A consistent message within critical care publications has been that a restrictive transfusion strategy is non-inferior, and possibly superior, to a liberal strategy for stable, non-bleeding critically ill patients. Translation into clinical practice has, however, been slow. Here, we describe the degree of adherence to UK best practice guidelines in a regional network of nine intensive care units within Wessex. All transfusions given during a 2-month period were included (n = 444). Those given for active bleeding or within 24 h of major surgery, trauma or gastrointestinal bleeding were excluded (n = 148). The median (IQR [range]) haemoglobin concentration before transfusion was 73 (68-77 [53-106]) g.l-1 , with only 34% of transfusion episodes using a transfusion threshold of < 70 g.l-1 . In a subgroup analysis that did not study patients with a history of cardiac disease (n = 42), haemoglobin concentration before transfusion was 72 (68-77 [50-98]) g.l-1 , with only 36% of transfusion episodes using a threshold of < 70 g.l-1 (see Fig. 3). Most blood transfusions given to critically ill patients who were not bleeding in this audit used a haemoglobin threshold > 70 g.l-1 . The reason why recommendations on transfusion triggers have not translated into clinical practice is unclear. With a clear national drive to decrease usage of blood products and clear evidence that a threshold of 70 g.l-1 is non-inferior, it is surprising that a scarce and potentially dangerous resource is still being overused within critical care. Simple solutions such as electronic patient records that force pause for thought before blood transfusion, or prescriptions that only allow administration of a single unit in non-emergency circumstances may help to reduce the incidence of unnecessary blood transfusions.
Keywords: blood transfusion; critical care; transfusion triggers; translational science.
© 2017 The Association of Anaesthetists of Great Britain and Ireland.