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

Disclosures

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

In This Article

Results

A total of 4,551 records were initially retrieved using our database search strategy (Figure 1). After reviewing article titles and abstracts, a total of 34 articles were included for full text review. Review of the bibliographies of retrieved publications identified five additional articles for possible inclusion.[13,19,22–24] After full text review, 28 articles were excluded because they were either review articles, did not provide data on outcomes of interest, did not verify diagnosis of IBD, did not have a control group, or used a surrogate marker for antibiotic use. Five of these articles were more closely considered for inclusion, but they were ultimately deemed to not meet the inclusion and exclusion criteria. Two articles were excluded because they used diagnosis of infection as a surrogate marker for antibiotic exposure.[9,24] Another study was excluded because it did not have a control group.[23] The fourth study was excluded because data regarding antibiotic exposure were not available.[8] A fifth study was excluded because it did not verify the diagnosis of IBD.[25] One meeting abstract of interest was identified during review of past major conferences.[26] However, this study was not included in the final analysis because we were unable to adequately evaluate the study quality. In the end, a total of 11 studies were included in the meta-analysis: 8 case–control and 3 cohort studies.[10–19,27]

Figure 1.

Article selection flow sheet. IBD, inflammatory bowel disease.

The case–control study characteristics are described in Table 1 and the cohort study characteristics are presented in Table 2. All studies were conducted using data from populations in Europe, New Zealand, or Canada. Of the 11 studies, 8 were deemed to be of high quality on the basis of the NOS score (Table 1 and Table 2). A breakdown of the specific elements of the NOS score is provided in Supplementary Table S1 online http://www.nature.com/ajg/journal/v109/n11/suppinfo/ajg2014246s1.html.

A total of 7,208 patients diagnosed with IBD were included in the analyses: 3,937 CD, 3,207 UC, and 64 IBD, type unclassified. The majority of studies (8 of 11) excluded some time period before new diagnosis of IBD in order to account for possible diagnostic delay. The exclusion time frame varied from 3.9 months to 4 years. We obtained extra data from the authors of six of the included studies. An overview of the type of data provided by these authors is outlined in Supplementary Table S2 http://www.nature.com/ajg/journal/v109/n11/suppinfo/ajg2014246s1.html.

Data on the interval between antibiotic exposure and new-onset IBD were available for a few of the included studies. Survey-based studies tended to evaluate more remote exposures and did not specify the average interval. For example, Van Kruiningen et al[19] surveyed about exposures anytime from birth to the age of 20 years in a patient population with a mean age of diagnosis of 24 years. Database studies provided more information on the interval between exposure and diagnosis. Both Card et al[16] and Shaw et al[12] studied antibiotic exposures 2 to 5 years before new diagnosis of IBD.[12,16] Shaw et al.[14] specifically investigated antibiotic exposure in the first year of life in children diagnosed on average at 8.4 years of age. Margolis et al.[17] noted that the average time between exposure and new IBD was 3.2 years for doxycycline and 4 years for tetracycline. Virta et al.[15] reported that 60% of antibiotic exposures analyzed were in the 2 years before diagnosis.

Overall, exposure to any antibiotic significantly increased the risk of new-onset IBD with an OR of 1.570 (95% CI 1.268–1.944; Figure 2). There was significant heterogeneity among the 11 studies included in the analysis (Q=156.3, degrees of freedom (d.f.)=11, P<0.001). According to the I 2 value, ~92% of the variability in effect estimates was due to heterogeneity between the studies rather than a sampling error or chance. The increased risk of new-onset IBD remained significant when including only high-quality studies in the analysis (OR 1.571, 95% CI 1.263–1.953) and when excluding survey-based studies (OR 1.556, 95% CI 1.203–2.013). We also performed an analysis without studies that did not exclude some time period between antibiotic exposure and diagnosis of IBD for their primary outcome. Antibiotics were still significantly associated with an increased risk of new-onset IBD when excluding these studies (OR 1.503, 95% CI 1.440–1.568). Results remained significant when stratifying by age, with children having a higher risk of new-onset IBD compared with adults (Figure 3a,b).

