Anemia of Critical Illness

Definition of Anemia

Anemia may be defined as a reduction in circulating red cell mass with an associated decrease in the oxygen-carrying capacity of blood. In clinical practice, this is determined by measurement of the hemoglobin concentration; hematocrit or red cell count with hemoglobin concentration and hematocrit are used most commonly. In 1968 a World Health Organization report outlined criteria required for the diagnosis of anemia: a hemoglobin concentration of 140 g/L in adult males and 120 g/L for (nonpregnant) adult females. Despite being derived from very limited data, these definitions were promulgated widely and adopted by the medical profession. Their validity has been challenged with several alternate definitions derived from large population-based studies. Given the lack of consensus regarding the definition of anemia, it is not surprising that no definition or grades severity exists for anemia observed in critically ill patients. Developing a specific grading system for the critically ill is difficult because the underlying illness and the therapies provided influence red cell mass and plasma volume. In addition, despite the publication of randomized controlled trials of transfusion strategies and guidelines for transfusion , variation in practice inevitably will still exist. Such variation will influence epidemiologic studies of anemia in the critically ill.

Epidemiology

Numerous studies have demonstrated that anemia is a common problem in the intensive care unit (ICU). In 1995 Corwin et al. demonstrated that more than 90% of critically ill patients have a subnormal hemoglobin level within 72 hours after ICU admission. In any individual ICU, the case mix, severity of illness, and local transfusion practice influence the prevalence of anemia. Accordingly, the most robust data come from multicenter observational cohort studies. In the past decade five such studies have been published. These studies demonstrate the prevalence of anemia on admission to ICU and also the development of anemia during stay in the ICU. These two subsets of the epidemiology of anemia of critical illness are addressed separately in this chapter.

Two large observational studies conducted last decade provide the best evidence of the epidemiology of this condition. The Anemia and Blood Transfusion in Critical Care (ABC) study is an observational cohort study of 3534 patients admitted to 146 European ICUs. Consecutive patients admitted to the participating ICUs during a 2-week period in 1999 were enrolled. The researchers' objective was to define the incidence of anemia in this group of patients. The mean (SD) hemoglobin concentration on admission to ICU was 113 (2.3) g/L, with 63% having a hemoglobin level of less than 120 g/L and 29% less than 100 g/L. The study enrolled a broad cross-section of elective surgical (41.4%), medical (32.6%), emergency surgical (16.6%), and trauma patients (7.6%) with a mean Acute Physiology and Chronic Health Evaluation (APACHE) II score of 14.8 (7.9), mean sequential organ failure assessment (SOFA) score of 5.2 (3.8), and length of stay of 4.5 (6.7) days. In this study only 13% of patients had a recent history of anemia. Of all patients admitted with a hemoglobin concentration of less than 100 g/L (36.6%), only 50% had a history of either acute bleeding or documented cause for their anemia. A weak negative correlation between hemoglobin concentration on ICU admission, severity of illness, and ICU length of stay was observed.

In the United States the CRIT study enrolled 4892 patients from 284 ICUs during a 9-month period from August 2000 to April 2001. It is not reported what proportion of total admissions this represents, and the risk of selection bias and the potential for an overestimation of the prevalence of anemia exist. There were fewer elective postoperative admissions in this cohort of patients (20% vs. 41.4%) compared with the ABC study. The mean hemoglobin level on admission was similar—110 (2.4) g/L, with just less than two thirds having a hemoglobin concentration less than 120 g/L. In contrast to the ABC study, the patient cohort was sicker, with a median APACHE II score of 19.7 (8.2), SOFA score of 6.2 (3.7), and ICU length of stay of 7.4 (7.3) days. These differences are most likely explained by selection bias. Despite the CRIT study having a sicker patient cohort, the mean admission hemoglobin and the distribution are remarkably similar. This suggests that severity of illness may not be associated with a higher prevalence of anemia on admission to the ICU. Observational cohort studies reporting the hemoglobin concentration on admission also have been done in Scotland and Australia, where the median hemoglobin concentrations on admission were 105 g/L and 98 g/L, respectively.

Although differences in methodology limit the generalizability of these data, it appears that up to 25% of critically ill patients will have a hemoglobin concentration of 90 g/L or less at some time during their admission. This makes anemia one of the most frequently encountered clinical syndromes that intensive care clinicians must manage.

The ABC ¹ and CRIT studies also describe the development and severity of anemia during ICU stay. Although hemoglobin concentrations measured during ICU stay obviously are associated with transfusion, these vary between institutions and jurisdictions. The TRICC trial demonstrated the equivalence of restrictive and liberal transfusion strategies in a heterogeneous group of critically ill patients. Improved outcome in patients randomized to the restrictive strategy (transfusion trigger hemoglobin concentration 70 g/L) was demonstrated in a secondary endpoint (hospital mortality) and a subgroup of less severely ill patients (APACHE II score of less than 25). Observational studies suggest that this study has had a significant impact on transfusion practice ; the observed transfusion trigger is approximately 80 g/L. This is in contrast to clinicians' attitudes toward transfusion practice published before the TRICC study. Although there inevitably will be variation in transfusion practice, it seems that the TRICC study has improved uniformity among jurisdictions.

The CRIT study is the only observational study to report longitudinal hemoglobin concentration data of transfused compared with nontransfused patients ( Fig. 77.2 ). Fig. 77.2 shows the hemoglobin concentration levels from day 1 through day 30 in all patients, those transfused, and those not transfused. In the nontransfused group, the mean admission hemoglobin concentration was approximately 120 g/L. This decreased acutely to approximately 110 g/L after 3 to 4 days. After this, only a slow decline in mean hemoglobin in concentration is observed.

In a single-center observational study, Nguyen et al. also investigated the time course of hemoglobin concentration in critically ill patients who were not transfused. This study excluded patients with recent evidence of bleeding, the majority of postsurgical patients, those with either acute or chronic renal failure, and those with an ICU length of stay less than 24 hours. Follow-up was for 7 days. Despite the differences in methodology, their results are similar to those of the CRIT study: a rapid decrease in hemoglobin (about 50 g/L) for the first 3 days followed by a slow decline in hemoglobin concentration (12 g/L).

Anemia is therefore frequently present on admission to ICU, and if not present on admission develops rapidly within the first 72 hours in the majority of patients. Thereafter, a slow decline in hemoglobin concentration is observed. The available data also suggest that the majority of patients with an ICU length of stay greater than 7 days have a hemoglobin concentration less than 90 g/L at some time during their ICU stay.