Developing and Validating a Trigger Tool to Improve Detection of Anticoagulation-Associated Bleeding in Hospitalized...

Anticoagulant-associated bleeding is an important patient safety concern in hospitalized patients. These events may be underdetected in routine clinical practice. The authors developed and validated a trigger-based surveillance approach to support anticoagulation-associated bleeding detection in hospitalized adult general medicine patients.

The trigger tool for identifying anticoagulation-associated bleeding was developed through a systematic review and a two-round modified Delphi consensus. A retrospective observational study was conducted in hospitalized adult patients receiving anticoagulants at a tertiary care hospital between March and June 2024 to evaluate performance and feasibility of the trigger tool. The tool was applied to identify potential bleeding events, which were then assessed by pharmacists for causality, severity, and preventability. Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of individual triggers and the overall tool were evaluated.

Fourteen triggers were included in the Delphi consensus process, resulting in 11 candidate triggers for validation. Among 395 admissions, 50 in-hospital bleeding events were identified in 43 patients (10.9%) on anticoagulation therapy. Most bleeding events caused temporary harm (60.0%), while 40.0% were serious (prolonged hospital stay or life-threatening situation). Overall, 40.0% of events were considered preventable. The trigger tool showed a sensitivity of 62.8% (95% confidence interval [CI] 46.7%-77.0%), specificity of 90.9% (95% CI 87.4%-93.7%), PPV of 45.8% (95% CI 32.7%-59.2%), and NPV of 95.2% (95% CI 92.4%-97.3%).

This anticoagulation-associated bleeding trigger tool is a promising approach to enhance the identification of potential bleeding events in hospitalized adult patients. Further evaluation is needed to assess its applicability across broader patient populations and clinical settings.