Famotidine and Mortality in Coronavirus Disease 2019

Dear Editors:

We read with great interest the study by Yeramaneni et al¹ in which the authors have retrospectively analyzed the effect of famotidine on 30-day mortality in hospitalized patients with Coronavirus Disease 2019 (COVID-19). In a matched cohort of 410 patients who received famotidine and 746 who did not, 30-day mortality was higher with famotidine (15.1% vs 9.8%, P = .007). A few points merit consideration. First, the authors adjusted the 2 groups for World Health Organization severity within 48 hours of admission. World Health Organization severity level 5 includes patients on mechanical ventilation or extracorporeal membrane oxygenation. Of all patients, 6.3% and 0.5% in the famotidine and nonfamotidine groups, respectively, were classified as World Health Organization severity level 5, leading to a mismatch. Even the postmatch famotidine group had a higher proportion of patients with concomitant steroids, antiviral, and tocilizumab use because of severe disease. The mortality in the famotidine group among patients on mechanical ventilation was extremely high: 63 patients required mechanical ventilation and 62 (99%) patients died. In such patients, any drug is unlikely to be of much benefit. Second, the use of steroids and tocilizumab in the cohort was associated with higher mortality. In contrast, prior studies suggest reduced mortality in patients receiving steroids and tocilizumab.^{2 , 3}

The imbalance in steroid use across groups may indicate that the severity of patients may not have been matched despite the multivariable model with Coarsened Exact Matching, because sicker patients are more likely to be administered steroids. In this case, it cannot be ruled out that the increase in mortality due to famotidine use observed in this study is a spurious association because of confounding. It has been reported earlier that failing to adjust for time-dependent variables may lead to a spurious increase in mortality. Therefore, a conventional analysis adjusted for baseline characteristics will not account for confounders, and an analysis that considers time-dependent variables would do better. However, even such a sophisticated analysis is no substitute for randomization. Even the best observational studies can be affected by residual confounding, and randomized controlled trials often reveal contrasting effects. Their Supplementary Table 1 mentions the number of patients intubated, receiving mechanical ventilation, and 30-day mortality as 21.4, 27.7, and 72.9, respectively, which needs to be rechecked.

Their results are in contrast to other published studies. However, it is difficult to directly compare their results with those of other studies because of heterogeneity at multiple levels. We conducted a systematic review and meta-analysis of the previously published studies.⁴ We searched the databases Medline, Embase, Cochrane CENTRAL, and Medrxiv for title, abstract, and full-text screening. We calculated pooled hazard ratios and 95% confidence intervals (CIs) for the composite outcome of death and intubation using the Generic Inverse Variance approach. The random-effects model was used to conduct the meta-analysis. We carried out the statistical analysis using Review Manager 5.3. Heterogeneity was assessed using visual inspection of forest plots and the I ² statistic. The risk of bias was assessed using the revised version of the Newcastle-Ottawa scale for cohort studies, and the Grading of Recommendations, Assessment, Development and Evaluations methodology was used to rate the certainty of evidence. Of the 13 studies identified, 5 studies were eligible for inclusion.⁴ These studies included 2643 patients with COVID-19, of whom 312 patients received famotidine. All except 10 patients were hospitalized with a moderate to severe illness. The dose of famotidine varied from 40–233 mg/day given for 5–21 days. Two cohort studies^{5 , 6} with matched control subjects that included 84 and 83 patients on famotidine and 1536 and 795 control subjects without famotidine, respectively, showed a significant reduction in mortality: 58% (hazard ratio, 0.42; 95% CI, 0.21–0.85; P = .02) and 63% (odds ratio, 0.37; 95% CI, 0.16–0.86; P = .02). A meta-analysis of 2 cohort studies showed a statistically significant decrease in the composite outcome of death and intubation with famotidine (hazard ratio, 0.44; 95% CI, 0.27–0.73).⁴ Heterogeneity regarding disease severity, inconsistency in severity classification, variation in the dose, timing and route of famotidine, confounding due to co-medications, and comorbidities are likely to explain the differences in the results by Yeramaneni et al and our meta-analysis.

We believe this study's results reporting no benefit of famotidine need to be interpreted cautiously. Famotidine is an over-the-counter drug with an excellent safety profile and can be a useful adjunct in patients with mild to moderate disease. The discrepancy in the results of Yeramaneni et al and our meta-analysis calls for a randomized trial during the ongoing pandemic.