Antibiotics Associated With Increased Risk of New-onset Crohn's Disease but not Ulcerative Colitis

A Meta-analysis

Ryan Ungaro MD; Charles N Bernstein MD; Richard Gearry MB ChB; PhD; Anders Hviid MSc; Kaija-Leena Kolho MD; PhD; Matthew P Kronman MD; MSCE; Souradet Shaw MSc; Herbert Van Kruiningen MD; Jean-Frédéric Colombel MD; PhD; Ashish Atreja MD; MPH


Am J Gastroenterol. 2014;109(11):1728-1738. 

In This Article


Identification and Retrieval of Primary Studies

We followed the MOOSE (Meta-analysis of Observational Studies in Epidemiology) guidelines for our study.[20] We performed a search of PUBMED/Medline, Embase, and Cochrane Library (Database of Abstracts of Reviews of Effect and Health Technology Assessment Database) for articles written in English that examined exposure to antibiotics before a new diagnosis of IBD. The following terms were used in our search strategy: Inflammatory bowel disease or Crohn's disease or ulcerative colitis and antibiotics or penicillin or cephalosporin or tetracycline or doxycycline or fluoroquinolone or macrolide or sulfonamide or metronidazole. All databases were searched from their inception through October 2013. Meeting abstracts from major gastroenterology conferences (Digestive Disease Week and American College of Gastroenterology Annual Meeting) from the past 2 years were searched to identify studies that were potentially missed in our database search. Articles were selected for full text review based on title and abstract. We also manually searched the bibliographies of retrieved publications (backward snowballing) to further increase the yield of potentially relevant articles.

Inclusion and Exclusion Criteria

We searched for original articles in English language on cohort studies and case–control studies that provided data on antibiotic exposure before new diagnoses of IBD. To be included, studies had to report on antibiotic exposure before new-onset IBD. Included studies needed to define IBD cases using independent validation or record linkage in order to verify diagnosis. Antibiotic exposure was determined by either medication prescription records or by patient survey. Studies that reported results for CD or UC individually were also included. Case–control studies must have compared IBD patients with a matched or random sample of individuals without IBD (control subjects, cohort without IBD, or general population). Studies were excluded if they did not report data on antibiotic exposure before IBD diagnoses, did not have a control group, or used a surrogate marker for antibiotic use (e.g., diagnosis of infection).

Study Selection and Data Extraction

Two reviewers (R.U. and A.A.) worked independently to determine whether a study met the inclusion criteria, collected information to assess the methodological validity of each included study, and extracted data with structured data extraction forms. The reviewers resolved discrepancies by jointly reviewing the study in question. If no consensus was reached on whether a study should be included, a third reviewer (J.-F.C.), unaware of the other reviewers' determinations, functioned as an arbitrator. For each publication, we extracted details on research design, country in which the study was conducted, average age of study participants, total number of subjects (cases, controls, and/or cohort size), number of subjects exposed to antibiotics, subjects with primary outcome (new diagnosis of IBD, CD, and/or UC), frequency of antibiotic exposure, and multivariate adjusted risk estimates (relative risk, odds ratio, or hazard ratio) with corresponding 95% confidence interval (CI) when available. We also recorded whether the study excluded any time between antibiotic exposure and IBD diagnosis in its primary analyses to account for possible diagnostic delay.

For studies that analyzed specific antibiotics, we extracted data on the frequency of exposure to different antibiotic classes in patients with newly diagnosed IBD vs. controls. In addition, we extracted the frequency of antibiotic exposure and risk estimates stratified by the magnitude of exposure. Magnitude was quantified by the number of courses of antibiotic treatment before new-onset IBD. For studies in which raw or adjusted data were not available in the published article for outcomes of interest, the corresponding author was contacted to see whether data could be provided. All authors who provided further data were offered coauthorship and were allowed to review the collected data and article.

Quality Assessment

The Newcastle–Ottawa Scale (NOS) was used to assess the quality of the observational studies included in our meta-analysis.[21] The NOS has two different instruments for assessing case–control and cohort studies. Each instrument includes measures of quality in three domains: selection, comparability, and exposure. A study can receive up to one point for each of four areas measured within the selection domain and for each of three areas measured within the exposure domain. A maximum of two points can be assigned within the comparability domain. The highest possible score is nine. High-quality studies were considered to have a score of seven or greater. Study quality was assessed independently by two of the investigators (R.U. and A.A.), and any discrepancies were addressed by a joint reevaluation of the original article.

Statistical Analysis

We derived estimates of pooled odds ratio (OR) and 95% CI for outcomes of interest using random-effects models given the presence of heterogeneity between studies. Data were pooled if three or more studies provided the appropriate information. Whenever possible, estimates adjusted for potential confounding variables were used (OR for case–control studies and relative risk or hazard ratio for cohort studies). Heterogeneity between studies was assessed using the Q and I 2 statistics. Publication bias was evaluated with the use of funnel plots and Egger's test for asymmetry.

The primary outcome was risk of new-onset IBD, CD, and UC for any antibiotic exposure. Other outcomes investigated were risk of new-onset IBD, CD, and UC for any antibiotic exposure stratified by adults vs. children, risk of new-onset IBD, CD, and UC stratified by magnitude of antibiotic exposure (measured by total number of antibiotic courses as defined in the primary studies), and risk of new-onset IBD with exposure to specific antibiotic classes. Studies were classified as investigating adult-onset IBD if the study population included patients ≥18 years of age. Childhood-onset IBD studies included patients who were newly diagnosed under the age of 18 years. With regard to specific antibiotic classes, there were insufficient data to stratify by CD and UC, and thus analyses were reported as risk of new-onset IBD. We also conducted a number of sensitivity analyses, including stratifying by study quality using NOS score and excluding survey studies. For analyses other than tests of heterogeneity and publication bias, any P value of <0.05 was considered to be statistically significant. All analyses were conducted using Comprehensive Meta-Analysis Version 2 software (Biostat, Englewood, NJ).