C O M M E N TA RY The AABB recommendations for the Choosing Wisely campaign of the American Board of Internal Medicine Jeannie L. Callum, Jonathan H. Waters, Beth H. Shaz, Steven R. Sloan, and Michael F. Murphy
C
hoosing Wisely is an initiative of the American Board of Internal Medicine Foundation designed to help physicians and patients engage in conversations to reduce overuse of tests and procedures and support physician efforts to help patients make smart and effective care choices. Blood transfusion is the commonest procedure performed in the hospitalized patient.1 Unnecessary use of blood transfusion in the hospitalized patient is common worldwide. Overuse of blood transfusion has also been listed as a Choosing Wisely statement by the American Society of Hematology, the Society of Hospital Medicine, and the Critical Care Societies Collaborative. To support this AABB Choosing Wisely initiative, the AABB developed a set of 10 recommendations with input from AABB committees and the AABB Board of Directors. This list was vetted by numerous AABB members to select the top five statements as required by the American Board of Internal Medicine. As required, all of these statements start with “Don’t.” The development of these statements and commentaries are intended to assist you with the promotion of better patient blood management at your local institution. These statements are intended to prompt non– transfusion medicine physicians to rethink their engrained culture of liberal transfusion practice and prompt patients to question why they are being prescribed blood.
ABBREVIATIONS: INR(s) = international normalized ratio(s); PCC(s) = prothrombin complex concentrate(s); RR = risk ratio. From the AABB, Bethesda, Maryland. Address reprint requests to: Professor Mike Murphy, NHS Blood & Transplant, John Radcliffe Hospital, Oxford OX3 9BQ, United Kingdom; e-mail: [email protected]. Received for publication June 19, 2014; and accepted June 20, 2014. doi: 10.1111/trf.12802 © 2014 AABB TRANSFUSION 2014;54:2344-2352. 2344
TRANSFUSION Volume 54, September 2014
1. Don’t transfuse more units of blood than absolutely necessary A restrictive threshold (7.0-8.0g/dL) should be used for the vast majority of hospitalized, stable patients without evidence of inadequate tissue oxygenation (evidence supports a threshold of 8.0g/dL in patients with existing cardiovascular disease). Transfusion decisions should be influenced by symptoms and hemoglobin (Hb) concentration. Singleunit red blood cell (RBC) transfusions should be the standard for nonbleeding hospitalized patients. Additional units should only be prescribed after reassessment of the patient and their Hb value. A total of 13.8 million units of whole blood and RBCs were transfused in the United States in 2011 equating to 44 units per 1000 population,2 which is considerably higher than in other developed countries such as Australia, Canada, the Netherlands, and the United Kingdom where the rates of RBC transfusion are at least 25% lower. In common with those countries, the use of RBC units is decreasing in the United States; the rate was 48.8 per 1000 population in 2008. In 2011, there were approximately 21 million blood components transfused in the United States. Each transfusion carries risks, although the number of transfusion-related fatalities reported to the US Food and Drug Administration (58 in 2013) and the number of transfusion-related adverse reactions reported to the National Blood Collection and Utilization Survey (51,000 in 2011) remain small in comparison to the total number of transfusions.2,3 There is considerable variation in the use of blood between countries, hospitals, and even clinical teams within the same hospital.4-7 This observation has been documented over many years and in several clinical settings and probably indicates that a substantial amount of blood is being transfused inappropriately. The precise reasons for this variation are uncertain, but they include lack of knowledge about the evidence for the restrictive use of blood and inadequate feedback of comparative data on blood utilization. The number of published clinical practice guidelines for RBC transfusion including those on behalf of the AABB8 and the American College of Physicians9 attest to the interest in appropriate blood utilization.10 The guidelines generally acknowledge the necessity of considering
AABB CHOOSING WISELY CAMPAIGN
patient covariables or other patient-specific criteria for making transfusion decisions, and that RBC transfusion is not of benefit when the Hb is greater than 10.0 g/dL, but when the Hb is less than 6.0 to 7.0 g/dL. However, the selection of a discrete Hb as a “trigger” for transfusion has been controversial. A Cochrane systematic review of prospective randomized trials up to 201111 compared “high” versus “low” Hb thresholds in 19 trials involving a total of 6264 patients. The authors found that restrictive RBC transfusion strategies did not increase the frequency of adverse events such as myocardial infarction, stroke, and pneumonia and were associated with a reduced in-hospital mortality (risk ratio [RR], 0.77; 95% confidence interval [CI], 0.62-0.95), but not 30-day mortality (RR, 0.85; 95% CI, 0.70-1.03). Restrictive transfusion strategies reduced the risk of receiving a RBC transfusion by 39% (RR, 0.61; 95% CI, 0.52-0.72) and the volume of transfusion by 1.19 units (RR, 0.61; 95% CI, 0.53-1.83 units). The authors of the review highlighted the fact that there were no trials in patients with acute coronary syndrome, and the appropriate threshold for patients with existing cardiovascular disease remains uncertain. Since the publication of the Cochrane systematic review, a single-center prospective study12 of patients with upper gastrointestinal bleeding demonstrated that patients randomized to a restrictive (Hb < 7.0 g/dL) versus a liberal (Hb < 9.0 g/dL) threshold for RBC transfusions had significantly improved outcomes, including mortality at 45 days and rates of rebleeding. Two recent systematic reviews provided further support for recommending a restrictive RBC transfusion strategy.13,14 A meta-analysis, including data from the recent trial in upper gastrointestinal bleeding, found that a restrictive RBC transfusion strategy using a Hb of 7.0 g/dL reduced cardiac events, rebleeding, bacterial infections, and in-hospital mortality.13 Another metaanalysis comparing serious infections in more than 7500 patients enrolled in 18 randomized trials found that the rate of hospital-acquired infection was 16.9% in the liberal RBC transfusion group and 11.8% in the restrictive transfusion group (RR, 0.82; 95% CI, 0.72-0.95).14 A recent editorial summarized these new findings and concluded that considerable progress has been made in developing the evidence base for guidelines on the use of RBC transfusions in most clinical settings where the patient is stable; further evidence is required in “high-risk” patients such as those with acute coronary syndrome, generalized vascular disease, and acute brain injury.15 The introduction of a policy for single-unit RBC transfusions in patients without active bleeding, usually combined with a restrictive RBC transfusion strategy, has been found in observational studies to reduce the use of RBC transfusion.16-19 Additional RBC units can be ordered after reassessment of the patient and their Hb concentration. Real-time clinical decision support systems for computer-
