Nearly Two Decades Using the Check-Type to Prevent ABO-Incompatible Transfusions One Institution s Experience

Coagulation and Transfusion Medicine / CHECK-TYPE POLICY FOR ABO CONFIRMATION Nearly Two Decades Using the Check-Type to Prevent ABO-Incompatible Transfusions One Institution s Experience Priscilla I. Figueroa, MD, 1 Alyssa Ziman, MD, 2 Christine Wheeler, MD, 3 Jeffrey Gornbein, DrPH, 4 Michael Monson, 2 and Loni Calhoun, MT(ASCP)SBB 2 Key Words: Check type; ABO-incompatible transfusion; Specimen collection errors; Miscollected sample; Transfusion errors; Sample identification; Blood typing Abstract To detect miscollected (wrong blood in tube [WBIT]) samples, our institution requires a second independently drawn sample (check-type [CT]) on previously untyped, non group O patients who are likely to require transfusion. During the 17-year period addressed by this report, 94 WBIT errors were detected: 57% by comparison with a historic blood type, 7% by the CT, and 35% by other means. The CT averted 5 potential ABO-incompatible transfusions. Our corrected WBIT error rate is 1 in 3,713 for verified samples tested between 2000 and 2003, the period for which actual number of CTs performed was available. The estimated rate of WBIT for the 17-year period is 1 in 2,262 samples. ABO-incompatible transfusions due to WBIT-type errors are avoided by comparison of current blood type results with a historic type, and the CT is an effective way to create a historic type. Mistransfusion of ABO-incompatible blood poses a greater risk for transfusion recipients than the risk of transmission of infectious disease, accounting for 37% of all reported transfusion-associated fatalities in the United States. 1 Published reports cite an incidence of ABO discrepancy due to inappropriately identified specimens ranging from 1 in 517 to 1 in 3,400 samples. 2-4 Considering that approximately two thirds of transfused units will be ABOcompatible by chance alone and that the true incidence of transfusion errors has been estimated to be as high as 5 times the number of detected errors, the risk of mistransfusion may be severely underestimated. 5,6 The first step in preventing mistransfusion is obtaining blood for pretransfusion testing from the right patient and ensuring that all labeling is correct. Errors in these critical steps are recognized as the primary source of mistransfusion. Of greatest concern are the errors that cannot be identified by visual inspection of the samples and associated requisitions, whether in paper or electronic form. These miscollected blood samples (wrong blood in tube [WBIT]) in which tubes and requisitions appear properly labeled but the sample is drawn from a different patient are the stealthiest of errors and easily may go undetected until the event of an incompatible transfusion. According to a large multinational study involving 62 hospitals by Dzik et al 2 in 2003, WBIT accounts for up to 0.09% of samples collected. As a means of preventing ABO-incompatible transfusions due to WBIT-type errors, our institution has a policy of requiring a check-type (CT) sample a second independently drawn blood type sample on previously untyped non group O patients who require transfusion or are likely to require transfusion, including all patients admitted to labor and delivery, the 422 Am J Clin Pathol 2006;126:422-426 Downloaded 422 from https://academic.oup.com/ajcp/article-abstract/126/3/422/1759935

Coagulation and Transfusion Medicine / ORIGINAL ARTICLE operating room, or an intensive care unit for whom a request for type and screen (T&S) is received. Exception is made for trauma patients for whom a dual banding system is used as a means to prevent identification errors. To ensure independent specimen collection, the specimen must be collected into a prelabeled specimen tube generated by the blood bank after review of records indicates a need for a CT sample. The CT policy, in effect, creates a historic type when one does not exist. This article reviews our 17-year experience with the CT system. Materials and Methods All reports of mislabeled specimens detected between January 1987 and December 2003 were reviewed. These reports, summarized on a Blood Bank Specimen Labeling Error form, documented specimens and requisitions with any type of labeling error. From these, WBIT-type errors were identified based on the following: (1) discrepant ABO or Rh typing results between the implicated sample and the patient s historic record or CT sample, (2) the service having notified the blood bank that a specimen was mislabeled, and (3) irregularities not related to name, identification number, or blood type that raised suspicions about the identity of the sample. Discrepancies were resolved by