PMD8 - Bayesian Network Meta-Analysis on Prevention of Stroke For Patent Foramen Ovale Closure

Abstract

Stroke is associated with high rates of morbidity and is the globally second leading cause of death. Up to 40% of ischaemic strokes are cryptogenic. A network meta-analysis to compare the effectiveness of patent foramen ovale closure in patients with cryptogenic stroke or embolism. A systematic literature search for randomized clinical trials for patent foramen ovale closure was undertaken for the databases Pubmed, Embase, Biosis, Google Scholar and Cochrane. Data was collected for the study type, methods, country and key findings. Extracted study data included study design, patient charatcertristcis and stroke related outcomes. A bayesian random effects network meta-analysis model was developed in WinBUGS14. We identified 1430 references and found 10 studies on four randomized trials in 2963 patients. The devices included in our study were Amplatzer (AMP),STARFlex (STF), and HELEX (HLX). The odds ratios (ORs) for AMP versus HLX were 0.54 (0.05-5.38), AMP versus Medical 0.37 (0.07-1.35), AMP versus SLX 0.36 (0.05-1.87), HLX versus Medical 0.68 (0.05-7.78), HLX versus SLX 0.66 (0.07-6.00) and Medical versus SLX 0.99 (0.18-5.53). The probability to be best in preventing strokes was 66.7% for AMP, 25.8% for HLX, 4.5% for STF, and 2.9% for medical therapy. Patent foramen ovale closure with Amplatzer appears superior to medical therapy in preventing strokes in patients with cryptogenic embolism.