Proton Pump Inhibitors Versus H2 Blockers

04 Mar

Posted by SCCM as Concise Critical Appraisal

Up to 25% of intensive care unit (ICU) patients may have gastrointestinal bleeding when stress ulcer prophylaxis is not administered. In a systematic review and meta-analysis by Alhazzani et al, the effect of proton pump inhibitors (PPI) versus histamine 2 receptor antagonists (H2RA) on the prevention of clinically important gastrointestinal bleeding was investigated. Their results were published in the February issue of Critical Care Medicine.

The authors conducted a systematic review that included critically ill adults in both medical and surgical ICUs. The primary outcome of interest was bleeding, with secondary outcomes including nosocomial pneumonia, all-cause ICU mortality, ICU length of stay, and Clostridium difficile infection. A comprehensive search strategy and rigorous Cochrane-based risk of bias assessment were executed. Data were analyzed with a random effects model, and pooled relative risks were calculated. Several subgroup and sensitivity analyses were performed.

After screening 1,215 titles and abstracts, fourteen randomized trials were found, including a total of 1,720 patients. PPIs were associated with a lower risk of clinically important bleeding compared to H2RAs (risk ratio [RR] 0.36; 95% confidence interval [CI], 0.19-0.68, P=0.002; I2=0%). The number needed for prophylaxis was estimated to be 78 using a control event rate of 2%. PPIs were also associated with a lower risk of overt bleeding compared to H2RAs, with a number needed for prophylaxis of 30 using a control event rate of 5%. No statistically significant differences between PPIs and H2RAs were found for the outcomes of nosocomial pneumonia, mortality or ICU length of stay. No studies reported the incidence of C. difficile infection. A sensitivity analysis that excluded abstracts yielded results that were consistent with the primary analysis.

The overall methodology employed for this study was outstanding, and this study should serve as a model of how to properly conduct and report a systematic review and meta-analysis. A few limitations were openly acknowledged by the authors. First, some degree of publication bias was evident; small negative trials appeared to be missing. Second, the overall quality as assessed by the Grading of Recommendations Assessment, Development and Evaluation approach was low to moderate for the primary and secondary outcomes, with serious risk of bias for all outcomes. Third, the treatment effect appeared to be smaller in trials of higher quality. Although all the included studies were randomized, elements of the randomized design, such as blinding, were missing in several of the trials, thus potentially inflating the observed benefit of PPIs. Nevertheless, based on this study, the highest quality available evidence suggests that PPIs may lower the risk of both clinically important and overt gastrointestinal bleeding in ICU patients without increasing the risk of nosocomial pneumonia.

Concise Critical Appraisal is a regular feature authored by SCCM member Samuel M. Galvagno Jr., DO, PhD. Each installment highlights journal articles most relevant to the critical care practitioner.

Be Sociable, Share!

One Response

Sean Kane

March 4th, 2013 at 2:08 pm
1

Regardless of the quality of the study methodology, I would be cautious in adopting these results into clinical practice. Keep in mind that even with the very best meta-analysis approach, the quality of the findings can only be as strong as the studies that are included.

As mentioned in the manuscript and the accompanying editorial, there are a few important limitations that deserve further mention. Most concerning is the lack of standard definition for clinically important GI bleeding. Based on the definition variety, I’m not sure these endpoints should have been pooled together. The definition spanned everything from “hemodynamic instability not explained by other causes” to EGD-confirmed bleeding. As demonstrated by the variety in definitions, the event rate was 0% in 5 of the 12 studies, and as high as 31% in one study.

There are a number of other issues, such as the variety in H2RA dosing strategies used (continuous infusion, BID, daily, etc.) and the overall small study sizes (nearly all with fewer than 50 patients in the H2RA arm) that are indicative of the limited amount of published data rather than the quality of this manuscript.

Now that four meta-analyses have been published, I think we’re still left with a paucity of data that no amount of data slicing and meta-analysis can solve.