The Safety of Device Registries for Endovascular Abdominal Aortic Aneurysm Repair: Systematic Review and Meta-regression

Objectives: New and re-designed stent grafts for endovascular aortic aneurysm repair (EVAR) are released regularly. Manufacturers use data from registries to assess stent graft performance, but little is known about the ability of such registries to detect rates of clinically relevant complications. The aim of this paper was to perform a systematic review and meta-analysis to determine pooled failure rates for EVAR stent grafts, to define an acceptable non-inferiority limit for these devices, and then to calculate the number of patients needed for a new device to achieve non-inferiority against published devices. Data sources and review methods: MEDLINE and EMBASE were searched for studies reporting outcomes of specific EVAR grafts being used for intact infrarenal abdominal aortic aneurysms, from inception to November 2016. Meta-regression was performed to pool data and calculate the patient numbers needed to detect non-inferiority of a future graft performance. An expert consensus was performed to define adequate standards for device safety. Results: One hundred and forty-seven moderate quality papers involving 27,058 patients were included. Multiple outcomes were pooled. Of these, the estimated rate (+/- standard error) of overall endoleak (excluding Type II) at 2 years was 5.7 +/- 0.6%. The pooled re-intervention rate was 11.1 +/- 0.7% at 2 years. There were differences in pooled endoleak rates between different stent graft types. Expert consensus defined non-inferiority as better performance than the worst performing 25% of stent grafts. The most popular outcome in the expert consensus was cumulative endoleak rate (excluding Type II). The number of patients who would need to be enrolled in a registry to demonstrate non-inferiority at this level was 525. Only two of 147 included studies achieved this. The second most popular choice in the expert consensus was re-intervention rate; 492 patients are required to demonstrate this. Conclusions: Five hundred and twenty-five patients need to be entered into a registry to demonstrate non-inferiority to previous stent grafts. Almost all previous publications have captured lower patient numbers. With performance varying between devices, and new devices being introduced regularly, there is an urgent need to capture higher quality long-term data on EVAR stent grafts. (C) 2017 European Society for Vascular Surgery. Published by Elsevier Ltd. All rights reserved.