RBC Transfusion Strategies in the ICU: A Concise Review

Abstract

Objectives:

To critically assess available high-level clinical studies regarding RBC transfusion strategies, with a focus on hemoglobin transfusion thresholds in the ICU.

Data Sources:

Source data were obtained from a PubMed literature review.

Study Selection:

English language studies addressing RBC transfusions in the ICU with a focus on the most recent relevant studies.

Data Extraction:

Relevant studies were reviewed and the following aspects of each study were identified, abstracted, and analyzed: study design, methods, results, and implications for critical care practice.

Data Synthesis:

Approximately 30–50% of ICU patients receive a transfusion during their hospitalization with anemia being the indication for 75% of transfusions. A significant body of clinical research evidence supports using a restrictive transfusion strategy (e.g., hemoglobin threshold < 7 g/dL) compared with a more liberal approach (e.g., hemoglobin threshold < 10 g/dL). A restrictive strategy (hemoglobin < 7 g/dL) is recommended in patients with sepsis and gastrointestinal bleeds. A slightly higher restrictive threshold is recommended in cardiac surgery (hemoglobin < 7.5 g/dL) and stable cardiovascular disease (hemoglobin < 8 g/dL). Although restrictive strategies are generally supported in hematologic malignancies, acute neurologic injury, and burns, more definitive studies are needed, including acute coronary syndrome. Massive transfusion protocols are the mainstay of treatment for hemorrhagic shock; however, the exact RBC to fresh frozen plasma ratio is still unclear. There are also emerging complimentary practices including nontransfusion strategies to avoid and treat anemia and the reemergence of whole blood transfusion.

Conclusions:

The current literature supports the use of restrictive transfusion strategies in the majority of critically ill populations. Continued studies of optimal transfusion strategies in various patient populations, coupled with the integration of novel complementary ICU practices, will continue to enhance our ability to treat critically ill patients.

RBC transfusions are an integral part of ICU treatment. Although it is often assumed that RBC transfusions will improve tissue oxygen delivery in adequately fluid resuscitated patients, questions still remain around the appropriate uses and goals of RBC transfusions in the critically ill (1).

ICU patients, like other patient populations, can experience transfusion-related reactions and other adverse effects of transfusions, and RBC transfusions are expensive (2–4). Diligent understanding and use of transfusions in the critically ill are essential. This review provides an overview of RBC transfusion epidemiology within the ICU, summarizes transfusion threshold evidence in specific populations, and looks at future directions regarding RBC transfusions in the critically ill.

EPIDEMIOLOGY OF RBC TRANSFUSIONS IN THE ICU

Approximately 30–50% of ICU patients receive a RBC transfusion during their hospitalization (5), receiving on average nearly five RBC units per ICU stay (6, 7). Anemia or low hemoglobin, as reported by the attending physician as the primary indication for transfusion, accounts for approximately 46–90% of RBC transfusions (2, 7, 8). Nearly 60% of patients will be anemic (hemoglobin < 12 g/dL) on ICU admission (7, 8), and 97% will be anemic (hemoglobin < 12 g/dL for women and < 14 g/dL for men) by ICU day 8 (5). Anemia in the ICU is commonly multifactorial with the most common etiology being anemia of chronic disease due to iron homeostasis dysregulation (9).

Historically, the transfusion threshold for anemia has varied widely and until recently was based on little clinical data. Adams and Lundy (10) introduced the “10/30” rule, referring to a hemoglobin threshold of 10 g/dL or a hematocrit of 30%, in 1942 to guide transfusions for surgical patients. Although this threshold was broadly accepted for many years, the benefit of more restrictive transfusion thresholds has been evaluated over the past 2 decades.

