- Email: [email protected]
Blood transfusion practice today
Crit Care Clin 20 (2004) 179 – 186
Blood transfusion practice today Nicholas S. Ward, MD, Mitchell M. Levy, MD* Rhode Island Hospital, Brown Medical School, 593 Eddy Street, Providence, RI 02903, USA
In the last decade or so, the practice of packed red blood cell (PRBC) transfusions in critically ill patients has become the subject of many investigations. Once regarded as a relatively safe and effective means of increasing oxygen delivery, it has become clear that there are several aspects of blood transfusion that are not understood completely. Many investigators and clinicians feel that more scrutiny of this practice is warranted given an increased awareness of bloodborne pathogens, evidence that allogeneic PRBCs may lower susceptibility to infection [1– 7], and the ever increasing consumption of blood products by a growing critical care population. Additionally, studies on the pathogenesis and treatment of septic shock have shown conflicting data on the benefits of augmenting oxygen delivery with allogeneic PRBCs [8 –12]. In the last several years, many important studies have been published that not only cast doubt on the benefits of PRBC transfusion but also have sought to redefine the optimal threshold value of hemoglobin (HGB) concentration that warrants transfusion. Recently, there have been two large descriptive studies that sought to describe the current practice of PRBC transfusion in intensive care units (ICUs) and correlate that information with clinical outcomes. This article reviews these studies of current transfusion practices and tries to use this information as a guide to effective transfusion practice.
Significant studies One of the first descriptive studies of ICU transfusion practice was published by Corwin et al in 1995 [13]. In this retrospective chart review of 142 ICU patients who had a length of stay greater than 1 week, 85% had received blood transfusions. They found that only 35% of the transfusion events were for acute blood loss, and 29% had no identifiable trigger. With the exception of a low hematocrit as a trigger, the pretransfusion hematocrit was generally 27%. The
* Corresponding author. E-mail address: [email protected] (M.M. Levy). 0749-0704/04/$ – see front matter D 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.ccc.2003.12.004
180
N.S. Ward, M.M. Levy / Crit Care Clin 20 (2004) 179–186
authors went on to note that while this patient population represented only 4% of inpatient days, they received 16% of all inpatient transfusions. Like others [14], they noted a correlation between blood phlebotomized and blood transfused. In their study, phlebotomy was estimated to be equal to one third of the blood transfused. Further insight into ICU transfusion practices occurred in 1998 when He´bert et al published the results of a survey of Canadian critical care physicians [15]. The authors used a scenario-based survey with changing variables to assess the usual practices of ICU physicians and what clinical factors would modify their practices. Their data collected from the responses of 193 physicians showed a wide degree of variability in transfusion practices, with most of the transfusion thresholds ranging from 8.0 to 10.0 g/dL hemoglobin (Hbg). Overall, 35% of physicians selected a transfusion threshold of 9.0 g/dL Hbg, and 40% selected a threshold of 10.0 g/dL Hbg. Once they decided to transfuse, more than 90% of physicians administered 2 units of PRBC. In 1999, a landmark study that attempted to clarify what is an appropriate transfusion level of HGB for patients in ICUs was published. He´bert and the Canadian Clinical Trials Group conducted a randomized controlled trial of patients in ICUs who were transfused at either 7.0 g/dL or 10.0 g/dL [16]. Using the primary endpoint of 30-day mortality of all causes and several other secondary endpoints, the authors showed that using the lower transfusion threshold was at least as safe as the higher value. In several subgroups, such as patients younger than 55 years and patients with lower severity scores (Acute Physiology and Chronic Health Evaluation [APACHE] II less than 20), they showed a statistically significant mortality benefit in the lower transfusion trigger group. Other mortality time points such as in-hospital mortality and 60-day mortality also showed a statistically significant benefit to the lower threshold. In addition to these clinical outcomes data, the blood use was dramatically different in the two groups. The low transfusion threshold group received about half as many units of blood (average of 2.6 units versus 5.6 units per patient). Furthermore, 30% of the low threshold group received no blood transfusions at all during the study, compared with 0% without a transfusion in the higher threshold group. The Anemia and Blood Transfusion in Critically Ill Patients trial Two large multi-center studies have been performed in the years since the publication of the randomized controlled trial of He´bert et al that have taken our understanding of current transfusion practice even further. The first study was published in 2002 by Vincent et al and titled the ABC study (Anemia and Blood Transfusion in Critically Ill Patients) [17]. The stated objectives of this study were to define the incidence of anemia in critically ill patients, describe transfusion practices, and to explore possible risks and benefits associated with transfusion. Data were collected at 146 ICUs in 15 countries throughout Europe and encompassed 3534 patients. Most of the ICUs (71%) had mixed medical and