ized physician order entry of blood orders have been used to support the implementation of single-unit and restrictive RBC transfusion policies19 and compliance with guidelines for the use of plasma and platelets (PLTs).20,21 Such policies can be further supported by using linked data from the blood bank, laboratory medicine, and hospital billing and physician order entry information systems and displaying it in a simple format for discussion with individual clinicians to encourage improvement in blood utilization.22,23
2. Don’t transfuse RBCs for iron deficiency without hemodynamic instability Blood transfusion has become a routine medical response despite cheaper and safer alternatives in some settings. Preoperative patients with iron deficiency and patients with chronic iron deficiency without hemodynamic instability (even with low Hb levels) should be given oral and/or intravenous (IV) iron. Iron deficiency anemia is the most common cause of anemia in the United States, affecting up to 3% of the population.24 Iron deficiency anemia generally results from chronic blood loss, poor dietary provision of iron, or poor absorption of iron. Chronic iron deficiency anemia occurs in many patient groups but is most common in women of childbearing age. One in 10 premenopausal women are iron deplete (1 in 25 women are also anemic).25 In menstruating females, chronic blood loss through heavy menses can lead to iron deficiency anemia. In pregnant women, the iron needs of the developing fetus as well as expansion of the maternal blood volume leads to exhaustion of the maternal iron reserves. After childbirth, the excess blood volume is reabsorbed, partially restoring iron stores. If hemorrhage occurs during birth, then this excess blood is not available to restore iron stores. Iron deficiency and iron deficiency anemia can persist for years after obstetric hemorrhage.26 Other populations commonly affected by iron deficiency anemia are the very young and the very old. In older individuals, iron deficiency anemia can be due to an iron-deplete diet or poor absorption; however, it is often related to occult gastrointestinal bleeding from an upper gastrointestinal source or colon polyps or cancer.27 Iron deficiency anemia is also common in the pediatric population with up to 4% to 8% of children between 12 and 36 months having iron deficiency anemia,28,29 typically related to an iron-poor diet where iron-poor milk is the primary caloric source. Iron deficiency in children can lead to poor growth and cognitive and motor developmental delay. Less than half of pediatric patients with iron deficiency anemia presenting to hospital are prescribed iron therapy.30,31 In adults, one study estimated one-quarter of transfusions for iron deficiency anemia were unnecessary Volume 54, September 2014 TRANSFUSION
2345
CALLUM ET AL.
and one-third of patients did not receive oral iron on discharge.32 In elderly patients presenting to hospital with anemia, only one-third undergo testing to determine if iron deficiency is a causative factor.33 The management of iron deficiency anemia with RBC transfusion is unfortunately all too common. One study reported that 39% of patients admitted to a tertiary care hospital with iron deficiency anemia were transfused.34 While patients presenting with severe iron deficiency anemia with symptoms of inadequate oxygen delivery (e.g., syncope, chest pain) will likely benefit from transfusion, patients with severe anemia (often Hb levels well below 7.0 g/dL) usually only have fatigue and do not require transfusion. While acute blood loss is frequently associated with hemodynamic instability and poor oxygen delivery, chronic anemia is not associated with hypovolemia, and oxygen delivery is facilitated by increases in 2,3-diphosphoglycerate and a shift in the oxygen dissociation curve. If a decision is made to transfuse a patient for iron deficiency anemia, often a single unit of RBCs is sufficient and further increases in Hb can be facilitated with oral and/or IV iron. While RBC transfusion transiently increases the Hb value, it neglects underlying mechanisms causing the anemia. Blindly transfusing a patient neglects the cause of anemia. The importance of understanding the mechanism by which the patient is anemic is illustrated by an example of the total joint replacement patient who has been chronically taking nonsteroidal anti-inflammatory drugs. Gastric ulcers are expected in 15% to 35% of this population.35 After joint replacement, anticoagulation is routine to prevent thrombotic complications. Anticoagulation in a patient with a bleeding gastric ulcer could be catastrophic. Patients with iron deficiency anemia need appropriate follow-up to ensure that they are adequately investigated,36 the therapy is tolerated, and the anemia is fully corrected. RBC transfusion for the iron-deficient patient is not only bad from a diagnostic perspective, but is also not cost-effective. The cost of transfusion varies, depending upon how it is calculated. Acquisition of a unit of RBCs typically costs $225;2 however, if all associated costs are accounted, a cost of $761 has been estimated.37 Oral iron replacement is cheap (ferrous sulfate costs as little as $0.05 per tablet). Anemia correction should take place with 50 to 60 mg of elemental iron two to three times daily.38 Oral iron is maximally absorbed if prescribed on an empty stomach (e.g., at bedtime) and with a source of vitamin C. A reticulocyte response can be anticipated within 3 to 4 days after the initiation of therapy.39 If a more rapid increase in Hb is required due either to severe anemia or to a planned surgical procedure in the next 4 to 6 weeks, IV iron can be prescribed. Newer formulations of IV iron are easy to administer and do not have the adverse effect profile that has led to the clinical reluctance to administer IV iron in the past. Additionally, some patients 2346
TRANSFUSION Volume 54, September 2014
cannot tolerate oral iron and other patients cannot absorb iron and require IV therapy. Fundamentally, transfusion of a hemodynamically stable patient with chronic iron deficiency anemia is bad for the patient and bad for health care economics. As such, transfusion is not recommended for iron deficiency or iron deficiency anemia in a hemodynamically stable patient irrespective of his or her Hb level.
