Review of available evidence supporting different transfusion thresholds in different patient groups with anemia
Abstract
In patients with anemia, transfusion of red blood cells (RBCs) can save lives and improve quality of life. The choice to transfuse should be cautiously made owing to risks of transfusion, economic costs, and limitations on the blood supply. Until the 1980s, the decision for RBC transfusion was guided by Hb threshold, with the aim of maintaining the patient's blood Hb level over 100 grams per liter. Since then, multiple randomized controlled trials and key systematic reviews have provided evidence-based guidelines as to appropriate transfusion thresholds in a number of clinical settings. Here, we aimed to address the outcome of defining different anemia criteria in specific clinical populations exclusively on the basis of the need for RBC transfusion based on Hb concentration. We focused on the patient populations, where there were the most available data on differing transfusion thresholds, which looked at transfusing to a higher or liberal transfusion threshold in comparison with a lower or restrictive transfusion threshold. These included patients in intensive care with or without septic shock, hip fracture surgery, cardiovascular surgery, and upper gastrointestinal bleeding, the pediatric population, and also those with malaria, by reviewing key randomized controlled trials and systematic reviews. Twenty-four randomized controlled studies and 12 systematic reviews have been included, and these are discussed below.
Introduction
Red blood cell (RBC) transfusions can improve oxygen transport to vital organs. Clinically severe anemia can result in impairment of oxygen delivery to tissues, and may result in symptoms and/or signs of myocardial ischemia or cardiac failure, features of hypovolemia, such as postural hypotension, and impaired development in children. In patients with anemia, transfusion of RBCs can save lives and improve quality of life.
The choice to transfuse should be cautiously made not only due to limitations on the blood supply, but also due to high costs associated with allogeneic blood product transfusion. Blood product transfusion carries with it largely underestimated high production and service costs, as well as dose-dependent increase in morbidity and mortality. This can add billions in unnecessary health care expenditure and loss of income due to disability-adjusted life years (DALYs).1 Therefore, it is of great importance to carefully assess the clinical appropriateness of blood transfusion.
Until the 1980s, the decision for RBC transfusion was generally guided by hemoglobin (Hb) threshold or the “10/30 rule”—with the aim of maintaining a patient's blood Hb level over 100 g/L (or 10 g/dL) and hematocrit over 30%, regardless of clinical status.2 Since then, multiple randomized controlled trials (RCTs) have demonstrated that transfusing at a lower target Hb level of 70–80 g/L can produce comparable or better outcomes in many patient populations, and can significantly reduce blood component usage. Increasing evidence from these studies has led to the development of many evidence-based and expert consensus clinical practice guidelines. They generally agree that transfusion is not indicated for patients with an Hb level over 100 g/L except in special situations, and transfusion may be considered appropriate when Hb is 70–90 g/L, depending on the clinical context. These guidelines emphasize that patient characteristics, comorbidities, symptoms, signs, and wishes need to be incorporated into the decision-making process of transfusion.3-5
Here, we aimed to review the outcome of defining different anemia criteria in specific clinical populations exclusively on the basis of the need for treatment based on Hb concentration. The aim of this study was to examine evidence underpinning the use of different Hb thresholds for RBC transfusion in the following patient groups. We focused on the patient populations where there were the most available data on differing transfusion thresholds, which looked at transfusing to a higher or liberal transfusion threshold in comparison with a lower or restrictive transfusion threshold. These included patients in intensive care and septic shock, hip fracture surgery, cardiovascular surgery, and acute upper gastrointestinal bleeding (UGIB), the pediatric population, and also those with malaria. The outcome was to scope the extent of available literature in this context.
Methodology
We performed a literature search using the Ovid platform in the MEDLINE database without timeline or language restrictions. The following search terms were used: a combination of Medical Subject Heading (MeSH) terms “transfusion trigger” or “transfusion threshold” and/or keyword and MeSH terms of “blood transfusion” or “RBC transfusion” or “erythrocyte transfusion.” Studies were included if they compared RBC transfusion strategies using differing transfusion thresholds with effect on clinical outcome. This study was aimed as a starting point to scope the available literature surrounding transfusion thresholds, and due to time constraints, only one search database was used.
We focused on these specific populations: intensive care and septic shock, hip fracture surgery, cardiovascular surgery, acute UGIB, and the pediatric population. We included RCTs (prospective studies with patients randomized to intervention groups) and systematic reviews. We excluded nonrandomized studies (to reduce the presence of confounding factors), retrospective reviews, case reports, audits, and letters to the editor, surveys, animal studies, and non-English articles. We also screened the reference lists of the most relevant articles for studies not captured in the preliminary search.
