Network meta-analysis on patent foramen ovale: is a stroke or atrial fibrillation worse?

Systematic reviews suggest that patent foramen ovale closure (PFOc) is performed percutaneously with low complication rates. We did a network meta-analysis (NMA) comparing devices for PFO closures, evaluating safety and efficacy of transcatheter PFOc in preventing neurological events in patients with stroke when compared with medical therapy (MT), and assessing risk of atrial fibrillation (AF).

We searched 3 databases (MEDLINE, EMBASE, CENTRAL/CCTR) identifying six randomized controlled trials from 2012 until December 2019. We performed a Bayesian NMA; number-needed-to-treat and number-needed-to-harm were derived by applying the estimated odds ratios (ORs). The likelihood of being helped or harmed (LHH) was evaluated to estimate the risk-effectiveness balance.

The 3560 patients allocated to PFOc were less subject to a stroke than patients with MT. The overall ORs of PFOc versus MT were 0.41 with fixed-effects, and 0.22 with random-effects model. NMA proves that PFOc induces AF episodes significantly higher than MT, even when analysis is limited to only new episodes of "serious AF." LHH (0.68 fixed-effects, 0.79 random-effects) showed that strokes saved are less than cases of AFs added. By considering only serious AF, strokes saved are higher than serious AFs induced by the PFOc (LHH was 3.46 and 4.00 respectively).

NMA supported PFOc in patients with cryptogenic stroke, confirming that devices are better than MT, but increase the risk of AF by over 2/4 times (serious or unserious AF). Considering serious AFs (real risky clinical condition), patients have more advantages in being treated, since LHH is ≥ 3-4.