Antibiotics and Risk of New Onset
Inflammatory Bowel Disease:
A Meta-Analysis
Ryan Ungaro1, Charles Bernstein2, Richard Gearry3, Anders Hviid4,
Kaija-Leena Kolho5, Matthew Kronman6, Souradet Shaw2, Herbert
Van Kruiningen7, Jean-Frédéric Colombel1, Ashish Atreja1
2013 CCFA Advances in Inflammatory Bowel Diseases Conference
1. Ichan School of Medicine at Mount Sinai, New York, NY
2. University of Manitoba, Winnipeg, Canada
3. University of Otago, Christchurch, New Zealand
4. Statens Serum Institut, Coopenhagen, Denmark
5. University of Helsinki, Helsinki, Finland
6. Seattle Children’s Hospital, Seattle, Washington
7. University of Connecticut, Storrs, Connecticut
Disclosures
Nothing to Disclose
Background
● Medications emerging as risk factor for IBD
● Antibiotics may increase risk of CD and UC1
● However some studies have found no
association2,3
● Few studies have reported risk associated
with specific antibiotic classes
1. De Vroey et al. Am J Gastroenterol. 2010;105(12)
2. Castiglione F et al. J Crohns Colitis. 2012;6(3)
3. Van Kruiningen et al. Inflamm Bowel Dis. 2005;11(4)
Study Aims
1. To examine antibiotic use as a risk factor for
new onset IBD, CD and UC
2. To evaluate if effect of antibiotics different in
children versus adults
3. To determine IBD risk with specific antibiotic
classes
Methodology
● Meta-analysis of Observational Studies in
Epidemiology (MOOSE)1
● Search strategy
○ Medline, Embase and Cochrane
databases
○ Search keywords:
■ Inflammatory bowel disease, Crohn’s
disease, ulcerative colitis, antibiotics,
penicillin, cephalosporin, tetracycline,
fluoroquinolone, macrolide,
sulfonamide, metronidazole
1. Stroup DF, et al. JAMA 2000;283.
Methodology
● Inclusion criteria
○ Cohort or case-control
○ Data on exposure to antibiotics prior to new
diagnosis of IBD (CD, UC, or both)
● Data collection
○ Two reviewers extracted data
○ Authors contacted if data not available in
published manuscript (7 studies)
● Data analysis
○ Random-effects model to determine overall
pooled estimates and 95% confidence intervals
(CI)
○ The Newcastle-Ottawa Scale (NOS) to assess
study quality1
1. Wells et al. Accessed at: http://www.ohri.ca/programs/clinical_epidemiology/oxford.asp.
Search Results
4508 Citations Retrieved
2558 from Pubmed
1798 from Embase
149 from Cochrane
3 from Backward Snowballing
1234 Review
Articles Excluded
3274 Original Articles
3259 Articles
Excluded by Title and
Abstract Review
16 Original Articles
11 Articles Included
● 8 Case-Control
● 3 Cohort
5 Studies Excluded
After Full Text Review
● 3 used surrogate
for antibiotic use
● 1 without
controls
● 1 insufficient
data available
Case-Control Studies
Study
Year
Country
Cases, n
Controls,
n
Population
Age, years
Quality
Score
(NOS)1
Card et al
2004
UK
587 CD
1460
41.64 (mean)
8
Van Kruiningen
et al*
2005
Belgium
74 CD
140
24 (mean)
7
Gearry et al*
2010
New Zealand
638 CD, 653
UC
600
20 or older (no
average given)
8
Han et al
2010
New Zealand
315 CD
536
45.6 (mean)
5
Shaw et al*
2010
Canada
27 CD, 9 UC
360
8.4 (mean)
7
Shaw et al*
2011
Canada
1025 CD, 1218
UC
22,346
43.4 (mean)
6
Castiglione et al
2012
Italy
468 CD, 527
UC
562
36 (median, CD)
37 (median, UC)
3
Virta et al*
2012
Finland
233 CD, 362
UC
2380
9.7 (mean, CD)
8.5 (mean, UC)