We identified 404 initial search results. A total of 22 RCTs and eight systematic reviews were identified, which are outlined below (see Fig. 1 and Appendix 1, online only).
For reviewing blood transfusion triggers in malaria, we performed a literature search using the Ovid platform in the MEDLINE database without timeline restrictions or language restrictions. The following search terms were used: a combination of MeSH terms “malaria–treatment” and/or keyword and MeSH terms of “blood transfusion” or “RBC transfusion” or “erythrocyte transfusion.” Studies were included if they examined the role of RBC transfusion on clinical outcome in patients with malaria (see Fig. 2).
We identified an initial 105 search results. Two systematic reviews and two RCTs were identified.
Results
Intensive care
Anemia is frequent in the intensive care setting and is associated with overall worse mortality and morbidity.6 RBC transfusion thresholds in the intensive care context have been studied in two large RCTs and two systematic reviews (see Table 1).
| Study | Clinical setting | RCT groups and/or outcomes | Conclusions |
|---|---|---|---|
| RCT | |||
| Transfusion Requirements in Critical Care (TRICC)7 | Critically ill euvolaemic patients admitted to ICU and Hb < 9.0 g/dL 72 h within ICU admission (n = 838) |
|
|
| Transfusion Requirements in Septic Shock (TRISS)8 | Patients in ICU with septic shock and Hb < 9.0 g/dL (n = 1005) |
|
No significant difference at 90-day mortality (43% versus 45%, P = 0.44), ischemic and adverse events, and use of life support |
| Systematic review | |||
| Impact of transfusion on patients with sepsis admitted in intensive care unit: a systematic review and meta‑analysis6 | RCTs and cohort studies in the comparison of transfusion thresholds in ICU patients with septic shock (n = 2762) | Mortality, nosocomial infection, acute lung injury, and acute kidney injury | Restrictive transfusion not associated with increased mortality; but was associated with increased nosocomial infection, acute lung injury, and acute kidney injury |
| Impact of more restrictive blood transfusion strategies on clinical outcomes: a meta-analysis and systematic review9 | RCTs evaluating transfusion thresholds in ICU patients (n = 2364) | Mortality, acute coronary syndrome, pulmonary edema, infections, rebleeding, the number of patients transfused, and units of blood transfused per patient | Restrictive transfusion threshold was associated with decreased ischemic cardiac events, infection, rebleeding, and mortality |
The first large RCT conducted was the 1999 TRICC trial, which reviewed 838 nonseptic patients with anemia in ICU and found that there was no significant difference in the 30-day mortality rate (primary outcome) between transfusion thresholds of 100 or 70 gram per liter. The in-hospital mortality rate was significantly lower in the restrictive group (22.2% versus 28.1%, P = 0.05), but there was no significant difference in rates of 30-day mortality.7
In addition, the liberal Hb threshold of 100 g/L was historically regular practice in the early phase of septic shock management. This carries with it potential complications of fluid overload, infection, and immunomodulatory effects.6 The TRISS trial in 2014 demonstrated that restrictive transfusion threshold (<70–75 g/L) was not associated with an increase in 90-day mortality (43% versus 45%, P = 0.44), and patients received fewer transfusions (1 unit of blood versus 4 units) in comparison with the liberal group (<100–105 g/L). However, there were more protocol violations in the restrictive group due to ischemic and bleeding complications.8
Two systematic reviews were identified. The 2014 systematic review by Salpeter et al. of 2364 critically ill patients concluded that restrictive transfusion threshold was associated with decreased ischemic cardiac events, infection, rebleeding, and mortality.9
The 2017 Dupuis et al. systematic review of 2762 patients with sepsis in the ICU identified an additional 12 cohort studies in addition to one RCT, and their overall conclusions were that a restrictive transfusion threshold was not associated with increased mortality.6 Note this review included just one RCT, and included 12 cohort studies that did not focus specifically on patients with sepsis; therefore, the authors concluded that no definite conclusions should be drawn from this review due to the heterogeneity of the studies included.
