Special Feature

A Specific Transfusion Threshold for Critically Ill Patients?

By Francisco Baigorri, MD, PhD

Maintenance of sufficient oxygen delivery (DO2) to meet tissue oxygen demands is the mainstay in the treatment of critically ill patients. Red blood cell transfusion (RBCt) increases oxygen content by raising hemoglobin levels and has been proposed as an effective therapeutic maneuver for increasing tissue oxygen availability. However, RBCt involves a number of risks: it may transmit viruses such as the human immunodeficiency virus and hepatitis, and the red cells also exert immunosuppressive and microcirculatory effects. Consequently, current guidelines for blood transfusion suggest that patients with hemoglobin levels less than 70 g/L should receive transfusions but recommend avoiding an empirical, automatic transfusion threshold and stress the need for individual patient assessment.1 However, these guidelines do not make specific recommendations for critically ill patients. This essay discusses the evidence available for making specific recommendations about transfusion requirements in critically ill patients.

Table 1

Physiologic adjustments occurring during anemia

    · Increase in cardiac output

    · Redistribution of the cardiac output

    · Increment of the extraction capabilities

Compensatory Mechanisms for Anemia in Normal Subjects and Critically Ill Patients

The following mechanisms at the systemic and capillary levels are involved in the maintenance of adequate tissue oxygenation during anemia. (See Table 1.2):

1) Reducing the hematocrit lowers blood viscosity, leading to an increased venous return and a decreased ventricular afterload. It provides a greater cardiac output. As a consequence of the increase in cardiac output when hematocrit is reduced, systemic DO2 reaches a peak value of about 110% of pre-anemic DO2 for a hematocrit of about 30%. However, as hematocrit declines further, DO2 begins to decrease and falls below normoxic pre-anemic levels for a hematocrit of 20-25%. (See Figure.)

2) The increase in cardiac output during normovolemic hemodilution is associated with a redistribution of blood flow between organs, which is not altered by beta-adrenergic blockade.

3) An increase in tissue oxygen capabilities has also been observed during hemorrhagic shock in hemodiluted anesthetized dogs. These physiologic adjustments allow oxygen consumption (VO2) to remain constant until hematocrit falls to about 10%.


    Relationship between oxygen delivery (DO2) and hematocrit (Htc) during

    normovolemic hemodilution. Adapted from Van der Linden P. The opti-

    mal hematocrit. In: Vincent JL, ed. Yearbook of Intensive Care and Emer-

    gency Medicine. Berlin: Springer-verlag; 1994:228-236.

There is a remarkable tolerance of low hematocrit values, even in subjects with coronary artery disease or carotid artery stenosis due to an increase of trans-stenotic flow during hemodilution.3 The efficacy of the mechanisms maintaining adequate tissue DO2 when blood oxygen content is reduced seems to depend mainly on the maintenance of normovolemia and on the integrity of the myocardial function.

In critical illnesses, most of the compensatory mechanisms for anemia are altered by the presence of hypovolemia, hypoxemia, depressed myocardial function, and/or altered tissue oxygen extraction capabilities. In addition, the oxygen demand is often increased. Although the information available does not allow us to define a generally applicable number for the lowest hematocrit tolerated without organ dysfunction, it has been shown that optimal survival in critically ill patients is associated with a hematocrit of around 33% (hemoglobin level above 100 g/L).4

Transfusion Threshold in Critically Ill Patients

A hemoglobin level of 100 g/L has been widely used as transfusion threshold in the resuscitation of critically ill patients. In fact, most randomized, controlled, clinical trials (RCTs) comparing supernormal with normal DO2 in critically ill patients maintain hemoglobin values greater than 100 g/L.4-10

However, the usefulness of this practice remains controversial. In a recently published cohort study evaluating the effect of transfusion practice on mortality rates in 4470 critically ill patients, it was again observed that patients who died had lower hemoglobin values (95 ± 26 vs 104 ± 23 g/L; P < 0.0001);11 however, only the subgroup of patients with anemia (hemoglobin < 95 g/L), a high APACHE II score (> 20), and a cardiac diagnosis (including all diagnoses related to ischemic heart disease, rhythm disturbances, and cardiac arrest, as well as cardiac and vascular surgical procedures) had a significantly lower mortality rate when given RBCt.

Moreover, most studies evaluating the effect of RBCt on oxygen transport values have shown that RBCt fails to increase global VO2 despite a significant increase in DO2,12 even in patients who showed an increase in VO2 during an infusion of dobutamine.13 The studies in which VO2 was determined by indirect calorimetry confirmed that VO2 did not increase after RBCt was given to increase hemoglobin above 100 g/L, even in those patients who had increased concentrations of plasma lactate.14,15

It has also been shown that RBCt worsens gastric tissue hypoxia measured by gastric tonometry in septic patients.15,16 This deterioration has been related to increased age of stored donor RBCs.15 Storage transitorily depresses the ability of RBCs both to unload oxygen in the periphery17 and to deform18 in relation to 2,3-DPG and ATP decrease. Consequently, the ability of RBCt to support oxygen uptake requires further investigation.

