Anticoagulation for pregnant individuals with mechanical heart valves: a methodological review of systematic reviews

We aimed to describe methodological variations in systematic reviews (SRs) that estimated clinical outcomes with different anticoagulant strategies for pregnant individuals with mechanical heart valves (MHVs), and to provide most reliable risk estimates. We identified eligible SRs through a search involving eight databases and critically appraised (a) study quality using A MeaSurement Tool to Assess systematic Reviews (AMSTAR-2) (b) search strategies using Peer Review of Electronic Search Strategies (PRESS) guidelines and operationalized criteria of SR searches, and (c) meta-analytic approaches using a self-designed checklist. We determined most reliable estimates for clinical outcomes using an algorithm considering AMSTAR-2 scores, quality of search strategy and recency of publication. Of the 12 eligible SRs, most (9/12) were of critically low quality based on AMSTAR-2. Of the 4 that published search strategies, 3 were of low quality based on PRESS guidelines. Meta-analytic approaches varied widely. The most reliable risk estimates with VKAs were 0.9% [95% Confidence Interval (CI 0.1–1.6%)] for maternal mortality, 2.7% (1.4–4.0%) for thromboembolism, 35.5% (19.8–51.2%) for fetal loss and 2.0% (0.3–3.7%) for congenital anomalies. These risks with sequential therapy were 2.0% (0.8–3.1%), 5.8% (3.8–7.7%), 20.1% (14.4–25.7%) and 1.4% (0.3–2.5%), and with LMWH, they were 2.9% (0.2–5.7), 8.7% (3.9–13.4), 8.0% (2.0–13.9) and 0.0% (0.0–0.0) respectively. SRs on anticoagulation for pregnant individuals with MHVs demonstrate considerable heterogeneity in terms of study quality, search strategies, and meta-analytical approaches. The provided risk estimates could inform shared decision-making and clinical practice guidelines.