| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Guest Access | Sign In via User Name/Password
|
|
|
|||||||
Guest Access | Sign In via User Name/Password |
|||||||||
Washington, DC
Dr. Shorr is affiliated with the Pulmonary, Critical Care, and Sleep Medicine Service, Department of Medicine and the Critical Care Medicine Service, Walter Reed Army Medical Center, and Dr. Jackson is affiliated with the Department of Surgery, Walter Reed Army Medical Center.
Correspondence to: Andrew F. Shorr, MD, MPH, Pulmonary, Critical Care, and Sleep Medicine, Walter Reed Army Medical Center, Washington, DC 20307; e-mail: afshorr{at}dnamail.com
Anemia is a common problem in the ICU. The vast majority of patients are anemic on admission to the ICU.12 Among the few patients who have normal hemoglobin levels at presentation, nearly all become anemic during the course of their ICU stay.12 The causes for anemia in critically ill patients are manifold. In some instances, anemia results from acute blood loss after trauma, GI hemorrhage, or surgery. For other individuals, anemia arises because of earlier treatment with chemotherapeutic agents or because of the patient’s chronic medical conditions. All patients, however, are exposed to the risk of frequent phlebotomy. Some estimates123 have suggested that we remove nearly 60 mL blood per day from those in the ICU.
Blunted erythropoiesis also contributes to the development of anemia in the ICU. For example, despite anemia and adequate iron stores, Rogiers and colleagues4 demonstrated that critically ill individuals fail to produce appropriate levels of erythropoietin (EPO). Although one would expect an inverse correlation between the hematocrit and EPO measurements, these investigators found no such relationship in a mixed ICU cohort.4 Furthermore, when compared to non-acutely ill individuals, critically ill subjects have a relatively limited reticulocytosis in response to their anemia.5 These observations have led to the development of the concept of "anemia of critical illness." Physiologically, this phenomenon appears to result from the inhibition of the gene regulating EPO production.5 Elevated levels of proinflammatory cytokines suppress not only the transcription of a key sensor of tissue hypoxemia, but also limit the endogenous production of EPO.6 These cytokines further contribute to anemia by directly hindering the production of RBCs in the bone marrow and altering iron metabolism.7
Irrespective of the cause of anemia, physicians must determine if and how to treat this problem. Despite the frequency of anemia in the ICU, no recent, formal guidelines exist to help the clinician. The recommendations made in older position statements89 suggest that transfusion is rarely indicated when the hemoglobin concentration is > 10 g/dL but may be indicated when the level falls below this threshold, especially if surgery is planned. However, such recommendations stem more from tradition rather than from the results of clinical trials. These guidelines also fail to acknowledge the evidence that certain patients tolerate severe anemia so long as they received adequate fluid resuscitation.10
This lack of strong data results in variable practice styles and certainly helps to explain the frequency of blood transfusion in the ICU. For example, a Canadian survey11 revealed that many intensivists perform transfusions at hemoglobin levels of approximately 9 g/dL. Moreover, this study noted a wide variability in practice. A more recent European prospective, observational study1 of 3,534 patients documented that anemia was common in critically ill patients, with nearly one third of subjects having hemoglobin concentrations of < 10 g/dL. Approximately 40% of the cohort received transfusions during the course of their ICU stay, and the mean pretransfusion hemoglobin level was 8.4 g/dL.1 A similar study12 in the United Kingdom confirmed these observations. In both reports,112 the most common reason for transfusion was "low hemoglobin." In other words, patients did not receive transfusions in response to acute hemorrhage or as part of resuscitation for shock. Rather, the intensivist determined that the degree of anemia posed a threat to the patient’s health and that this necessitated a response.
In this issue of CHEST (see page 928), Levy and colleagues describe the results of a subgroup analysis of US patients who were enrolled in the CRIT study. The CRIT study2 initially represented a large multicenter observational investigation into transfusion practices in the ICU. In their report, Levy et al focus on subjects requiring mechanical ventilation (MV). Appreciating process of care and outcomes in the mechanically ventilated patient is a crucial aspect of health services research in critical care, since MV disproportionately contributes to cost and length of stay in the ICU. MV also represents a technology that is unique to the ICU environment. Not surprisingly, Levy et al found that individuals requiring MV were more severely ill than those not requiring MV and more often received transfusions while in the ICU. Strikingly, though, MV patients accounted for > 75% of all the units of blood administered, and the typical MV subject received nearly 5 U packed RBCs during care in the ICU. One might expect that this disproportionate use of transfusion reflects the higher severity of illness among patients receiving MV. However, when queried as to the reason for transfusion, the most common reason for transfusion in those patients who were receiving MV was not hemodynamic instability but, rather, low hemoglobin. How low was the hemoglobin level that triggered transfusion, and the attendant costs and risks of transfusion? The mean (± SD) pretransfusion hemoglobin level was 8.4 ± 1.4 g/dL. Furthermore, 40% of the transfusions performed in the MV population were done after day 4 of an ICU stay (ie, during the extended phase of the patient’s illness).