3. Don’t routinely use blood products to reverse warfarin Patients requiring reversal of warfarin can often be reversed with vitamin K alone. Prothrombin complex concentrates (PCCs) or plasma should only be used for patients with serious bleeding or requiring emergency surgery. The risk of major bleeding on warfarin, a commonly prescribed drug, is estimated at 1% to 3% per year.40,41 Hence, the reversal of warfarin for supratherapeutic international normalized ratios (INRs), before surgical procedures (urgent or elective), and for hemorrhagic symptoms is a commonly encountered clinical situation. Vitamin K is the foundation of therapy for the vast majority of patients requiring reversal and frequently the only treatment required. Reversal with the human-derived blood product PCCs, or plasma when PCCs are unavailable, should be restricted to use in patients with major hemorrhage or requiring medically urgent surgery within 6 hours. In the setting of emergency reversal, PCCs are the preferred product due their smaller volume (reduced risk of circulatory overload), faster infusion time, and superior INR correction.42 PCCs (or plasma) should not be used for elective reversal or nonurgent surgical procedures due to the risk of adverse reactions, risk of thromboembolic complications, and cost. A substantial proportion of plasma (or PCCs) continues to be transfused unnecessarily for warfarin reversal,43 of which a substantial proportion could be mitigated with prospective monitoring by the transfusion service of all plasma (or PCCs) requests.44 It should be noted that vitamin K antagonists only inhibit the gamma-carboxylation of Factor (F)II, FVII, F IX, and FX, thus preventing these proteins from binding calcium ions. The factors are still manufactured in the presence of vitamin K antagonists, and only this final step in the assembly of these factors is blocked (i.e., these proteins are already synthesized and do not require transcription for reversal of the anticoagulant effect). Hence, the administration of IV vitamin K in urgent situations allows for substantial reversal within 6 hours. There are three situations where vitamin K should be used alone in the reversal of warfarin effect: 1) supratherapeutic INR without bleeding, 2) before nonemergency surgical procedures, and 3) for minor
AABB CHOOSING WISELY CAMPAIGN
bleeding events. Only oral and IV vitamin K routes are appropriate. Both subcutaneous and intramuscular routes should be avoided. Asymptomatic supratherapeutic INRs are commonly encountered with warfarin due to its narrow therapeutic window, unpredictable dose response, and numerous drug and food interactions. Due to a lack of well-designed clinical trials, there is a lack of consensus on the level of INR at which a patient should be offered vitamin K (and the dose and route of vitamin K that should be prescribed). It is clear that an INR in excess of 5 to 6 is associated with an increased risk of bleeding,45,46 although the absolute risk is small (single INR between 5 and 9 is associated with a 0.95%/month absolute risk47). There are also logistic issues that must be considered: lack of availability of an oral vitamin K tablet for patients to keep on hand for such events and hence the need for the patient to travel to a hospital or local emergency department to obtain vitamin K, which can be risky for an elderly overanticoagulated patient who may be at risk for falls, especially with inclement weather. Vitamin K is recommended by the Australian and New Zealand guidelines if the INR exceeds 10 (at a dose of 3-5 mg orally or IV) or the INR is 4.5 to 10 and the patient is considered high risk for bleeding (major bleed in the last 4 weeks, major surgery in the past 2 weeks, PLT count below 50 × 109/L, liver disease or concurrent anti-PLT therapy) at a lower dose (1-2 mg orally or 0.5-1 mg IV).48 The American College of Chest Physicians recommends against the use of vitamin K for asymptomatic patients if the INR is under 10 and recommends the use of oral vitamin K (dose not specified) above an INR of 10.49 The British Committee for Standards in Haematology took a middle road and recommended oral vitamin K (1-5 mg) if the INR exceeds 8.50 The INR should be checked the following day to determine if additional doses of vitamin K are required. There is consensus from the existing literature and across current guidelines that PCCs (or plasma) should not be prescribed for asymptomatic patients with supratherapeutic INRs. The second common scenario requiring warfarin reversal involves patients requiring nonemergency surgical procedures. IV vitamin K is faster at reversing the INR than oral vitamin K, with significant reversal seen at 6 hours in most patients.51,52 When faced with an urgent surgical procedure, first decide if reversal is required for the planned procedure. Second, decide if there is a medical need for the procedure to be performed within the next 6 hours (not just an opening in the operating room that is convenient for the surgical team). Warfarinized patients requiring medically emergent procedures, such as removal of a ruptured spleen after traumatic injury or ruptured aneurysm, require both IV vitamin K and PCCs (or plasma) for immediate surgery. There are numerous patients requiring nonemergency procedures where an immediate surgical procedure is medically unnecessary
and there is time to wait for the INR to be reversed with IV vitamin K. Examples include hip fracture, bowel obstruction, and incarcerated hernia. If the procedure is not an emergency, but should occur as soon as possible, administer IV vitamin K as soon as it is identified that the patient needs to proceed to the operating room. There is no solid evidence to guide what dose should be administered.53 However, if complete (or near-complete) reversal of warfarin is required, administration of 5 to 10 mg is more likely to achieve success than a lower dose. Furthermore, the common misconception that subsequent “warfarin resistance” will develop with vitamin K is not commonly encountered in clinical practice. The time to administration (fast) and route of administration (IV) are likely more important than the dose prescribed in expediting a surgical procedure.54 Retrospective reviews of the management of warfarin reversal have identified substantial room for improvement, with unnecessary delays and huge variability in care.53-56 It is clear from these studies that a substantial number of patients could completely avoid blood product exposure with better planning by the clinical team. Patients booked for elective surgery should never require reversal with PCCs (or plasma) with appropriately planned and monitored reversal. Patients who present to the emergency department with noncritical bleeding that does not warrant immediate reversal with PCCs (or plasma) can be managed with IV vitamin K. Examples of noncritical bleeding include epistaxis, bleeding from dental procedures, hematuria, and hemorrhoid bleeding. First, determine if there is a need to reverse warfarin. For example, bleeding from the oral cavity can often be managed with tranexamic acid mouth wash. If reversal is required, the British Committee for Standards in Haematology recommends a dose of 1 to 3 mg IV,50 and the Australian and New Zealand guidelines recommend a dose of 1 to 2 mg orally or 0.5 to 1 mg IV.48 Neither recommends the use of PCCs (or plasma) for this clinical situation. Some clinicians continue to actively avoid the use of IV vitamin K (and hence rely heavily on plasma or PCCs) due to concerns regarding the risk of anaphylaxis. This concern is not warranted and results in delays in INR reversal (due to the use of the oral route) and hence time to surgery. The risk of anaphylaxis is very low (0.04-11 per 10,000 doses).57 Historically, the vitamin K formulations contained polyethoxylated castor oil, which may have triggered the early reports of anaphylaxis that is no longer a clinical concern.
