DOI: 10.1093/bjs/znag093.035 ISSN: 0007-1323

Identification of patients with idiopathic colon ischemia in the Swedish ESPRESSO cohort: a validation study

Helene Klinglöf, Jonas Söderling, Charlotte R H Hedin, Thomas Frisell, Anna Löf Granström, Jonas F Ludvigsson, Åsa H Everhov

Abstract

Introduction

Colon ischemia (CI) is the most common form of intestinal ischemia, yet case identification in epidemiological research remains challenging due to heterogeneous diagnostic criteria and absence of a specific ICD code. The use of SNOMED (Systematized Nomenclature of Medicine) for gastrointestinal biopsies and surgical specimens enables a unique possibility to identify CI cases.

Methods

In this validation study, 231 individuals with SNOMED code M54 (necrosis) and colorectal topography codes (T67-T68) were randomly selected from five Swedish regions. Medical charts were retrieved for 204 cases, where 143 contained sufficient information for evaluation. Cases were classified according to Brandt & Boley criteria and assigned an etiological group. Positive predictive values (PPVs) were calculated for overall IC based on histopathology alone and for etiological subsets where histopathology was combined with ICD codes.

Results

Histopathology confirmed ischemia of any cause in 99% (141/143) cases. According to Brandt & Boley criteria, 14 (10%) patients had definite CI, 0 had probable CI, 78 (54%) had possible CI, and 51 (36%) no CI. Common triggers included postoperative complications after treatment for colorectal cancer, intestinal obstruction, inflammatory disorders, arterial disease. Idiopathic CI was the most common etiological group (44%). Combining SNOMED with ICD codes yielded a PPV of 84% for idiopathic CI, increasing to 92% when using stricter ICD-10 criteria (K55.x or K52.8/9).

Discussion

Histopathology-based identification using SNOMED codes provides high diagnostic accuracy for CI of any cause. Combining SNOMED with ICD codes offers a feasible approach to identify patients with idiopathic CI, for future population-based research.

More from our Archive