Between May 28 and August 27, 2020, a total of 448 patients were assessed for inclusion criteria at 12 participating centers, and 334 patients were enrolled. One patient withdrew informed consent before receiving the intervention. Consequently, 228 patients were assigned to convalescent plasma and 105 to placebo (Figure 1), and each patient received the assigned infusion.
The median age of the patient population was 62 years (interquartile range, 52 to 72); 67.6% of the patients were men, and 64.9% had a coexisting condition at entry into the trial. The median time from the onset of Covid-19 symptoms to enrollment was 8 days (interquartile range, 5 to 10). An oxygen saturation below 93% while the patient was breathing ambient air was the most common severity criterion for enrollment, and more than 90% of the patients were receiving oxygen and glucocorticoids at the time of entry into the trial (Table 1).
The median volume of infused convalescent plasma was 500 ml (interquartile range, 415 to 600). Of the 215 patients from whom a baseline total anti–SARS-CoV-2 IgG antibody level could be obtained, the median titer was 1:50 (interquartile range, 0 to 1:800); 46.0% of patients had no detectable antibody level.
Total IgG and neutralizing SARS-CoV-2 antibody titers were also analyzed in the infused convalescent plasma pools, using the COVIDAR assay. The total IgG antibody median value of all pools was 1:3200 (interquartile range, 1:800 to 1:3200). Analysis of SARS-CoV-2 neutralizing antibody titers was available for 125 of the infused convalescent plasma doses (56%), with an 80% inhibitory concentration median titer of 1:300 (interquartile range, 1:136 to 1:511). The correlation analysis between the total SARS-CoV-2 antibody titer and the neutralizing antibody titer in the convalescent plasma pools is provided in the Figure S1.
At day 30, no significant difference was noted between the convalescent plasma group and the placebo group in the distribution of clinical outcomes according to the ordinal scale (odds ratio, 0.83; 95% confidence interval [CI], 0.52 to 1.35; P=0.46) (Table 2 and Figure 2). The assumption of the proportional odds ratio for the primary outcome was supported by the nonsignificant results of the Brant test (P=0.34). After adjustment for sex, history of COPD, and history of tobacco use, the odds ratio for the score on the ordinal scale between the convalescent plasma and placebo groups was 0.92 (95% CI, 0.59 to 1.42; P=0.70).
The 30-day mortality was 10.96% (25 of 228 patients) in the convalescent plasma group and 11.43% (12 of 105) in the placebo group, for a risk difference of −0.46 percentage points (95% CI, −7.8 to 6.8). No significant between-group differences in clinical status on the ordinal scale were seen either at day 7 (odds ratio, 0.88; 95% CI, 0.58 to 1.34) or at day 14 (odds ratio, 1.00; 95% CI, 0.65 to 1.55) (Figure 2 and Table S2). The median time from enrollment to hospital discharge was 13 days (interquartile range, 8 to 30) in the convalescent plasma group and 12 days (interquartile range, 7 to 30) in the placebo group (subhazard ratio, 0.99; 95% CI, 0.75 to 1.32). Throughout the trial, the proportion of ICU admissions and invasive ventilatory support requirements was 53.9% (123 of 228 patients) and 26.8% (61 of 228 patients), respectively, in the convalescent plasma group and 60% (63 of 105 patients) and 22.9% (24 of 105 patients), respectively, in the placebo group. No significant differences were noted in the time to death or in the time to clinical improvement of at least two categories on the ordinal scale or hospital discharge (Figure 3 and Table 2).
No differences in ferritin and d-dimer levels were noted between the patient groups at day 14. Although baseline median titers were identical, patients receiving convalescent plasma had SARS-CoV-2 total antibody levels that were higher at day 2 than levels in patients receiving placebo. No differences in antibody titers were noted at days 7 or 14 (Table S3).
The prespecified subgroup analyses failed to suggest any credible subgroup effects. Convalescent plasma appeared to be associated with a worse clinical outcome in the subgroup of patients younger than 65 years of age. However, the rest of the outcome analyses for this subgroup did not show similar results (Fig. S2 and S3). Analyses of the primary outcome and of clinical improvement of at least two ordinal categories in relation to total and neutralizing antibody titers in the infused plasma pools are provided in the Supplementary Appendix.
Infusion-related adverse events were slightly more common in the convalescent plasma group (4.8%; 11 of 228 patients) than in the placebo group (1.9%; 2 of 105 patients) (odds ratio, 2.62; 95% CI, 0.57 to 12.04). Five patients in the convalescent plasma group and none in the placebo group had nonhemolytic febrile reactions. No significant differences were found in the overall incidence of adverse events (odds ratio, 1.21; 95% CI, 0.74 to 1.95) or serious adverse events (Table 2 and Table S4).