4. Don’t perform serial blood counts on clinically stable patients Blood counts should only be obtained on hospitalized patients when there is reason to believe that a new clinically significant abnormality will be detected. For stable Volume 54, September 2014 TRANSFUSION
2347
CALLUM ET AL.
patients, serial blood counts are unlikely to detect clinically significant abnormalities but can contribute to iatrogenic anemia. In a stable patient, serial Hb and hematocrit (Hct) measurements are not useful for determining whether a RBC transfusion is indicated because stable patients do not need RBC transfusions based on laboratory values alone. Additionally, frequent laboratory blood tests contribute to anemia in hospitalized patients. One study of internal medicine patients at Toronto General Hospital found that the mean volume of blood phlebotomized during a hospitalization was 74.6 mL.58 After adjusting for other factors, the volume of phlebotomy strongly predicted decreases in Hb and Hct with every 100 mL of phlebotomy being associated with decreases in Hb and Hct of 0.7 g/dL and 1.9%, respectively. Another study of patients treated for acute myocardial infarction at 57 hospitals found phlebotomy blood losses to be an independent risk factor for clinically significant anemia.59 Indeed, 20.1% of these patients developed moderate to severe anemia that was associated with worse outcomes. While more phlebotomy blood loss occurred on the first 2 days of hospitalization, significant phlebotomy blood loss occurred each subsequent day of hospitalization, which was likely due to “routine, scheduled laboratory draws.” There was significant variability in hospital practice with phlebotomy blood losses at some hospitals being up to twice those of other hospitals. Hospitals that have introduced strategies to reduce blood tests have not seen any adverse changes in clinical outcomes while saving money. Vanderbilt University Hospital modified their hospital computer system to reduce laboratory test ordering.60 They reduced several common laboratory tests including a significant 15% reduction in complete blood counts. With this reduction, there was no change in the median length of stay or percentages of patients experiencing readmission, intensive care unit transfer, or mortality. Similarly, a large Israeli academic medical center found no change in readmission rates or diagnoses of anemia when they used educational efforts to reduce Hb measurements from 1.97 per admission to 1.26 per admission.61 Reductions of this type also yield significant cost savings.61,62 While unstable patients may require serial blood counts, these should also be minimized. In particular, one group of unstable patients, those with hypoproliferative thrombocytopenia, usually require only one PLT count per day. Most studies of prophylactic PLT transfusions have used a single-morning PLT count each day to determine whether a PLT transfusion was indicated.63-65 No studies have tested the utility of a second PLT count during the day for clinically stable patients with thrombocytopenia. Additionally, using a PLT count later in the day would be difficult to interpret since PLT counts, like most blood cell counts, have a diurnal variation.66 Hence, when 2348
TRANSFUSION Volume 54, September 2014
a sample is not collected in the morning, it is not clear what PLT count value should be used to trigger a PLT transfusion.
5. Don’t transfuse O– blood except to O– patients and in emergencies for women of childbearing potential with unknown blood group O– blood units are in chronic short supply due in part to overutilization for patients who are not O–. O– RBCs should be restricted to: 1) O– patients or 2) women of childbearing potential with unknown blood group who require emergency transfusion before blood group testing can be performed. Group O D– RBCs are in short supply. Group O RBC units are the universal type and can be transfused to patients of all ABO blood types. Furthermore, group O D– are often used as the universal type to prevent anti-D formation. However, the majority of transfused individuals do not form anti-D. The only population with significant consequences after anti-D formation are women of childbearing potential because anti-D can result in hemolytic disease of the fetus and newborn. Appropriate use of this rare blood type will enable its use in the appropriate patients when it is needed. In the United States, 49.4% of the 13,785,000 RBC transfusions were group O; 39.8% O+ and 9.6% O–.2 The demand for group O RBCs in general and O– RBCs specifically is highlighted by the number of days of available supply, O+ 2 to 8 days and O– 2 to 4 days, and that only 0.5% of group O RBC units processed were outdated compared to 2.1% of all allogeneic units. A total of 45% of whites and 49% of African Americans are group O and 16% of whites and 7% of African Americans are D–, leading to significant shortages since 87% of donors are white.67,68 Although blood suppliers target recruitment of group O donors, particularly O– donors, these units continue to be in short supply due to increased demand because many institutions use O– units for emergency transfusion when the patient’s blood type is unknown rather than using O+ units for males and females who are not of childbearing potential, typically defined as older than 45 to 50 years old.69,70 Institutions should establish policies for emergency release and delivery of blood products. There should also be protocols for administration of D+ RBCs to a D– or unknown patient.69 The frequency of anti-D formation after transfusion of D+ blood products to a D– patient is approximately 20% for RBCs and less than 4% for PLTs (recently reported as 0% for apheresis PLTs).71-73 It is especially important to prevent anti-D formation in females of childbearing potential because anti-D can cause hemolytic disease of the fetus and newborn in future pregnancies. Therefore, females of childbearing potential should receive D– RBCs. However, institutions may transfuse
AABB CHOOSING WISELY CAMPAIGN
D+ RBCs to a D–/D-unknown female of childbearing potential after a certain number (such as 8 units) of D– RBCs have already been transfused given the balance of available inventory and the likelihood of survival. In addition, RhIG might be considered to prevent anti-D alloimmunization in D– patients receiving D+ blood products. The practice of giving RhIG is more common when D+ PLTs are given to D– patients. The risk of hemolysis through the administration of RhIG needs to be weighed against the benefit of preventing alloimmunization, especially when RhIG is given to a patient receiving more than 1 or 2 units of D+ RBC.74 Supporting this Choosing Wisely statement is the National Health Service’s guideline from the United Kingdom, “The appropriate use of group O D– RBCs,” which determined that D– RBCs were mandatory indicated in O D– patients with anti-D, O D– females with childbearing potential, and in emergency to premenopausal (
C
hoosing Wisely is an initiative of the American Board of Internal Medicine Foundation designed to help physicians and patients engage in conversations to reduce overuse of tests and procedures and support physician efforts to help patients make smart and effective care choices. Blood transfusion is the commonest procedure performed in the hospitalized patient.1 Unnecessary use of blood transfusion in the hospitalized patient is common worldwide. Overuse of blood transfusion has also been listed as a Choosing Wisely statement by the American Society of Hematology, the Society of Hospital Medicine, and the Critical Care Societies Collaborative. To support this AABB Choosing Wisely initiative, the AABB developed a set of 10 recommendations with input from AABB committees and the AABB Board of Directors. This list was vetted by numerous AABB members to select the top five statements as required by the American Board of Internal Medicine. As required, all of these statements start with “Don’t.” The development of these statements and commentaries are intended to assist you with the promotion of better patient blood management at your local institution. These statements are intended to prompt non– transfusion medicine physicians to rethink their engrained culture of liberal transfusion practice and prompt patients to question why they are being prescribed blood.