N.S. Ward, M.M. Levy / Crit Care Clin 20 (2004) 179–186
181
surgical patients. There was a mix of institution sizes, with 16.5% of ICUs from community hospitals, 34% from city or regional centers, and 49% from academic medical centers. The investigators collected data on demographic information, medical procedures, illnesses, and severity scores on all patients. They then followed the patients for 28 days (or discharge), collecting data on HGB levels, transfusions, reasons for transfusion, and mortality. In addition to these, a substudy done the week before the main study recorded the frequency and amount of blood drawn from ICU patients. Many important points were demonstrated by the data from this study. First, the substudy on phlebotomy showed that blood withdrawal was considerable, averaging 41 mL per day. The rest of the data gives a very good picture of the average transfusion requirements and practices in these ICUs. Overall, the transfusion rate for the whole 28-day period (including post-ICU time) was 41% of 3534 patients included. The average HGB level was 11.3 at admission, and the average pretransfusion HGB was 8.4 g/dL. There was no difference in the average pretransfusion hemoglobin for patients who were thought to be actively bleeding (8.4 g/dL) and those who were not (8.5 g/dL). This is surprising, in that physicians generally are more likely to transfuse a patient at a higher HGB level if they know of active bleeding. One factor that may explain this was that patients transfused for the indication of coronary artery disease alone had a higher than average transfusion threshold of 8.7 g/dL. The authors then looked at the relationship between blood transfusions and other patient characteristics. The data showed a positive correlation between receiving any blood transfusions and 28-day mortality. The mortality risk increased with the number of units transfused. The authors then divided the patients by organ dysfunction scores and showed an increase in mortality in the transfused versus nontransfused patients in all but the most severely ill patients. Finally, a logistic regression model was used to generate propensity scores for two groups of matched patients in an effort to adjust for characteristics other than transfusion. The data, which compared 516 transfused versus 516 nontransfused patients matched for patient characteristics, showed a significantly higher mortality in transfused patients (22.7% versus 17.1%; P = 0.02) (Fig. 1).
The CRIT trial A recent multi-center study called the CRIT study that sought to quantify the transfusion practices of United States ICUs and describe associated clinical outcomes and complications was completed [28]. Patients who were admitted to any medical, surgical, or mixed ICU were enrolled and followed for 30 days or until death or discharge from the hospital. The study took place at 284 ICUs in 213 hospitals. The ICUs were 31% medical, 29% surgical, and 40% mixed. Residents were present in 76% of the ICUs, and 71% were open ICUs. Nineteen percent of the ICUs had an institutional transfusion protocol.
182
N.S. Ward, M.M. Levy / Crit Care Clin 20 (2004) 179–186
Fig. 1. Increase in mortality attributable to blood transfusion in ICU patients. (From Vincent JL, Baron JF, Reinhart K, Gattinoni L, Thijs L, Webb A, et al. Anemia and blood transfusion in critically ill patients. JAMA 2002;288(12):1499 – 507; with permission.)
The study included 4892 patients with an average age of 60 years. Overall, ICU and in-hospital mortality rates were 13% and 18%, respectively. The baseline HGB values recorded were similar to the European ICUs, with the average HGB at entry of 11.0 g/dL and approximately two thirds of patients with HGB less than 12 g/dL. Like previous studies, the average hemoglobin declined steadily over time. Overall, 44.1% of patients were transfused while in the ICU, and 48.2% were transfused during the total 30-day observation period. The average pretransfusion HGB was 8.6 g/dL. Subgroup analysis showed a higher threshold for surgical (8.8 g/dL) than for medical (8.2 g/dL) patients. All patients had severity scores calculated (APACHE II and sequential organ failure assessment [SOFA]), and there was a positive correlation between severity of illness and the number of units transfused. The average age of the blood transfused in this study was 21.3 plus or minus 11.4 days, but the authors reported no significant difference in clinical outcomes that correlated with the median age of the blood transfused to a given patient. Like previous studies, the authors looked for an association of number of units transfused and clinical outcomes. Multivariate analysis showed that the number of RBC units transfused was associated with increased ICU and hospital length of stay (LOS) compared with nontransfused patients. Furthermore, RBC transfusion also was associated independently with higher mortality rates. Like the ABC trial, the investigators used adjusted case matching analysis to look for any additional mortality risk conveyed by the transfusion of RBCs independently. The investigators matched 1059 transfused patients to 1059 nontransfused patients and adjusted for the propensity for receiving a blood transfusion. Blood transfusion remained statistically significantly associated with an increased risk for death (adjusted mortality ratio, 1.65; 95% CI, 1.35 –2.03; log-rank P < 0.001).
