Over the past decade, a number of studies have suggested that use of proton pump inhibitors is associated with increased risk of several adverse health outcomes including cardiovascular disease, acute kidney injury, chronic kidney disease, gastric cancer, dementia, pneumonia, osteoporotic fractures, Clostridioides difficile infections, and others. Some of these adverse health outcomes are associated with increased risk of death. These observations led us to ask whether the use of proton pump inhibitors is associated with increased risk of death. In 2017, we examined this question and reported increased risk of all-cause mortality among users of proton pump inhibitors.
Following the publication of the 2017 paper, we received overwhelmingly encouraging feedback from patients, healthcare providers, research scientists, medical societies, deprescription groups, not-for-profit organizations, mainstream media, scientific media, and others. Our discussions with stakeholders centered around a common question: What are the causes of death attributable to the use of proton pump inhibitors? Answering this question will help inform the public about the cause-specific hazards of long-term use of proton pump inhibitors and may inform risk stratification, risk mitigation strategies, and may help curb overuse of these medications.
The biggest challenge we needed to overcome was that some adverse health outcomes associated with use of proton pump inhibitors may take years to develop and progress to mortality. In addition, looking into each specific cause of death, especially for those causes of death that are uncommon, would require a large number of study participants and a long follow-up time. We assessed that a randomized clinical trial may not be the most practical or feasible approach to study this research question. Thus, we sought to leverage the power of big data and used large-scale electronic health record data from the US Department of Veterans Affairs (a US veteran specific national health service) to emulate a target randomized trial and used a causal inference approach to more accurately estimate the effect of proton pump inhibitors use on cause-specific mortality. In addition, we tested negative controls to rule out possible non-causal interpretation of the results.
To better communicate the risk, we estimated the absolute risk based on the difference in survival rates in addition to reporting hazard ratios. The absolute risk not only provides information on ratio differences between two groups, but also considers the baseline risk, and provides a more meaningful description of risk to the wider public.
In a new research paper published in The BMJ , our findings suggest excess burden of death due to cardiovascular disease, chronic kidney disease, and upper gastrointestinal cancer that was attributable to use of proton pump inhibitors. We also found that increased duration of exposure is associated with greater risk.
We also found that 51.42% of participants in the study were on proton pump inhibitors for more than 90 days without clear indication (or documented need for proton pump inhibitors use)—our research suggests that these patients may be exposed to potential harm when it is unlikely for them to derive any medical benefit from use of proton pump inhibitors.
Overuse of proton pump inhibitors is not devoid of harm. The totality of the results suggests the need to avoid proton pump inhibitors when not medically indicated; for those who have a medical indication for proton pump inhibitors, use should be limited to the lowest effective dose and shortest duration possible. Since the overuse of proton pump inhibitors appears to be abundant, a concerted effort to develop deprescription interventions is needed. Mechanisms to curb un-indicated initiation of proton pump inhibitors should also be developed.
Also, very importantly—and beyond the findings on proton pump inhibitors—we appreciated feedback from patients, healthcare providers, researchers, and others, which helped us sharpen the focus and relevance of our research inquiry. This work may also provide an example of how to make use of large-scale electronic health records and advanced statistical methodologies to help identify adverse event signals associated with drug use.
Ziyad Al-Aly is a clinical epidemiologist at Clinical Epidemiology Center, Department of Veterans Affairs St Louis Health Care System
Yan Xie is a biostatistician at Clinical Epidemiology Center, Department of Veterans Affairs St Louis Health Care System
Competing interests: Please see research paper