Convalescent Plasma Therapy for COVID-19: A Graphical Mosaic of the Worldwide Evidence

Abstract

Convalescent plasma has been used worldwide to treat patients hospitalized with coronavirus disease 2019 (COVID-19) and prevent disease progression. Despite global usage, uncertainty remains regarding plasma efficacy, as randomized controlled trials (RCTs) have provided divergent evidence regarding the survival benefit of convalescent plasma. Here, we argue that during a global health emergency, the mosaic of evidence originating from multiple levels of the epistemic hierarchy should inform contemporary policy and healthcare decisions. Indeed, worldwide matched-control studies have generally found convalescent plasma to improve COVID-19 patient survival, and RCTs have demonstrated a survival benefit when transfused early in the disease course but limited or no benefit later in the disease course when patients required greater supportive therapies. RCTs have also revealed that convalescent plasma transfusion contributes to improved symptomatology and viral clearance. To further investigate the effect of convalescent plasma on patient mortality, we performed a meta-analytical approach to pool daily survival data from all controlled studies that reported Kaplan-Meier survival plots. Qualitative inspection of all available Kaplan-Meier survival data and an aggregate Kaplan-Meier survival plot revealed a directionally consistent pattern among studies arising from multiple levels of the epistemic hierarchy, whereby convalescent plasma transfusion was generally associated with greater patient survival. Given that convalescent plasma has a similar safety profile as standard plasma, convalescent plasma should be implemented within weeks of the onset of future infectious disease outbreaks.

Keywords: COVID-19; Kaplan–Meier analysis; SARS-CoV-2; convalescent plasma therapy; passive antibody transfer.

Figure 1

Kaplan–Meier survival plots from worldwide controlled studies investigating convalescent plasma therapy for coronavirus disease 2019 (COVID-19). Available Kaplan–Meier survival data from worldwide controlled studies [n = 7 randomized clinical trials (11, 27, 41, 43, 46, 47, 71) and n = 9 matched-control studies (, , –62, 64, 69, 70)] was extracted and replotted. Each panel represents data from one study, where the orange line represents the survival of the cohort transfused with convalescent plasma, and the blue line represents the survival of the cohort that received standard of care or placebo transfusion. The label for each study provides the first author name, location, and study type, where matched-control studies are denoted by (M) and randomized clinical trials are denoted by (R). For each study, sample sizes from the number at risk at study onset are presented for the treatment and control cohorts. No statistical analyses were performed. Interpretation: Qualitative inspection of the Kaplan–Meier survival data from 16 controlled studies revealed a directionally consistent pattern whereby convalescent plasma transfusion was associated with greater patient survival compared to non-transfused patients.

Figure 2

Aggregate Kaplan–Meier survival plots from all available controlled studies investigating convalescent plasma therapy for COVID-19. (A) Pools all controlled studies [n = 6 randomized clinical trials (11, 27, 41, 43, 47, 71, 74) and n = 9 matched-control studies (, , –62, 64, 69, 70)] with available Kaplan–Meier survival data except for the large randomized clinical trial, the RECOVERY Trial. (B) Pools randomized clinical trials (n = 6) with available data except for the RECOVERY Trial. (C) Pools matched-control studies (n = 9) with available data. (D) Shows the RECOVERY Trial data from the overall cohort, and (E) shows the RECOVERY Trial data from the subgroup of patients that were seronegative at the time of plasma transfusion. Available 28-day Kaplan–Meier survival data from worldwide controlled studies was extracted, pooled, and replotted. For each panel, the orange line represents the survival of the cohort transfused with convalescent plasma, and the blue line represents the survival of the cohort that received standard of care or placebo transfusion. Error bands represent 95% confidence intervals. The 28-day survival (%) for the aggregate convalescent plasma cohort and the aggregate control cohort is presented. The table at the bottom of each panel presents the number at risk on the day of enrollment or randomization, day 10, day 20, and day 28 for the aggregate convalescent plasma cohort and the aggregate control cohort. Interpretation: The aggregate survival data from (A) all controlled studies except for the RECOVERY Trial, (B) all randomized clinical trials except for the RECOVERY Trial, and (C) all matched-control studies revealed a consistent pattern whereby convalescent plasma transfusion was associated with greater patient survival compared to non-transfused patients. (D) The overall RECOVERY Trial cohort demonstrated no survival benefit of convalescent plasma. (E) The modest efficacy signal from the RECOVERY Trial subgroup of seronegative patients supports the concept that early plasma transfusion reduces patient mortality.