These results from the CRIT investigators are particularly distressing, since the trial was conducted between August 2000 and April 2001. The findings from the Transfusion Requirements in Critical Care (TRICC) study were published in 1999.13 In that landmark trial, Hebert et al randomized critically ill patients either to a liberal transfusion strategy, with a hemoglobin level goal of 10 g/dL, or to a restrictive protocol, with a hemoglobin level goal of 7 g/dL. The study included > 800 persons, many of whom required MV. At 30 days post-study enrollment, the mortality rate was similar between the two arms of the study, indicating that a restrictive transfusion strategy was at least as safe as a liberal approach.13 More importantly, the results from this project suggested that the greater use of transfusion might actually result in harm to our patients. The hospital mortality rate was higher in those persons randomized to the liberal transfusion strategy arm (28.1% vs 22.2%, respectively; p = 0.05).13 In both younger patients (ie, those < 55 years) and in less severely ill subjects (ie, acute physiology and chronic health evaluation [APACHE] II score, < 20), mortality was significantly lower in the restrictive transfusion cohort.13 Because of continuing controversy regarding the "optimal hemoglobin" level that would facilitate liberation from MV, Hebert and coworkers14 separately analyzed outcomes in subjects requiring MV. Although not designed to expressly explore transfusion and MV, the study by Hebert et al14 reported that the higher hemoglobin goal did not result in shorter durations of MV. In short, nearly a year after the publication of a well-done clinical trial that provided important insight into the management of ICU patients, Levy et al reveal that few intensivists had modified their clinical practice. Failure to alter our treatment algorithms despite clinical studies focused on our patients only underscores the amount of work that remains to be done if critical care is to become an evidence-based specialty.
It is important to note that the TRICC trial13 only enrolled hemodynamically stable patients. The findings from the TRICC trial cannot and should not be applied to individuals undergoing short-term resuscitation. The goals and objectives of resuscitation are unique, and hence the decision to transfuse in this scenario must be based on how the patient responds to other interventions, including early, aggressive fluid resuscitation. Nonetheless, more liberal transfusion criteria during acute resuscitation cannot fully explain the significant transfusion rates documented by Levy et al. For example, the vast majority of the blood given to those needing MV was given after what many would consider to have been the acute phase of the subject’s illness.
The analysis by Levy et al has several limitations. First, they did not provide sufficient data regarding the physician’s reasoning as to why he/she chose to transfuse. "Low hemoglobin" as a potential choice on a list of options may not adequately capture more nuanced features in the decision-making process. Second, Levy et al did not describe the incidence of ongoing shock in the cohort or the number of persons who initially received therapy with vasopressors. They opted not to perform any multivariate analyses to explore the relationships among transfusion, MV, and other confounders such as severity of illness. The connection between greater lengths of stay in the ICU and more transfusions in the MV cohort may reflect only an association, and not necessarily a causative relationship.
Beyond health services research and descriptive analyses like that presented by Levy and colleagues, another area of active investigation in transfusion and intensive care medicine has been an effort to better understand the risks related to transfusion in the critically ill patient. Several research teams have noted that transfusion is immunomodulatory, and increases the production of proinflammatory cytokines and alters T-cell function both in vitro and in vivo.1516 Correspondingly, transfusion heightens the potential for nosocomial infection.1718 Similarly, there is now greater appreciation of the potential for transfusion-related acute lung injury (TRALI). TRALI represent a form of acute lung injury that is thought to arise from the activation of primed neutrophils and is the third most common cause of transfusion-related death.1920 Epidemiologic reviews1920 have suggested that subclinical forms of TRALI exist but that these cases may not be appreciated by the clinician. Hence the true incidence of the disease is unknown because of underreporting.
Thus, in the last 5 years the potential benefits of transfusion have come into question, while the impact of the relative dangers of transfusion have become more evident. Nonetheless, it remains unclear whether we have made any effort to reevaluate the balance of these risks and benefits in our practice.
It is important to note that issues of benefit and harm do not necessarily provide insight into what represents the "optimal" hemoglobin level. Needless to say, that will be a function of a patient’s clinical status at the time that one considers whether to perform the transfusion. As newer alternatives to transfusion become available, though, we will have to reexamine our transfusion goals. The opportunity to raise a patient’s hemoglobin level without exposing them to the risks of transfusion remains appealing. At present, multiple phase III trials2122 are underway investigating the role for both artificial hemoglobins and exogenous EPO. Although the early results of investigations employing these products have been encouraging, as clinicians we should demand that future studies be well designed and target important clinical end points such as survival, rates of nosocomial infection, and length of stay.2122 Showing that any new approach only alters some surrogate or laboratory marker without a commensurate improvement in outcomes should not prompt us to modify our practice. However, let us hope that if such trials are successful, we adopt those evidence-based interventions more rapidly than we have responded to the current data regarding the value of and role for transfusion.
Footnotes
The opinions expressed herein are not to be construed as official or as reflecting the policies of either the Department of the Army or the Department of Defense.
References
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