8
* provided extra data
Pediatric studies highlighted
1. Wells et al. Accessed at: http://www.ohri.ca/programs/clinical_epidemiology/oxford.asp.
Cohort Studies
Study
Year
Country
Cases, n
Cohort Size,
n
Population
Age, years
Quality
Score
(NOS) 1
Margolis et
al
2010 UK
CD 71, UC 94,487
99, 37 IBDU
22.1 (mean)
9
Hviid et al*
2011 Denmark
50 CD, 67
UC
577,627
3.4 (mean)
8
Kronman
et al*
2012 UK
449 CD,
272 UC,
27 IBDU
1,072,426
17 or
9
younger
(no average
given)
* provided extra data
Pediatric studies highlighted
IBDU = IBD, type unclassified
1. Wells et al. Accessed at: http://www.ohri.ca/programs/clinical_epidemiology/oxford.asp.
Included Studies
● 11 total studies with 7,208 patients diagnosed
with IBD
○ 3,937 patients with CD
○ 3,207 patients with UC
○ 64 patients with IBDU
● Majority studies (9 of 11) explicitly excluded
some time period prior to new diagnosis of IBD
○ Range from 3.9 months to 4 years
○ Limit confounding by diagnostic delay
IBDU = IBD, type unclassified
Antibiotic exposure associated with increased
overall risk of new onset IBD
(CD)
(UC)
I2 = 82.35, p = 0.00
Antibiotic exposure and risk of
new onset IBD in adults and children
Adults
OR 1.431
(95% CI 1.116-1.834)
I2 = 90.07, p = 0.00
Children
OR 1.894
(95% CI 1.237-2.989)
(CD)
(UC)
I2 = 48.78, p = 0.099
I2 = 48.78, p = 0.099
Antibiotic exposure associated with increased
risk of new onset CD
I2 = 84.87, p = 0.00
Antibiotic exposure and risk of
new onset CD in adults and children
Adults
OR 1.565
(95% CI 1.177-2.081)
I2 = 89.97, p = 0.00
Children
OR 2.747
(95% CI 1.723-4.379)
I2 = 0.00, p = 0.63
Antibiotic exposure NOT associated with
increased risk of new onset UC
I2 = 47.41, p = 0.07
Antibiotic exposure and risk of
new onset UC in adults and children
Adults
OR 1.058
(95% CI 0.851-1.316)
I2 = 72.71, p = 0.012
Children
OR 1.112
(95% CI 0.773-1.602)
I2 = 0.00, p = 0.87
Most classes of antibiotics are associated with
increased risk of new onset IBD
Antibiotic Class
Number of
Studies
Providing Data
Pooled OR (95% CI)
P value
Heterogeneity
I2 (p value)
Metronidazole
3
5.010 (1.646 - 15.245)
0.005
I2 = 91.018 (0.000)
Quinolones
3
1.789 (1.027 - 3.118)
0.040
I2 = 85.901 (0.001)
Broad-spectrum
penicillins
3
1.313 (1.143 - 1.509)
0.000
I2 = 70.824 (0.032)
Tetracyclines
4
1.301 (1.173 - 1.443)
0.000
I2 = 0.000 (0.794)
Cephalosporins
4
1.268 (1.143 - 1.407)
0.000
I2 = 6.703 (0.360)
Macrolides
4
1.231 (1.112 - 1.363)
0.000
I2 = 20.012 (0.290)
Sulfonamides
4
1.174 (1.058 - 1.302)
0.003
I2 = 0.000 (0.396)
Penicillin
3
1.117 (0.761 - 1.639)
0.572
I2 = 74.191 (0.021)
Limitations
● Data compiled from retrospective studies
○ Questionnaire studies subject to recall
bias
● Range in the quality of studies (based on
QOS scale)
● Significant heterogeneity in certain analyses
● Different antibiotic exposure exclusion time
periods
Summary
● Antibiotic exposure increases odds of new
diagnosis IBD
● Antibiotic use increases risk new onset CD
but not UC
● CD risk greatest in children
● Most antibiotic classes associated with IBD
○ Penicillin not associated
○ Metronidazole and quinolones most
strongly associated
Discussion
• Antibiotics may increase IBD risk
by altering microbiome
• Alternatively, association may be
surrogate marker for increased risk
infections in CD
1. Manichanh C et al.Nat Rev Gastroenterol Hepatol. 2012;;9(10).
2. Perez-Cobas et al. Gut. 2013; 62(11).
3. Gradel KO et al. Gastroenterology. 2009;137.
Thank You
● Ashish Atreja
● Jean-Frederic Colombel
●
●
●
●
●
●
●
Charles Bernstein
Richard Gearry
Anders Hviid
Kaija-Leena Kolho
Matthew Kronman
Souradet Shaw
Herbert Van Kruiningen
● Crohn’s and Colitis Foundation of America