ratio was 0.67 (95%-confidence interval 0.31–1.44). A funnel plot was constructed to indicate the presence of publication bias; this plot showed that publication bias favoring the results after endovascular repair could not be ruled out.

Conclusion: In our systematic review, endovascular repair for patients with a ruptured AAA resulted in lower 30-day mortality compared to open surgery. This benefit, however, diminished in magnitude and was no longer significant after adjustment for patients' hemodynamic condition. Furthermore, studies comparing endovascular repair and open surgery in patients with a ruptured AAA were subject to publication bias.