N.S. Ward, M.M. Levy / Crit Care Clin 20 (2004) 179–186
183
Table 1 Transfusion practice in North America and Europe
ABC trial CRIT trial
Percent transfused
Average admission HGB
Average pre-transfusion HGB
In ICU
ICU + post-ICU
8.4 8.6
11.3 11.0
37.0 44.1
41.6 48.2
Common themes These studies go a long way to illustrating the current extent of ICU anemia and transfusion therapy. Notable similarities in the major parameters between the two studies are summarized in Table 1. In addition to these variables, both studies showed a positive correlation between number of units transfused and poor outcomes. Also, risk-adjusted models from both patient data sets suggested there might be an increased risk of mortality just from blood transfusion. Finally, it should be noted that the pretransfusion thresholds of these two studies of 8.4 and 8.6 are not very different than the threshold of 9.0 seen in the study by Corwin et al in 1994, before the results of the large randomized controlled trial that supported lower transfusion thresholds. What is good transfusion practice? Although these studies are very comprehensive, they are only observational studies and therefore do not assist clinicians greatly in deciding when to transfuse. Nevertheless, some important lessons can be learned from these studies. The first is that anemia is extremely common in critically ill patients and therefore is almost to be expected. In all likelihood, much of the laboratory anemia seen is actually a reflection of fluid shifts that result from hydration of patients, so common in the ICU. This phenomenon has been demonstrated in two previous studies that showed hematocrit is an inaccurate measurement of red cell volume when there are fluid shifts [18,19]. The second lesson is that very often the only reason for giving a blood transfusion to patients is that they have a low HGB value. One must assume that these transfusions are given with the hope of improving oxygen delivery in some way. Unfortunately, there is much evidence to the contrary of this fact. There have been several studies that have shown that RBC transfusions in critically ill or septic patients may not, in fact, improve oxygen delivery [10 – 12,20]. A study done by Marik et al seemed to indicate that the failure of stored RBCs to improve oxygen delivery may be related to their duration of storage [11]. Further confirmation of this phenomenon came from a later study by Fitzgerald et al in 1997, which showed that blood transfused shortly after removal improved oxygenation, while blood stored 28 days did not [9]. The third lesson is that allogeneic blood transfusion may be injurious to patients. The associations found in these studies between more units transfused
184
N.S. Ward, M.M. Levy / Crit Care Clin 20 (2004) 179–186
Fig. 2. Percentage of patients developing a postoperative bacterial infection after coronary bypass surgery. (From Chelemer SB, Prato BS, Cox Jr PM, O’Connor GT, Morton JR. Association of bacterial infection and red blood cell transfusion after coronary artery bypass surgery. Ann Thorac Surg 2002;73(1):138 – 42; with permission.)
and evidence of worse outcome is not unexpected. Over the years, there has been increasing data that seem to indicate that there are significant harmful effects of blood transfusions. For example, many studies have found correlations between the number of red cell transfusions and postoperative infections [2– 4, 7,21]. In a recent prospective cohort study of 533 coronary artery bypass patients, the authors found infection rates of 4.8% for no transfusions, 15.2% for 1 to 2 units, 22.1% for 3 to 5 units, and 29.0% with greater than or equal to 6 units
Fig. 3. Incremental increase in hospital charges associated with increasing number of allogeneic red blood cell transfusions (From Blumberg N, Kirkley SA, Heal JM. Cost analysis of autologous and allogeneic transfusions in hip replacement surgery. Am J Surg 1996;171(3):324 – 30; with permission).
N.S. Ward, M.M. Levy / Crit Care Clin 20 (2004) 179–186
185
(Fig. 2). Other data published in recent years have shown an association between the number of allogeneic red cell units transfused and increased risk of cancer recurrence [22 –24]. In addition to these data suggesting possible harmful qualities of allogeneic RBC transfusions, there are further data showing that hospital costs rise with the number of transfusions [25 –27]. A recent paper by Blumberg et al showed that there was significant difference in incremental hospital costs of $1000 to $1500 per unit transfused when comparing patients who got the same number of autologous versus allogeneic units (Fig. 3) [25].
Summary In trying to develop a cohesive and logical transfusion strategy, one must try to keep all of the previously mentioned information in mind. While recognizing that RBC transfusions are an integral part of care for critically ill patients, one also must recognize that there are large amounts of data that associate multiple transfusions of allogeneic blood with increased risk of morbidity and mortality. Some data exist showing causation and association, such as studies demonstrating the immunomodulatory effects of allogeneic transfusions. In addition, the blood supply is a limited resource that should not be used indiscriminately. It therefore behooves clinicians to subject each unit of blood transfused to scrutiny. In all likelihood, patients who are not actively bleeding and who are not hypovolemic probably get little to no benefit from allogeneic blood transfusions while their hemoglobin is greater than 7g/dL. They do, however, get needless exposure to a potentially toxic substance. Although it is the authors’ hope that more research will be performed to clarify the risks and benefits of blood transfusion, the authors also hope that knowledge of already published studies will continue to spread and replace the unfounded practices of the past.
References [1] Vamvakas EC, Blajchman MA. Deleterious clinical effects of transfusion-associated immunomodulation: fact or fiction? Blood 2001;97(5):1180 – 95. [2] Vamvakas EC, Carven JH, Hibberd PL. Blood transfusion and infection after colorectal cancer surgery. Transfusion 1996;36:1000 – 8. [3] Vamvakas EC, Moore SB, Cabanela M. Blood transfusion and septic complications after hip replacement surgery. Transfusion 1995;35(2):150 – 6. [4] Tartter PI. Blood transfusion and infectious complications following colorectal cancer surgery. Br J Surg 1988;75(8):789 – 92. [5] Blumberg N, Heal JM. Immunomodulation by blood transfusion: an evolving scientific and clinical challenge. Am J Med 1996;101(3):299 – 308. [6] Blajchman MA, Dzik S, Vamvakas EC, Sweeney J, Snyder EL. Clinical and molecular basis of transfusion-induced immunomodulation: summary of the proceedings of a state-of-the-art conference. Transfus Med Rev 2001;15(2):108 – 35. [7] Agarwal N, Murphy JG, Cayten CG, Stahl WM. Blood transfusion increases the risk of infection after trauma. Arch Surg 1993;128(2):171 – 6 [discussion 176 – 7].