Acute gastrointestinal bleeding
There is high mortality in upper gastrointestinal hemorrhage, and RBC transfusion is frequently administered in this setting for correction of anemia and volume expansion.10 In the hemodynamically unstable bleeding patient, transfusion is generally guided by clinical parameters rather than Hb thresholds. In bleeding but hemodynamically stable patients, two large RCTs by Villaneuva et al. and Jairath et al. have contributed to data regarding transfusion thresholds in mortality and rebleeding rates (see Table 2).
| Study | Clinical setting | RCT groups and/or outcomes | Conclusions |
|---|---|---|---|
| RCT | |||
| Transfusion strategies for acute UGIB,11 RCT | Severe acute UGIB (n = 921) |
|
Improved 6-week mortality (95% versus 91%, P = 0.02), reduced bleeding (10% versus 16%, P = 0.01), and reduced adverse events (40% versus 48%, P = 0.02) with liberal versus restrictive threshold |
| Transfusion in Gastrointestinal Bleeding Trial (TRIGGER),13 feasibility trial | Acute GIB and Hb < 10 g/dL (n = 936) |
|
|
| Systematic review | |||
| Red cell transfusion for the management of upper gastrointestinal hemorrhage10 | Randomized and quasi-randomized studies comparing RBC transfusion and standard care in patients with upper GIH (n = 126) | Main outcomes: death and rebleeding rates | More deaths and rebleeding in the transfusion arms, but it was difficult to draw conclusions due to the number of underpowered studies and large volume of missing data |
| Restrictive versus liberal blood transfusion for gastrointestinal bleeding: a systematic review and meta-analysis of randomized controlled trials14 | Randomized controlled trials comparing restrictive and liberal RBC transfusion strategies for acute UGIB (n = 1965) | Main outcomes: mortality, rebleeding, ischemic events, and mean RBC transfusion |
|
- UGIB, upper gastrointestinal bleeding; GIB, gastrointestinal bleeding; GIH, gastrointestinal hemorrhage; RBC, red blood cell; RR, relative risk.
A 2013 single-center RCT by Villaneuva et al. demonstrated that patients with acute UGIB who were assigned to a restrictive transfusion threshold (<70 g/L) in comparison with a liberal threshold (of <90 g/L) had significantly superior outcomes with improved mortality at 6 weeks (95% versus 91%), reduced secondary bleeding (10% versus 16%), and reduced adverse events (40% versus 48%).11 It is important to note that this study excluded patients with major cardiovascular morbidity and patients had ready access to endoscopic treatment. It is postulated that this mechanism may be due to increase in portal pressures from “overtransfusion” in patients with variceal bleeding.12
Following on from that, the 2015 multicenter cluster randomized feasibility transfusion (TRIGGER) trial examined transfusion thresholds in all patients with acute gastrointestinal bleeding, except in cases of exsanguinating hemorrhage, which was defined as patients who received emergency O-negative blood; had features of shock and were transfused RBCs within 2 h of presentation; or received first endoscopy in emergency, intensive care, or operating theater due to severity of bleeding.13 Four hundred and three patients were enrolled to the restrictive threshold (<80 g/L) and 533 patients were enrolled to the liberal threshold (<100 g/L). No significant differences in clinical outcome were noted, including 28-day mortality, rebleeding rates or cardiac events, infection, or quality of life. There were less RBC units transfused in the restrictive threshold, but this difference was not significant. It is important to note that this was designed as a feasibility trial to assess the practicality of implementation of differing transfusion thresholds on larger scales, and clinical outcomes derived from this trial were not meant for use in directing clinical practice.
A 2010 Cochrane review reviewed three RCTs in patients with acute upper gastrointestinal hemorrhage, comparing RBC transfusion versus no transfusion (standard care with other intravenous fluids), and concluded that there were more deaths and rebleeding in the transfusion arms.10 However, it was difficult to draw conclusions due to the number of underpowered studies and a large volume of missing data.
A recent 2017 systematic review and meta-analysis reviewed five RCTs with acute UGIB, comparing liberal and restrictive transfusion threshold.14 Restrictive transfusion was associated with a lower risk of all-cause mortality and rebleeding. There was no difference in risk of stroke, acute coronary events, or acute kidney injury. There was a nonstatistically significant trend to a higher risk of mortality in the restrictive group in patients with ischemic heart disease (RR = 4.4), and lower mortality in patients without ischemic heart disease (RR = 0.58). It is noted that a substantial proportion of patients in this review were recruited from patients from the Villaneuva et al. trial that excluded patients with major cardiovascular disease. While the restrictive transfusion threshold appears favorable in this clinical group of patients, exceptions may include patients with coronary comorbidities, such as ischemic heart disease, who may benefit from higher transfusion thresholds.