retyping an additional sample drawn by a different staff member or at a different time from the implicated sample. Group O Rh blood was issued if the transfusion could not be delayed for retyping of an additional specimen. For each WBIT-type error, the following data were extracted from the Blood Bank Specimen Labeling Error forms and available follow-up documentation: location of collection; identity of responsible phlebotomist; initial (historic) blood type and discrepant blood type; how the error was found or identified; whether the miscollected specimen was the specimen requested as the CT, and if yes, whether it was collected in the blood bank generated prelabeled tube; and whether the miscollected specimen could have resulted in ABO and/or Rh incompatibility. Workload data, obtained through a computer query for applicable test codes, provided the number of T&S tests performed during the 1987-2003 period. A true count of the number of CT tests performed was available only by computer query beginning in 2000 when a specific CT computer test code was created. Before 2000, CT tests were coded as ABO- Rh tests, and numbers of tests can only be estimated. By using the aforementioned information, we calculated the estimated true rate of WBIT corrected for the rate of verified samples and for the rate of silent errors by the method reported by Dzik et al. 2 The raw rate of WBIT was determined by using the ratio of the number of samples whose result did not match the previous record on the patient divided by the total number of samples for which a previous result was on record (verified samples). This rate then was corrected for silent WBIT errors, which can occur when the ABO group of the blood in the tube matches the ABO group on record although the wrong patient s blood is collected. The correction factor was based on the distribution of ABO groups in the University of California, Los Angeles (UCLA) patient population. Results Between January 2000 and December 2003, the time frame when T&S and CT numbers could be accurately counted, 107,835 T&S and 16,164 CT specimens were tested. Fifteen specimens were identified as miscollected (or WBIT-type errors). The rate of miscollected specimens (WBIT) between 2000 and 2003 with correction for silent errors was 1 in 3,713 (based on 15 WBITs of 80,026 verified samples with a correction factor of 1.437). Table 1 shows yearly corrected rates of WBIT for 2000 through 2003 and the number of specimens tested, verified (CT or historic type as part of a previous blood bank workup) and nonverified (group O patients and patients with only 1 blood type on record). Table 1 Rate of WBIT for 2000-2003 2000 2001 2002 2003 2000-2003 Total T&S performed 26,553 26,864 26,808 27,610 107,835 Verified types 21,058 19,442 19,273 20,253 80,026 CT performed 3,212 4,256 4,342 4,354 16,164 Historic blood type available 17,684 15,186 14,931 15,899 63,862 Nonverified blood types 5,495 7,422 7,535 7,357 27,809 WBIT 7 3 2 3 15 Raw WBIT rate 3,008 6,481 9,636 6,751 5,335 Corrected WBIT* 10 4 3 4 21 Corrected WBIT rate* 2,094 4,510 6,706 4,698 3,713 CT, check type; T&S, type and screen; WBIT, wrong blood in tube. * Corrected for silent errors. Correction factor of 1.437 (1 Q = 0.696). Based on blood group frequencies of the University of California, Los Angeles patient population: O Rh+, 44.23%; O Rh, 4.95%; A Rh+, 29.90%; A Rh, 4.13%; B Rh+, 11.63%; B Rh, 1.27%; AB Rh+, 3.47%; and AB Rh, 0.43%. Downloaded from https://academic.oup.com/ajcp/article-abstract/126/3/422/1759935 Am J Clin Pathol 2006;126:422-426 423 423 423

Figueroa et al / CHECK-TYPE POLICY FOR ABO CONFIRMATION During the 17-year period addressed by this report (1987-2003), 411,705 T&Ss were performed and a total of 94 WBITtype errors were identified. Of 94 errors, 61 (65%) were detected by discrepant typing results (~57% by comparison with a historic blood type, ~7% by the CT), 26 (28%) when the clinical service informed the blood bank of the error, 3 (3%) by other specimen labeling errors (second label discovered under first, date of birth inconsistent with cord blood requests), and 4 (4%) by other means (eg, inquiry regarding a blood order or availability of blood products for a particular patient). Of the 61 WBIT-type errors detected by discrepant typing results, 27 cases (44%) could have resulted in an ABO-incompatible transfusion and 6 (10%) could have resulted in an Rhincompatible transfusion, representing 1.6 and 0.4 transfusions per year, respectively, at our institution. Of 61 discrepant results, 40 (66%) were detected by comparing the current type with an established historic blood type. Of these, 14 initially typed as O Rh+ and 8 as O Rh. Of the O Rh+ types, 12 subsequently were shown to be non