Population
Adult patients in any clinical setting (30 studies, n= 17,334).
Intervention
Restrictive red blood cell transfusion.
Comparison
Liberal red blood cell transfusion.
Outcome
The pooled risk ratio (RR) for thromboembolic events was 0.65 (very low-quality evidence), favouring the restrictive strategy. There were no significant differences in cerebrovascular accidents (RR= 0.83, very low-quality evidence) or myocardial infarction (RR= 1.05, low-quality evidence). Subgroup analyses showed that a restrictive (relative to liberal) strategy reduced thromboembolic events in trials conducted in North America, and myocardial infarctions in the subgroup of trials where the restrictive transfusion threshold was 7 g/dl but not in the 8 g/dl subgroup (with a liberal transfusion threshold of 10 g/dl in both subgroups).