Use of non-randomised evidence alongside randomised trials in a systematic review of endovascular aneurysm repair: strengths and limitations.

OBJECTIVE: To assess whether limitations of randomised controlled trials (RCTs) of endovascular aneurysm repair (EVAR) can be addressed by evidence from non-randomised studies.

DESIGN: Analysis of data from a systematic review.

METHODS: We conducted a review of EVAR versus open repair or non-surgical management of abdominal aortic aneurysms. In addition to RCTs, we included pre-specified registries of EVAR and open repair.

RESULTS: The six included RCTs randomised patients in 2003 and earlier. Of the three registries included, one contributed data on a large (>8000) sample of patients treated with newer generation EVAR devices and followed up for up to 8 years. However, treatment dates of these patients overlapped with those of the RCTs. The other registries were of limited usefulness. A large (>45,000) controlled observational study published while the review was in progress broadly supported the findings of RCTs comparing EVAR with open surgery. A comparison of outcomes across all studies did not support the hypothesis that the findings of the RCTs are no longer representative of clinical practice.

CONCLUSIONS: Both randomised and non-randomised sources of evidence have strengths and weaknesses for assessing the effectiveness of EVAR. Further research should explore the optimum use of registry data, including patient-level analyses.