Hip fracture/revision surgery patients
Hip fracture patients are often elderly with multiple comorbidities and require surgical intervention. These patients frequently require transfusion due to blood loss from primary fracture or intraoperative losses.15
There were six RCTs addressing different transfusion thresholds in hip fracture patients (see Table 3). Earlier studies with small patient numbers suggested that there may be benefit from a liberal transfusion threshold.16, 17
| Study | Clinical setting | RCT groups and/or outcomes | Conclusions |
|---|---|---|---|
| RCT | |||
| A pilot randomized trial comparing symptomatic versus hemoglobin-level−driven transfusions following hip fracture16 | Patients with hip fracture surgery and Hb < 100 g/L (n = 84) |
|
|
| The effects of transfusion thresholds on ambulation after hip fracture surgery17 | Patients with hip surgery aged >65 years with intact cognition (n = 120) |
|
|
| Functional Outcomes in cardiovascular patients undergoing surgical hip fracture repair (FOCUS)18 | Patients undergoing surgery for hip fracture, aged >50 years and either history or risk factor for cardiovascular disease (n = 2016) |
|
No significant difference at 30- or 60-day mortality, ischemic events, duration of hospitalization, and performance status |
| Functional Outcomes in cardiovascular patients undergoing surgical hip fracture repair (FOCUS)—3-year follow up20 | Patients undergoing surgery for hip fracture, aged >50 years and either history or risk factor for cardiovascular disease (n = 2016) |
|
Three-year follow up—no difference in mortality (P = 0.21) |
| Randomized trial of blood transfusion after hip fracture surgery21 | Patients post hip surgery aged >60 years with Hb 8.0–9.5 g/dL (n = 200) |
|
No significant difference between mortality at 1 year (P = 1.0), hospital duration (P = 0.64), or regaining of mobility (P = 0.34) |
| Transfusion requirements in frail elderly (TRIFE)19 | Patients undergoing surgery for hip fracture, aged >65 years (n = 284) |
|
|
| Systematic review | |||
| Red blood cell transfusion for people undergoing hip fracture surgery15 | Randomized controlled trials comparing restrictive and liberal RBC transfusion strategies in patients undergoing hip fracture surgery (n = 2722) | Main outcomes: mortality, functional recovery, or postoperative morbidity | No difference in 30- and 60-day mortality, or functional recovery at 60 days |
However, follow-up larger RCTs, in particular, the 2011 FOCUS trial, which reviewed the largest patient cohort of 2016 elderly patients (>50 years old) with history or risk factors for cardiac disease undergoing hip fracture surgery, showed that elderly patients who underwent hip fracture surgery were able to tolerate low Hb levels to 80 g/L, and there were no significant differences in mortality, morbidity, or functional performance at 60 days post-surgery between the two transfusion groups (restrictive threshold was <80 g/L and liberal threshold was <100 g/L).18 Three-year follow-up assessment did not show any difference in long-term mortality between the two groups. Most of the patients were relatively fit, with only 10% of patients from nursing homes in this study.
The 2015 TRIFE study reviewed transfusion thresholds in frail elderly patients (>65 years old) from nursing homes or sheltered housing undergoing hip fracture surgery and again did not find significant differences in 90-day mortality, and also no difference in functional status.19 However, 90-day mortality was higher in the restrictive group in nursing home residents, suggesting that some patients may benefit from liberal transfusion. In this study, note that higher transfusion thresholds were used in both the liberal (<113 g/L) and restrictive (<97 g/L) groups.
A 2015 Cochrane review of five RCTs in patients undergoing hip fracture surgery concluded that there was no difference between transfusion groups in 30- and 60-day mortality, or functional recovery at 60 days.15 Major complications following surgery were found to be similar between the two groups, with the exception for myocardial infarct, which may be lower in liberal transfusion threshold (RR = 0.59).
Oncology
Overall, there is limited evidence to support specific transfusion thresholds in patients with anemia and patients with malignancies undergoing active treatment.
Four RCTs were identified (see Table 4).