Figure 2.

Risk of new-onset inflammatory bowel disease (IBD) with any antibiotic exposure across all studies. Forest plot showing study-specific and summary odds ratios (ORs). CD, Crohn's disease; CI, confidence interval; UC, ulcerative colitis.

Figure 3.

Risk of new-onset inflammatory bowel disease (IBD) stratified by age. (a) Risk of new-onset IBD with any antibiotic exposure in children. (b) Risk of new-onset IBD with any antibiotic exposure in adults. Forest plots showing study-specific and summary odds ratios (ORs). CD, Crohn's disease; CI, confidence interval; UC, ulcerative colitis.

When combining all studies that provided CD-specific data, there was an increased risk of new-onset CD (OR 1.735, 95% CI 1.354–2.225; Figure 4). Children appeared to have a higher risk of new-onset CD compared with adults (OR 2.747, 95% CI 1.723–4.379 vs. OR 1.565 95% CI 1.177–2.081; Figure 5a,b). In contrast, antibiotic exposure was not associated with new-onset UC (Figure 6). This finding was the same when stratifying by pediatric (OR 1.112, 95% CI 0.773–1.602) and adult (OR 1.058, 95% CI 0.913–1.268) populations.

Figure 4.

Risk of new-onset Crohn's disease (CD) with any antibiotic exposure across all studies. Forest plot showing study-specific and summary odds ratios (OR). CI, confidence interval.

Figure 5.

Risk of new-onset Crohn's disease (CD) stratified by age. (a) Risk of new-onset CD with any antibiotic exposure in children. (b) Risk of new-onset CD with any antibiotic exposure in adults. Forest plots showing study-specific and summary odds ratios (ORs). CI, confidence interval.

Figure 6.

Risk of new-onset ulcerative colitis (UC) with any antibiotic exposure across all studies. Forest plot showing study-specific and summary odds ratios (ORs). CI, confidence interval.

We then pooled data on classes of antibiotics. Six studies reported on the risk associated with specific antibiotics. Sufficient data were available on eight antibiotic classes. We found that all antibiotic classes, with the exception of penicillin, were associated with a new diagnosis of IBD (Table 3). The risk of new-onset IBD was most marked after exposure to either fluoroquinolones (OR 1.789, 95% CI 1.027–3.118) or metronidazole (OR 5.010, 95% CI 1.646–15.245).

Six of the included studies stratified their data by number of antibiotic courses to measure the effect of the magnitude of antibiotic exposure.[10–16] However, we were unable to pool these data because of the differing cutoff points used to stratify the number of antibiotic courses. Five of the six studies did find that increasing numbers of antibiotic dispensations increased the risk of new-onset IBD, suggesting a possible dose-dependent relationship (Supplementary Table S3 http://www.nature.com/ajg/journal/v109/n11/suppinfo/ajg2014246s1.html).

In addition, two studies specifically reported data that stratified the risk of IBD following antibiotic exposure by sex.[14,15] Both studies assessed pediatric patients and found that boys with IBD were significantly more likely to have used antibiotics than girls. Shaw et al.[14] reported that 75% of boys with IBD had an antibiotic exposure compared with 38% of girls with IBD. Virta et al[15] found that boys exposed to antibiotics were more likely to be diagnosed with CD (OR 11.86, 95% CI 1.61–87.37).

A funnel plot was used to evaluate publication bias (Figure 7). Egger's regression asymmetry test was not significant (P=0.41, intercept=0.45, and 95% CI −3.7 to 4.6), suggesting that there was no significant bias.

Figure 7.

Funnel plot for publication bias for primary outcome of included studies.

Comments

3090D553-9492-4563-8681-AD288FA52ACE

processing....