Approach to Improving Safety
- Audit and Feedback
- Continuous Quality Improvement
- Forcing Functions
- Bar Coding and Radiofrequency ID Tagging
Setting of Care
A married couple, Mr. and Mrs. M, was brought to the emergency department (ED) of a Level 1 trauma center after a half-ton truck that had skidded out of control struck their car. Mr. M appeared hemodynamically stable, but had bilateral femoral fractures. Mrs. M had been the driver. Her blood pressure remained low despite wide-open crystalloid infusions, and she had signs of peritoneal irritation on exam. Both patients were typed and crossed, although only Mrs. M appeared to need packed red blood cells urgently.
The husband and wife patients had been placed in a large trauma bay with two beds. In the commotion of stabilizing and assessing both patients, the blood typing tube for Mr. M was labeled with the sticker for Mrs. M. Once the specimen was labeled and sent to the lab, this error would normally have been undetectable based on the standard protocols for handling transfusion products. By coincidence, however, Mrs. M had previously undergone a Cesarean section at the same hospital. She had been typed and crossed at that time. She and her husband did not share the same blood type (she was Type O and he Type A). The alert technologist in the blood bank noticed the change in blood type and inferred that a mistake must have been made. She called the ED immediately. They agreed to redraw her blood sample for re-typing, but also requested that O-negative blood be sent the ED immediately in case the patient deteriorated. Mrs. M thus never received the wrong blood.
This case represents a very serious near miss.
But for the coincidence of Mrs. M's blood type being on file at the
same hospital, she would have received a potentially fatal
incompatible transfusion matched for her husband's blood type (A)
and not her own (O).
Defects in the process of transfusion currently pose a greater hazard to patient safety than does transfusion-associated infectious disease.(1) As is well demonstrated in this case, safe transfusion begins with the correct collection and labeling of a blood sample from the right patient.
Miscollected blood samples—the wrong blood in the tube—are a common problem in blood transfusion. In a 10-nation study of more than 690,000 blood samples from 62 hospitals, the median rate of sample miscollection was 0.5/1,000; interquartile range <0.3 - 0.9/1000.(2) Notably, both Sweden and Finland, countries with national identification systems, had the lowest miscollection rates (both were too low to estimate reliably).
Mislabeled samples—samples with labels not meeting locally accepted standards—had an even higher but more variable rate of error: 1.2 - 17/1000.(2) This variability was most probably due to local hospital differences in laboratory guidelines for acceptance of samples. A different study of tested samples rejected because of mislabeling showed that these samples had a 40-fold higher rate of blood type discrepancies than correctly labeled samples.(3) These data indicate that the usual strict rejection policy for inadequately labeled samples is prudent and should be supported, despite any perceived inconvenience.
Paradoxically, since inadequately labeled samples give the lab some degree of warning, they afford less risk than a correctly labeled, but miscollected sample. In a 10-year study of transfusion error in New York State, 13% of ABO incompatible transfusions originated from such samples.(4) Unlike laboratory tests in which there is a clinical frame of reference (eg, serum bilirubin levels), an incorrect blood type raises no red flags. Because of this, some transfusion services do not release blood, other than group O, for routine transfusion without a prior consistent blood type.(5) In the case under discussion, the routine procedure of comparing test results of a new sample with the prior blood type test record was the key to preventing a potentially disastrous miscollection error.
The disquieting facts that 1/38,000 transfusions are ABO incompatible, and that half of these mistransfusions result in an hemolytic transfusion reaction, are obscured by a low fatality rate (1/600, 000 to 1/1,800,000).(4) Apparently, patients benefit more from good fortune, and possibly rescue (early investigation, discontinuance of transfusion, and prompt treatment), than from prevention.
The relatively low fatality rate may contribute to a lack of mindfulness in blood sample collection. Well practiced routines such as phlebotomy, although having little cognitive error potential, are often performed in an "autopilot" or "rehearsed" mode of behavior. Such autopilot behaviors can be expected to be compromised by periodic errors due to faulty execution even in the hands of competent and well intended personnel. Risk factors for such "slips" include interruption, changed circumstances, or momentary lapses in concentration.(6)
Feedback to staff of the potential dangers of phlebotomy error, especially in emergency circumstances, may help to reinforce a more mindful approach. Although education and reinforcement are potentially useful and necessary, they may not be sufficient.(7) Concern about the reliability of patient and sample identification has led to promulgation of national standards (8) and proposed federal regulation.(9)
The frequency and potential risk from flawed patient identification is well recognized as a general medical issue. The Joint Commission on Accreditation of Healthcare Organizations (JCAHO) has included "Improve the accuracy of patient identification" as one of its "National Patient Safety Goals" for the last 2 years.(8) JCAHO has recommended the use of at least two patient identifiers (exclusive of the patient's room number) whenever taking blood samples or administering medication or blood transfusion.
The hectic environment of an emergency department increases the risks of patient/sample misidentification. For example, patients may be unresponsive and unknown, clinical urgency may lead to samples being collected by pressured staff, and blood tubes may be handed off frequently or labeled away from the patient. The simultaneous arrival of two or more patients with the same surname (as in this case) adds to the potential for error. ED staffs are acutely sensitive to these issues and have adopted strategies to help achieve the correct linkage of patient and sample identification. One approach has been the use of pre-made matched labels and ID bands.