group O, and 4 were shown to be truly Rh. Of the O Rh types, 5 later were shown to be non group O, and only 1 was truly Rh. The remaining 21 (34%) of 61 discrepant samples were detected by comparing the type of the CT sample with the T&S sample. In 7 of these (11% of all discrepant typings and 33% of discrepancies involving a CT), the CT sample revealed an incorrect initial T&S sample. Transfusion based on the initial sample would have resulted in an ABO-incompatible transfusion in 5 of 7 cases. Of the remaining 14 cases involving a CT sample (23% of all discrepant typings, 66% of those involving a CT sample), the prelabeled blood bank generated CT sample was the miscollected sample. All WBIT-type errors were attributed to personnel not adhering to patient identification protocols. In 85 cases (90%), the identification error occurred at the time of phlebotomy. In 9 cases (10%), the identification error occurred at the time of patient admission or registration. Samples involving WBITtype errors were drawn from patients in all inpatient and outpatient areas; non intensive care unit inpatient units and labor and delivery had a disproportionately high number of occurrences Table 2. Personnel responsible for obtaining the samples involved in the 94 WBIT-type errors included the following: nurses, 45 (48%); phlebotomists, 26 (28%); physicians, 19 (20%); clinical partners (noncredentialed nursing-affiliated staff), 3 (3%); and medical student, 1 (1%). Discussion Most ABO-incompatible transfusions result not from laboratory errors but from mistakes in patient identification, including sample labeling errors that result in the wrong blood collected in the tube or misidentification of the patient at the time of Table 2 Location of Specimen Collections for 94 Wrong Blood in Tube Errors No. (%) of Sites of Distribution of Specimen Location Miscollected Specimens Collection at UCLA (%) Inpatient units 59 (63) 38.8 Delivery room 11 (12) 2.5 Intensive care units 8 (9) 18.4 Outpatient clinics 7 (7) 20.2 Outpatient laboratory 4 (4) 3.0 Operating rooms 3 (3) 13.6 Emergency department 2 (2) 3.4 UCLA, University of California, Los Angeles. transfusion. The risk of ABO-incompatible transfusion is estimated to be 3 times greater than the combined risks of transfusion transmission of hepatitis B, hepatitis C, and HIV. 1 This high risk is receiving increasingly greater attention from the blood banking community and the public. As of December 29, 2004, the College of American Pathologists Transfusion Medicine Checklist has required that facilities have a documented program to ensure that the risk of pretransfusion sample misidentification is monitored and subjected to continual process improvement (TRM.30550, phase II). Methods of compliance provided in the checklist include the following: (1) obtaining a second separately collected sample for repeated ABO grouping, (2) mechanical barrier or electronic identification verification systems, (3) issuing only group O units for transfusion, and (4) documentation in laboratory records of a previous ABO determination from a correctly labeled specimen. The UCLA CT policy, which satisfies examples 1 and 4 of the College of American Pathologists requirement, was instituted in an effort to prevent ABO-incompatible transfusions resulting from miscollected specimens. Our data represent the longest experience using a check-type system reported by a US institution. Our WBIT-type error rate corrected for silent errors is 1 in 3,713 for verified samples submitted for pretransfusion testing between 2000 and 2003. Based on the data and the total number of WBIT errors detected between 1987 and 2003, we estimate that the rate of WBIT during the 17-year period was 1 in 2,262 samples tested (based on 94 WBIT errors of an estimated 305,532 verified samples [using the ratio of verified specimens/t&ss performed between 2000 and 2003]) with a silent error correction factor of 1.437. Because our CT policy is applicable only to patients who are likely to require transfusion and do not initially type as group O, this number underestimates the incidence of WBIT-type errors at our institution. Chiaroni et al 4 reported the use of a similar system applied across all blood groups by a laboratory performing pretransfusion testing for 35 French hospitals. As in our study, ABO discrepancies were detected by comparing 2 current 424 Am J Clin Pathol 2006;126:422-426 Downloaded 424 from https://academic.oup.com/ajcp/article-abstract/126/3/422/1759935