186
N.S. Ward, M.M. Levy / Crit Care Clin 20 (2004) 179–186
[8] Blumberg N, Heal JM. Mortality risks, costs, and decision making in transfusion medicine. Am J Clin Pathol 2000;114(6):934 – 7. [9] Fitzgerald RD, Martin CM, Dietz GE, Doig GS, Potter RF, Sibbald WJ. Transfusing red blood cells stored in citrate phosphate dextrose adenine-1 for 28 days fails to improve tissue oxygenation in rats. Crit Care Med 1997;25(5):726 – 32. [10] Lorente JA, Landin L, De Pablo R, Renes E, Rodriguez-Diaz R, Liste D. Effects of blood transfusion on oxygen transport variables in severe sepsis. Crit Care Med 1993;21(9):1312 – 8. [11] Marik PE, Sibbald WJ. Effect of stored-blood transfusion on oxygen delivery in patients with sepsis. JAMA 1993;269(23):3024 – 9. [12] Shah DM, Gottlieb ME, Rahm RL, Stratton HH, Barie PS, Paloski WH, et al. Failure of red blood cell transfusion to increase oxygen transport or mixed venous PO2 in injured patients. J Trauma 1982;22(9):741 – 6. [13] Corwin HL, Parsonnet KC, Gettinger A. RBC transfusion in the ICU. Is there a reason? Chest 1995;108(3):767 – 71. [14] Smoller BR, Kruskall MS. Phlebotomy for diagnostic laboratory tests in adults. Pattern of use and effect on transfusion requirements. N Engl J Med 1986;314(19):1233 – 5. [15] He´bert PC, Wells G, Martin C, Tweeddale M, Marshall J, Blajchman M, et al. A Canadian survey of transfusion practices in critically ill patients. Transfusion Requirements in Critical Care Investigators and the Canadian Critical Care Trials Group. Crit Care Med 1998;26(3):482 – 7. [16] He´bert PC, Wells G, Blajchman MA, Marshall J, Martin C, Pagliarello G, et al. A multicenter, randomized, controlled clinical trial of transfusion requirements in critical care. Transfusion Requirements in Critical Care Investigators, Canadian Critical Care Trials Group. N Engl J Med 1999;340(6):409 – 17. [17] Vincent JL, Baron JF, Reinhart K, Gattinoni L, Thijs L, Webb A, et al. Anemia and blood transfusion in critically ill patients. JAMA 2002;288(12):1499 – 507. [18] Cordts PR, LaMorte WW, Fisher JB, DelGuercio C, Niehoff J, Pivacek LE, et al. Poor predictive value of hematocrit and hemodynamic parameters for erythrocyte deficits after extensive elective vascular operations. Surg Gynecol Obstet 1992;175(3):243 – 8. [19] Jones JG, Holland BM, Hudson IR, Wardrop CA. Total circulating red cells versus haematocrit as the primary descriptor of oxygen transport by the blood. Br J Haematol 1990;76(2):288 – 94. [20] Babineau TJ, Dzik WH, Borlase BC, Baxter JK, Bistrian BR, Benotti PN. Reevaluation of current transfusion practices in patients in surgical intensive care units. Am J Surg 1992; 164(1):22 – 5. [21] Chelemer SB, Prato BS, Cox Jr PM, O’Connor GT, Morton JR. Association of bacterial infection and red blood cell transfusion after coronary artery bypass surgery. Ann Thorac Surg 2002; 73(1):138 – 42. [22] Vamvakas E, Moore SB. Perioperative blood transfusion and colorectal cancer recurrence: a qualitative statistical overview and meta-analysis. Transfusion 1993;33(9):754 – 65. [23] Vamvakas EC. Transfusion-associated cancer recurrence and postoperative infection: meta-analysis of randomized, controlled clinical trials. Transfusion 1996;36(2):175 – 86. [24] Tartter PI. The association of perioperative blood transfusion with colorectal cancer recurrence. Ann Surg 1992;216(6):633 – 8. [25] Blumberg N, Kirkley SA, Heal JM. A cost analysis of autologous and allogeneic transfusions in hip replacement surgery. Am J Surg 1996;171(3):324 – 30. [26] Vamvakas EC. The cost-effectiveness of autologous transfusion revisited: implications of an increased risk of bacterial infection with allogeneic transfusion. Transfusion 2000;40(3):384 – 7. [27] Vamvakas EC, Carven JH. RBC transfusion and postoperative length of stay in the hospital or the intensive care unit among patients undergoing coronary artery bypass graft surgery: the effects of confounding factors. Transfusion 2000;40(7):832 – 9. [28] CRIT: The CRIT study: Anemia and blood transfusion in the critically ill—current clinical practice in the United States. Crit Care Med 2004;32(1):290 – 1.