ABBREVIATIONS: INR(s) = international normalized ratio(s); PCC(s) = prothrombin complex concentrate(s); RR = risk ratio. From the AABB, Bethesda, Maryland. Address reprint requests to: Professor Mike Murphy, NHS Blood & Transplant, John Radcliffe Hospital, Oxford OX3 9BQ, United Kingdom; e-mail: [email protected]. Received for publication June 19, 2014; and accepted June 20, 2014. doi: 10.1111/trf.12802 © 2014 AABB TRANSFUSION 2014;54:2344-2352. 2344
TRANSFUSION Volume 54, September 2014
1. Don’t transfuse more units of blood than absolutely necessary A restrictive threshold (7.0-8.0g/dL) should be used for the vast majority of hospitalized, stable patients without evidence of inadequate tissue oxygenation (evidence supports a threshold of 8.0g/dL in patients with existing cardiovascular disease). Transfusion decisions should be influenced by symptoms and hemoglobin (Hb) concentration. Singleunit red blood cell (RBC) transfusions should be the standard for nonbleeding hospitalized patients. Additional units should only be prescribed after reassessment of the patient and their Hb value. A total of 13.8 million units of whole blood and RBCs were transfused in the United States in 2011 equating to 44 units per 1000 population,2 which is considerably higher than in other developed countries such as Australia, Canada, the Netherlands, and the United Kingdom where the rates of RBC transfusion are at least 25% lower. In common with those countries, the use of RBC units is decreasing in the United States; the rate was 48.8 per 1000 population in 2008. In 2011, there were approximately 21 million blood components transfused in the United States. Each transfusion carries risks, although the number of transfusion-related fatalities reported to the US Food and Drug Administration (58 in 2013) and the number of transfusion-related adverse reactions reported to the National Blood Collection and Utilization Survey (51,000 in 2011) remain small in comparison to the total number of transfusions.2,3 There is considerable variation in the use of blood between countries, hospitals, and even clinical teams within the same hospital.4-7 This observation has been documented over many years and in several clinical settings and probably indicates that a substantial amount of blood is being transfused inappropriately. The precise reasons for this variation are uncertain, but they include lack of knowledge about the evidence for the restrictive use of blood and inadequate feedback of comparative data on blood utilization. The number of published clinical practice guidelines for RBC transfusion including those on behalf of the AABB8 and the American College of Physicians9 attest to the interest in appropriate blood utilization.10 The guidelines generally acknowledge the necessity of considering
AABB CHOOSING WISELY CAMPAIGN
patient covariables or other patient-specific criteria for making transfusion decisions, and that RBC transfusion is not of benefit when the Hb is greater than 10.0 g/dL, but when the Hb is less than 6.0 to 7.0 g/dL. However, the selection of a discrete Hb as a “trigger” for transfusion has been controversial. A Cochrane systematic review of prospective randomized trials up to 201111 compared “high” versus “low” Hb thresholds in 19 trials involving a total of 6264 patients. The authors found that restrictive RBC transfusion strategies did not increase the frequency of adverse events such as myocardial infarction, stroke, and pneumonia and were associated with a reduced in-hospital mortality (risk ratio [RR], 0.77; 95% confidence interval [CI], 0.62-0.95), but not 30-day mortality (RR, 0.85; 95% CI, 0.70-1.03). Restrictive transfusion strategies reduced the risk of receiving a RBC transfusion by 39% (RR, 0.61; 95% CI, 0.52-0.72) and the volume of transfusion by 1.19 units (RR, 0.61; 95% CI, 0.53-1.83 units). The authors of the review highlighted the fact that there were no trials in patients with acute coronary syndrome, and the appropriate threshold for patients with existing cardiovascular disease remains uncertain. Since the publication of the Cochrane systematic review, a single-center prospective study12 of patients with upper gastrointestinal bleeding demonstrated that patients randomized to a restrictive (Hb < 7.0 g/dL) versus a liberal (Hb < 9.0 g/dL) threshold for RBC transfusions had significantly improved outcomes, including mortality at 45 days and rates of rebleeding. Two recent systematic reviews provided further support for recommending a restrictive RBC transfusion strategy.13,14 A meta-analysis, including data from the recent trial in upper gastrointestinal bleeding, found that a restrictive RBC transfusion strategy using a Hb of 7.0 g/dL reduced cardiac events, rebleeding, bacterial infections, and in-hospital mortality.13 Another metaanalysis comparing serious infections in more than 7500 patients enrolled in 18 randomized trials found that the rate of hospital-acquired infection was 16.9% in the liberal RBC transfusion group and 11.8% in the restrictive transfusion group (RR, 0.82; 95% CI, 0.72-0.95).14 A recent editorial summarized these new findings and concluded that considerable progress has been made in developing the evidence base for guidelines on the use of RBC transfusions in most clinical settings where the patient is stable; further evidence is required in “high-risk” patients such as those with acute coronary syndrome, generalized vascular disease, and acute brain injury.15 The introduction of a policy for single-unit RBC transfusions in patients without active bleeding, usually combined with a restrictive RBC transfusion strategy, has been found in observational studies to reduce the use of RBC transfusion.16-19 Additional RBC units can be ordered after reassessment of the patient and their Hb concentration. Real-time clinical decision support systems for computer-
