Anemia in the ICU (Review)
Anemia in the ICU: Review and Update (More PulmCCM Topic Updates)
This document will be updated periodically to incorporate new research findings. To suggest an article for inclusion, please comment below.
Anemia is almost inevitable during many episodes of critical illness, and has been associated with worsened outcomes and prognosis. However, the preponderance of evidence suggests that correcting anemia (by transfusion or with erythropoetin) does not help most critically ill patients. Although a restrictive transfusion threshold of hemoglobin 7 g/dL is appropriate for most ICU patients, controversy remains as to whether certain patients might benefit from transfusion to higher hemoglobin levels, such as patients with acute coronary syndromes, GI bleeds, or who are unable to wean from mechanical ventilation.
Anemia is Associated with Worsened Outcomes
>95% of patients in the ICU are anemic after their first week in the ICU. The healthy replacement rate for red blood cells is 15-20 mL/day, but daily phlebotomy in the ICU steals 40-70 mL (of which only about 1 mL is actually tested inside the machines); this modern-day "bloodletting" may be responsible for a third of blood transfusions. Critical illness itself mediates most of the fall in red blood cell mass, through a variety of incompletely understood mechanisms. A variety of observational studies have established an association between anemia and worsened outcomes, including mortality, failure to wean from mechanical ventilation, and myocardial infarction. However, no study has clearly shown that anemia is directly causative of poor outcomes, or if it is merely a marker for more severe disease; eliminating confounding from such studies is difficult or impossible.
RBC Transfusion in the ICU
More than one-third of critically ill patients receive blood transfusions, rising to >70% who are in the ICU for one week or longer, owing mainly to tradition and outdated practice patterns. There are harms associated with liberal red blood cell transfusion. Generally speaking, observed adverse events are rare and/or self limited, including infections, transfusion-related acute lung injury (TRALI) or transfusion related immunomodulation (TRIM), and transfusion-related circulatory overload (TACO). These harms may partially account for the observed association between transfusion and worsened outcomes (pooled odds ratios: mortality 1.7, ARDS 2.5, nosocomial infection 1.8 for transfused patients). A large 2014 meta-analysis provided convincing evidence that liberal blood transfusion (to >8 g/dL) results in nosocomial infections, with a number needed to harm of 20-40 patients transfused liberally to cause a hospital-acquired infection.
Age of Transfused Blood: Does it Affect Outcomes?
Blood transfusion's harms are hypothesized to result from changes in blood that occur during storage. Most U.S. blood is stored for 16-21 days, and must be discarded at 42 days, per FDA rules. During storage, concentrations of ATP and 2,3-DPG fall; pro-inflammatory cytokines rise, and cell membranes break down, possibly leading to vasoconstriction and immunosuppression. The studies performed to date on whether the age of transfused blood carries a higher risk of adverse outcomes have been too heterogeneous, and their findings too discordant, to allow for any conclusions. There are 2 large randomized trials underway studying the question of the age of stored blood's effects on outcomes (ABLE, in Canadian ICUs and RECESS, in U.S. cardiac patients).
Transfusion Thresholds or "Triggers" in the ICU: TRICC Trial
The TRICC trial is the only randomized trial testing transfusion thresholds in critically ill adults. Among 838 euvolemic patients with normal baseline hemoglobin and no active cardiac ischemia or bleeding, there was no difference in all-cause 30-day mortality between patients transfused to a restrictive threshold or trigger of hemoglobin 7 g/dL, as compared to those transfused liberally to 10 g/dL. In post hoc subgroup analyses, the youngest (<55 years) and least ill (APACHE < 20) patients receiving restrictive transfusions (who received much less blood) had decreased mortality compared to age- and illness-matched patients transfused liberally.
Controversies in Transfusion Triggers
TRICC was one of the few landmark, practice-changing randomized trials in critical care, and a restrictive transfusion strategy is believed by many to be the standard of care for most patients in the ICU (without bleeding or myocardial ischemia). Notwithstanding its clearly positive results, controversy continues as to whether subgroups of critically ill patients exist for whom a more liberal transfusion strategy might be beneficial. Authors suggest these might include:
patients with cardiovascular disease,
advanced age,
those dependent on mechanical ventilation after days, and
those in the early stages of septic shock.
Cardiovascular Disease: Liberal or Restrictive Transfusion? Among the 257 patients with coronary artery disease in TRICC, those receiving a restrictive transfusion strategy had higher mortality of 26% vs. 21%: not statistically significant, but enough to raise concern about extending a restrictive transfusion strategy to all patients with ischemic heart disease. Two randomized trials have further supported the safety of a restrictive transfusion strategy: one in patients undergoing cardiac surgery with a transfusion threshold of hemoglobin ~8 g/dL (hematocrit 24%); the other among elderly patients with cardiovascular disease (or risk factors for same) undergoing hip fracture repair, transfused to a restrictive hemoglobin of 8 g/dL. In both, the restrictive transfusion strategy was noninferior to a transfusion threshold of hemoglobin 10 g/dL. Four observational studies among patients with acute coronary syndromes (1, 2, 3, 4) suggest that transfusions are associated with worsened short-term mortality (comparing patients with the same hemoglobin, transfused or not). A highly cited observational study (by Wu NEJM 2001) suggested a hemoglobin of 11 was ideal for elderly patients with myocardial infarction. However, it is impossible to exclude confounding in these studies.
Extubation Failure and Anemia: Could Transfusion Help? Anemia is independently associated with respiratory failure after extubation, and one case series (highly prone to selection and publication bias) suggested transfusion could help anemic vent-dependent COPD patients escape the ventilator. The only significant body of randomized data -- TRICC -- did not show that a liberal strategy reduced extubation failure in the mechanically ventilated patients, but had power only to detect a 25% difference between groups for this parameter. This remains an unsettled question with advocates on either side.
Early Sepsis: Is Transfusion to Hb 10 Helpful or Harmful as a "Goal?" A single-center trial of 263 patients with severe sepsis (Rivers NEJM 2001) included in its bundle the goal of transfusion to hemoglobin of 10 g/dL for patients with low central or mixed venous oxygen saturation. This severe sepsis "early goal directed therapy" bundle became standard of care due to inappropriate enthusiasm and inadequate vetting. A large randomized multicenter trial (ProCESS) refuted the need for blood transfusion as a necessary part of treatment of sepsis or septic shock.
Preventing Anemia in the ICU
Iatrogenic blood loss in critically ill patients should be minimized by:
Using small-volume (2 mL) phlebotomy tubes;
Using a closed-system arterial catheter -- standard arterial catheters result in 44% more blood being taken from the patient thanks to convenience and frequent discarding of flushed blood.
Avoiding bleeding the patient with daily blood draws, which are usually unnecessary and are advised against by Choosing Wisely guidelines for critical care. I call this "don't ask, don't tell, don't need to transfuse later").
Erythropoeitin, blood substitutes, and intravenous iron supplementation are all quite controversial, with risks and costs that potentially far outweigh any possible benefits, and are therefore not recommended for the vast majority of ICU patients.
This document will be updated periodically to incorporate new research findings. To suggest an article for inclusion, please comment below.
Prakash D. Crit Care Clin. 2012 Jul;28(3):333-43, v. doi: 10.1016/j.ccc.2012.04.012.
Hayden SJ et al. Am J Respir Crit Care Med. 2012 May 15;185(10):1049-57. doi: 10.1164/rccm.201110-1915CI. Epub 2012 Jan 26.
Shander A. Crit Care Clin. 2004 Apr;20(2):159-78. Fowler RA et al. Crit Care Clin. 2004 Apr;20(2):313-24, x.