TRANSFUSION THRESHOLDS IN SPECIFIC POPULATIONS

General ICU

Mounting evidence supports restrictive transfusion strategies for general critically ill patients and specific ICU sub-populations (Table 1). Published in 1999, the landmark Transfusion Requirements in Critical Care (TRICC) trial found no difference in 30-day mortality for critically ill patients with euvolemia and a hemoglobin less than 9 g/dL within 72 hours after admission managed using a restrictive (hemoglobin < 7 g/dL) compared with liberal (hemoglobin < 10 g/dL) transfusion threshold (11). Consequently, evidence favoring restrictive transfusion strategies is being translated into clinical practice for over a decade, with varying degrees of uniformity across different hospital systems (12, 13). The 2018 Patient Blood Management International Consensus Conference recommends a hemoglobin transfusion threshold of 7 g/dL for critically ill but clinically stable ICU patients (14).

TABLE 1.

Key Studies of Restrictive Versus Liberal Transfusion Strategies by Patient Population

References	Study Design	Patient Population (Sample Size)	Transfusion Strategy, Hemoglobin		Primary Outcome and Results
			Restrictive	Liberal
General ICU
Hébert et al (11)	Multicenter RCT	ICU patients (n = 838)	7 g/dL	10 g/dL	No difference in overall 30-d mortality (18.7% restrictive vs 23.3 liberal; p = 0.11). Less acutely ill subgroup favored the restrictive group (p = 0.03)
Sepsis
Holst et al (15)	Multicenter RCT	Septic shock (n = 998)	7 g/dL	9 g/dL	No difference in 90-d mortality (43% restrictive vs 45% liberal; p = 0.44). The restrictive group received fewer RBC units than liberal group (median 1 vs 4 U, respectively)
ACS
Cooper et al (16)	Multicenter RCT	Acute MI (n = 45)	8 g/dL	10 g/dL	Better composite outcome (in-hospital death, recurrent MI, or new or worsening congestive heart failure) in restrictive (13%) vs liberal groups (38%; p = 0.046)
Carson et al (17)	Multicenter RCT	Acute MI or stable ACS undergoing cardiac catheterization (n = 110)	8 g/dL	10 g/dL	Trend of higher all-cause mortality, MI, or unscheduled coronary revascularization up to 30 d for restrictive strategy compared with liberal group (25.5% vs 10.9%, respectively; p = 0.054)
CVD
Docherty et al (18)	Meta-analysis	CVD (11 studies)	8 g/dL	10 g/dL	No difference in 30-d mortality (risk ratio, 1.15; 95% CI, 0.88–1.50). Restrictive group had increased risk of new ACS (risk ratio, 1.78; 95% CI, 1.18–2.70)
GIB
Villanueva et al (19)	Single-center RCT	Acute upper GIB (n = 921)	7 g/dL	9 g/dL	Lower all-cause mortality at 45 d in restrictive group (5%) compared with liberal group (9%; p = 0.02)
Hematologic malignancy
Estcourt et al (20)	Meta-analysis	Malignant hematologic disorder or hemopoietic stem cell transplant (six studies)	7–9 g/dL	8–12 g/dL	Low-quality evidence that restrictive strategy has little or no effect on all-cause mortality
DeZern et al (21)	Single-center RCT	Acute leukemia (n = 90)	7 g/dL	8 g/dL	Established feasibility for a larger trial. Restrictive group transfused on average 8 U RBCs vs the liberal group 11.7 U RBCs (p = 0.0003), no difference in bleeding events or neutropenic fever
Cardiac surgery
Hajjar et al (22)	Single-center RCT	Cardiac surgery with cardiopulmonary bypass (n = 502)	Hematocrit < 24%	Hematocrit < 30%	No difference in 30-d all-cause mortality and severe morbidity (11% restrictive vs 10% liberal; p = 0.85)
Murphy et al (23)	Multicenter RCT	Nonemergent postcardiac surgery (n = 2,003)	7.5 g/dL	9 g/dL	No difference in serious infection or ischemic event at 3 mo (35.1% restrictive vs 33.0% liberal; p = 0.30)
Mazer et al (24)	Multicenter RCT	Cardiac surgery (n = 5,243)	7.5 g/dL	9.5 g/dL (operating room or ICU) 8.5 g/dL (ward)	Restrictive strategy was noninferior to liberal strategy with regards to all-cause mortality, MI, stroke, new-onset renal failure requiring dialysis by hospital discharge or by day 28 (11.4% restrictive vs 12.5% liberal)
Acute neurologic injury
Robertson et al (25)	Multicenter RCT	Closed head injury (n = 200)	7 g/dLa	10 g/dL	No difference in dichotomized Glasgow Outcome Scale 6 mo post-injury (42.5% restrictive vs 33.0% liberal; p = 0.28). Lower thromboembolism frequency (secondary outcome) in restrictive (8.1%) vs liberal group (21.8%; p = 0.009)
McIntyre et al (26)	Multicenter RCT secondary analysis	Closed head injury subgroup of Transfusion Requirements in Critical Care (n = 67)	7 g/dL	9 g/dL	No difference in 30-d mortality (17% restrictive vs 13% liberal; p = 0.64)
Burns
Palmieri et al (27)	Multicenter RCT	Burn patients (≥ 20% total body surface area) (n = 345)	7 g/dL	10 g/dL	No difference in blood stream infection frequency, approximately 24% in both restrictive and liberal groups (p = 0.904). Fewer mean RBC units transfused in the restrictive (7 U) vs liberal group (15 U; p < 0.0001)