Blood transfusion practice today Nicholas S. Ward, MD, Mitchell M. Levy, MD* Rhode Island Hospital, Brown Medical School, 593 Eddy Street, Providence, RI 02903, USA
In the last decade or so, the practice of packed red blood cell (PRBC) transfusions in critically ill patients has become the subject of many investigations. Once regarded as a relatively safe and effective means of increasing oxygen delivery, it has become clear that there are several aspects of blood transfusion that are not understood completely. Many investigators and clinicians feel that more scrutiny of this practice is warranted given an increased awareness of bloodborne pathogens, evidence that allogeneic PRBCs may lower susceptibility to infection [1– 7], and the ever increasing consumption of blood products by a growing critical care population. Additionally, studies on the pathogenesis and treatment of septic shock have shown conflicting data on the benefits of augmenting oxygen delivery with allogeneic PRBCs [8 –12]. In the last several years, many important studies have been published that not only cast doubt on the benefits of PRBC transfusion but also have sought to redefine the optimal threshold value of hemoglobin (HGB) concentration that warrants transfusion. Recently, there have been two large descriptive studies that sought to describe the current practice of PRBC transfusion in intensive care units (ICUs) and correlate that information with clinical outcomes. This article reviews these studies of current transfusion practices and tries to use this information as a guide to effective transfusion practice.
Significant studies One of the first descriptive studies of ICU transfusion practice was published by Corwin et al in 1995 [13]. In this retrospective chart review of 142 ICU patients who had a length of stay greater than 1 week, 85% had received blood transfusions. They found that only 35% of the transfusion events were for acute blood loss, and 29% had no identifiable trigger. With the exception of a low hematocrit as a trigger, the pretransfusion hematocrit was generally 27%. The
* Corresponding author. E-mail address: [email protected] (M.M. Levy). 0749-0704/04/$ – see front matter D 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.ccc.2003.12.004
180
N.S. Ward, M.M. Levy / Crit Care Clin 20 (2004) 179–186
authors went on to note that while this patient population represented only 4% of inpatient days, they received 16% of all inpatient transfusions. Like others [14], they noted a correlation between blood phlebotomized and blood transfused. In their study, phlebotomy was estimated to be equal to one third of the blood transfused. Further insight into ICU transfusion practices occurred in 1998 when He´bert et al published the results of a survey of Canadian critical care physicians [15]. The authors used a scenario-based survey with changing variables to assess the usual practices of ICU physicians and what clinical factors would modify their practices. Their data collected from the responses of 193 physicians showed a wide degree of variability in transfusion practices, with most of the transfusion thresholds ranging from 8.0 to 10.0 g/dL hemoglobin (Hbg). Overall, 35% of physicians selected a transfusion threshold of 9.0 g/dL Hbg, and 40% selected a threshold of 10.0 g/dL Hbg. Once they decided to transfuse, more than 90% of physicians administered 2 units of PRBC. In 1999, a landmark study that attempted to clarify what is an appropriate transfusion level of HGB for patients in ICUs was published. He´bert and the Canadian Clinical Trials Group conducted a randomized controlled trial of patients in ICUs who were transfused at either 7.0 g/dL or 10.0 g/dL [16]. Using the primary endpoint of 30-day mortality of all causes and several other secondary endpoints, the authors showed that using the lower transfusion threshold was at least as safe as the higher value. In several subgroups, such as patients younger than 55 years and patients with lower severity scores (Acute Physiology and Chronic Health Evaluation [APACHE] II less than 20), they showed a statistically significant mortality benefit in the lower transfusion trigger group. Other mortality time points such as in-hospital mortality and 60-day mortality also showed a statistically significant benefit to the lower threshold. In addition to these clinical outcomes data, the blood use was dramatically different in the two groups. The low transfusion threshold group received about half as many units of blood (average of 2.6 units versus 5.6 units per patient). Furthermore, 30% of the low threshold group received no blood transfusions at all during the study, compared with 0% without a transfusion in the higher threshold group. The Anemia and Blood Transfusion in Critically Ill Patients trial Two large multi-center studies have been performed in the years since the publication of the randomized controlled trial of He´bert et al that have taken our understanding of current transfusion practice even further. The first study was published in 2002 by Vincent et al and titled the ABC study (Anemia and Blood Transfusion in Critically Ill Patients) [17]. The stated objectives of this study were to define the incidence of anemia in critically ill patients, describe transfusion practices, and to explore possible risks and benefits associated with transfusion. Data were collected at 146 ICUs in 15 countries throughout Europe and encompassed 3534 patients. Most of the ICUs (71%) had mixed medical and
