This study aimed to estimate and compare risk of incidence and assess outcomes after COVID-19 in Korean patients prescribed PPI or H2RA. Specifically, we evaluated the incidence of COVID 19 in subjects prescribed PPI or H2RA for more than 7 days and show that short-term use of PPI was not associated with incidence of COVID-19 compared to short-term H2RA users or non-users. Moreover, among COVID-19 patients, PPI use for ≥ 7 days was not significantly associated with risk of complications compared to H2RA use except for all-cause mortality in PSM analysis, and it was not associated with complications in hospitalized patients.
Two recent studies have addressed the association between PPI use and incidence of COVID-19 infection.11,12 Lee et al have reported that patients taking PPIs are at increased risk for severe clinical outcomes with COVID-19 but that they are not more susceptible to SARS- CoV-2 infection.12 They defined current PPI users as patients who took PPIs 1–30 days before the first SARS-CoV-2 test date.12 Further, they also showed that there was no significant difference in SARS-CoV-2 positivity rates between PPI users and non-users, irrespective of short-term (<30 days) or long-term (>30 days) use.12 Another study, an online survey by Almario et al., reported that individuals using PPIs up to once daily (aOR 2.15; 95%CI, 1.90–2.44) or twice daily (aOR 3.67; 95%CI, 2.93–4.60) had significantly higher odds for testing COVID-19 positive compared to those not taking PPIs.11 In contrast, we show that PPI use was not associated with the higher risk of COVID-19 infection compared to H2RA use or no acid-suppressant use. This could be due to our use of the Korean national claims database wherein data was converted to the OMOP-CDM format, and this permitted adjustment for many more covariates than previous studies. Additionally, large-scale propensity matching was used to overcome potentially unmeasured confounding factors and we also performed multiple sensitivity analyses. We also calibrated our analysis using 123 falsification end points to detect and reduce confounding factors, selection bias, and systematic errors. Thus, of the 48 analyses performed (9 primary and 39 secondary), most results were consistent with the calibrations.
In the secondary analysis, we compared the complication of COVID-19 between PPI and H2RA using multiple sensitivity analyses. The result showed no significant association between PPI and H2RA. Only PSM analysis measuring the association between PPI and all-cause mortality, compared to H2RA, showed significant results, however, the other analyses (i.e., unadjusted and stratification) showed opposite results. We could not perform large-scale PSM in the secondary analysis due to small number of included COVID-19 patients, therefore, there might be biases in the result.
To date, several studies have addressed clinical outcomes in COVID-19; however, most studies only included a small number of patients and were limited by the presence of confounding factors.12–14, 23 Lee et al found that PPI use led to greater risk of severe clinical outcomes in COVID-19, including intensive care unit admission, requirement of invasive ventilation, or death.12 However, that study did not consider PPI use after COVID-19 diagnosis, and the comparator group comprised non-PPI users, which could have led to indication bias, i.e., patients in the PPI group could have experienced a more severe course of COVID-19 compared to non-users and the difference might have led to more severe outcomes. Therefore, to avoid indication bias, we compared clinical outcomes between PPI and H2RA users, and consistent with our results, Zhang et al also reported that PPI use had no effect on the clinical course of COVID-19.13 Additionally, Taştemur et al. have suggested that PPIs may be used for both prophylaxis and treatment because hydroxychloroquine and azithromycin may prevent viral spread by accumulating in organelles with acidic content and raising their pH. Thus, given their effects on pH, they concluded that PPIs show similar effects on viral entry and intracellular distribution.15 Such inconsistent results imply that the risk and benefits of PPI use in viral infection have remained controversial to date.23
Our study has certain limitations. First, although we used large-scale PSM in the primary analysis, there were a few relatively unmatched covariates that may have led to selection bias, showing standardized mean difference greater than 0.1. Additionally, there might have been residual indication bias between PPI and H2RA users. Nonetheless, we measured and adjusted the systematic error in this study through empirical calibration by employing 123 falsification end points in the primary analysis. Second, we only included PPI use for 7 days, and therefore, we could not evaluate the effects of long-term PPI use, and as the HIRA database also had data only pertaining to a short period, we could not analyze the long-term effects of acid-suppressants. Third, this was an observational study; therefore, it is not possible to establish causality. The effects of acid-suppressants on viral infection, especially COVID-19, require further clarification.