Episodes and Transfusion Errors
Accurate matching of patients with appropriate blood units is imperative for safe blood transfusion practice. Transfusion of blood to the wrong patient with a subsequent acute hemolytic transfusion reaction is one of the most frequent avoidable causes of blood transfusion-associated morbidity and mortality. Acute hemolytic transfusion reactions were the second leading reported cause of death from transfusion in the United States from 2005 to 2009. A total of 158 deaths were reported to the US Food and Drug Administration between 1976 and 1985, approximately 1 death per 250,000 red blood cell transfusions, and an estimated 1 near-miss episode occurred per 1000 red blood cell transfusions. A near-miss transfusion episode is a narrowly avoided attempt to administer blood to the wrong patient. Medical errors that result in adverse events are unfortunately common and underreported in the United States and worldwide. Two studies in New York State estimated that transfusion errors occurred once for every 12,000 U transfused before 1992 and once for every 19,000 U transfused from 1990 to 1998.
In 1998 Mayo Clinic contracted with Picis to install their anesthesia information management system (AIMS) Chart+. In 2002, we initiated the bar code patient identification project as added functionality to AIMS. Adding a bar code to the patient identification armband, which consisted of the patient’s unique Mayo Clinic number (MCN), was the first step. The user still enters the MCN and the patient’s last name and then uses a bar code reader to capture the MCN on the patient’s hospital identification bracelet. This information is sent to the admission, discharge, and transfer computer system for reconciliation, and if successful the anesthesia record is launched. In late 2005, we expanded the bar code identification to blood products in the operating rooms and intensive care units. Our blood product bag labeling includes the patient’s name, MCN, and a bar code with the patient’s MCN. In 2006, the blood bank increased the bar-coded information on each blood product bag and expanded its use throughout the institution. These bags now include the patient’s MCN, blood product unit number, and the blood product type.
The blood product verification process requires the user to scan the patient’s identification band and the 3 bar codes on the blood bag before administration. This scanning process checks the unit into AIMS and communicates with the blood bank computer system SafeTrace Tx (SSTx; Wyndgate Technologies), identifying that it is the right patient, the right unit, and the right blood product. The user then documents when the unit has been administered and records the reason for the transfusion and its results. Finally, the bar-coded blood component identification number is scanned a second time to document the unit number in the fluids documentation of AIMS.
We performed a retrospective study to determine the incidence of transfusion errors before and after institution of our computerized bar code-based blood identification system. Our hypothesis was that the computerized bar code-based blood identification system would result in a reduction in transfusion errors.
After institutional review board approval, we performed a retrospective study to determine the incidence of transfusion errors the 4 years before and the 4 years after institution of our bar code-based blood identification system. We excluded the year 2006 because of the phase in the computerized bar code-based blood identification system.
For the 4 years before 2006, all patients who reported to the blood bank with transfusion errors were reviewed. From January 1, 2002, to December 31, 2005, there was manual verification of the patient’s identity via the patient’s clinic number. The person administering the blood was supposed to check the unit identification against the patient’s clinic number. The reporting of mistransfusion events and near-miss transfusion episodes was voluntary on the part of the person administering the blood.
For the 4 years after 2006 (January 1, 2007, to December 31, 2010), all the records of patients who were identified by the blood identification database as having scanning errors were reviewed. The blood product verification (BPV) system interfaces with SSTx to determine whether the blood product delivered to the bedside is the correct unit for the patient. The computer system creates a message for the person transfusing the blood that alerts him/her to proceed with transfusion when the clinic number on the blood bag and patient’s armband match. The blood unit number is also scanned. If the query fails, the BPV system alerts the user not to administer blood without further verification. The error is documented in AIMS and summarized in a monthly error report.
Our investigation involved reviewing 106 scanning transactions with the following 2 error messages:
- ## NoMatch-Date/time SSTx does not recognize the scanned blood bag unit number as the appropriate unit for the scanned armband of patient.
- Error-Clinic number MISMATCH-BPV alerts the user that the patient’s armband does not match the scanned clinic number on the blood bag.
Each scanning error event required a separate investigation to trace the blood product bag and the patient at the bedside for date and time stamps in both SSTx and AIMS. Reviewing the time stamps associated with the request and issue of blood product and administration verification at the bedside, we were able to provide a short synopsis of the potential of harm to the patient.
The retrospective investigation confirmed 43 BPV-flagged events, whereby the scanned clinic number on the armband and compatibility label on the blood bag did not match. Nine events were merged clinic numbers, where the patient’s clinic number was updated by registration from the time the blood order was placed, assigned, and issued and BPV flagged by the user scanning the patient’s new armband. Of the 43 flagged events, 34 near-miss events were identified.
Statistical analysis was performed using SAS software (SAS Institute, Inc). The effect on categorical variables was analyzed using the Fisher exact test. P<.05 was taken as statistically significant. We calculated 95% CIs using the Poisson approximation to the binomial.
A total of 388,837 U were transfused during the 4-year period (2002-2005) before the institution of our computerized bar code-based blood identification system. There were 6 misidentification episodes of a blood product transfused to the wrong patient during that period (incidence of 1 in 64,806 U or 1.5 per 100,000 transfusions; 95% CI, 0.6-3.3 per 100,000 transfusions) (Table 1). There was one acute hemolytic transfusion reaction in 2004. The blood was stopped after 20 mL had been infused due to flank pain and facial flushing. The posttransfusion serum sample was light orange. The posttransfusion urine sample showed trace amounts of hemoglobin. Posttransfusion laboratory test results were negative for posttransfusion coagulopathy or impaired renal function.