| Study | Clinical setting | RCT groups and/or outcomes | Conclusions |
|---|---|---|---|
| RCT | |||
| Feasibility of augmented red blood cell transfusion for patients with induction chemotherapy for acute leukemia or stem cell transplantation20 | Patients with acute leukemia receiving induction chemotherapy or stem cell transplantation (n = 60) |
|
|
| Transfusion requirements in surgical oncology patients22 | Patients undergoing major cancer surgery and in ICU (n = 198) |
|
Major complications and mortality were more common in the restrictive group (36% versus 20%, P = 0.012) favoring liberal transfusion threshold |
| Liberal versus restrictive transfusion thresholds in leukemia patients: feasibility pilot study21 | Patients with acute leukemia with plans for inpatient myelosuppressive chemotherapy (n = 90) |
|
|
| Transfusion requirements in Critically Ill Oncologic Patients (TRICOP)23 | Cancer patients with septic shock, aged >18 years (n = 300) |
|
Survival trend favoring liberal transfusion strategy in patients with septic shock. 28-day mortality better in liberal group (45% versus 56%, P = 0.08) and 90-day mortality better in liberal group (59% versus 70%, P = 0.03) |
| Systematic review | |||
| Restrictive versus liberal red blood cell transfusion strategies for people with hematological malignancies treated with intensive chemotherapy or radiotherapy, or both, with or without hematopoietic stem cell support24 | RCTs and prospective nonrandomized studies comparing restrictive and liberal RBC transfusion strategies for people with hematological malignancies undergoing chemotherapy or radiotherapy (n = 240) | Main outcomes: mortality, bleeding, hospital stay, and amount of RBC transfusion | Insufficient evidence at this stage to favor a restrictive threshold |
Two smaller trials reviewed transfusion thresholds in patients with acute leukemia. It is important to note that these were small studies conducted as feasibility trials, again to provide guidance for future larger studies and not meant for the derivation of clinical outcomes in routine clinical practice. Webert et al. recruited patients with acute leukemia undergoing myelosuppressive chemotherapy and found no clinically significant differences in bleeding rates and time to first bleed in patients with thrombocytopenia in comparing standard transfusion therapy (<80 g/L) to augmented (<120 g/L) transfusion therapy.20 DeZern et al. reviewed 90 patients with acute leukemia and found no significant differences in duration of hospitalization, bleeding rates, and duration of fever between the two transfusion thresholds (<80 versus < 70 g/L).21
Two studies found that liberal transfusion thresholds were better in critically ill oncology patients and in patients undergoing major cancer surgery. Almeida et al. reviewed 198 patients undergoing major cancer surgery and found that major complications, including cardiovascular events and intraabdominal infection, and mortality were higher in the restrictive group (<90 g/L) (36% versus 20%), and overall favored liberal transfusion threshold (<70 g/L).22 There was no significant difference in the occurrence of respiratory distress syndrome, acute kidney injury, duration of ICU admission, or use of vasopressors. The 2017 Bergamin Transfusion Requirements in Critically Ill Oncologic Patients (TRICOP) study concluded that there was a survival trend favoring liberal transfusion strategy in cancer patients with septic shock, with overall better 28- and 90-day mortality in the liberal group (<90 versus <70 g/L).23
A 2017 Estcourt Cochrane review on transfusion strategies in patients with hematological malignancies reviewed four studies (240 patients), and concluded that there is insufficient evidence at this stage to favor either a restrictive or liberal threshold due to limited patient numbers and small numbers of events.24
Cardiac surgery
The 2010 TRACs trial demonstrated that different transfusion thresholds (transfusion at hematocrit <24% versus hematocrit <30%) post elective cardiac surgery in 512 patients did not have an impact on 30-day mortality or severe inpatient morbidity.25 The number of transfused RBC units was found to be associated with poor outcome and mortality. Transfusion rate was significantly lower in the restrictive group (47% versus 78%) (see Table 5).
| Study | Clinical setting | RCT groups and/or outcomes | Conclusions |
|---|---|---|---|
| RCT | |||
| Lowering the Hb threshold for transfusion in coronary artery bypass procedures: effect on patient outcome27 | Elective cardiac surgery (coronary artery bypass procedure) (n = 428) |
|
No significant difference in inpatient mortality (P = 0.321) and morbidity |
| Transfusion requirements after cardiac surgery (TRACS)25 | Elective cardiac surgery (n = 512) |
|
No significant difference at 30-day mortality or severe complications (P = 0.85) |
| Liberal or restrictive transfusion after cardiac surgery (TITRe2)28 | Patients undergoing nonemergency cardiac surgery (n = 2003) |
|
No significant difference in serious infection, ischemic event (35% versus 33%, P = 0.30) |
| Systematic review | |||
| Transfusion triggers for guiding RBC transfusion for cardiovascular surgery: a systematic review and meta-analysis26 | RCTs comparing restrictive and liberal RBC transfusion strategies in cardiac and vascular surgery (n = 1262) | Main outcomes: units of RBC transfused, mortality, morbidity (including myocardial infarct, stroke, kidney injury, and infection), and duration of hospital stay | Restrictive thresholds did not have significant differences in mortality or risk of coronary events, stroke, kidney injury, infection, or duration of admission |
The multicenter parallel-group RCT (TITRe2) has been the largest trial in this patient population to date, and it assessed the effect of different transfusion thresholds on over 2000 patients undergoing elective cardiac surgery with regard to serious complications and health care costs.25 No significant difference in morbidity, including stroke, myocardial infarct, acute gut or kidney injury, or serious infection, or health care costs at 3 months, was found. Thirty-day mortality was slightly higher in the restrictive group (<75 g/L) (2.6% versus 1.9%) in comparison with the liberal group (<90 g/L), and 90-day mortality was again higher in the restrictive group (4.2% versus 2.6%). Transfusion rate was significantly lower in the restrictive group (53% versus 93%). Overall, restrictive transfusion demonstrated no clinical benefit over the liberal threshold with a statistically significant increase in 90-day mortality that is not fully understood.