ized physician order entry of blood orders have been used to support the implementation of single-unit and restrictive RBC transfusion policies19 and compliance with guidelines for the use of plasma and platelets (PLTs).20,21 Such policies can be further supported by using linked data from the blood bank, laboratory medicine, and hospital billing and physician order entry information systems and displaying it in a simple format for discussion with individual clinicians to encourage improvement in blood utilization.22,23
2. Don’t transfuse RBCs for iron deficiency without hemodynamic instability Blood transfusion has become a routine medical response despite cheaper and safer alternatives in some settings. Preoperative patients with iron deficiency and patients with chronic iron deficiency without hemodynamic instability (even with low Hb levels) should be given oral and/or intravenous (IV) iron. Iron deficiency anemia is the most common cause of anemia in the United States, affecting up to 3% of the population.24 Iron deficiency anemia generally results from chronic blood loss, poor dietary provision of iron, or poor absorption of iron. Chronic iron deficiency anemia occurs in many patient groups but is most common in women of childbearing age. One in 10 premenopausal women are iron deplete (1 in 25 women are also anemic).25 In menstruating females, chronic blood loss through heavy menses can lead to iron deficiency anemia. In pregnant women, the iron needs of the developing fetus as well as expansion of the maternal blood volume leads to exhaustion of the maternal iron reserves. After childbirth, the excess blood volume is reabsorbed, partially restoring iron stores. If hemorrhage occurs during birth, then this excess blood is not available to restore iron stores. Iron deficiency and iron deficiency anemia can persist for years after obstetric hemorrhage.26 Other populations commonly affected by iron deficiency anemia are the very young and the very old. In older individuals, iron deficiency anemia can be due to an iron-deplete diet or poor absorption; however, it is often related to occult gastrointestinal bleeding from an upper gastrointestinal source or colon polyps or cancer.27 Iron deficiency anemia is also common in the pediatric population with up to 4% to 8% of children between 12 and 36 months having iron deficiency anemia,28,29 typically related to an iron-poor diet where iron-poor milk is the primary caloric source. Iron deficiency in children can lead to poor growth and cognitive and motor developmental delay. Less than half of pediatric patients with iron deficiency anemia presenting to hospital are prescribed iron therapy.30,31 In adults, one study estimated one-quarter of transfusions for iron deficiency anemia were unnecessary Volume 54, September 2014 TRANSFUSION
2345
CALLUM ET AL.
and one-third of patients did not receive oral iron on discharge.32 In elderly patients presenting to hospital with anemia, only one-third undergo testing to determine if iron deficiency is a causative factor.33 The management of iron deficiency anemia with RBC transfusion is unfortunately all too common. One study reported that 39% of patients admitted to a tertiary care hospital with iron deficiency anemia were transfused.34 While patients presenting with severe iron deficiency anemia with symptoms of inadequate oxygen delivery (e.g., syncope, chest pain) will likely benefit from transfusion, patients with severe anemia (often Hb levels well below 7.0 g/dL) usually only have fatigue and do not require transfusion. While acute blood loss is frequently associated with hemodynamic instability and poor oxygen delivery, chronic anemia is not associated with hypovolemia, and oxygen delivery is facilitated by increases in 2,3-diphosphoglycerate and a shift in the oxygen dissociation curve. If a decision is made to transfuse a patient for iron deficiency anemia, often a single unit of RBCs is sufficient and further increases in Hb can be facilitated with oral and/or IV iron. While RBC transfusion transiently increases the Hb value, it neglects underlying mechanisms causing the anemia. Blindly transfusing a patient neglects the cause of anemia. The importance of understanding the mechanism by which the patient is anemic is illustrated by an example of the total joint replacement patient who has been chronically taking nonsteroidal anti-inflammatory drugs. Gastric ulcers are expected in 15% to 35% of this population.35 After joint replacement, anticoagulation is routine to prevent thrombotic complications. Anticoagulation in a patient with a bleeding gastric ulcer could be catastrophic. Patients with iron deficiency anemia need appropriate follow-up to ensure that they are adequately investigated,36 the therapy is tolerated, and the anemia is fully corrected. RBC transfusion for the iron-deficient patient is not only bad from a diagnostic perspective, but is also not cost-effective. The cost of transfusion varies, depending upon how it is calculated. Acquisition of a unit of RBCs typically costs $225;2 however, if all associated costs are accounted, a cost of $761 has been estimated.37 Oral iron replacement is cheap (ferrous sulfate costs as little as $0.05 per tablet). Anemia correction should take place with 50 to 60 mg of elemental iron two to three times daily.38 Oral iron is maximally absorbed if prescribed on an empty stomach (e.g., at bedtime) and with a source of vitamin C. A reticulocyte response can be anticipated within 3 to 4 days after the initiation of therapy.39 If a more rapid increase in Hb is required due either to severe anemia or to a planned surgical procedure in the next 4 to 6 weeks, IV iron can be prescribed. Newer formulations of IV iron are easy to administer and do not have the adverse effect profile that has led to the clinical reluctance to administer IV iron in the past. Additionally, some patients 2346
TRANSFUSION Volume 54, September 2014
cannot tolerate oral iron and other patients cannot absorb iron and require IV therapy. Fundamentally, transfusion of a hemodynamically stable patient with chronic iron deficiency anemia is bad for the patient and bad for health care economics. As such, transfusion is not recommended for iron deficiency or iron deficiency anemia in a hemodynamically stable patient irrespective of his or her Hb level.
