A NATIONWIDE cohort study from Iceland has validated an ICD-10 coding algorithm for identifying gastrointestinal (GI) bleeding in patients receiving oral anticoagulation. The findings demonstrate high accuracy for detecting clinically significant bleeding events and highlight how registry-based research can be strengthened through validated diagnostic codes.
High accuracy and clinical relevance
Researchers analysed data from all patients receiving oral anticoagulants in Iceland between 2014 and 2019, using a national prescription database and hospital records from the country’s five major hospitals. GI bleeding events were identified through comprehensive searches of ICD-10 codes and were cross-referenced with endoscopy results and the national death registry.
Each potential case was manually confirmed by chart review, serving as the gold standard for validation.
The resulting ICD-10 coding algorithm achieved 61.3% sensitivity, 99.6% specificity, 90.0% positive predictive value (PPV), and 98.0% negative predictive value (NPV). Events identified by the algorithm were typically major bleeds, often requiring hospital admission, anticoagulant reversal, treatment discontinuation, or endoscopic intervention.
Strengths and limitations
Although the algorithm reliably detected severe cases, its sensitivity for less serious bleeds was modest. The authors suggest that integrating endoscopy databases and death registry information could further improve detection and data completeness.
Implications for research
The study supports the use of this validated ICD-10 algorithm in observational and pharmacoepidemiological studies, ensuring greater accuracy when assessing bleeding risks in anticoagulated populations and improving comparability across future research.
Reference
Ingason AB et al. ICD-10 code validation for gastrointestinal bleeding in patients receiving oral anticoagulation: nationwide cohort study. BMC Gastroenterol. 2025;DOI: 10.1186/s12876-025-04325-9.