A 2014 systematic review looking at transfusion triggers in cardiac and vascular surgery identified seven trials utilizing transfusion thresholds and concluded that restrictive thresholds did not have significant differences in mortality or risk of coronary events, stroke, kidney injury, infection, or duration of admission.26
Pediatrics
There have been relatively few studies on transfusion threshold in children. Two RCTs found that a restrictive threshold did not result in adverse complications in those who were critically ill but stable, and those who had elective cardiac surgery for congenital heart disease.27, 28 More studies in this population are required (see Table 6).
| Study | Clinical setting | RCT groups | Conclusions |
|---|---|---|---|
| Transfusion requirements in pediatric intensive care units (TRIPICU)27 | Stable critically ill children and infants (n = 648) |
|
No significant difference in or progressive multiorgan dysfunction syndrome (P = 0.28–1.00 based on age range), mortality (P = 0.50), or adverse events (P = 0.44) between the groups |
| Safety and effects of two red blood cell transfusion strategies in pediatric cardiac surgery patients28 | Children and infants aged 6 weeks to 6 years undergoing elective surgery for noncyanotic congenital heart defect (n = 107) |
|
Shorter duration of hospitalization in restrictive group (8 versus 9 days, P = 0.047). No difference in duration of stay in PICU, duration of ventilation, and adverse events |
Preterm infants
The 2005 RCT was a single-center RCT, where infants of 500–1300 g in weight were randomized into two transfusion thresholds (transfusion if hematocrit <34% versus hematocrit <46%).29 More neurological events (intraparenchymal brain hemorrhage or periventricular leukomalacia (PVL)) and episodes of apnea were found in the restrictive group. There was no difference in mortality. The 2006 PINT study found no significant difference in morbidity between the two groups, but slightly higher death rates before discharge in the restrictive group (<75 g/L) versus liberal group (<90 g/L).30
However, a 2012 systematic review reviewed six RCTs in preterm infants with different transfusion triggers and concluded that there was no difference in mortality, chronic lung disease, brain hemorrhage, or PVL between the two groups.31 However, a long-term follow-up report of the cohort of infants in the Bell study at 8–15 years of age showed significant difference in verbal function, memory, and reading skills, favoring the restrictive group. This was conflicting with the follow-up Whyte report that showed no difference in neurodevelopmental impairment at 18–21 months.32 They concluded that there is currently insufficient evidence to recommend either transfusion threshold in preterm infants (see Table 7).
| Study | Clinical setting | RCT groups and/or outcomes | Conclusions |
|---|---|---|---|
| RCT | |||
| Randomized trial for red blood cell transfusion in preterm infants32 | Preterm infants 500–1300 g (n = 100) |
|
Major adverse neurological events (brain hemorrhage and periventricular leukomalacia) (12% versus 0%, P = 0.012) were more frequent in restrictive group |
| Premature infants in need of transfusion study (PINT)33 | ELBW infants <1000 g (n = 451) |
|
Death before discharge was slightly higher in the restrictive group (74.0% versus 69.7%, P = 0.25), no significant difference in morbidity |
| Premature infants in need of transfusion study (PINT) follow-up on neurodevelopmental outcome34 | ELBW infants <1000 g (n = 451) |
|
No significant difference in death or neurodevelopmental impairment at 18–21 months corrected age (P = 0.09) |
| Systematic review | |||
| The safety and efficacy of red cell transfusions in neonates: a systematic review of RCTs35 | RCTs where the intervention was transfusion of RBCs (n = 679) | Main outcomes: mortality and neurodevelopmental and respiratory endpoints | No difference in mortality, chronic lung disease, brain hemorrhage, or periventricular leukomalacia between the two groups |
Malaria
Malaria is a major health problem in many parts of the world, with an estimated one million deaths annually in severe cases.33 Malarial infection can result in anemia and shock. Blood transfusion may alleviate symptoms in malarial anemia but carries with it various potential risks and can be difficult in areas of limited supply.