N.S. Ward, M.M. Levy / Crit Care Clin 20 (2004) 179–186
181
surgical patients. There was a mix of institution sizes, with 16.5% of ICUs from community hospitals, 34% from city or regional centers, and 49% from academic medical centers. The investigators collected data on demographic information, medical procedures, illnesses, and severity scores on all patients. They then followed the patients for 28 days (or discharge), collecting data on HGB levels, transfusions, reasons for transfusion, and mortality. In addition to these, a substudy done the week before the main study recorded the frequency and amount of blood drawn from ICU patients. Many important points were demonstrated by the data from this study. First, the substudy on phlebotomy showed that blood withdrawal was considerable, averaging 41 mL per day. The rest of the data gives a very good picture of the average transfusion requirements and practices in these ICUs. Overall, the transfusion rate for the whole 28-day period (including post-ICU time) was 41% of 3534 patients included. The average HGB level was 11.3 at admission, and the average pretransfusion HGB was 8.4 g/dL. There was no difference in the average pretransfusion hemoglobin for patients who were thought to be actively bleeding (8.4 g/dL) and those who were not (8.5 g/dL). This is surprising, in that physicians generally are more likely to transfuse a patient at a higher HGB level if they know of active bleeding. One factor that may explain this was that patients transfused for the indication of coronary artery disease alone had a higher than average transfusion threshold of 8.7 g/dL. The authors then looked at the relationship between blood transfusions and other patient characteristics. The data showed a positive correlation between receiving any blood transfusions and 28-day mortality. The mortality risk increased with the number of units transfused. The authors then divided the patients by organ dysfunction scores and showed an increase in mortality in the transfused versus nontransfused patients in all but the most severely ill patients. Finally, a logistic regression model was used to generate propensity scores for two groups of matched patients in an effort to adjust for characteristics other than transfusion. The data, which compared 516 transfused versus 516 nontransfused patients matched for patient characteristics, showed a significantly higher mortality in transfused patients (22.7% versus 17.1%; P = 0.02) (Fig. 1).
The CRIT trial A recent multi-center study called the CRIT study that sought to quantify the transfusion practices of United States ICUs and describe associated clinical outcomes and complications was completed [28]. Patients who were admitted to any medical, surgical, or mixed ICU were enrolled and followed for 30 days or until death or discharge from the hospital. The study took place at 284 ICUs in 213 hospitals. The ICUs were 31% medical, 29% surgical, and 40% mixed. Residents were present in 76% of the ICUs, and 71% were open ICUs. Nineteen percent of the ICUs had an institutional transfusion protocol.
182
N.S. Ward, M.M. Levy / Crit Care Clin 20 (2004) 179–186
Fig. 1. Increase in mortality attributable to blood transfusion in ICU patients. (From Vincent JL, Baron JF, Reinhart K, Gattinoni L, Thijs L, Webb A, et al. Anemia and blood transfusion in critically ill patients. JAMA 2002;288(12):1499 – 507; with permission.)
The study included 4892 patients with an average age of 60 years. Overall, ICU and in-hospital mortality rates were 13% and 18%, respectively. The baseline HGB values recorded were similar to the European ICUs, with the average HGB at entry of 11.0 g/dL and approximately two thirds of patients with HGB less than 12 g/dL. Like previous studies, the average hemoglobin declined steadily over time. Overall, 44.1% of patients were transfused while in the ICU, and 48.2% were transfused during the total 30-day observation period. The average pretransfusion HGB was 8.6 g/dL. Subgroup analysis showed a higher threshold for surgical (8.8 g/dL) than for medical (8.2 g/dL) patients. All patients had severity scores calculated (APACHE II and sequential organ failure assessment [SOFA]), and there was a positive correlation between severity of illness and the number of units transfused. The average age of the blood transfused in this study was 21.3 plus or minus 11.4 days, but the authors reported no significant difference in clinical outcomes that correlated with the median age of the blood transfused to a given patient. Like previous studies, the authors looked for an association of number of units transfused and clinical outcomes. Multivariate analysis showed that the number of RBC units transfused was associated with increased ICU and hospital length of stay (LOS) compared with nontransfused patients. Furthermore, RBC transfusion also was associated independently with higher mortality rates. Like the ABC trial, the investigators used adjusted case matching analysis to look for any additional mortality risk conveyed by the transfusion of RBCs independently. The investigators matched 1059 transfused patients to 1059 nontransfused patients and adjusted for the propensity for receiving a blood transfusion. Blood transfusion remained statistically significantly associated with an increased risk for death (adjusted mortality ratio, 1.65; 95% CI, 1.35 –2.03; log-rank P < 0.001).
