'Risk of bias' assessment of Li 2020
Domain	Assessed outcomes	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Mortality Clinical improvement Adverse events	Low	Quote: "Patients were randomly assigned via computer‐generated random numbering (1:1) to receive standard treatment coupled with convalescent plasma transfusion or standard treatment alone (control group) (Figure 1). The randomization was stratified based on the severity of COVID‐19 (severe or life‐threatening) and a randomization schedule was generated using block randomization with block size of 4 for each type of COVID‐19 by SAS software."
Allocation concealment  (selection bias)	Mortality Clinical improvement Adverse events	Low	Quote: "This random number will connect the subject to the designated treatment group (experimental group or control group) for treatment. [...] Staff responsible for randomization will only be responsible for the assignment of random groups and will not be involved in any specific trial operations."
Blinding of participants and personnel (performance bias)	Mortality Clinical improvement Adverse events	High	Quote: "open‐label" Co‐interventions not balanced across arms
Blinding of outcome assessment (detection bias)	Mortality	Low	Quote: "To avoid assessment bias, the evaluation of clinical outcomes was performed by an investigator who was blind to the study group allocation."
	Clinical improvement Adverse events	Low	Quote: "To avoid assessment bias, the evaluation of clinical outcomes was performed by an investigator who was blind to the study group allocation."
Selective reporting (reporting bias)	Mortality	Low	Reported as determined at protocol stage
	Clinical improvement	Unclear	Quote: "A post hoc analysis was added to compare rates of improvement at days 7, 14, and 28."
	Adverse events	High	Only transfusion‐related adverse events reported
Incomplete outcome data (attrition bias)	Mortality	Low	ITT population reported
	Clinical improvement	Low	ITT population reported
	Adverse events	High	No safety data for control group available
Other bias	Mortality Clinical improvement Adverse events	Unclear	Quote: "Due to the containment of the COVID‐19 epidemic in Wuhan, China, the numbers of patients with COVID‐19 decreased in late March 2020. [...] The trial was terminated early after 103 of a planned 200 patients were enrolled." The study expressed effect estimates as odds ratios. Therefore we recalculated relative effects as risk ratios. We noticed that our calculation arrived at the same numerical values, and therefore highlight that effect estimates, which are indicated as odds ratios in the primary study, are in fact risk ratios.