3. Don’t routinely use blood products to reverse warfarin Patients requiring reversal of warfarin can often be reversed with vitamin K alone. Prothrombin complex concentrates (PCCs) or plasma should only be used for patients with serious bleeding or requiring emergency surgery. The risk of major bleeding on warfarin, a commonly prescribed drug, is estimated at 1% to 3% per year.40,41 Hence, the reversal of warfarin for supratherapeutic international normalized ratios (INRs), before surgical procedures (urgent or elective), and for hemorrhagic symptoms is a commonly encountered clinical situation. Vitamin K is the foundation of therapy for the vast majority of patients requiring reversal and frequently the only treatment required. Reversal with the human-derived blood product PCCs, or plasma when PCCs are unavailable, should be restricted to use in patients with major hemorrhage or requiring medically urgent surgery within 6 hours. In the setting of emergency reversal, PCCs are the preferred product due their smaller volume (reduced risk of circulatory overload), faster infusion time, and superior INR correction.42 PCCs (or plasma) should not be used for elective reversal or nonurgent surgical procedures due to the risk of adverse reactions, risk of thromboembolic complications, and cost. A substantial proportion of plasma (or PCCs) continues to be transfused unnecessarily for warfarin reversal,43 of which a substantial proportion could be mitigated with prospective monitoring by the transfusion service of all plasma (or PCCs) requests.44 It should be noted that vitamin K antagonists only inhibit the gamma-carboxylation of Factor (F)II, FVII, F IX, and FX, thus preventing these proteins from binding calcium ions. The factors are still manufactured in the presence of vitamin K antagonists, and only this final step in the assembly of these factors is blocked (i.e., these proteins are already synthesized and do not require transcription for reversal of the anticoagulant effect). Hence, the administration of IV vitamin K in urgent situations allows for substantial reversal within 6 hours. There are three situations where vitamin K should be used alone in the reversal of warfarin effect: 1) supratherapeutic INR without bleeding, 2) before nonemergency surgical procedures, and 3) for minor
AABB CHOOSING WISELY CAMPAIGN
bleeding events. Only oral and IV vitamin K routes are appropriate. Both subcutaneous and intramuscular routes should be avoided. Asymptomatic supratherapeutic INRs are commonly encountered with warfarin due to its narrow therapeutic window, unpredictable dose response, and numerous drug and food interactions. Due to a lack of well-designed clinical trials, there is a lack of consensus on the level of INR at which a patient should be offered vitamin K (and the dose and route of vitamin K that should be prescribed). It is clear that an INR in excess of 5 to 6 is associated with an increased risk of bleeding,45,46 although the absolute risk is small (single INR between 5 and 9 is associated with a 0.95%/month absolute risk47). There are also logistic issues that must be considered: lack of availability of an oral vitamin K tablet for patients to keep on hand for such events and hence the need for the patient to travel to a hospital or local emergency department to obtain vitamin K, which can be risky for an elderly overanticoagulated patient who may be at risk for falls, especially with inclement weather. Vitamin K is recommended by the Australian and New Zealand guidelines if the INR exceeds 10 (at a dose of 3-5 mg orally or IV) or the INR is 4.5 to 10 and the patient is considered high risk for bleeding (major bleed in the last 4 weeks, major surgery in the past 2 weeks, PLT count below 50 × 109/L, liver disease or concurrent anti-PLT therapy) at a lower dose (1-2 mg orally or 0.5-1 mg IV).48 The American College of Chest Physicians recommends against the use of vitamin K for asymptomatic patients if the INR is under 10 and recommends the use of oral vitamin K (dose not specified) above an INR of 10.49 The British Committee for Standards in Haematology took a middle road and recommended oral vitamin K (1-5 mg) if the INR exceeds 8.50 The INR should be checked the following day to determine if additional doses of vitamin K are required. There is consensus from the existing literature and across current guidelines that PCCs (or plasma) should not be prescribed for asymptomatic patients with supratherapeutic INRs. The second common scenario requiring warfarin reversal involves patients requiring nonemergency surgical procedures. IV vitamin K is faster at reversing the INR than oral vitamin K, with significant reversal seen at 6 hours in most patients.51,52 When faced with an urgent surgical procedure, first decide if reversal is required for the planned procedure. Second, decide if there is a medical need for the procedure to be performed within the next 6 hours (not just an opening in the operating room that is convenient for the surgical team). Warfarinized patients requiring medically emergent procedures, such as removal of a ruptured spleen after traumatic injury or ruptured aneurysm, require both IV vitamin K and PCCs (or plasma) for immediate surgery. There are numerous patients requiring nonemergency procedures where an immediate surgical procedure is medically unnecessary
and there is time to wait for the INR to be reversed with IV vitamin K. Examples include hip fracture, bowel obstruction, and incarcerated hernia. If the procedure is not an emergency, but should occur as soon as possible, administer IV vitamin K as soon as it is identified that the patient needs to proceed to the operating room. There is no solid evidence to guide what dose should be administered.53 However, if complete (or near-complete) reversal of warfarin is required, administration of 5 to 10 mg is more likely to achieve success than a lower dose. Furthermore, the common misconception that subsequent “warfarin resistance” will develop with vitamin K is not commonly encountered in clinical practice. The time to administration (fast) and route of administration (IV) are likely more important than the dose prescribed in expediting a surgical procedure.54 Retrospective reviews of the management of warfarin reversal have identified substantial room for improvement, with unnecessary delays and huge variability in care.53-56 It is clear from these studies that a substantial number of patients could completely avoid blood product exposure with better planning by the clinical team. Patients booked for elective surgery should never require reversal with PCCs (or plasma) with appropriately planned and monitored reversal. Patients who present to the emergency department with noncritical bleeding that does not warrant immediate reversal with PCCs (or plasma) can be managed with IV vitamin K. Examples of noncritical bleeding include epistaxis, bleeding from dental procedures, hematuria, and hemorrhoid bleeding. First, determine if there is a need to reverse warfarin. For example, bleeding from the oral cavity can often be managed with tranexamic acid mouth wash. If reversal is required, the British Committee for Standards in Haematology recommends a dose of 1 to 3 mg IV,50 and the Australian and New Zealand guidelines recommend a dose of 1 to 2 mg orally or 0.5 to 1 mg IV.48 Neither recommends the use of PCCs (or plasma) for this clinical situation. Some clinicians continue to actively avoid the use of IV vitamin K (and hence rely heavily on plasma or PCCs) due to concerns regarding the risk of anaphylaxis. This concern is not warranted and results in delays in INR reversal (due to the use of the oral route) and hence time to surgery. The risk of anaphylaxis is very low (0.04-11 per 10,000 doses).57 Historically, the vitamin K formulations contained polyethoxylated castor oil, which may have triggered the early reports of anaphylaxis that is no longer a clinical concern.