N.S. Ward, M.M. Levy / Crit Care Clin 20 (2004) 179–186
183
Table 1 Transfusion practice in North America and Europe
ABC trial CRIT trial
Percent transfused
Average admission HGB
Average pre-transfusion HGB
In ICU
ICU + post-ICU
8.4 8.6
11.3 11.0
37.0 44.1
41.6 48.2
Common themes These studies go a long way to illustrating the current extent of ICU anemia and transfusion therapy. Notable similarities in the major parameters between the two studies are summarized in Table 1. In addition to these variables, both studies showed a positive correlation between number of units transfused and poor outcomes. Also, risk-adjusted models from both patient data sets suggested there might be an increased risk of mortality just from blood transfusion. Finally, it should be noted that the pretransfusion thresholds of these two studies of 8.4 and 8.6 are not very different than the threshold of 9.0 seen in the study by Corwin et al in 1994, before the results of the large randomized controlled trial that supported lower transfusion thresholds. What is good transfusion practice? Although these studies are very comprehensive, they are only observational studies and therefore do not assist clinicians greatly in deciding when to transfuse. Nevertheless, some important lessons can be learned from these studies. The first is that anemia is extremely common in critically ill patients and therefore is almost to be expected. In all likelihood, much of the laboratory anemia seen is actually a reflection of fluid shifts that result from hydration of patients, so common in the ICU. This phenomenon has been demonstrated in two previous studies that showed hematocrit is an inaccurate measurement of red cell volume when there are fluid shifts [18,19]. The second lesson is that very often the only reason for giving a blood transfusion to patients is that they have a low HGB value. One must assume that these transfusions are given with the hope of improving oxygen delivery in some way. Unfortunately, there is much evidence to the contrary of this fact. There have been several studies that have shown that RBC transfusions in critically ill or septic patients may not, in fact, improve oxygen delivery [10 – 12,20]. A study done by Marik et al seemed to indicate that the failure of stored RBCs to improve oxygen delivery may be related to their duration of storage [11]. Further confirmation of this phenomenon came from a later study by Fitzgerald et al in 1997, which showed that blood transfused shortly after removal improved oxygenation, while blood stored 28 days did not [9]. The third lesson is that allogeneic blood transfusion may be injurious to patients. The associations found in these studies between more units transfused
184
N.S. Ward, M.M. Levy / Crit Care Clin 20 (2004) 179–186
Fig. 2. Percentage of patients developing a postoperative bacterial infection after coronary bypass surgery. (From Chelemer SB, Prato BS, Cox Jr PM, O’Connor GT, Morton JR. Association of bacterial infection and red blood cell transfusion after coronary artery bypass surgery. Ann Thorac Surg 2002;73(1):138 – 42; with permission.)
and evidence of worse outcome is not unexpected. Over the years, there has been increasing data that seem to indicate that there are significant harmful effects of blood transfusions. For example, many studies have found correlations between the number of red cell transfusions and postoperative infections [2– 4, 7,21]. In a recent prospective cohort study of 533 coronary artery bypass patients, the authors found infection rates of 4.8% for no transfusions, 15.2% for 1 to 2 units, 22.1% for 3 to 5 units, and 29.0% with greater than or equal to 6 units
Fig. 3. Incremental increase in hospital charges associated with increasing number of allogeneic red blood cell transfusions (From Blumberg N, Kirkley SA, Heal JM. Cost analysis of autologous and allogeneic transfusions in hip replacement surgery. Am J Surg 1996;171(3):324 – 30; with permission).
N.S. Ward, M.M. Levy / Crit Care Clin 20 (2004) 179–186
185
(Fig. 2). Other data published in recent years have shown an association between the number of allogeneic red cell units transfused and increased risk of cancer recurrence [22 –24]. In addition to these data suggesting possible harmful qualities of allogeneic RBC transfusions, there are further data showing that hospital costs rise with the number of transfusions [25 –27]. A recent paper by Blumberg et al showed that there was significant difference in incremental hospital costs of $1000 to $1500 per unit transfused when comparing patients who got the same number of autologous versus allogeneic units (Fig. 3) [25].
Summary In trying to develop a cohesive and logical transfusion strategy, one must try to keep all of the previously mentioned information in mind. While recognizing that RBC transfusions are an integral part of care for critically ill patients, one also must recognize that there are large amounts of data that associate multiple transfusions of allogeneic blood with increased risk of morbidity and mortality. Some data exist showing causation and association, such as studies demonstrating the immunomodulatory effects of allogeneic transfusions. In addition, the blood supply is a limited resource that should not be used indiscriminately. It therefore behooves clinicians to subject each unit of blood transfused to scrutiny. In all likelihood, patients who are not actively bleeding and who are not hypovolemic probably get little to no benefit from allogeneic blood transfusions while their hemoglobin is greater than 7g/dL. They do, however, get needless exposure to a potentially toxic substance. Although it is the authors’ hope that more research will be performed to clarify the risks and benefits of blood transfusion, the authors also hope that knowledge of already published studies will continue to spread and replace the unfounded practices of the past.
References [1] Vamvakas EC, Blajchman MA. Deleterious clinical effects of transfusion-associated immunomodulation: fact or fiction? Blood 2001;97(5):1180 – 95. [2] Vamvakas EC, Carven JH, Hibberd PL. Blood transfusion and infection after colorectal cancer surgery. Transfusion 1996;36:1000 – 8. [3] Vamvakas EC, Moore SB, Cabanela M. Blood transfusion and septic complications after hip replacement surgery. Transfusion 1995;35(2):150 – 6. [4] Tartter PI. Blood transfusion and infectious complications following colorectal cancer surgery. Br J Surg 1988;75(8):789 – 92. [5] Blumberg N, Heal JM. Immunomodulation by blood transfusion: an evolving scientific and clinical challenge. Am J Med 1996;101(3):299 – 308. [6] Blajchman MA, Dzik S, Vamvakas EC, Sweeney J, Snyder EL. Clinical and molecular basis of transfusion-induced immunomodulation: summary of the proceedings of a state-of-the-art conference. Transfus Med Rev 2001;15(2):108 – 35. [7] Agarwal N, Murphy JG, Cayten CG, Stahl WM. Blood transfusion increases the risk of infection after trauma. Arch Surg 1993;128(2):171 – 6 [discussion 176 – 7].