ACS = acute coronary syndrome, CVD = coronary vascular disease, GIB = gastrointestinal bleed, MI = myocardial infarction, RCT = randomized controlled trial.

^aErythropoietin or placebo.

Sepsis

Sepsis is a life-threatening organ dysfunction caused by a dysregulated host response to infection, often accompanied by decreased oxygen delivery and organ hypoperfusion (28). Historically, liberal transfusion thresholds (hematocrit < 30%) were included in early goal-directed therapy and recommended in early Surviving Sepsis guidelines (29, 30). Published in 2014, the Trauma and Injury Severity Score trial was the largest trial to randomize patients with septic shock to restrictive (7 g/dL) or liberal (9 g/dL) transfusion thresholds (15). Ninety-day mortality or ischemic events did not differ between study arms with the restrictive group receiving a median of three RBC units per patient less than the liberal group. The 2016 Surviving Sepsis Campaign guidelines have removed the higher hemoglobin targets for acute sepsis resuscitation and now recommend a hemoglobin threshold of less than 7.0 g/dL, targeting a hemoglobin of 7–9 g/dL (31).

Acute Coronary Syndrome

Decreased myocardial oxygen delivery is the major concern in acute coronary syndrome (ACS) due to impaired coronary artery blood flow (32, 33). The Conservative versus Liberal Red Cell Transfusion in Acute Myocardial Infarction (CRIT) trial, enrolled 45 patients to either a liberal or restrictive hemoglobin transfusion threshold, 8 g/dL or 10 g/dL, respectively (16). There was a lower composite frequency of in-hospital death, recurrent myocardial infarction (MI), or new or worsening congestive heart failure in the restrictive group (13%) compared with the liberal group (38%; p = 0.046). However, the subsequent Myocardial Ischemia and Transfusion Trial (MINT) pilot study of 110 patients showed an unexpected trend toward higher combined mortality and major cardiac events in the restrictive transfusion arm (seven deaths in the restrictive group and one death in the liberal group; p = 0.03) (17). The ongoing MINT trial aims to randomize 3,500 patients with ACS and a baseline hemoglobin less than 10 g/dL to a liberal (hemoglobin < 10 g/dL) or restrictive (hemoglobin < 8 g/dL) transfusion strategy (34). Current guidelines choose not to make specific recommendations for or against liberal or restrictive transfusion thresholds in patients with ACS (35) with more reliable evidence needed.

