Risk of bias assessment of Liu 2020
Domain	Assessed outcomes	Authors' judgement	Support for judgement
Bias due to confounding	Mortality Clinical improvement	Serious	Only adjusted for hydroxychloroquine and azithromycin, intubation status and duration, length of hospital stay, and oxygen requirement on the day of transfusion. Not adjusted for e.g. age and gender
	Adverse events	Not applicable	Paper only reports transfusion‐related adverse events for intervention group
Bias in selection of participants into the study	Mortality Clinical improvement	Moderate	Selection into the study may have been related to intervention and outcome, but the study authors used appropriate methods to adjust for the selection bias. Quote: "propensity score‐matched analysis using The Mount Sinai Hospital’s COVID‐19 confirmed patient pool from the same calendar period (24 March 2020 105 to 8 April 2020). A logistic regression was fit to predict the potential for plasma therapy based on time series data obtained at baseline upon admission, prior to transfusion, and the day of 107 transfusion."
	Adverse events	Not applicable	Paper only reports transfusion‐related adverse events for intervention group
Bias in classification of interventions	Mortality Clinical improvement	Critical	Assignment to control group was done retrospectively. Treatment details of control group are not provided. Knowledge of participants' outcomes at the time of assignment to the control group could have had a major impact on the selection.
	Adverse events	Not applicable	Paper only reports transfusion‐related adverse events for intervention group
Bias due to deviations from intended interventions	Mortality Clinical improvement	Low	All participants received the intended intervention. Most common co‐interventions (hydroxychloroquine and azithromycin, intubation status and duration, length of hospital stay, and oxygen requirement on the day of transfusion) were propensity score‐matched. Other co‐interventions were administered too infrequently to enforce exact matching
	Adverse events
Bias due to missing data	Mortality Clinical improvement	Low	Data were reasonably complete
	Adverse events	Critical	No safety data for control group available
Bias in measurement of outcomes	Mortality Clinical improvement	Moderate	Median follow‐up comparable between groups. However, outcome assessors were not blinded to intervention and the study was performed retrospectively.
	Adverse events	Critical	Only transfusion‐related adverse events reported
Bias in selection of the reported results	Mortality Clinical improvement	Critical	Retrospective study; selection of all reported results are likely biased
	Adverse events	Critical	Observation period unclear, non‐occurrence of transfusion‐related adverse events only reported in discussion section
Overall bias	Mortality	Critical	The study is too problematic to provide any useful evidence, however better evidence is yet insufficient.
	Clinical improvement	Critical	The study is too problematic to provide any useful evidence, however better evidence is yet insufficient.
	Adverse events	Critical	The study is too problematic to provide any useful evidence, however better evidence is yet insufficient.