A variety of other approaches for assuring accurate identification have been utilized in EDs as well as in other settings, including: mechanical generation of labels from the wristband itself; a forcing function requiring reference to a unique number located only on a patient's wristband for both specimen labeling and for opening a barrier combination lock prior to transfusion (10); and automation using barcode technology to secure both the initial as well as subsequent phases of the transfusion (and medication) process.(10,11)
Bar code identification has received the most attention. The FDA has proposed point-of-care bar code requirements for medication and blood product administration.(9) Automation has clear advantages in decreasing total reliance on human vigilance, but it may introduce a new set of concerns. Technology is fully effective only when introduced into a properly designed system. Attempting to implement bar coding without careful system review and appropriate redesign may merely automate bad practice. For example, one hospital had a dangerous routine of allowing documentation in advance of medication administration, specifically, at the time of removing the medication from a dispensing device or medication cart. Bar codes were introduced to correct this flawed practice, but the solution provided merely an illusion of safety and perpetuated the flaw: extra bar coded patient wristbands were printed in advance to allow easy scanning of the wristband, medication, and person intending to administer it, regardless of who or where the recipient happened to be.(12) A recent article has documented the various new error modes attributable to the introduction of bar coding systems.(13) As is often the case, the introduction of a new technology, intended to decrease a target risk, may create a new contravening one. This underscores not only the importance of careful design, but the cautious monitoring of the effects of change as well.(14)
Just as intended corrections may create an
illusion of safety, so may recovered near misses. The "save" in
this case was due to a combination of luck and an alert technician;
both cannot be counted on to protect future patients. Although the
robust ability to recover from error is of critical importance, the
"glass is half-empty" mindset characteristic of a high reliability
reminds us that an initial failure may still need to be addressed.
In this way, near miss events provide an important "canary in the
mine" early warning of a potential system weakness.
- Systems providing confirmation of blood sample typing results against prior test records provide an important safety barrier for transfusion.
- Well-practiced routines, such as phlebotomy, may be particularly vulnerable to error when interrupted, or when unanticipated signals are encountered.
- The use of automation, including bar code technology, can provide a significant increase in reliability of patient and sample identification, but this improvement can only be realized when the automation is combined with careful system design.
Harold S. Kaplan, MD
Professor of Clinical Pathology
Director of Transfusion Medicine
New York Presbyterian Hospital
Columbia Presbyterian Medical Center
1. Dzik WH, Corwin H, Goodnough LT, et al. Patient safety and blood transfusion: new solutions. Transfus Med Rev. 2003;17:169-80.[ go to PubMed ]
2. Dzik WH, Murphy MF, Andreu G, et al. An international study of the performance of sample collection from patients. Vox Sang. 2003;85:40-7.[ go to PubMed ]
3. Lumadue JA, Boyd JS, Ness PM. Adherence to a strict specimen-labeling policy decreases the incidence of erroneous blood grouping of blood bank specimens. Transfusion. 1997;37:1169-72.[ go to PubMed ]
4. Linden JV, Wagner K, Voytovich AE, Sheehan J. Transfusion errors in New York State: an analysis of 10 years' experience. Transfusion. 2000;40:1207-13.[ go to PubMed ]
5. Prevention of
mislabelling of pre-transfusion compatibility specimens. e-Network
Forum. The California Blood Bank Society (CBBS) Web site. August
15, 2001. Available at:
[ go to related site ]. Accessed January 28, 2004.
6. Langer EJ. Mindfulness. April Reprint ed. New York, NY: Perseus Publishing; 1990.
7. Callum JL, Kaplan HS, Merkley LL, et al. Reporting of near-miss events for transfusion medicine: improving transfusion safety. Transfusion. 2001;41:1204-11.[ go to PubMed ]
8. National patient
safety goals. Joint Commission on Accreditation of Healthcare
Organizations Web site. Available at:
[ go to related site ]. Accessed February 9, 2004.
9. FDA proposes drug
bar code regulation. U.S. Food and Drug Administration. FDA News
Web site. March 13, 2003. Available at:
[ go to related site ]. Accessed January 28, 2004.
10. Wenz B, Mercuriali F, AuBuchon JP. Practical methods to improve transfusion safety by using novel blood unit and patient identification systems. Am J Clin Pathol. 1997;107:S12-6. [ go to PubMed ]
11. Turner CL, Casbard AC, Murphy MF. Barcode technology: its role in increasing the safety of blood transfusion. Transfusion. 2003;43:1200-9[ go to PubMed ]
12. Shea B. The Patient Safety Discussion Forum. NPSF Listserv. August 8, 2002. Available at: [ go to related site ]. Accessed January 13, 2004.
13. Patterson ES, Cook RI, Render ML. Improving patient safety by identifying side effects from introducing bar coding in medication administration. J Am Med Inform Assoc. 2002;9:540-53.[ go to PubMed ]
14. Wilde G. Target risk 2: a new psychology of safety and health. Toronto, ONT: PDE Publications; 2001.
15. Weick K, Sutcliffe K. Managing the unexpected: assuring high performance in an age of complexity. San Francisco, CA: Josey-Bass; 2002.