Classification of posttransfusion adverse events using a publicly available artificial intelligence system.

BACKGROUND: Correct classification of transfusion reactions is important not only for effective patient care and donor management but also for accurate tracking of events in hemovigilance systems. We compared the ability of a generative artificial intelligence (AI) system to correctly diagnose hypothetical clinical situations as transfusion reactions in comparison to previous studies reporting the accuracy of transfusion medicine (TM) specialists in assessing these cases. METHODS: An AI system was requested to assess 36 case scenarios to provide a diagnosis, severity, and imputability of the transfusion reactions using the CDC National Healthcare Safety Network (NHSN) criteria. Responses were compared to an expert panel's classifications and to the published responses of a panel of TM specialists. Additionally, the AI's responses were compared to the TM specialists' prior attempts to use the TrDDx web-based algorithm for the five most challenging cases. RESULTS: The AI's classification accuracy varied widely depending on the NHSN category. The AI accurately classified all transfusion-associated circulatory overload and transfusion-related acute lung injury cases, exceeding TM specialists' assessments. Conversely, it did not correctly identify any cases in select NHSN categories such as DSTR. Overall accuracy among all diagnostic categories was 48.7% for AI responses versus 72.1% for prior TM specialist responses (p = 0.005). AI-generated responses included non-standard terminology, limited severity assessments, and no imputability determinations. DISCUSSION: A generative AI system may have a role in helping healthcare providers to consider transfusion reaction categories that might be missed, but caution is advised in applying the AI's output to transfusion reaction classification at present.

Evaluation of a new West Nile virus nucleic acid test for screening of blood donations.

BACKGROUND: West Nile virus (WNV) is transmitted to humans through mosquito bites and can be further transmitted to humans through transfusion or transplantation. Because most infected individuals are asymptomatic, blood donor screening is important in areas where WNV is endemic. These studies evaluated the performance of a new test for detection of WNV RNA in blood donations. STUDY DESIGN AND METHODS: Analytical performance evaluation included sensitivity, specificity, inclusivity, and correlation. A clinical specificity study was conducted at four blood donor testing laboratories in parallel with the cobas TaqScreen WNV Test (Roche Molecular Systems, Inc.). RESULTS: The 95% and 50% limit of detection for cobas WNV was 12.9 copies/mL (95% confidence interval [CI], 10.8-16.3) and 2.1 copies/mL (95% CI, 1.9-2.4) for WNV lineage 1, respectively, and 6.2 copies/mL (95% CI, 4.8-8.9) and 1.1 copies/mL (95% CI, 0.8-1.3) for WNV lineage 2, respectively. Clinical specificity was 100% in 10,823 donor samples tested individually (95% CI, 99.966%-100%) and 63,243 tested in pools of 6 (95% CI, 99.994%-100%). Samples of other members of the Japanese encephalitis virus serocomplex, including St Louis encephalitis, Japanese encephalitis, Murray Valley encephalitis, Usutu, and Kunjin viruses were detected by cobas WNV. CONCLUSION: The cobas WNV test for use on the cobas 6800/8800 System, a fully automated test system, demonstrated high sensitivity and specificity and is suitable for the detection of WNV in blood donors.

Sensitivity and specificity of a new automated system for the detection of hepatitis B virus, hepatitis C virus, and human immunodeficiency virus nucleic acid in blood and plasma donations.

BACKGROUND: Use of nucleic acid testing (NAT) in donor infectious disease screening improves transfusion safety. Advances in NAT technology include improvements in assay sensitivity and system automation, and real-time viral target discrimination in multiplex assays. This article describes the sensitivity and specificity of cobas MPX, a multiplex assay for detection of human immunodeficiency virus (HIV)-1 Group M, HIV-2 and HIV-1 Group O RNA, HCV RNA, and HBV DNA, for use on the cobas 6800/8800 Systems. STUDY DESIGN AND METHODS: The specificity of cobas MPX was evaluated in samples from donors of blood and source plasma in the United States. Analytic sensitivity was determined with reference standards. Infectious window periods (WPs) before NAT detectability were calculated for current donor screening assays. RESULTS: The specificity of cobas MPX was 99.946% (99.883%-99.980%) in 11,203 blood donor samples tested individually (IDT), 100% (99.994%-100%) in 63,012 donor samples tested in pools of 6, and 99.994% (99.988%-99.998%) in 108,306 source plasma donations tested in pools of 96. Seven HCV NAT-yield donations and one seronegative occult HBV infection were detected. Ninety-five percent and 50% detection limits in plasma (IU/mL) were 25.7 and 3.8 for HIV-1M, 7.0 and 1.3 for HCV, and 1.4 and 0.3 for HBV. The HBV WP was 1 to 4 days shorter than other donor screening assays by IDT. CONCLUSION: cobas MPX demonstrated high specificity in blood and source plasma donations tested individually and in pools. High sensitivity, in particular for HBV, shortens the WP and may enhance detection of occult HBV.

