Derivation of a Microscopic Colitis Disease Activity Index (MCDAI)

Abstract

Purpose: Microscopic colitis is a common cause of chronic diarrhea, often with additional symptoms. No validated instruments exist to assess disease activity in microscopic colitis, making it difficult to draw meaningful conclusions regarding efficacy between treatments in clinical series and controlled trials. We aimed to identify the clinical features which independently predicted physician global assessment (PGA) of disease severity and create a microscopic colitis disease activity index (MCDAI). Methods: A cohort of patients with microscopic colitis was prospectively administered a survey assessing a number of GI symptoms, as well as their global assessment (GA) of disease activity on a 10-point scale. A single physician also scored a physician global assessment (PGA) on a 10-point scale, considering the individual symptoms recorded by the patient, but blinded to the patient's GA. The primary analysis utilized a multiple linear regression model to identify which individual patient-reported symptoms were independently associated with the PGA using a backward variable selection approach. Due to some outlier values, individual data points were first transformed to rank scale. We also assessed the degree of agreement between the patient's GA and the PGA. Results: Of 165 patients enrolled, 14 (8.5%) did not complete the survey. The remaining 151 patients had a median age of 66 years (range, 1-87); 74% were female. Five symptoms were independently associated with the PGA (in order of strength of correlation: # unformed bowel movements (BM) per day, incontinence, abdominal pain, # nocturnal BM, weight loss). When combined, these scores strongly predicted the PGA (R2 = 0.84): X1(#BM) + X2 (#incontinence episodes per month) + X3 (abdominal pain [scored 1-10]) + X4 (# nocturnal BM) + X5(weight loss in pounds). Urgency and the use of anti-diarrheal medications were not associated with PGA. The patient's GA had good agreement with the PGA (concordance correlation coefficient = 0.63); 60% of the time, these scores were within +/- 1 point. PGA was >= 2 points higher than the patient's score in 11% and >= 2 points lower in 29%. Conclusion: Five symptoms independently correlated with disease activity as measured by PGA. Scores for these symptoms were combined to create an MCDAI that strongly predicted the PGA. PGA had good correlation with patient's GA. This index can be used to objectively and consistently score disease severity in clinical trials and case series. This work was supported by a clinical research grant from the American College of Gastroenterology.