RCTs Evaluating Transfusion Strategies

Six RCTs19-24 evaluating transfusion strategies on a variety of outcomes were identified in a recent review.25 (See Table 2.) Most of these studies included a small number of patients.

Table 2

Randomized clinical trials evaluating transfusion strategies

    Author/ Patients No. of Interventions

    Year published patients

    Weisel, et al.19/ CABG* 27 Crystalloid vs. blood 1984 or colloid solutions

    Blair, et al.20/ GI_ hemorrhage 50 Immediate RBCt vs.1886 no transfusion

    Fortune, et al.21/ Trauma 25 Hct = 30% vs. 1987 Hct = 40%

    Johnson, et al.22/ CABG 38 Hct maintained at 1992 25% vs. Hct maintained at 32%

    Vichinsky, et al.24/ Sickle cell 604 Decreasing hemo-1995 disease + surgery globin S < 30% or Hb > 100 g/L

    Hébert, et al.23,25/ ICU 69 Hb = 70-90 g/L vs. 1995 Hb = 100-120 g/L * CABG = coronary artery bypass grafting _ GI = gastrointestinal

Only one of these studies included normovolemic critically ill patients and evaluated mortality and organ failure.23 Patients enrolled in that study had been admitted to the intensive care unit with hemoglobin values less than 90 g/L within 72 hours of admission. Patients were randomly allocated to maintain hemoglobin values between 100 and 120 g/L (n = 36) and between 70 and 90 g/L (n = 33). This small pilot RCT did not detect any difference in mortality (including 30-day all-cause mortality, ICU mortality, 120-day mortality, and survival times) or in organ dysfunction scores between the two groups. On the other hand, maintaining hemoglobin levels between 70 and 90 g/L decreased the average number of units transfused from 4.8 to 2.5 units (a 48% reduction). This study continued enrolling patients until November 1997, and Hébert et al hope to demonstrate that a simple and cheap intervention of this kind is not only efficacious in decreasing RBCt but is also safe in high-risk patients.25


Larger clinical trials are needed for the implementation of treatment guidelines regarding transfusion thresholds in critically ill patients. No automatic transfusion threshold can be proposed at present, given the data available. In patients with a hematocrit between 20% and 30% (or hemoglobin value between 70 and 100 g/L), the suitability of DO2 requires careful assessment. The need for RBCt should be evaluated on an individual basis, considering additional factors such as the age of the patient, cause of the anemia, chronicity of the anemia, hemodynamic status, and presence of coexisting cardiac, pulmonary, or vascular disease.


    1. American College of Physicians. Ann Intern Med 1992; 116:403-406.

    2. Van der Linden P. The optimal hematocrit. In: Vincent JL, ed. Yearbook of Intensive Care and Emergency Medicine. Berlin: Springer-Verlag; 1994:228-236.

    3. Spahn DR, Pasch T. Critical hematocrit. In: Vincent JL, ed. Yearbook of Intensive Care and Emergency Medicine. Berlin: Springer-Verlag; 1996:635-642.

    4. Shoemaker WC, et al. Chest 1988;94:1176-1186.

    5. Tuchschmidt J, et al. Chest 1992;102:216-220.

    6. Gutierrez G, et al. Lancet 1992;339:195-199.

    7. Yu M, et al. Crit Care Med 1993;21:830-838.

    8. Boyd O, et al. JAMA 1993;270:2699-2707.

    9. Hayes MA, et al. N Engl J Med 1994;330:1717-1722.

    10. Yu M, et al. Crit Care Med 1995;23:1025-1032.

    11. Hébert PC, et al. Am J Respir Crit Care Med 1997;155: 1618-1623.

    12. Dietrich KA, et al. Crit Care Med 1990;18:940-944.

    13. Lorente JA, et al. Crit Care Med 1993;21:1312-1318.

    14. Ronco JJ, et al. Am Rev Respir Dis 1991;143:1267-1273.

    15. Marik PE, Sibbald WJ. JAMA 1993;269:3024-3029.

    16. Silverman HJ, Tuma P. Chest 1992;102:184-188.

    17. Apstein CS, et al. Am J Physiol 1985;285:H508-H515.

    18. Stuart J, Nash GB. Blood Rev 1990;4:141-147.

    19. Weisel RD, et al. J Thorac Cardiovasc Surg 1984;88:26-38.

    20. Blair SD, et al. Br J Surg 1986;73:783-785.

    21. Fortune JB, et al. J Trauma 1987;27:243-249.

    22. Johnson RG, et al. J Thorac Cardiovasc Surg 1992; 104:307-314.

    23. Hébert PC, et al. JAMA 1995;273:1439-1444.

    24. Vichinsky EP, et al. N Engl J Med 1995;333:206-213.

    25. Hébert PC. Transfusion requirements in critical care: A multicenter controlled clinical trial. In: Vincent JL, ed. Yearbook of Intensive Care and Emergency Medicine. Berlin: Springer; 1998:202-217.