The decision whether to transfuse anemia remains difficult. The 2015 WHO malaria guidelines are largely based on expert consensus recommendations and recommend blood transfusion in children with Hb < 50 g/L in high transmission areas or Hb < 70 g/L in low transmission areas.34 Adherence to transfusion guidelines has been found to be generally poor.35
Two RCTs were identified from Tanzania and Gambia in children with malaria. In the first study by Holzer et al., 116 children with malaria who were clinically stable were randomized to receive a blood transfusion versus standard care.36 All patients were treated with chloroquine and mebendazole. Outcomes measured were mortality, increase in packed cell volume, need for additional transfusion readmission, and adverse events. There were fewer deaths in transfused patients, but this was not statistically significant due to small numbers. One adverse effect of transfusion (fever and chills) was reported in a patient who later died (see Table 8).
| Study | Clinical setting | RCT groups and/or outcomes | Conclusions |
|---|---|---|---|
| RCT | |||
| Childhood anemia in Africa36 |
|
|
3.8% of children died in the no transfusion group and 1.9% of children died in the transfused group (P = 0.9) |
| Predictors of mortality in Gambian children with severe malaria anemia37 |
|
|
One child died in the no transfusion group and none in the transfused group |
| Systematic review | |||
| Blood transfusion for treating malarial anemia33 | RCTs comparing transfusion versus conservative treatment/iron therapy in children with malaria who are stable (n = 230) | Main outcomes: death and adverse outcomes |
|
| Malaria: severe, life-threatening38 | RCTs comparing transfusion versus conservative treatment/iron therapy in children with malaria who are stable (n = 230) | Main outcomes: death and adverse outcomes | Studies were small and loss to follow-up was greater than 10%, both of which are potential sources of bias |
In the second RCT by Bojang et al., 114 children with malaria and severe anemia who were clinically stable were randomized to receive RBC transfusion versus standard care with the addition of oral iron.37 The method of randomization was unclear. All patients were treated with chloroquine and sulfadoxine/pyrimethamine. Outcomes measured were mortality, increase in packed cell volume, duration of admission, additional need for transfusion, adverse events, respiratory distress, and effect on parasite load. There were fewer deaths in transfused patients, but this was not statistically significant due to small numbers. Seven patients developed adverse effects attributed to transfusion (convulsion and coma). Ten children from the nontransfused group eventually required blood transfusion. The mean duration of admission was shorter in the transfusion group (2.36 days in the transfusion group versus 4.24 days in the iron group).
Two systematic reviews were identified, which reviewed the two RCTs.38, 33 Both concluded that there was no statistical significance in mortality between the groups, and there was increase in convulsions and coma in the transfused group (7% versus 0%) that was not statistically significant. In both studies, overall numbers were small, not adequately powered, and children with severe anemia (defined as hematocrit <12%, hemorrhage or features of congestive cardiac failure as per expert consensus) were excluded. There was no information regarding the species of malarial parasites in either study. Rates of transfusion-transmitted infection were not discussed. Patients lost to follow-up was high (>10%) and follow-up duration was relatively short (8 weeks).36, 37 The conclusion was that there was insufficient information to recommend routine RBC transfusion in children with malarial anemia who were clinically stable.
General
A 2015 systematic review by Holst et al. comparing transfusion thresholds in adults or children found no difference in mortality, morbidity, and coronary events.39 An updated 2016 Cochrane systematic review by Carson et al. compared liberal versus restrictive transfusion thresholds in 12,587 patients and found no increase or decrease in the risk of 30-day mortality (RR = 0.97), cerebrovascular events, and infection.40 The exposure of patients to RBC transfusion in the restrictive group was reduced by 43% (RR = 0.57). The majority of participants analyzed were from trials involving hip surgery, critical care, gastrointestinal bleeding, and cardiac surgery; and there was only one trial recruiting pediatric participants and two trials recruiting patients with hematological malignancies. Limitations identified in these studies include variations in the definitions of restrictive and liberal values and lack of blinding of physicians due to the low feasibility from clinical requirements in monitoring during the administration of blood components (see Table 9).
| Systematic review | Clinical setting | Outcomes | Conclusions |
|---|---|---|---|
| Restrictive versus liberal transfusion strategy for red blood cell transfusion: systematic review of randomized trials with meta-analysis and trial sequential analysis39 | RCTs comparing transfusion thresholds in all patient groups (adult and children), excluding neonates and children with low birth weight (n = 9813) | Main outcomes: mortality, overall morbidity, or myocardial infarction | No difference in mortality, morbidity, and myocardial infarctions |
| Transfusion thresholds and other strategies for guiding allogeneic red blood cell transfusion40 | RCTs comparing transfusion thresholds in all patient groups (adult and children) (n = 12,587) |
|
No difference in 30-day mortality, cerebrovascular events, and infection |
Discussion
Multiple trials and reviews have attempted to guide appropriate transfusion thresholds in many clinical settings. A variety of randomized trials and systematic reviews have favored the restrictive transfusion threshold approach, but more good quality studies are required in many clinical settings.