Stable Cardiovascular Disease

Patients with cardiovascular disease (CVD) have impaired compensatory mechanisms to adequately increase oxygen delivery in times of acute illness or anemia. Consequently, the appropriate and safe hemoglobin trigger for transfusion is critically important while balancing the risks with receiving a transfusion. In a recent meta-analysis of adult patients with CVD, excluding those who had undergone cardiac surgery, 30-day mortality did not differ between restrictive (hemoglobin < 8 g/dL) versus liberal (hemoglobin < 10 g/dL) transfusion strategies (18). There was a slightly higher risk of ACS in the restrictive transfusion group; however, the actual diagnosis of ACS was given a low Grading of Recommendations, Assessment, Development, and Evaluations quality of evidence. The authors ultimately recommended considering a transfusion threshold hemoglobin less than 8 g/dL in patients with CVD until more robust clinical trials can be performed. Current guidelines maintain higher transfusion thresholds (hemoglobin < 8 g/dL) for patients with CVD (14, 35).

Gastrointestinal Bleeds

In one large trial, patients with severe acute upper gastrointestinal bleeding (UGIB) were randomized to restrictive (hemoglobin < 7 g/dL) versus liberal (hemoglobin < 9 g/dL) transfusion strategies (19). Patients were randomized, and in both groups, one RBC unit was transfused initially. The hemoglobin level was assessed after the transfusion and used for subsequent transfusion decisions based on the assigned strategy (restrictive vs liberal). The restrictive strategy had significantly better 6-week survival, less rebleeding, no increase in portal-pressure gradient, and reduced need for additional transfusion. The restrictive versus liberal blood transfusion for acute upper gastrointestinal bleeding (TRIGGER) trial (36) was a pragmatic, cluster-randomized trial of 936 patients with acute UGIBs comparing hemoglobin transfusion thresholds of 8–10 g/dL. The restrictive policy group had fewer RBC units transfused; however, this was not significant. The 2012 American College of Gastroenterology (ACG) UGIB guidelines recommend a hemoglobin greater than or equal to 7 g/dL, with higher targets in patients with intravascular volume depletion or comorbidities such as coronary artery disease (37). Citing limited evidence for lower gastrointestinal bleeding (LGIB), the ACG made the same restrictive transfusion recommendations in 2016 LGIB guidelines, relying on the UGIB evidence (38). Similar transfusion targets (hemoglobin 7–9 g/dL) are recommended for UGIBs by the European Society of Gastrointestinal Endoscopy (39).

Hematologic Malignancies

RBC transfusions play an important supportive role in the management of hematologic malignancies. A recent Cochrane Review aimed to determine the efficacy and safety of restrictive (goal hemoglobin: 7–9 g/dL) versus liberal RBC transfusion strategies in patients with hematologic malignancies undergoing intensive chemotherapy and/or radiation (20). Four small studies were analyzed (three randomized controlled trials [RCTs], n = 156 and one nonrandomized trial, n = 84) and the findings, deemed low-quality evidence, found a restrictive transfusion strategy reduced the total number of RBCs per patient, but had little impact on mortality, bleeding, or hospital stay. A feasibility trial for a larger trial randomized 90 patients with acute leukemia compared a restrictive hemoglobin trigger (hemoglobin 7 g/dL) to a higher hemoglobin trigger (hemoglobin 8 g/dL) (21). Although the primary outcome of establishing feasibility for a larger trial was achieved, the restrictive group on average were transfused nearly four fewer units of RBCs compared with the higher group, with no significant difference in bleeding events or neutropenic fever. Ongoing trials are investigating transfusion strategies in hematologic malignancies, one in hematopoietic stem cell transplant patients and the other in hematologic ICUs (40, 41).

Cardiac Surgery

Cardiac surgery accounts for the largest proportion of surgically related transfused RBCs with more than 50% of patients receiving at least one RBC transfusion (42). The Transfusion Requirements After Cardiac Surgery (TRACS) study enrolled 502 patients undergoing cardiac surgery with cardiopulmonary bypass (22). Thirty-day mortality or severe morbidity did not differ between restrictive (hematocrit 24%) and liberal (hematocrit 30%) groups. Regardless of group assignment, the number of RBC units transfused increased mortality and hospital length of stay (43).

