Triggers for identifying anticoagulation-associated adverse drug events in hospitalized patients: a systematic review and meta-analysis.

Abstract

Background: Anticoagulation therapy presents a high risk of adverse drug events (ADEs) in hospitalized patients, highlighting the need for effective detection strategies in clinical practice.

Aim: The review aimed to identify triggers for detecting anticoagulation-associated ADEs in hospitalized patients and describe the performance of these triggers.

Method: PubMed, Cochrane, and Embase were queried until April 19, 2024. We included studies on trigger tools for detecting anticoagulation-related ADEs in hospitalized patients. The study quality was assessed using the Quality Assessment Trigger framework based on the QUADAS-2 tool for diagnostic accuracy. We performed random-effects meta-analyses to pool positive predictive values (PPV), with heterogeneity assessed via I² statistic and Cochran's Q test.

Results: Twenty-three studies were included. Seventeen triggers were reported and categorized into three modules: abnormal laboratory values (11 triggers), medications or antidotes (three triggers), and care (three triggers). Elevated international normalized ratio (INR) was the most prevalent trigger (16 studies) with a PPV of 0.539 (95% Confidence Interval [CI] 0.310-0.768). Other triggers exhibited lower pooled PPVs than elevated INR, including vitamin K administration (15 studies, PPV 0.222, 95% CI 0.153-0.290), abrupt medication cessation (14 studies, PPV 0.418, 95% CI 0.169-0.667), and activated partial thromboplastin time > 100 s (11 studies, PPV 0.336, 95% CI 0.173-0.498). Importantly, in-hospital stroke (three studies) had the highest pooled PPV of 0.964 (95% CI 0.803-1.000).

Conclusion: This review demonstrated the variability in PPVs of triggers for anticoagulation-associated ADEs, emphasizing the necessity for a specialized trigger tool for hospitalized patients on anticoagulants.