1. Gastrointestinal Safety of Cyclooxygenase-2 Inhibitors: A Cochrane Collaboration Systematic Review 
Alaa Rostom, Katherine Muir, Catherine Dubé, Emilie Jolicoeur, Michel Boucher, Janet Joyce, Peter Tugwell, George W. Wells 
Clinical Gastroenterology and Hepatology 
Volume 5, Issue 7, Pages e5 (July 2007)
DOI: /j.cgh
Copyright © 2007 AGA Institute Terms and Conditions

2. Figure 1 Description of study selection.
Clinical Gastroenterology and Hepatology 2007 5, e5DOI: ( /j.cgh )
Copyright © 2007 AGA Institute Terms and Conditions

3. Figure 2
Endoscopically detected ulcers. Summary RR associated with the use of COX-2s (left) vs nonselective NSAIDs (right). Overall, the use of low- or high-dose COX-2s is associated with significantly reduced RR of endoscopic ulcers.
Clinical Gastroenterology and Hepatology 2007 5, e5DOI: ( /j.cgh )
Copyright © 2007 AGA Institute Terms and Conditions

4. Figure 3
Endoscopically detected gastroduodenal ulcers. Summary RR associated with the use of COX-2s (left) vs placebo (right). Overall, the use of low-dose COX-2s is not statistically different than placebo (fixed effects). High-dose COX-2s demonstrated a trend toward a greater RR of endoscopic ulcers than placebo (heterogeneity – random effects).
Clinical Gastroenterology and Hepatology 2007 5, e5DOI: ( /j.cgh )
Copyright © 2007 AGA Institute Terms and Conditions

5. Figure 4
Clinical ulcer complications with COX-2s vs nonselective NSAIDs. COX-2s are associated with a lower RR of POBs and PUBs (random effects) compared with traditional NSAIDs.
Clinical Gastroenterology and Hepatology 2007 5, e5DOI: ( /j.cgh )
Copyright © 2007 AGA Institute Terms and Conditions