Perhaps the most essential element in implementing a patient blood management (PBM) programme in a hospital is the dissemination of the highest quality transfusion threshold evidence to physicians so as to inform their transfusion decisions every time one is being made.  Ideally this would be done in a one-on-one consultation with a transfusion medicine specialist who is well versed in the current literature.  Unfortunately this is only likely possible at the smallest of hospitals.  However, using electronics to disseminate this evidence can be a helpful way to engage clinicians when they are deciding if a transfusion is appropriate for their patients. This brief report will detail the experience in implementing electronic enhancements for PBM at the author’s institution.

Electronic crossmatch

Perhaps the single most important electronic enhancement in the field of transfusion medicine is the electronic crossmatch (EXM).  In order to implement the EXM, the transfusion service must have a computer system with the following features:

  • Can recognize when a recipient has a valid type and screen
  • Can recognize when an ABO compatible unit has been selected for transfusion, and permit the issuing of that unit
  • Can recognize when an ABO incompatible unit has been selected for transfusion, and can block the issuing of that unit
A recipient must meet the following criteria for the use of EXM:
  • No current or historical antibodies
  • A valid type and screen (might include a requirement for a check type sample)
The benefits of using the EXM over the serological crossmatch are as follows:
  • Can be performed in <5 minutes
  • Can be performed by virtually any member of the laboratory staff, i.e., does not require specialist knowledge of immunohematology
  • Since RBCs can be crossmatched and issued in a few minutes, fewer RBCs need to be crossmatched and reserved on the transfusion service’s shelves. Permits fluid use of inventory
  • Facilitates use of remote RBC unit issue, i.e., crossmatched RBC units can be stored near the operating rooms, intensive care units or emergency departments and accessed by the ward staff on demand 1,2


Electronics to reduce non-evidence based transfusions

The first systematic effort to give feedback to clinicians on their transfusion decisions at this multi-hospital, regional health care network was a manual process.  The transfusion medicine physicians would receive a daily list showing all of the transfusions that occurred at their hospitals on the previous day, as well as the antecedent and consequent laboratory values.  This list would be scrutinized and if an RBC transfusion was administered at a Hb level of > 9 g/dl, a personalized letter was emailed to the ordering physician asking them to justify why the transfusion was administered (such as bleeding), and reminding them about the institutional transfusion criteria.

This approach was very successful in reducing the number of transfusions that were administered at this high level, with reductions of between 38%-45% between the implementation of this program in August 2009 and its termination in April 2011 (unpublished data).  The clinicians also appreciated the opportunity to justify why they were administering transfusions in this apparently non-evidence based manner (often because of acute hemorrhage), and to learn about the institution’s transfusion thresholds.

Modification of the computerized physician order entry system to show alerts when non-evidence based transfusions are ordered

The email campaign was terminated when it became feasible to adapt the computerized physician order entry (CPOE) system to automatically generate alerts when an ordered transfusion was not substantiated by the laboratory evidence. Table 1 demonstrates the evidence based indications for transfusion that were implemented across this healthcare system.  

Previous research had demonstrated that including several factors in the CPOE alert system, such as providing the alert at the time the order is being placed, providing evidence for the recommendation, and not requiring the physician to enter additional information beyond what is normally required to order products, would increase the likelihood that the alerts would be heeded.3 Thus these factors were borne in mind when the alert system was being designed.  

Through a close collaboration with the institution’s information technology department, a drop down menu of evidence based indications for transfusions were generated. The ordering physician had to select from one of these indications (an Other category was also provided for orders that did not meet any of the specified criteria), and the CPOE system would check the patient’s antecedent laboratory values; if the values were consistent with the indication that the physician had ordered, the order would be transmitted to the blood bank. If the laboratory value was inconsistent with the selected indication (for example, a patient’s actual Hb value was 7.8 g/dl and the selected indication was that the patient was anemic with a Hb value <7.0 g/dl), an alert would appear on the screen informing the physician that the transfusion did not appear to be necessary based on the laboratory parameter.

The alert could be overridden if the physician felt that the transfusion was still indicated.  Alerts for all blood products have been implemented across the author’s institutions and the results have been impressive – between 10-15% of the alerted orders are cancelled, and the number of orders that generate an alert is also down.4-6 These alerts have been very successful in helping to decrease the number of blood products that are transfused across the author’s hospital system. For example the largest hospital in the system, an ~800 bed tertiary care hospital now transfuses <20,000 RBCs per year whereas only 7-8 years ago that figure was >30,000 RBCs per year.7  


Table 1. Indications for transfusion in the CPOE drop down menu (example).

