The trusted source for
healthcare information and
By Jane Guttendorf, DNP, CRNP, ACNP-BC, CCRN
Assistant Professor, Acute & Tertiary Care, University of Pittsburgh, School of Nursing
Dr. Guttendorf reports no financial relationships relevant to this field of study.
In critical care, there has been a focus on adhering to a restrictive transfusion strategy since the Transfusion Requirements in Critical Care (TRICC) trial was published in 1999. In that trial, the authors evaluated a restrictive (hemoglobin [Hgb] < 7 g/dL) vs. liberal (Hgb < 10 g/dL) transfusion trigger in ICU patients and found reduced in-hospital mortality in the restrictive compared to the liberal transfusion group (23% vs. 28%, respectively; P = 0.05).1 Additional studies conducted in various populations of patients then followed with a growing body of evidence in support of restrictive transfusion thresholds. What follows is a review of the evidence supporting restrictive transfusion strategies, a definition of appropriate patient populations, and specific approaches for clinicians to use in developing effective blood management programs.
The most recent transfusion guidelines, published in 2016 by the AABB (formerly known as the American Association of Blood Banks), were based on 31 trials that included 12,587 patients and offered two primary recommendations: 1) a restrictive Hgb threshold of < 7.0 mg/dL (compared to a liberal transfusion threshold of < 10 mg/dL) for hemodynamically stable hospitalized adult patients, including critical care patients, and 2) a Hgb threshold of < 8.0 mg/dL for patients undergoing orthopedic or cardiac surgery or presenting with existing cardiovascular (CV) disease. Excluded from these recommendations because insufficient evidence are patients with acute coronary syndrome (ACS), patients with severe thrombocytopenia at risk for bleeding (hematology/oncology patients), and patients with chronic transfusion-dependent anemia.2
Multiple additional studies have been conducted comparing restrictive vs. liberal transfusion thresholds in different populations of patients. Patients undergoing orthopedic surgery, specifically older patients with hip fractures, were studied in the FOCUS trial. In this study, 2,016 patients ≥ 50 years of age with either a history of CV disease or risk factors for CV disease with Hgb < 10 mg/dL after hip fracture surgery were randomized to a restrictive (Hgb < 8.0 or symptomatic anemia) or liberal (Hgb < 10) strategy. The liberal group did not demonstrate reduced rates of death at 60-day follow up.3 In a long-term follow-up study of those same patients, three-year survival did not differ significantly between the liberal and restrictive transfusion groups (hazard ratio [HR], 1.09; 95% confidence interval [CI], 0.95-1.25; P = 0.21).4 The authors of an additional randomized, controlled trial (RCT) of 284 frail, elderly, anemic patients with hip fracture (TRIFE) found no statistically significant difference in recovery from physical disabilities, but higher 30-day and 90-day mortality in the restrictive (Hgb < 9.7) compared to liberal (Hgb < 11.3) groups. Notably in this trial, the restrictive Hgb threshold was almost equal to the liberal Hgb threshold in the earlier FOCUS trial.5 The authors of a 2017 systematic review and meta-analysis of transfusion thresholds in hip and knee surgery reviewed 10 RCTs that included 3,788 patients. They noted no differences in mortality or other complications, including myocardial infarction (MI), congestive heart failure, and pneumonia between restrictive and liberal transfusion groups.6
A systematic review and meta-analysis of nine RCTs (5,780 patients) evaluating transfusion strategies in older adults (including both FOCUS and TRIFE) found that the 30-day risk of death (relative risk [RR], 1.36; 95% CI, 1.05-1.74; P = 0.17) and 90-day risk of death (RR, 1.45; 95% CI, 1.05-1.98; P = 0.22) were both higher in older patients with a restrictive transfusion strategy vs. a liberal strategy, suggesting that a more liberal transfusion strategy may be favored in geriatric patients.7
In cardiac surgery patients, a 2010 RCT (TRACS) conducted in 505 patients at a single center in Brazil demonstrated no significant difference in composite endpoint of 30-day mortality and severe morbidity (cardiogenic shock, acute respiratory distress syndrome, acute renal injury requiring dialysis, or hemofiltration) in the liberal (targeting hematocrit [Hct] > 30%) group vs. the restrictive (targeting Hct > 24%) group.8 Following this, a 2014 meta-analysis of seven RCTs (including TRACS; 1,262 patients) evaluating transfusion triggers in cardiac surgery demonstrated no difference in mortality between restrictive and liberal transfusion groups (RR, 1.12; 95% CI, 0.65-1.95; P = 0.60).9 The TITRe2 trial, reported in 2015, included 2,004 cardiac surgery patients and found that a restrictive transfusion strategy (Hgb < 7.5 mg/dL) was not superior to a liberal strategy (Hgb < 9 mg/dL) regarding morbidity. The restrictive group exhibited more deaths (4.2% vs. 2.6%; HR, 1.64; 95% CI, 1.00-2.67; P = 0.045), suggesting that restrictive transfusion may affect mortality negatively.10 The authors of the recently published TRICS III study randomized 5,000 cardiac surgery patients prior to surgery to a restrictive (Hgb < 7.5 g/dL, either intraoperatively or postoperatively) or to a liberal strategy (Hgb < 9.5 g/dL, intraoperatively or postoperatively in the ICU, or Hgb < 8.5 g/dL postoperatively on the ward). The restrictive strategy was noninferior to the liberal strategy for a composite outcome of death, MI, stroke, or new renal failure with dialysis. Mortality was 3% in the restrictive group compared to 3.6% in the liberal group (odds ratio [OR], 0.85; 95% CI, 0.62-1.16).11
