Validation and modification of a diagnostic scoring system to predict microscopic colitis

Objective: many patients with diarrhoea undergo colonoscopy. If this is macroscopically normal, random biopsies are obtained to rule out microscopic colitis (MC), but most patients have functional disease. Accurate predictors of MC could avoid the need to take biopsies in a substantial proportion of patients, saving money for the health service. We validated a previously described diagnostic scoring system for MC, and incorporated further variables to assess whether this improved performance. Material and methods: consecutive adults with loose stools undergoing colonoscopy in Leeds, UK were included. Demographic and symptom data were collected prospectively. The diagnostic scoring system described previously was applied. In addition, the incorporation of further variables, including drugs associated with MC, number of stools, nocturnal passage of stools, and duration of loose stools, into the scoring system was assessed. Sensitivities, specificities, and positive and negative predictive values were calculated. Results: among 242 patients (mean age 51.0 years, 163 (67.4%) female), 26 (10.7%) of whom had MC, a cut off of ≥4 on the original scoring system had a sensitivity of 92.3% and specificity of 35.2%. Nocturnal passage of stools and duration of loose stools <6 months were significant predictors of MC. Incorporating these variables in a new scoring system with a cut off of ≥6 identified MC with 95.7% sensitivity and 46.0% specificity. Conclusions: incorporating nocturnal passage of stools and duration of loose stools into the scoring system may improve ability to predict MC, and avoid random colonic biopsies in a greater proportion of patients with loose stools.