A total of 304,136 U were transfused during the 4-year period (2007-2010) after the institution of our bar code-based blood identification system. There was one misidentification episode of a blood product transfused to the wrong patient during that period (incidence of 1 in 304,136 U or 0.3 per 100,000 transfusions; 95% CI, <0.1-1.8 per 100,000 transfusions; P=.14) (Table 2). The system was not used properly in this case because the unit of blood was scanned after it was administered. Fortunately, the blood bag labeled for another patient was the same blood type.
There was one reported near-miss transfusion episode in the 4-year period before institution of our computerized bar code-based blood identification system (incidence of 1 in 304,136 U or 0.3 per 100,000 transfusions; 95% CI, <0.1-0.4 per 100,000 transfusions). There were 34 reported near-miss transfusion episodes in the 4-year period after institution of our bar code-based blood identification system or 11.2 per 100,000 transfusions (95% CI, 7.7-15.6 per 100,000 transfusions; P<.001) (Table 2). Of these near misses, 6 could have resulted in acute hemolytic transfusion reactions.
Our incidence of blood being given to the wrong patient was much lower than the published incidence, which estimated that blood units were given to the wrong patients once for every 12,000 U transfused before 1992 and once for every 19,000 U from 1990 to 1998. In fact, the upper bound of our 95% CI is below the published rates. The incidence of prevented transfusion errors after institution of the computerized bar code-based blood identification system, however, is similar to those estimates of 34 of 304,136 U (incidence of 1 per 8,945 U).
Our results are similar to those of Askeland et al. They found 3 documented instances of misidentification errors regarding transfusions in the 7.5 years before implementation of a comprehensive computerized bar code-based tracking system. After implementation of their comprehensive computerized bar code-based tracking system, they found 113 prevented identification errors, with 10 being the patient’s barcoded wristband not matching the bar code label on the blood product bag. They identified one patient in the intensive care unit who was A positive but who received an O negative unit of red blood cells because of misidentification of an initial blood type sampling. A review of recipient identification published in 2006 found that the use of the BloodLoc and I-TRAC systems prevented 10 transfusion errors of 98,339 U in 6 studies. A fingerprint and bar code-based transfusion identification system, the Securblood system, was instituted at Ragusa Hospital in Italy. Overall, 1777 patients received transfusions of 5606 U of blood with no transfusion errors. The bar code-based I-TAC Plus system was instituted in Mantova, Italy. During 5 years 71,400 U of blood were transfused to 15,430 patients. The system prevented transfusion of blood to the wrong patient 12 times.
The large increase in discovered transfusion errors that were prevented by our computerized bar code-based blood identification system was most likely due to the automatic reporting within the system. Before institution of our system, transfusion errors were self-reported. It is known that medical errors that result in adverse events are unfortunately common and underreported in the United States and worldwide. Another possibility is that before institution of our computerized bar code-based blood identification system the errors in transfusion were not recognized because most errors found by our computerized bar code-based blood identification system were compatible blood. If the patient did not have an acute hemolytic transfusion reaction, it may not have been regarded as a transfusion error. An alternative explanation would be that our computerized bar code-based blood identification system was somehow causing an increase in transfusion errors, although this seems unlikely. Our study has some limitations. One limitation is that the blood identification system was not used in all blood transfusions in the later period. We implemented the system first in the operating rooms and intensive care units. It was then extended to other areas. In 2007, 73% of the units administered used the system. The rate of use increased to 81% in 2008, 80% in 2009, and 93% in 2010. Another limitation is that the computerized bar code-based blood identification system is one component in our institution’s process to provide safe blood transfusion therapy. The process begins with standardized patient identification processes for blood collection, the generation of bar code-based specimen labels that are placed at the bedside, a requirement for 2 independently determined and verified ABO and Rh types before the provision of type-specific blood components, complex computer-based specimen tracking, a formalized blood issue process, and patient identification procedures that require either an automated system plus one person as described or 2 individuals in the absence of the automated system. All steps are monitored, audited, and analyzed as part of our Quality Management Program. The data are tracked and trended, and corrective and preventive actions are implemented as necessary. It is difficult to determine when each of these processes was put in place because different groups implemented the processes over time. Murphy et al describe the evolution of the “end-to-end electronic management system” at Oxford University Hospitals during 4 years. They found results similar to ours. Finally, it is possible that other factors resulted in the changes that we have seen unrelated to the use of the bar code system.
Institution of a computerized bar code-based blood identification system was associated with a large increase in discovered near-miss events. It was also associated with a reduction in blood products administered to the wrong patient that was not statistically significant.
Abbreviations and Acronyms: AIMS = anesthesia information management system; BPV = Blood Product Verification; MCN = Mayo Clinic number; SSTx = SafeTrace Tx
Used with permission from Elsevier. This article was published in Mayo Clinic Proceedings, 88, Nuttall GA, Abenstein JP, Stubbs JR, et al: Computerized Bar Code-Based Blood Identification Systems and Near-Miss Transfusion Episodes and Transfusion Errors, 354-359, copyright Elsevier (April 2013). References omitted. The complete article is available online at www.mayoclinicproceedings.org