4. Don’t perform serial blood counts on clinically stable patients Blood counts should only be obtained on hospitalized patients when there is reason to believe that a new clinically significant abnormality will be detected. For stable Volume 54, September 2014 TRANSFUSION
2347
CALLUM ET AL.
patients, serial blood counts are unlikely to detect clinically significant abnormalities but can contribute to iatrogenic anemia. In a stable patient, serial Hb and hematocrit (Hct) measurements are not useful for determining whether a RBC transfusion is indicated because stable patients do not need RBC transfusions based on laboratory values alone. Additionally, frequent laboratory blood tests contribute to anemia in hospitalized patients. One study of internal medicine patients at Toronto General Hospital found that the mean volume of blood phlebotomized during a hospitalization was 74.6 mL.58 After adjusting for other factors, the volume of phlebotomy strongly predicted decreases in Hb and Hct with every 100 mL of phlebotomy being associated with decreases in Hb and Hct of 0.7 g/dL and 1.9%, respectively. Another study of patients treated for acute myocardial infarction at 57 hospitals found phlebotomy blood losses to be an independent risk factor for clinically significant anemia.59 Indeed, 20.1% of these patients developed moderate to severe anemia that was associated with worse outcomes. While more phlebotomy blood loss occurred on the first 2 days of hospitalization, significant phlebotomy blood loss occurred each subsequent day of hospitalization, which was likely due to “routine, scheduled laboratory draws.” There was significant variability in hospital practice with phlebotomy blood losses at some hospitals being up to twice those of other hospitals. Hospitals that have introduced strategies to reduce blood tests have not seen any adverse changes in clinical outcomes while saving money. Vanderbilt University Hospital modified their hospital computer system to reduce laboratory test ordering.60 They reduced several common laboratory tests including a significant 15% reduction in complete blood counts. With this reduction, there was no change in the median length of stay or percentages of patients experiencing readmission, intensive care unit transfer, or mortality. Similarly, a large Israeli academic medical center found no change in readmission rates or diagnoses of anemia when they used educational efforts to reduce Hb measurements from 1.97 per admission to 1.26 per admission.61 Reductions of this type also yield significant cost savings.61,62 While unstable patients may require serial blood counts, these should also be minimized. In particular, one group of unstable patients, those with hypoproliferative thrombocytopenia, usually require only one PLT count per day. Most studies of prophylactic PLT transfusions have used a single-morning PLT count each day to determine whether a PLT transfusion was indicated.63-65 No studies have tested the utility of a second PLT count during the day for clinically stable patients with thrombocytopenia. Additionally, using a PLT count later in the day would be difficult to interpret since PLT counts, like most blood cell counts, have a diurnal variation.66 Hence, when 2348
TRANSFUSION Volume 54, September 2014
a sample is not collected in the morning, it is not clear what PLT count value should be used to trigger a PLT transfusion.
5. Don’t transfuse O– blood except to O– patients and in emergencies for women of childbearing potential with unknown blood group O– blood units are in chronic short supply due in part to overutilization for patients who are not O–. O– RBCs should be restricted to: 1) O– patients or 2) women of childbearing potential with unknown blood group who require emergency transfusion before blood group testing can be performed. Group O D– RBCs are in short supply. Group O RBC units are the universal type and can be transfused to patients of all ABO blood types. Furthermore, group O D– are often used as the universal type to prevent anti-D formation. However, the majority of transfused individuals do not form anti-D. The only population with significant consequences after anti-D formation are women of childbearing potential because anti-D can result in hemolytic disease of the fetus and newborn. Appropriate use of this rare blood type will enable its use in the appropriate patients when it is needed. In the United States, 49.4% of the 13,785,000 RBC transfusions were group O; 39.8% O+ and 9.6% O–.2 The demand for group O RBCs in general and O– RBCs specifically is highlighted by the number of days of available supply, O+ 2 to 8 days and O– 2 to 4 days, and that only 0.5% of group O RBC units processed were outdated compared to 2.1% of all allogeneic units. A total of 45% of whites and 49% of African Americans are group O and 16% of whites and 7% of African Americans are D–, leading to significant shortages since 87% of donors are white.67,68 Although blood suppliers target recruitment of group O donors, particularly O– donors, these units continue to be in short supply due to increased demand because many institutions use O– units for emergency transfusion when the patient’s blood type is unknown rather than using O+ units for males and females who are not of childbearing potential, typically defined as older than 45 to 50 years old.69,70 Institutions should establish policies for emergency release and delivery of blood products. There should also be protocols for administration of D+ RBCs to a D– or unknown patient.69 The frequency of anti-D formation after transfusion of D+ blood products to a D– patient is approximately 20% for RBCs and less than 4% for PLTs (recently reported as 0% for apheresis PLTs).71-73 It is especially important to prevent anti-D formation in females of childbearing potential because anti-D can cause hemolytic disease of the fetus and newborn in future pregnancies. Therefore, females of childbearing potential should receive D– RBCs. However, institutions may transfuse
AABB CHOOSING WISELY CAMPAIGN
D+ RBCs to a D–/D-unknown female of childbearing potential after a certain number (such as 8 units) of D– RBCs have already been transfused given the balance of available inventory and the likelihood of survival. In addition, RhIG might be considered to prevent anti-D alloimmunization in D– patients receiving D+ blood products. The practice of giving RhIG is more common when D+ PLTs are given to D– patients. The risk of hemolysis through the administration of RhIG needs to be weighed against the benefit of preventing alloimmunization, especially when RhIG is given to a patient receiving more than 1 or 2 units of D+ RBC.74 Supporting this Choosing Wisely statement is the National Health Service’s guideline from the United Kingdom, “The appropriate use of group O D– RBCs,” which determined that D– RBCs were mandatory indicated in O D– patients with anti-D, O D– females with childbearing potential, and in emergency to premenopausal (
