Critically ill patients are frequently transfused, with 40–50 % of patients receiving a red blood cell transfusion during their stay in the intensive care unit (ICU) [1].

Current red blood cell transfusion practice in the ICU has largely been shaped by a landmark trial published in 1999, which taught us that a restrictive transfusion trigger is well tolerated in the critically ill and of particular benefit in the young and less severely ill [2]. Following this trial, a restrictive trigger has been widely adopted [3–5]. Nevertheless, transfusion rates in the ICU remain high, rendering blood transfusion part of everyday practice in the ICU.

Red blood cell transfusion rates in the ICU are high because many patients suffer moderately to severe anemia. Anemia is a hallmark of critical illness, occurring in up to 90 % of patients. The cause of anemia is multifactorial, but the presence of inflammation is an important contributor. As anemia usually develops early in the course of critical illness, the term “anemia of inflammation” has become interchangeable with the term “anemia of chronic disease,” which may better describe the critically ill patient population. Transfusion of fresh frozen plasma (FFP) is also common practice in the ICU, with estimates of 12–60 % of patients receiving plasma during their stay [6, 7]. Frequent transfusion of FFP is due to a large proportion of patients with a coagulopathy and/or patients who experience or are considered at risk for bleeding [7, 8]. The reported wide variation in the practice of FFP transfusion suggests clinical uncertainty about best practice [7–9].

Similarly, thrombocytopenia is a prevalent, occurring in up to 30 %, triggering platelet transfusion in 10 % of patients [7]. Taken together, transfusion of blood products is one of the most common therapies in the ICU.

It is increasingly clear that an association between transfusion and adverse outcome exists, including the occurrence of lung injury, multiple organ failure, thromboembolic events, and nosocomial infections. These associations are not restricted to the critically ill patient population, but the relation between blood transfusion and adverse outcome seems most apparent in this group [10], suggesting that critically ill patients may have specific features which render them susceptible to possible detrimental effects of a blood transfusion. Thereby, ICU physicians are advised to be restrictive with transfusion [11, 12]. A challenge in understanding the optimum use of blood products in the critically ill is delineating whether this association is causative or simply a result of the residual confounding and bias by indication which influences observational studies.

The dark side of these efforts to adhere to a restrictive practice to mitigate adverse effects of blood transfusion may be under-transfusion, which may be particularly relevant to the correction of anemia with red blood cells. Multiple studies have shown an association between anemia and adverse outcome, in a wide variety of patients, including brain injury and myocardial infarction [11, 13–16]. Thereby, both anemia and transfusion are unwanted conditions, posing a challenge to the treating physician, who wonders what to do with a low hemoglobin level. Transfuse, not transfuse, or consider an alternative treatment?

These observations underline the need for a careful assessment of whether risks of transfusion outweigh the perceived benefit. In other words, can a particular patient tolerate anemia? Tolerance to anemia differs between different populations, depending on physiologic state, diagnosis, comorbidity, and cause of anemia. Although guidelines advise taking age and other physiologic variables into consideration in the decision to transfuse [11, 12], studies which have compared different triggers in different settings have been limited, and the overall evidence base is weak. Red cell transfusion, in particular, is still strongly influenced by the landmark “TRICC” trial and applied in a “one-size-fits-all” fashion. This despite changes to the red cell product in many countries (the introduction of leucodepletion), improvements in other aspects of critical care (which might change the “signal-to-noise” ratio associated with blood transfusion), and the fact that the original trial was underpowered and stopped early having reached only half of the intended sample size.