Coagulation and Transfusion Medicine / ORIGINAL ARTICLE blood types or 1 current blood type and a historic type. 4 They demonstrated that during a 5-year period (September 1998 to July 2003), the incidence of ABO discrepancies was 1 per 3,400 tests performed. In their experience, most discrepancies (58%) were secondary to phlebotomy errors in which the sample was obtained from the wrong patient, whereas the second most common cause of discrepancy was error during patient registration or identification (30%). In comparison, 90% of our errors were due to patient identification at the time of phlebotomy, and only 10% were due to admission or registration errors. During the 17-year period (1986-2003), our policy of requiring 2 independent samples for blood typing before issuing type-specific blood for non group-o patients potentially averted an estimated 1.6 incompatible transfusions per year at our institution. In the majority of cases, the 2-sample requirement was met by having a historic type on record, but when no historic type was available, the CT sample was an effective means of establishing a historic type. The CT sample detected an incorrect initial type in 11% of phlebotomy errors (potentially averting 5 ABO-incompatible transfusions). In an additional 23% of phlebotomy errors, the CT itself was the source of error. These CT samples were miscollected in the prelabeled blood bank generated CT specimen tubes, raising the question of whether this practice alone increases the risk of miscollected samples. Given 2 known WBIT-type errors in 31,806 CTs collected between 2000 and 2003 (0.0629%), and 14 errors from an estimated 121,433 CT samples (0.0115%) obtained during the 17-year period, UCLA continues to use the prelabeled tube system. We have concluded that the inconvenience and cost of needing to draw a third specimen owing to a WBIT error involving the CT sample is small relative to the risk incurred when staff, through lack of knowledge or deliberate noncompliance, draw the CT sample at the same time as the initial specimen. The CT policy seems to be a cost-effective approach for reducing WBIT-type errors, especially, if legal costs are considered. A recent cost analysis for a CT drawn by a phlebotomist at our institution was $2.39, resulting in costs of approximately $20,000 per year. Although several other innovations have been developed to address this problem and reduce transfusion errors, including mechanical barrier systems, unique transfusion identification numbers with or without barcode scanning, and radiofrequency scanners that read information encoded on chips on blood units and patient identification bands, they are much more costly. 7-20 For example, AuBuchon and Littenberg 16 concluded that a mechanical barrier system could be cost-effective, if legal costs were considered, and that such a system was successful in reducing the risk of fatal acute hemolytic transfusion reactions due to mistransfusion by more than 99.99%. A mechanical barrier such as the Bloodloc Safety System (Novatek Medical, Altamont, IL) would cost our institution $100,408 to $133,878 per year based on 234,286 RBC units transfused from 1998 through 2004 (each with locks costing between $3.00 and $4.00 per unit). 9 Another proposed safety measure involves barcoded labels with a unique transfusion identification number attached to the patient s wristband, the sample tubes, blood request, and compatible blood units, which at the time of transfusion are scanned with a portable barcode scanner to verify the match between patient identification and the blood unit. UCLA recently studied the feasibility of implementing such a system; a quote obtained from a leading manufacturer was approximately $200,000 for 20 users including 1 year of technical support. The disadvantages of requiring 2 independent blood samples before issuing type-specific blood include costs, inconvenience to the patient and patient care services, potential delays in providing blood, and possible increased use of group O blood. In addition, when the requirement for type confirmation is limited to non group O samples, the risk of Rh-incompatible transfusion remains. The extent to which these disadvantages may be minimized depends on the degree of institutional support for a second sample (CT) policy, which in turn is dependent on the effectiveness of educational campaigns to increase awareness about ABO transfusion errors. Perhaps the most effective policy for patient safety and convenience is one that rather than deferring all blood typing until transfusion is probable, requires establishing a historic blood type as part of the patient s medical record. The financial impact of such a policy could be minimized by targeting patients with greater risk of requiring blood transfusion (eg, patients requiring surgery, patients with hematologic or oncologic disorders or gastrointestinal bleeding, and geriatric patients), by obtaining blood type samples with other routine laboratory blood