Here is a recently published evidence summary:
Conclusions: Many stress ulcer prophylaxis recommendations are based on older studies at risk of bias, which may not be applicable to modern practice. Stress ulcer prophylaxis should be limited to patients considered to be at high risk for clinically important bleeding. When evaluating only the trials at low risk for bias, the evidence does not clearly support lower bleeding rates with proton pump inhibitors over histamine 2 receptor antagonists; however, proton pump inhibitors appear to be the dominant drug class used worldwide today. The current rate of upper gastrointestinal bleeding and the relative adverse effects of acid suppression on infectious risk may drive not only the effectiveness, but also the cost-effectiveness of stress ulcer prophylaxis today. Research is currently underway to better address these issues.