186
N.S. Ward, M.M. Levy / Crit Care Clin 20 (2004) 179–186
[8] Blumberg N, Heal JM. Mortality risks, costs, and decision making in transfusion medicine. Am J Clin Pathol 2000;114(6):934 – 7. [9] Fitzgerald RD, Martin CM, Dietz GE, Doig GS, Potter RF, Sibbald WJ. Transfusing red blood cells stored in citrate phosphate dextrose adenine-1 for 28 days fails to improve tissue oxygenation in rats. Crit Care Med 1997;25(5):726 – 32. [10] Lorente JA, Landin L, De Pablo R, Renes E, Rodriguez-Diaz R, Liste D. Effects of blood transfusion on oxygen transport variables in severe sepsis. Crit Care Med 1993;21(9):1312 – 8. [11] Marik PE, Sibbald WJ. Effect of stored-blood transfusion on oxygen delivery in patients with sepsis. JAMA 1993;269(23):3024 – 9. [12] Shah DM, Gottlieb ME, Rahm RL, Stratton HH, Barie PS, Paloski WH, et al. Failure of red blood cell transfusion to increase oxygen transport or mixed venous PO2 in injured patients. J Trauma 1982;22(9):741 – 6. [13] Corwin HL, Parsonnet KC, Gettinger A. RBC transfusion in the ICU. Is there a reason? Chest 1995;108(3):767 – 71. [14] Smoller BR, Kruskall MS. Phlebotomy for diagnostic laboratory tests in adults. Pattern of use and effect on transfusion requirements. N Engl J Med 1986;314(19):1233 – 5. [15] He´bert PC, Wells G, Martin C, Tweeddale M, Marshall J, Blajchman M, et al. A Canadian survey of transfusion practices in critically ill patients. Transfusion Requirements in Critical Care Investigators and the Canadian Critical Care Trials Group. Crit Care Med 1998;26(3):482 – 7. [16] He´bert PC, Wells G, Blajchman MA, Marshall J, Martin C, Pagliarello G, et al. A multicenter, randomized, controlled clinical trial of transfusion requirements in critical care. Transfusion Requirements in Critical Care Investigators, Canadian Critical Care Trials Group. N Engl J Med 1999;340(6):409 – 17. [17] Vincent JL, Baron JF, Reinhart K, Gattinoni L, Thijs L, Webb A, et al. Anemia and blood transfusion in critically ill patients. JAMA 2002;288(12):1499 – 507. [18] Cordts PR, LaMorte WW, Fisher JB, DelGuercio C, Niehoff J, Pivacek LE, et al. Poor predictive value of hematocrit and hemodynamic parameters for erythrocyte deficits after extensive elective vascular operations. Surg Gynecol Obstet 1992;175(3):243 – 8. [19] Jones JG, Holland BM, Hudson IR, Wardrop CA. Total circulating red cells versus haematocrit as the primary descriptor of oxygen transport by the blood. Br J Haematol 1990;76(2):288 – 94. [20] Babineau TJ, Dzik WH, Borlase BC, Baxter JK, Bistrian BR, Benotti PN. Reevaluation of current transfusion practices in patients in surgical intensive care units. Am J Surg 1992; 164(1):22 – 5. [21] Chelemer SB, Prato BS, Cox Jr PM, O’Connor GT, Morton JR. Association of bacterial infection and red blood cell transfusion after coronary artery bypass surgery. Ann Thorac Surg 2002; 73(1):138 – 42. [22] Vamvakas E, Moore SB. Perioperative blood transfusion and colorectal cancer recurrence: a qualitative statistical overview and meta-analysis. Transfusion 1993;33(9):754 – 65. [23] Vamvakas EC. Transfusion-associated cancer recurrence and postoperative infection: meta-analysis of randomized, controlled clinical trials. Transfusion 1996;36(2):175 – 86. [24] Tartter PI. The association of perioperative blood transfusion with colorectal cancer recurrence. Ann Surg 1992;216(6):633 – 8. [25] Blumberg N, Kirkley SA, Heal JM. A cost analysis of autologous and allogeneic transfusions in hip replacement surgery. Am J Surg 1996;171(3):324 – 30. [26] Vamvakas EC. The cost-effectiveness of autologous transfusion revisited: implications of an increased risk of bacterial infection with allogeneic transfusion. Transfusion 2000;40(3):384 – 7. [27] Vamvakas EC, Carven JH. RBC transfusion and postoperative length of stay in the hospital or the intensive care unit among patients undergoing coronary artery bypass graft surgery: the effects of confounding factors. Transfusion 2000;40(7):832 – 9. [28] CRIT: The CRIT study: Anemia and blood transfusion in the critically ill—current clinical practice in the United States. Crit Care Med 2004;32(1):290 – 1.