Development and Validation of a Scoring System to Identify Patients With Microscopic Colitis

BACKGROUND & AIMS: Diarrhea is a common indication for colonoscopy. Biopsies are collected and analyzed from patients with a macroscopically normal colon to exclude microscopic colitis (MC), but the diagnostic yield is low because most patients have functional disease. We developed and validated a diagnostic scoring system to identify patients with MC to reduce the need to collect biopsies from all patients. METHODS: We performed a retrospective study, which analyzed demographic and symptom data from adult patients with chronic diarrhea evaluated by colonoscopy and biopsy at 3 endoscopy centers in Leeds, United Kingdom. To derive the scoring system, we analyzed data from 476 adult patients (mean age, 53.6 years; 63.7% female) examined in 2011. Factors significantly associated with the presence of MC were assigned item scores, and total scores were determined for each patient. To validate the system, we used it to assess data from 460 patients (mean age, 52.9 years; 59.8% female) examined in 2012. The primary aim of the study was to determine the performance of the diagnostic scoring system in identifying patients with MC by using histologic findings as a reference. RESULTS: In the derivation cohort, 85 patients were diagnosed with MC on the basis of histologic analysis. Age >= 50 years, female sex, use of proton pump inhibitors or nonsteroidal anti-inflammatory drugs, weight loss, and absence of abdominal pain were significantly associated with MC. We created a scoring system for diagnosis of MC, with scores ranging from -8 to +38; scores >= 8 were used to identify the presence of MC. This cutoff value identified patients with MC in the validation cohort (74 patients, 16.1%) with 90.5% sensitivity and 45.3% specificity (area under the receiver operating characteristic curve value, 0.76). Because of its ability to exclude MC and therefore avoid the need for routine collection of colonic biopsies, this scoring system reduced the cost of evaluation by >7000 pound in the cohort. CONCLUSIONS: We collected data on risk factors for MC to create a scoring system that identifies patients with MC with more than 90% sensitivity. This system can also reduce costs by identifying patients who are unlikely to have MC who do not require biopsy analysis.