Contrasting results were reported in the Transfusion Indication Threshold Reduction (TITRe2) trial, where 2003 nonemergent cardiac surgery patients with postoperative hemoglobin less than 9 g/dL were randomized to restrictive (hemoglobin < 7.5 g/dL) or liberal (hemoglobin < 9 g/dL) strategies (23). No difference was seen in the primary composite outcome within 3 months; however, mortality was higher in the restrictive group (4.2%) compared with the liberal group (2.6%; p = 0.045).

The subsequent Transfusion Requirements in Cardiac Surgery (TRICS)–III trial (24) was the largest RCT of 5,243 cardiac surgery patients and confirmed that a restrictive strategy (hemoglobin < 7.5 g/dL) was noninferior to a liberal strategy (hemoglobin < 9.5 g/dL intraoperative and ICU, < 8.5 g/dL non-ICU ward). The 2016 American Association of Blood Banks recommends a hemoglobin trigger of 8 g/dL for patients undergoing cardiac surgery (35); however, these guidelines were published prior to the publication of the TRICS III trial. The 2018 Patient Blood Management International Consensus Conference recommends a hemoglobin threshold of 7.5 g/dL for patients undergoing cardiac surgery (14).

Acute Neurologic Injury

Acute neurologic injury including traumatic brain injury (TBI) and subarachnoid hemorrhage (SAH) (44) have high morbidity and mortality. Given the brain’s vulnerability to hypoxia, transfusions have been considered an important intervention to improve oxygen delivery in this population (45). Anemia is associated with worse outcomes in TBI (46). However, RBC transfusions are also associated with worse outcomes (47). A study randomizing 200 closed head injury patients to restrictive (hemoglobin < 7 g/dL) or liberal (hemoglobin < 10 g/dL) transfusion thresholds found no significant difference in neurologic recovery at 6 weeks (25). Secondary analysis saw more brain tissue hypoxia events (brain tissue oxygenation < 10 mm Hg) in normal brain tissue in the restrictive group compared with the liberal group, but no difference in intracranial pressure, long-term neurologic outcomes, or mortality (48). Similarly, 30-day mortality did not differ in a TRICC trial subgroup analysis of patients with moderate to severe TBI (26). In patients with TBI, current guidelines agree there is no benefit of liberal transfusion strategies (hemoglobin < 10 g/dL) and recommend a target hemoglobin of 7–9 g/dL (49, 50).

Transfusion strategy evidence in SAH is even more limited, with guidelines recommending individualized transfusion practices (50). Ongoing studies are evaluating transfusion strategies in TBI (HEMOglobin transfusion threshold in Traumatic brain Injury OptimizatioN [HEMOTION] trial) (51), acute brain injury (Transfusion strategies in Acute brain INjured patients [TRAIN] trial) (52), and SAH (aneurysmal SubArachnoid Hemorrhage- Red blood cell transfusion And outcome [SAHaRA] trial) (53).

Burns

Severe burns require aggressive fluid resuscitation and, commonly, blood transfusions (54). The Transfusion Requirements in Burn Evaluation study randomized patients with greater than or equal to 20% of total body surface area burns to restrictive versus liberal groups (hemoglobin < 7 g/dL or < 10 g/dL, respectively) (27). The restrictive arm received significantly fewer RBC units than the liberal arm (mean: 7 vs 15 U, respectively). There was no difference in bloodstream infection frequency, or any secondary outcomes including organ dysfunction, ventilator days, time to wound healing, or 30-day mortality. There are currently no burn-specific transfusion threshold guidelines.

