key: cord-0820487-zivz0r7b
authors: Hajifathalian, Kaveh; Katz, Philip O.
title: Regarding “Increased Risk of COVID-19” in Patients Taking Proton Pump Inhibitors
date: 2020-10-08
journal: Am J Gastroenterol
DOI: 10.14309/ajg.0000000000000920
sha: f42e2066fe906e126d1c2109bdcc32fa5a5bee4a
doc_id: 820487
cord_uid: zivz0r7b

nan

We read with interest the revised and updated study by Almario et al. (1) in a prepublication article suggesting that proton pump inhibitor (PPI) use, particularly twice daily dosing, is associated with an increased odds ratio (OR) for a reported positive coronavirus disease 2019 (COVID-19) test. The data came from an internet study, termed a national health survey, of a voluntary subject selfreporting of demographics, gastrointestinal (GI) symptoms, PPI and H2 blocker use, and reported COVID-19 test results. The authors performed a multivariable analysis, reportedly verified by independent statisticians, concluding that there is an increase OR for a positive COVID-19 test in those who self-report taking PPIs once or twice daily. We note that once daily dose of an H2 blocker was associated with a statistically significant reduction in OR for a positive test, a finding that we did not see discussed in the study, whereas twice a day H2 blocker had no effect. The COVID-19 positive participants have a substantially different demographics (young, little college, surprisingly high household income, predominantly Latino) than the overall study participants. Although the authors provide a potential explanation for the demographics, the distribution of the study population remains problematic. In fact, the authors' explanations present more concerns. The authors compare their study sample with that of the general population of the United States and show similarity; however, it is unclear why a sample of participants with GI symptoms should be similar to the US general population. They also report that the demographics of the COVID-19 positive respondents changed dramatically during the study period. Rather than being reassuring, this creates more concerns about the reliability of the online survey. It is not adequately explained why this change happened. The authors report that younger age groups and Latinos have high "case proportions" for COVID-19 in some states, but what determines the likelihood of being selected in a representative population sample is the absolute number of cases within an age and demographic group compared with the overall reference population, and not the proportion of positive results in that specific age and demographic group. The authors explanation that a minority of COVID-19 positive subjects report having gastroesophageal reflux disease (3.2%) because they are either selfmedicating for symptoms or they do not recognize that the medical term "gastroesophageal reflux disease" is plausible but begs the issue that ultimately we do not know the reason for PPI use, compliance, or, in fact, the actual dose. The issues of misclassification and recall bias are very concerning in this study. The authors explain anomalies in data by suggesting that participants might be dishonest about their education or income level or might not remember their correct diagnosis, then downplay the effect of the same biases later on by postulating that it is unlikely for participants to be wrong about COVID-19 diagnosis or PPI use.

It is not clear from the study whether the control group participants were tested and were COVID-19 negative or if they are a mix of tested and untested participants. In the absence of this crucial piece of information, it remains unclear whether the study demonstrates an association between PPI use and the risk of contracting COVID-19 or simply the likelihood of seeking a test or being tested for COVID-19 in a very unusual stratum of the US population. Data regarding relevant comorbidities of the study groups (e.g., hypertension, cardiovascular disease, and chronic respiratory or kidney disease) are importantly absent leading to higher likelihood of confounding. The increase in gastric pH may indeed be beneficial to this virus, but this is theoretical and not proven. Although PPIs certainly elevate intragastric pH, so can H2 blockers. It is still therefore entirely unclear why once daily H2 blockers were associated with a significant decreased risk of COVID-19 in the same multivariable analysis. It is unexplained why PPI and H2 blockers should have opposite effects.

Finally, the effect estimates presented by PPI dose and duration of use show surprising variabilities that are not explained by the authors in the revised study. It is not clear why once daily PPI use for .6 months should have a significantly smaller effect than once daily PPI use of ,6 months. PPI twice daily for ,6 months was no different than once daily PPI, whereas the same dose for .6 months had a significantly larger effect than once daily dosing. Given that there is no tachyphylaxis with PPI and steady state is achieved in 5-7 days, the effects of acid control should be consistent across the presented durations of use. The lack of consistency of the results based on dose and duration of use is troubling.

In our view, the methodologic issues and result inconsistencies remain after revision and expanded discussion and prevent any substantial clinical interpretation of these results regarding PPIs and risk of COVID-19. Although we have no argument with any recommendation to use the lowest effective dose required for individual patients, we are concerned that these data will result in needless anxiety and perhaps discontinuation of PPIs in patients who need them for appropriate clinical indications, which, in some cases, can be associated with significant risk of harm.

Increased risk of COVID-19 among users of proton pump inhibitors