Does the long-term administration of proton pump inhibitors increase the risk of adverse cardiovascular outcomes? A ChatGPT powered umbrella review

Background Proton pump inhibitors (PPIs) are commonly prescribed for the treatment of acid-related disorders. In the context of coronary artery disease (CAD), PPIs are commonly prescribed along with antiplatelet medications. In fact, the potential interaction between these two classes of medications has been subject to much debate. This review aimed to summarise the findings from systematic reviews and meta-analyses on the casual relationship between PPI use (alone) and major adverse cardiovascular events (MACE). Furthermore, the recent release of ChatGPT has provided reviewers with a powerful natural language processing tool. We therefore aimed to assess the utility of ChatGPT in the systematic review process.Methods A comprehensive search of PubMed was conducted to identify relevant systematic reviews and meta-analyses published up to March 2023. Two independent reviewers assessed the eligibility of the studies, extracted the data, and assessed the methodological quality using AMSTAR 2.0. The population of interest was adults that received the medications of interest (PPIs) for a minimum of three months, regardless of indication. Control groups were defined as placebo or active comparators. The outcomes of interest were described under the general term MACE, which include cardiovascular death, non-fatal myocardial infarction, and non-fatal stroke. There were no restrictions with regards to time, but we only included reports in English. A different group of independent reviewers simultaneously ran the same process using ChatGPT. The results were then compared with the human generated results.Results Seven systematic reviews and meta-analyses were included, involving a total of 46 randomised controlled trials and 33 observational studies. The studies examined the association between PPI use and MACE, including stroke, myocardial infarction, and all-cause mortality. The results of the individual studies were conflicting, with some showing a positive association between PPI use and MACE, some showing no association, and others showing mixed results. However, the majority of the studies that included observational data reported a positive association between PPI use and MACE. Sensitivity analyses conducted in some studies did not significantly alter the primary results, suggesting that the findings were robust. Furthermore, ChatGPT was successfully prompted to execute most tasks involved in this review. We therefore present text that was generated by ChatGPT, including the abstract, introduction, results, and discussion sections.Conclusion The findings of this umbrella review suggest that a causal relationship between PPI use and an increased risk of MACE cannot be ruled out. Further research is needed to better understand this relationship, particularly the underlying mechanisms and potential confounding factors. Healthcare professionals should consider the long-term use of PPIs and carefully weigh the risks and benefits for each patient. Finally, ChatGPT was successfully prompted to execute most of the tasks involved in this review. We therefore feel that this tool will be of great assistance in the field of evidence synthesis in the near future.