samples, and by batch testing these typings. Avoidance of ABO-incompatible transfusion and detection of WBIT-type errors in the absence of reliable electronic methods for bedside patient identification depends primarily on comparison of current blood type results with a historic type. Therefore, until such technology is implemented, the use of the CT, although subject to erroneous collection itself, seems to be an effective and relatively inexpensive method for detecting errors in blood sample collection and is an additional means of preventing ABO-incompatible transfusions by creating a historic type. From the Departments of Pathology and Laboratory Medicine, 1 the Cleveland Clinic, Cleveland, OH; 2 Division of Transfusion Medicine, University of California, Los Angeles; 3 University of California, Irvine; and the 4 Department of Biomathematics, University of California, Los Angeles. Address reprint requests to Dr Figueroa: Cleveland Clinic, Section of Transfusion Medicine, Desk L20, 9500 Euclid Ave, Cleveland, OH 44195. Downloaded from https://academic.oup.com/ajcp/article-abstract/126/3/422/1759935 Am J Clin Pathol 2006;126:422-426 425 425 425

Figueroa et al / CHECK-TYPE POLICY FOR ABO CONFIRMATION Acknowledgments: We thank and acknowledge the assistance of Rebecca Davis, MT(ASCP), Tracey Allen, MT(ASCP)SBB, Marianne Silva, MS, MT(ASCP)SBB, CQA(ASQ), Mary Anne Anthony, MT(ASCP)SBB, and Carma Lizza for their support and help in obtaining the data for this study. References 1. Aubuchon JP, Kruskall MS. Transfusion safety: realigning efforts with risks. Transfusion. 1997;37:1211-1216. 2. Dzik WH, Murphy MF, Andreu G, et al. An international study of the performance of sample collection from patients. Vox Sang. 2003;85:40-47. 3. Murphy MF, Steam BE, Dzik WH. Current performance of patient sample collection in the UK. Transfus Med. 2004;14:113-121. 4. Chiaroni J, Legrand D, Dettori I, et al. Analysis of ABO discrepancies occurring in 35 French hospitals. Transfusion. 2004;44:860-864. 5. Ibojie J, Urbaniak SJ. Comparing near misses with actual mistransfusion events: a more accurate reflection of transfusion errors. Br J Haematol. 2000;108:458-460. 6. Callum JL, Kaplan HS, Merkley LL, et al. Reporting of nearmiss events for transfusion medicine: improving transfusion safety. Transfusion. 2001;41:1204-1211. 7. Marconi M, Sirchia G. Increasing transfusion safety by reducing human error. Curr Opin Hematol. 2000;7:382-386. 8. Lau FY, Wong R, Chui CH, et al. Improvement in transfusion safety using a specially designed transfusion wristband. Transfus Med. 2000;10:121-124. 9. US Food and Drug Administration. Best practices for reducing transfusion errors: OBRR/CBER/FDA Workshop. February 14, 2002. Available at http://www.fda.gov/cber/minutes/ 0215bloo.htm. Accessed July 25, 2004. 10. Dzik WH, Corwin H, Goodnough LT, et al. Patient safety and blood transfusion: new solutions. Transfus Med Rev. 2003;17:169-180. 11. Discussion paper of the American Association of Blood Banks for the 2nd National Summit on Patient Safety Research. November 7, 2003. Available at http://www.aabb.org/content/ Members_Area/Members_Area_Regulatory/Patient_Safety/pss 101003.htm. Accessed July 25, 2004. 12. Wenz B, Burns ER. Improvement in transfusion safety using a new blood unit and patient identification system as part of safe transfusion practice. Transfusion. 1991;31:401-403. 13. Mercuriali F, Inghilleri G, Colotti MT, et al. One-year use of the Bloodloc system in an orthopedic institute. Transfus Clin Biol. 1994;1:227-230. 14. Mercuriali F, Inghilleri G, Colotti MT, et al. Bedside transfusion errors: analysis of 2 years use of a system to monitor and prevent transfusion errors. Vox Sang. 1996;70:16-20. 15. Linden JV, Paul B, Dressler KP. A report of 104 transfusion errors in New York State. Transfusion. 1992;32:601-606. 16. AuBuchon JP, Littenberg B. A cost-effectiveness analysis of the use of a mechanical barrier system to reduce the risk of mistransfusion. Transfusion. 1996;36:222-226. 17. Ehrlich A. Simple method helps to detect transfusion medicine errors. Hospitals. September 1976;50:89-90. 18. Jensen NJ, Crosson JT. An automated system for bedside verification of the match between patient identification and blood unit identification. Transfusion. 1996;36:216-221. 19. Marconi M, Langeberg AF, Sirchia G, et al. Improving transfusion safety by electronic identification of patients, blood samples and blood units. Immunohematology. 2000;16:82-85. 20. Turner CL, Casbard AC, Murphy MF. Barcode technology: its role in increasing the safety of blood transfusion. Transfusion. 2003;43:1200-1209. 426 Am J Clin Pathol 2006;126:422-426 Downloaded 426 from https://academic.oup.com/ajcp/article-abstract/126/3/422/1759935