In this retrospective cohort study, we identified patients aged 18 years and older who underwent colonoscopy between January 1, 2007, and December 31, 2016 at Columbia University Medical Centre, New York City, NY and Mayo Clinic, Rochester, MN. Cases were defined as patients who had a new diagnosis of biopsy-proven MC. Controls were defined as those who underwent colonoscopy for evaluation of diarrhoea with biopsies negative for MC. Cases were matched with up to two controls by age (±3 years), sex, and date of colonoscopy (±6 months). Patients with documentation of a prior histologically proven diagnosis of MC at the time of colonoscopy were excluded. Patients with a prior history of Crohn's disease or ulcerative colitis, identified by ICD-9 and ICD-10 codes (55–6, 558, K50-K52), were also excluded. In total, we identified 185 patients at Columbia University and 159 patients at Mayo Clinic who had MC and matched them with 371 and 297 controls respectively.
Data were collected on patient sociodemographic information including age at time of colonoscopy, gender, smoking history (current, former, never and unknown) and presence of coeliac disease. Coeliac disease was identified by chart review of clinical notes, laboratory parameters and histology findings at the Mayo Clinic and based on a prospectively maintained database at Columbia University, as previously described.
Concomitant medication use at time of colonoscopy was identified for patients in both groups. At Columbia University, medication use was predominantly gathered from the standardised pre-colonoscopy interview with a nurse on the day of the procedure, where the patient was directly asked to list all current outpatient medications, as per the protocol of the endoscopy unit. For 2% of patients at Columbia University, medication lists from pre-procedure interviews were not available, and manual electronic medical record review within 6 months of index colonoscopy date was performed. At Mayo Clinic, medication use was entirely gathered based on electronic medical record review either from clinical notes within 1 month from index colonoscopy date (80%) or clinical notes from the day of the index colonoscopy procedure (20%). The following medication classes were obtained: angiotensin converting enzyme inhibitors (ACE-Is), angiotensin II receptor blockers (ARBs), aspirin, B-blockers, benzodiazepines, bisphosphonates, diuretics, H2 blockers, PPIs, NSAIDs, oral diabetes medications, serotonin norepinephrine reuptake inhibitors (SNRIs), SSRIs and statins. For the purposes of this analysis, we grouped SNRIs and SSRIs into one combined category and included aspirin in the larger NSAIDs category.
Conditional logistic regression was used to measure the association between MC and medication use and multivariate conditional logistic regression was used to measure this association adjusted for smoking, which included a category for unknown smoking history. We then tested for interaction between coeliac disease and medication use with regard to their association with MC. We also stratified the medications found to be significant by histologic type of MC. Post-hoc analyses were performed that assessed the association between PPI use and MC by cohort site and also the association between MC and PPIs, adjusted for other medications that were found to be significant on univariate analysis and/or based on relevance in other studies. Associations were calculated as odds ratios (OR) and corresponding 95% confidence intervals (CI). SAS version 9.4 (Cary, NC) was used for all analyses. The Institutional Review Boards at Columbia University Medical Centre and Mayo Clinic approved this study.
Aliment Pharmacol Ther. 2021;53(11):1209-1215. © 2021 Blackwell Publishing