RBC CPOE order form menu


  • Hgb ≤ 7.0g/dL and/or Hct ≤ 22% in a hemodynamically stable ICU patient.
  • Hgb ≤ 8.0g/dL and/or Hct ≤ 24% in a non-ICU patient with symptomatic anemia.
  • Hgb ≤ 10g/dL and/or Hct ≤ 30% in a patient experiencing acute ischemic cardiovascular disease (angina pectoris, acute myocardial infarction).
  • Acute bleeding with hemodynamic instability requiring urgent RBC transfusion.
  • Other

   *The alert does not trigger if the acute bleeding indication is selected regardless of the antecedent Hgb level.


Cryoprecipitate CPOE order form menu:

  • Fibrinogen <100 mg/dl AND active bleeding OR pre-procedure
  • Other

   *This alert would trigger if the antecedent fibrinogen level is >100 mg/dl and the congenital factor XIII deficiency indication was not selected

Platelet CPOE order form menu

  • Prophylactic transfusion to prevent spontaneous bleeding in stable patient with platelet count <10,000/ul
  • Prophylactic transfusion to prevent spontaneous bleeding in patient with consumptive state (eg high fever, sepsis, DIC, or splenomegaly) and platelet count <20,000/ul
  • Active bleeding or Pre-procedure with platelet count <50,000/ul
  • Pre-procedure or bleeding patient who has taken a recent dose of anti-platelet medications, or with documented platelet dysfunction (abnormal closure time, abnormal TEG result)
  • Massive bleeding requiring multiple RBC transfusions
  • Other

    *The alert does not trigger if the platelet dysfunction or the massive bleeding indications are selected regardless of the antecedent platelet count.  


Plasma CPOE order form menu:

  • INR ≥1.6 and the patient is currently bleeding or pre-procedure and NOT a candidate for vitamin K
  • Massive bleeding requiring multiple RBC transfusions
  • Plasma required for therapeutic exchange
  • Other

    *Note that that alert does not fire if the massive bleeding indication or plasma exchange indications are selected regardless of the antecedent INR value


It is important to note that

the transfusion thresholds for the alerts at this institution were arrived at by a consensus method involving the transfusion physicians, and the largest users of the blood products such as the physicians in the hematology/oncology, intensive care, surgery, and emergency departments.  For example, the TRICC study indicated that a Hb <7.0 g/dl would be a safe threshold for stable ICU patients,8 but the ICU physicians at this institution were uncomfortable with that level and so it was agreed that that the alert would be generated if the Hb <7.5 g/dl.  After several years of working with this threshold, the ICU physicians were reapproached and agreed that it could be lowered to the evidence-based 7.0 g/dl.  Thus, by working in a consensus-based manner, greater acceptance of the alerts could be achieved, for the benefit of all patients.


Electronics to reduce blood product waste

Electronic enhancements can also be applied to reduce blood product waste. At the author’s institution, whole blood platelets (PLT) are the predominant PLT product.  As these PLTs must be pooled to create an adult dose and subsequently stored at room temperature, the resulting 4 hour shelf life can lead to wastage if they are not transfused within that time.  To this end, the method by which the nursing staff communicates with the blood bank when a blood product is to be issued to a patient was recently modified.  Previously, a hard copy patient identification form (PIF) was sent to the blood bank indicating when the product should be issued.  However, frequently the patient was not actually prepared to receive the transfusion, and the product would be wasted.  A new electronic PIF (ePIF) was designed and incorporated into the electronic medical record.  This new ePIF screen requires the nurse to verify that the patient had given consent for the transfusion, that the required vascular access was present and that any pre-transfusion medications had been administered.  In the year following the ePIF implementation, PLT waste was significantly reduced compared to the preceding 19 months.9  Somewhat unexpectedly, RBC waste was also decreased after ePIF implementation. 

Electronics to enhance the management of pre-operative anemia

Pre-operative anemia management is an essential part of a broad-based PBM program. Anemia is a risk factor for many diseases and itself can be a marker of serious illness. At the author’s institution, all surgical patients are scheduled using a single computer program.10 As part of the institution’s comprehensive PBM program, when a patient is scheduled for surgery, an automated script begins to search through multiple databases looking for Hb values.  When an anemic Hb value is detected (using the WHO classification of anemia), an email and/or facsimile report is sent to the patient’s surgeon and/or family doctor indicating that the patient is anemic and its cause should be investigated before the date of surgery. These notifications have been successful in alerting surgeons and physicians to patients with anemia, and over 45% of the patients on whom an alert was sent received some form of treatment or intervention to resolve their anemia. One limitation to this alert is that it was discovered that the surgeons were only ordering pre-operative Hb testing a mean of 15 days before the surgery date, thereby limiting the nature of the interventions that could be implemented to correct the anemia. Encouraging surgeons to evaluate their patients for anemia more in advance of the surgery should help to improve the efficacy of anemia management, when required.




The author

Mark Yazer

Mark Yazer

Clinical Transfusion Working Party Vice Chair, Professor of Pathology, University of Pittsburgh, USA