Remdesivir and mortality reduction in COVID-19 patients: a systematized subgroup analysis of clinical trials.

Alegre-Del Rey, Emilio Jesús; Gil-Sierra, Manuel David; Alarcón de la Lastra-Romero, Catalina; Sánchez-Hidalgo, Marina

OBJECTIVE Remdesivir has not shown survival benefit for patients with severe COVID-19. However, subgroup analysis of ACTT-1 Study Group showed an apparent reduction in mortality for patients who required non‑high-flow oxygen. Presentation of SOLIDARITY study results were associated by a meta-analysis combining mortality results by subsets rom randomized clinical trials. The aim is a methodological assessment of reliability and clinical applicability about findings by subgroups on the effect of remdesivir on mortality in patients with COVID-19. METHOD A validated tool was used to evaluate the findings of subgroup analyses in randomized clinical trials, including meta-analysis atached to SOLIDARITY study. It is structured in preliminary questions to reject subset analyses without relevant minimum conditions, and a specific checklist. The latter considers certain criteria: statistical association, which encompassed p of interaction, prespecification of subgroups, sample size, number of factors analyzed, and overall study result; biological plausibility of observed differences; and consistency between results of similar studies. A score was assigned to each criterion and the tool related global summation to a recommendation on the applicability of subset results in clinical decision making. RESULTS Preliminary questions had positive answers, so checklist was applied. Statistical association obtained "null" assessment (-3 points), including a "doubtful" p of interaction (p = 0.0650) among subgroups and mortality reached no statistical significance for global population. These findings reduced the reliability of subset analysis. Biological plausibility was considered "probable" (+3 points) because antiviral could have a greater effect before the inflammatory process and clinical worsening. Consistency between results of similar studies was evaluated as "possible" (+2 points) analysis for compatibility of ACTT-1 and SOLIDARITY study results. The recommendation about application of subset analysis results according to the risk of patients was "null". CONCLUSIONS This structured interpretation of subgroup analysis suggested too much uncertainty in hypothesis about remdesivir could reduce mortality in patients with severe COVID-19 who required non-high- flow oxygen. It was probably a random finding. Therefore, a randomized clinical trial about effect of remdesivir in mortality in patients with COVID‑19 and non-high-flow oxygen is essential.