Table 1.

Anticoagulants (all types) compared with no treatment for people hospitalized (ICU and ward) with COVID-19: summary of findings

Outcome	Impact	No. of participants (studies)	Certainty of the evidence (GRADE)
All-cause mortality	One study reported reduction in mortality by OR adjusted for confounding (58% reduction in chance of death; 2075 participants)	5685	⊕⊝⊝⊝
Follow-up: range 8–28 days	One study reported reduction in mortality only in a subgroup of severely ill participants (HR 0.86, 0.82 to 0.89; 395 participants)	(6 retrospective NRS)	Very low*†‡
	Three studies reported no differences by adjusted OR (1.64, 0.92 to 2.92; 449 participants), unadjusted OR (1.66, 0.76 to 3.64; 154 participants) or adjusted RR (1.15, 0.29 to 2.57; 192 participants)
	One study reported zero events in both intervention groups
Need for additional respiratory support	No study measured this outcome
Mortality related to COVID-19	No study measured this outcome
Deep vein thrombosis	No study measured this outcome
Pulmonary embolism	No study measured this outcome
Major bleeding  Follow-up duration not reported	One study reported 24 bleeding events (3%) in the intervention group and 38 (1.9%) in the control group (OR 1.62, 0.96 to 2.71)	2773 (1 retrospective NRS)	⊕⊕⊝⊝ Low‡§

Odds ratios (ORs), hazard ratios (HRs), and risk ratios (RRs) are shown with 95 per cent confidence intervals. GRADE Working Group grades of evidence: high certainty—the authors are very confident that the true effect lies close to that of the estimate of the effect; moderate certainty—the authors are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different; low certainty—the authors’ confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect; very low certainty—the authors have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect. *Downgraded one level owing to study limitations; overall critical/serious risk of bias across studies, especially related to confounding. ^†Downgraded one level owing to inconsistency; the authors decided not to pool data because of the heterogeneity of studies (especially differences in interventions). ^‡Downgraded one level owing to imprecision; narrative synthesis was conducted with imprecise estimates. ^§Downgraded one level owing to study limitations; overall serious risk of bias, especially related to confounding. NRS, non-randomized studies.