4. ROBINS‐I assessments: anticoagulants (all types) versus no treatment for people hospitalised with COVID‐19 (all‐cause mortality).

Study	Bias due to confounding	Bias in selection of participants into the study	Bias in classification of interventions	Bias due to deviations from the intended intervention	Bias due to missing data	Bias in measurement of outcomes	Bias in selection of the reported result	Overall risk of bias
Ayerbe 2020	Critical risk	No information	Serious risk	Low risk	Critical risk	Low risk	Low risk	Critical risk
Judgement	One or more prognostic variables are likely to be unbalanced between the compared groups. There is no baseline characteristics table comparing the two groups. Essential characteristics, such as participants already using anticoagulants, participants who underwent surgery during the hospitalisation, concomitant antiplatelet use, and history of venous thromboembolism, were not considered.	Participants included in both groups were selected from 17 hospitals, and the study was retrospective, therefore it is not possible to know whether the selection was free from bias.	As this was a retrospective study, there is a high risk that the interventions received by participants in the same group were not standardised. The type and doses of heparin in the intervention group were not described.	No deviations from the intended intervention were reported in the study, and if any deviation occurred from usual practice, it was unlikely to impact on the outcome.	There were missing outcome data for 56 participants with no specific information or appropriate analyses. These missing data could cause a critical impact on the estimates.	It is unlikely that the outcome assessment (death) was influenced by the knowledge of the intervention received by the study participants.	The study protocol was not identified but all reported results corresponded to the intended outcome.	The study is too problematic to provide useful evidence.
Liu 2020	Critical risk	Critical risk	Serious risk	Low risk	Low risk	Low risk	Serious risk	Critical risk
Judgement	One or more prognostic variables are likely to be unbalanced between the compared groups. There is no baseline characteristics table comparing the two groups. Essential characteristics, such as participants already using anticoagulants, participants who underwent surgery during the hospitalisation, concomitant antiplatelet use, and history of venous thromboembolism, were not considered.	Participants included in both groups were selected from a single hospital, and the study was retrospective, therefore it is not possible to know whether the selection was free from bias. The selection for the study was strongly related to both the intervention and the outcome of interest. We cold not adjust the analyses for this selection bias.	As this was a retrospective study, there is a high risk that the interventions received by participants in the same group were not standardised. The type and doses of heparin in the intervention group were not described.	No deviations from the intended intervention were reported in the study, and if any deviation occurred from usual practice, it was unlikely to impact on the outcome.	No missing data was reported for this outcome.	It is unlikely that the outcome assessment (death) was influenced by the knowledge of the intervention received by the study participants.	The study protocol was not identified or was not available or both (only a preprint was available), and it is not possible to exclude bias in selection of reported effect estimate, based on the results, from different subgroups analyses.	The study is too problematic to provide useful evidence.
Paranjpe 2020	Serious risk	Moderate risk	Serious risk	Low risk	Low risk	Low risk	Low risk	Serious risk
Judgement	To minimise the impact of the absence of randomisation, an adjusted analysis with propensity scores was performed considering confounding demographic, clinical, and medication use. However, the confounding factors 'participants who underwent surgery during the hospitalisation', 'active cancer treatment', 'concomitant antiplatelet use' and 'history of venous thromboembolism' were not considered.	The included participants in both groups were selected from the same hospital, and selection may have been related to intervention and outcome, but the study authors used appropriate methods to adjust for selection bias.	There is a high risk that the interventions received by participants in the same group were not standardised. There is a high risk of differential classification errors because the information on the status of the interventions was obtained retrospectively.	No deviations from the intended intervention were reported in the study, and if any deviation occurred from usual practice, it was unlikely to impact on the outcome.	No missing data were reported for this outcome.	It is unlikely that the outcome assessment (death) was influenced by the knowledge of the intervention received by the study participants.	The study protocol was not identified but all reported results corresponded to the intended outcome.	The study has some important problems
Russo 2020	Serious risk	Moderate risk	Serious risk	No information	Low risk	Low risk	Low risk	Serious risk
Judgement	To minimise the impact of the absence of randomisation, we performed an analysis with propensity scores, considering confounding demographic and clinical factors, and medication use. However, the study did not consider confounding factors 'participants who underwent a surgery during the hospitalisation', 'active cancer treatment' and 'history of venous thromboembolism'.	The included participants in both groups were selected from the same hospital, and selection may have been related to intervention and outcome, but the study authors used appropriate methods to adjust for selection bias.	There is a high risk that the interventions received by participants in the same group were not standardised. There is a high risk of differential classification errors because the information on the status of the interventions was obtained retrospectively.	Insufficient information to judge. No information is reported on whether there was deviation from the intended intervention.	No missing data were reported for this outcome.	It is unlikely that the outcome assessment (death) was influenced by the knowledge of the intervention received by the study participants.	The study protocol was not identified but all reported results corresponded to the intended outcome.	The study has some important problems
Shi 2020	Critical risk	Critical risk	Serious risk	Low risk	Low risk	Low risk	Low risk	Critical risk
Judgement	One or more prognostic variables are likely to be unbalanced between the compared groups. There is a baseline characteristics table comparing the two groups with limited items. However, the study did not compare essential characteristics, such as participants already using anticoagulants, participants who underwent surgery during the hospitalisation, concomitant antiplatelet use, and history of venous thromboembolism.	The participants of the two groups (intervention and comparator) were selected from the same hospital, but as the study was retrospective, it is not possible to know if the selection was free from bias. The selection for the study was strongly related to both the intervention and the outcome of interest. We could not adjust the analyses for this selection bias.	There is a risk that the interventions received by participants in the same group were not standardised. There is a high risk of differential classification errors because the information on the status of the interventions was obtained retrospectively.	No deviations from the intended intervention were reported in the study, and if any deviation occurred from usual practice, it was unlikely to impact on the outcome.	No missing data were reported for this outcome.	It is unlikely that the outcome assessment (death) was influenced by the knowledge of the intervention received by the study participants.	The study protocol was not identified but all reported results corresponded to the intended outcome.	The study is too problematic to provide useful evidence.
Tang 2020	Critical risk	Critical risk	Serious risk	No information	Low risk	Low risk	Critical risk	Critical risk
Judgement	One or more prognostic variables are likely to be unbalanced among the compared groups. There was no table comparing the characteristics of the two groups at baseline. The comparator group included participants who used heparin for less time or did not use heparin. These participants may be less severely ill than those in the intervention group.	Participants included in both groups were selected from the same hospital, but as the study was retrospective, it is not possible to know whether the selection was free from bias. The selection for the study was strongly related to both the intervention and the outcome of interest. We could not adjust the analyses for this selection bias.	As this was a retrospective study, there is a high risk that the interventions received by participants in the same group were not standardised. Besides, the comparator group also included participants who used heparin for less than seven days. This proximity to the case definition for the intervention group increases the risk of error in the classification of participants. Also, the comparator group considered two very different types of intervention.	Insufficient information to judge. No information is reported on whether there was deviation from the intended intervention.	No missing data were reported for this outcome.	It is unlikely that the outcome assessment (death) was influenced by the knowledge of the intervention received by the study participants.	The study protocol was not identified or was not available or both, and it is not possible to exclude bias in selection of reported effect estimate, based on the results, from multiple measurements within the outcome domain, multiple analyses of the intervention‐outcome relationship, and different subgroups analyses.	The study is too problematic to provide useful evidence.