In patients with sepsis, a subgroup analysis of the Transfusion Requirements in Septic Shock (TRISS) trial evaluated higher (9.0 g/dL) vs. lower (7.0 g/dL) Hgb thresholds for transfusion, with a primary endpoint of 90-day mortality in pre-specified subgroups of patients (chronic lung disease, hematologic malignancy, metastatic cancer, surgery patients, and septic shock by SEPSIS-3 criteria). No significant differences were found between the two Hgb threshold groups in terms of 90-day mortality in any of the subgroups.12 In the primary TRISS trial, 90-day mortality was 43% in the lower threshold group vs. 45% in the higher threshold group (RR, 0.94; 95% CI, 0.78-1.09; P = 0.44).13 A systematic review and meta-analysis of transfusion in critically ill sepsis patients comparing restrictive vs. liberal transfusion thresholds found only one RCT (TRISS) and nine cohort studies that included mortality as the primary endpoint. The restrictive strategy was not associated with either harm or benefit compared to a liberal strategy (as reported above for TRISS).14 For the cohort studies, red blood cell transfusion was not associated with increased mortality.14 Overall, there are few RCTs evaluating transfusion thresholds in sepsis, and further study is needed.
In critically ill patients, the authors of a pilot RCT that included six ICUs in the United Kingdom evaluated a restrictive (Hgb > 7 g/dL) vs. liberal (Hgb > 9 g/dL) transfusion strategy and randomized 100 patients. Mortality at 180 days trended toward higher rates in the liberal group than in the restrictive group (55% vs. 37%; RR, 0.68; 95% CI, 0.44-1.05; P = 0.073).15 A 2016 meta-analysis evaluated restrictive vs. liberal transfusion targets in critically ill patients and in patients with ACS. The analysis included six RCTs involving 2,156 patients. The authors found no differences in mortality between groups, and they concluded that a restrictive strategy was at least equivalent to a liberal strategy in critically ill patients, but noted that there was inconclusive evidence to recommend a restrictive strategy in ACS patients.16
In a context-specific systematic review and meta-analysis of RCTs in the perioperative and acute care settings, a critical care subset of patients showed no increased risk of 30-day complications (inadequate oxygen supply, mortality, composite of both). For perioperative patients, there was an increased risk of adverse events in the CV procedures restrictive strategy group.17 Two recent meta-analyses in perioperative and critical care patients both identified a possible increased risk for perioperative patients with restrictive transfusion strategies. Fominskiy et al reviewed 17 trials for perioperative patients (nine orthopedics, five cardiac surgery, one vascular, one oncology surgery, and one obstetrics study, for a total of 7,552 patients) and 10 trials in critically ill patients (3,469 patients). There was no difference in mortality in critically ill patients treated with restrictive vs. liberal strategies. However, in perioperative patients, a liberal strategy was associated with improved survival over a restrictive transfusion strategy (OR, 0.81; 95% CI, 0.66-1.00; P = 0.05).18 In a meta-analysis that included 27 RCTs or quasi-experimental studies evaluating perioperative and critical care patients, a restrictive transfusion strategy (Hgb 7-8 g/dL) significantly reduced 30-day mortality compared with a liberal (Hgb 9-10 g/dL) strategy (OR, 0.82; 95% CI, 0.70-0.97) in the critical care subpopulation. However, the restrictive strategy was associated with the opposite effect in the surgical patients, indicating either a potentially increased mortality risk or no difference between groups (OR, 1.31; 95% CI, 0.94-1.82).19
Since the 2016 AABB clinical practice guidelines for transfusion were published, additional information is available regarding oncology patients. A systematic review and meta-analysis that included six studies (both randomized and non-randomized) involving 983 patients concluded there was no difference in mortality between the restrictive and liberal transfusion groups (RR, 1.00; 95% CI, 0.32-3.18) and no difference in adverse events.20 One single-center, double-blind, RCT evaluating patients with solid tumors who presented with septic shock noted trends toward improved survival in the liberal transfusion group at both 28 days from randomization and at 90 days from randomization.21 These recent studies offer results that conflict with those previously reported and deserve further study, namely in perioperative and oncology patients.