Hemorrhagic Shock

Hemorrhagic shock accounts for over 60,000 U.S. deaths annually with the vast majority related to trauma (55). Nearly 30% of prehospital trauma deaths and 20% of all hospital trauma deaths are due to hemorrhage (56–60). Massive transfusion protocols (MTPs) are the standard of care for managing hemorrhagic shock. Traumatic injuries, ruptured abdominal aortic aneurysms, gastrointestinal, and obstetric bleedings represent the most frequent causes of MTP activation (61–63). Although a formal definition of massive transfusion is lacking, the traditional definition of 10 RBC units within 24 hours is used in many centers.

Landmark MTP studies, including the Prospective, Observational, Multicenter, Major Trauma Transfusion (PROMMTT) (64) and Pragmatic Randomized Optimal Platelet and Plasma Ratios (PROPPR) (65), support a 1:1:1 strategy (1 U of packed RBCs, 1 U of fresh frozen plasma [FFP], and 1 U of platelets). However, the optimal RBC, plasma, platelet, and cryoprecipitate ratio is still unknown. Many centers now emphasize earlier platelet and FFP use and include delivery of 6 U of type O RBCs, 6 U of FFP and a unit of apheresis platelets (66). Nontrauma MTP activations represent a significant proportion of MTP activations, as much as 30% and upwards depending on the center (61–63, 65, 67). Data regarding MTP use in nontrauma patients is limited, an area that may benefit from more research.

MTP use in postpartum hemorrhage helps to expedite the transfusion process (68, 69). Most protocols have been adapted from trauma evidence, with the criteria for activating the MTP an anticipation of replacing 50% or more of blood volume within 2 hours or continued bleeding after 4 U RBC in less than 2 hours (70). Although the 1:1:1 ratio strategy is recommended, one recent systematic review concluded a FFP/RBC ratio of greater than or equal to 1 was associated with improved patient outcomes (71). The American College of Obstetricians and Gynecologists currently recommends MTPs as part of a comprehensive postpartum hemorrhage management plan (72).

FUTURE DIRECTIONS

In addition to continuing to develop the evidence base for transfusion practices in specific patient populations, complimentary strategies are emerging including whole blood (WB) transfusions and nontransfusion strategies to avoid and treat anemia.

WB transfusions, which began in World War I and II, were almost entirely replaced in the 1970s by contemporary MTPs utilizing conventional component therapy (e.g., RBC, plasma, platelet, and cryoprecipitate transfusions). However, recent military experiences are prompting a reconsideration of civilian WB transfusions in trauma patients and others with massive hemorrhage (73, 74). WB, frequently low titer group O WB (LTOWB), is currently being used as first-line transfusions in over a dozen U.S. trauma centers (75, 76). To date, most civilian studies have been retrospective; however, prospective randomized studies are underway (77). In one retrospective analysis, patients who received WB plus conventional components did as well as or better than those who only received conventional components (78). Others are investigating the feasibility of using cold-stored LTOWB in prehospital transfusion protocols (79).

Iatrogenic causes of anemia in the ICU can account for 40–70 mL of blood loss daily (5). Anemia prevention strategies have typically focused on blood conservation, including reducing phlebotomy frequency and the volume removed with each blood draw (80). A blood-conserving bundle (including a closed-loop arterial blood sampling system) reduced iatrogenic blood loss in mechanically ventilated patients (81).

Although iron supplementation in the critically ill has not been shown to increase infections or adverse events, evidence is mixed regarding any effect on discharge hemoglobin or RBC transfusions (82, 83). Similarly, erythropoietin (84) supplementation did not improve survival or RBC transfusion frequency; furthermore, this treatment may increase the risk of thromboembolism (85). Although some studies suggest a mortality benefit from using erythropoietin-stimulating agents (ESAs) in trauma patients, further studies are warranted (86, 87). Currently, ESAs are only indicated in patients with chronic kidney disease (88).

CONCLUSIONS

Treating anemia, massive hemorrhage, and optimizing oxygen delivery in the critically ill with RBC transfusions remain a key component of ICU care. A maturing literature is available to better guide transfusion decisions in the majority of critically ill patients. We must couple this mounting evidence base with complementary practices to reliably provide high quality care to critically ill patients.