Currently, there is reasonably good clinical trial evidence that suggests that a restrictive policy of transfusion at an Hb concentration of 70–80 g/L may be reasonable to guide transfusion decisions in some patient populations. These population groups include most medically and surgically hospitalized and hemodynamically stable patients, patients in the ICU or with septic shock, fit elderly patients post hip surgery, and hemodynamically stable patients with acute gastrointestinal bleeding without cardiovascular comorbidity and with ready access to endoscopic treatment.
More definitive RCTs are required in many other clinical settings to establish best transfusion strategies, including patients post cardiac surgery, pediatric patients, patients with oncological malignancies undergoing active therapy, and patients with malaria. Current evidence is insufficient in these patient populations with regard to recommendations for Hb thresholds. There is insufficient evidence from RCTs in these populations, and variations in transfusion thresholds utilized and heterogeneity of other studies in these populations, which make it difficult for definitive conclusions at this stage.
Of interest are large ongoing RCTs that include the TRACT (Transfusion and Treatment of severe anemia in African children) multicenter RCT aiming to recruit >3000 children with severe uncomplicated malarial anemia, comparing liberal transfusion (30 mL/kg whole blood) versus conservative transfusion (20 mL/kg) versus no transfusion (control); and the Transfusion of Prematures (TOP) Trial, which aims to enroll >1800 extremely low birth weight infants to a liberal or restrictive threshold.41, 42
In addition, a recent study reviewed over 400,000 patient records between 2010 and 2014 in relation to long-term outcomes in hospitalized patients discharged with moderate anemia. There was a rise in the prevalence of anemia at hospital discharge, which was not associated with an increase in mortality, readmission to hospital, and subsequent RBC transfusions within 6 months of discharge.43 The authors considered these findings to support strategies to restrict RBC transfusion and tolerate anemia during and after hospitalization.43
Having useful transfusion thresholds is helpful to improve clinical outcomes for patients, reduce unnecessary transfusion, and improve blood inventory supplies; as well as reduce the economic burden of overtransfusion that includes costs associated with health care expenditure due to transfusion-related adverse outcomes, blood processing costs, and loss of income due to DALYs.
However, these transfusion trials were designed to assist in our understanding of the numerical indications for transfusion in specific patient subgroups and whether performing transfusion at a certain Hb versus another would lead to better patient outcomes, and they do not specifically address evaluation or treatment of the underlying causes of anemia.44 Clearly, the cause or causes of underlying anemia still need to be addressed in individual patients, and targeted treatment for these causes instituted. These may include alternatives to RBC transfusion, such as the replacement of essential cofactors (iron, folate, and vitamin B12), growth factors (erythropoietin), and the treatment of concurrent infection or inflammatory conditions.
While work in establishing appropriate transfusion thresholds in specific patient populations is required, it is important to remember that other factors apart from Hb triggers are important in the clinical setting when deciding whether or not to transfuse a patient. The underlying mechanism for the patient's anemia (whether it is due to chronic or acute blood loss; or whether marrow function is impaired due to malignancy or other intercurrent illness) as well as the tempo of change in the Hb level is important. Importantly, patient characteristics, comorbidities, symptoms, signs, and wishes also need to be incorporated into the clinical decision-making process. The ultimate clinical appropriateness of transfusion requires a multifactorial balancing assessment of Hb concentration, clinical acumen, and patient-related factors.
Acknowledgments
This work was commissioned and financially supported by the Evidence and Programme Guidance Unit, Department of Nutrition for Health and Development of the World Health Organization (WHO), Geneva, Switzerland.
Statement
This manuscript was presented at the World Health Organization (WHO) technical consultation “Use and Interpretation of Haemoglobin Concentrations for Assessing Anaemia Status in Individuals and Populations,” held in Geneva, Switzerland on November 29–30 and December 1, 2017. This paper is being published individually but will be consolidated with other manuscripts as a special issue of Annals of the New York Academy of Sciences, the coordinators of which were Drs. Maria Nieves Garcia-Casal and Sant-Rayn Pasricha. The special issue is the responsibility of the editorial staff of Annals of the New York Academy of Sciences, who delegated to the coordinators preliminary supervision of both technical conformity to the publishing requirements of Annals of the New York Academy of Sciences and general oversight of the scientific merit of each article. The workshop was supported by WHO, the Centers for Disease Control and Prevention (CDC); the United States Agency for International Development (USAID); and the Bill & Melinda Gates Foundation. The authors alone are responsible for the views expressed in this paper; they do not necessarily represent the views, decisions, or policies of the WHO. The opinions expressed in this publication are those of the authors and are not attributable to the sponsors, publisher, or editorial staff of Annals of the New York Academy of Sciences.
Competing interests
The authors declare no competing interests.
