OTU-15 Intermittent administration of PPI is not inferior to infusion following acute non-variceal upper gastrointestinal bleeding

Introduction Despite the ubiquitous use of PPI therapy following acute upper gastrointestinal bleeding, the optimal dose and method of administration remains controversial. Aim To determine whether PPI therapy delivered intermittently was non-inferior to continuous infusion; with regard to rebleeding, mortality, need for surgery and length of hospital stay, in acute upper gastrointestinal bleeding. Study design Systematic Review and Meta-analysis Method The MEDLINE database was search with a pre-defined search strategy. RCTs were considered eligible if they included a group treated with a stat dose of PPI followed by 8 mg/hr infusion for 72 hours compared to a group treated with an intermittent PPI regimen, following endoscopy for acute upper gastrointestinal bleeding. Abstracts were screened against eh eligibility criteria. Two reviewers trained in critical appraisal reviewed the full text articles, extracted data and applied the GRADE tool to assess for risk of bias. Data was synthesised using Mantel-Haenszel fixed-effects method. Heterogenicity was assessed with the I² statistic and examined as per the study protocol. Non-inferiority margins were pre-defined as 0.962, 0.934 and 0.914 for re-bleeding, need for surgery and mortality respectively. This was determined as 20% the difference from the odds ratio to the lower bound of a 2-sided 95% CI taken from infusion versus placebo meta-analysis. Results 26 RCTs were included (n=4368). Significant heterogeneity was found for the outcome measure length of hospital stay and was not resolved by sub-group analysis or the use of a random-effects model. The risk of rebleeding was lower in the intermittent administration group compared to infusion, although was not statistically significant at either 72 hours (OR 1.03 95%CI 0.77–1.37) or 30 days (OR 1.05 95%CI 0.85–1.29). This revealed non-inferiority as per the pre-defined margin. The risk of needing surgery was higher in the intermittent group (OR 0.87 95%CI 0.63–1.20), whilst this was not statistically significant it did not meet the strict margin set to determine non-inferiority. The risk of mortality also favoured intermittent administration (OR 1.13 95%CI 0.81–1.58) and showed non-inferiority. The overall risk of bias was high in 5, undetermined in 6 and low in 15 of the RCTs. Conclusion This study showed there was no statistically significant difference in the outcomes of rebleeding, need for surgery, mortality or length of hospital stay when PPIs were administered intermittently compared to by infusion. Conservative margins of non-inferiority were used due to the potential clinical implications, despite this non-inferiority was shown with regard to re-bleeding and mortality. The assessment of the quality of the evidence supports the validity of the findings. Given the delivery of PPI via infusion is more costly, timely and inconvenient for patients, the determination of non-inferiority supports a change in practice.