Despite a wealth of evidence supporting the benefits of a restrictive transfusion strategy in many patient populations, the guidelines often still are not followed with consistency. Implementing transfusion triggers or guidelines can effectively lower the number of transfusions and forms the basis of most blood management programs. The effect of implementing a transfusion trigger was evaluated in a systematic review and meta-analysis of 19 RCTs that included 6,264 patients. Patients had been randomized to a restrictive or liberal transfusion threshold. Clinical outcomes and blood use were evaluated. Patients in the restrictive transfusion strategy group demonstrated a 39% lower risk of receiving a transfusion compared to those in the liberal group (RR, 0.61; 95% CI, 0.52-0.72). Restrictive transfusion was associated with significantly lower hospital mortality, but not 30-day mortality.22
Several other strategies have been identified for improving compliance with recommended transfusion guidelines and are summarized in Table 1. Most start with developing a formal blood management program. This requires an organizational structure and support from executive leadership to implement a multidisciplinary process improvement program. Norgaard et al reported on the successful implementation of a blood management program at a large Danish hospital involving a hospital-wide effort to improve evidence-based transfusion practices. At three-year follow-up, sustained improvement was demonstrated with a reduction in percent transfusions above the Hgb threshold from 23% to 10% (P < 0.001), an improvement in transfusions at or below the trigger Hgb of 7.3 mg/dL from 7% to 19% (P < 0.001), and an increase in single-unit transfusions from 72% to 76% (P < 0.001).23
Quality improvement frameworks, such as the Institute for Healthcare Improvement’s Model for Improvement using Plan-Do-Study-Act cycles,24 or development and implementation of an evidence-based care bundle for transfusion, such as those used successfully to address ventilator-associated pneumonia or central line-associated blood stream infection, may be helpful. One Dutch ICU implemented a transfusion bundle that included process measures (consent, correct patient, and product identification) with a transfusion threshold (per individual patient baseline). After bundle implementation, the Hgb trigger decreased from 7.3 to 7.1 g/dL (P = 0.04), and the number of inappropriate transfusions decreased significantly from 25% to 10% (P = 0.001).25
Clinical decision support tools within the electronic medical record are used to inform clinicians at the time of order entry when orders fall outside the established evidence-based standard, prompting an action to either cancel or continue with ordering. One institution implemented a multimodal project to improve adherence to transfusion guidelines in two surgical ICUs for low-risk anemic patients, defined as not actively bleeding and hemodynamically stable.26 The multimodal approach included an educational lecture, active monthly auditing of each transfusion for appropriateness, and individualized peer-to-peer feedback delivered via email from a surgeon intensivist colleague to clinicians when a packed red blood cell transfusion was associated with a Hgb trigger of > 8 mg/dL or if multiple units were ordered without repeating the Hgb after transfusion of the first unit. Over six months, in comparison to a baseline pre-intervention group, the multimodal intervention resulted in a lowering of the mean transfusion trigger from Hgb 7.6 to 7.1 mg/dL (P < 0.001) and a marked reduction in percent of transfusions for a trigger Hgb > 8 mg/dL from 25% to 2% (P < 0.001). Transfusions decreased by 36% during the intervention period, suggesting that the personalized peer-to-peer intervention and active auditing with feedback were successful strategies to incorporate in a blood management program.26
In addition to transfusion guidelines, compliance monitoring and feedback programs and more strategies to minimize blood loss, particularly intraoperatively, are important. Some of these are included in a “blueprint” for implementing a blood management program outlined by Sadana et al, including use of antifibrinolytics, cell salvage and autologous transfusion, managing preoperative anemia for elective patients with iron or erythropoietin, use of point-of-care testing to minimize delays in response, and reducing phlebotomy (number and amount of blood draws).27
There is extensive evidence supporting a restrictive transfusion approach in critically ill patients and cardiac surgery patients. The data favor restrictive strategies in hip and knee surgery, except in geriatric patients for whom evidence favors a more liberal transfusion strategy. Evidence is sparse in sepsis, oncology, and acute coronary syndrome patients, warranting further study. Given the data from the two recent systematic reviews and meta-analyses linking restrictive transfusion and potentially poorer outcomes in surgical and perioperative patients, additional trials are needed to answer the question about whether a more restrictive or a more liberal transfusion strategy is most appropriate for these groups.
Despite good intentions, clinicians still are implementing the current guidelines inconsistently. Targeted multidisciplinary efforts to implement blood management programs are needed, incorporating evidence-based transfusion triggers, blood conservation strategies, and continuous quality improvement processes to evaluate and promote compliance with transfusion practices that have been shown to affect patient outcomes positively.
Financial Disclosure: Critical Care Alert’s Physician Editor Betty Tran, MD, MSc, Nurse Planner Amy Johnson, MSN, RN, Peer Reviewer William Thompson, MD, Executive Editor Leslie Coplin, Editor Jonathan Springston, and Editorial Group Manager Terrey L. Hatcher report no financial relationships relevant to this field of study.