International validation of a dithiothreitol (DTT)-based method to resolve the daratumumab interference with blood compatibility testing.

BACKGROUND: Daratumumab (DARA) consistently interferes with routine blood bank serologic testing by directly binding to CD38 expressed on reagent red blood cells (RBCs). Treating RBCs with dithiothreitol (DTT) eliminates the DARA interference. We conducted an international, multicenter, blinded study aimed at validating the DTT method for use by blood bank laboratories worldwide. STUDY DESIGN AND METHODS: Paired plasma sample unknowns were sent to 25 participating blood bank laboratories. Sample 1 was spiked with DARA only (10 µg/mL), and Sample 2 with DARA plus a clinically significant RBC antibody (anti-D [n = 6], anti-Fya [n = 9], or anti-s [n = 10]). Sites were instructed to perform an antibody screen with and without DTT-treated RBCs and to use a DTT-treated RBC panel for antibody identification. Qualitative data about the DTT method were collected by online survey. The primary outcome was the proportion of study sites able to identify the antibody unknown in the presence of DARA. RESULTS: All sites observed the DARA interference with the antibody screen. The DARA interference was seen with all testing methods (gel, tube, or solid phase). Using the DTT method, 25 of 25 sites (100%) successfully identified the antibody unknown in the presence of DARA. Feedback on the DTT method was positive, with 17 of 19 (90%) sites responding to the survey indicating that they planned to use the DTT method to test clinical samples from DARA-treated patients. CONCLUSION: The DTT method is robust and reproducible and can be implemented by transfusion services worldwide to help provide safe blood products to patients treated with DARA.

Serological blind spots for variants of human IgG3 and IgG4 by a commonly used anti-immunoglobulin reagent.

BACKGROUND: Human immunoglobulin G (IgG) includes four different subtypes (IgG1, IgG2, IgG3, and IgG4), and it is also now appreciated that there are genetic variations within IgG subtypes (called isoallotypes). Twenty-nine different isoallotypes have been described, with 7, 4, 15, and 3 isoallotypes described for IgG1, IgG2, IgG3, and IgG4, respectively. The reactivity of anti-IgG with different isoallotypes has not been characterized. STUDY DESIGN AND METHODS: A novel monoclonal anti-K antibody (PugetSound Monoclonal Antibody 1 [PUMA1]) was isolated and sequenced, and a panel of PUMA1 variants was expressed, consisting of the 29 known IgG isoallotypes. The resulting panel of antibodies was preincubated with K-positive red blood cells (RBCs) and then subjected to testing with currently approved anti-IgG by flow cytometry, solid phase systems, gel cards, and tube testing. RESULTS: A US Food and Drug Administration (FDA)-approved monoclonal anti-IgG (gamma-clone) failed to recognize 2 of 15 IgG3 isoallotypes (IgG3-03 and IgG3-13) and 3 of 3 IgG4 isoallotypes (IgG4-01, IgG4-02, and IgG4-03). In contrast, an FDA-approved rabbit polyclonal anti-IgG recognized each of the known human IgG isoallotypes. CONCLUSION: These findings demonstrate "blind spots" in isoalloantibody detection by a monoclonal anti-IgG. If a patient has anti-RBC antibodies predominantly of an IgG3 subtype (the IgG3-03 and/or IgG3-13 variety), then it is possible that a clinically significant alloantibody would be missed. IgG-03 and IgG-13 have an estimated frequency of 1% to 3% in Caucasian populations and 20% to 30% in certain African populations. Nonreactivity with IgG4 is a known characteristic of this monoclonal anti-IgG, but IgG4 isoallotypes have not been previously reported.

The journey toward safer and optimized blood service in China: national strategy and progress.

BACKGROUND: The objective was to analyze and evaluate the effects of strategic measures the Chinese government attempted and undertook throughout the fast-changing and difficult transition periods of the blood service system. STUDY DESIGN AND METHODS: We systematically reviewed data and information regarding policy, blood collection, service system, and clinical transfusion practice, which were the most representative indicators on a national level from 1978 to 2015. RESULTS: Blood donation in China has successfully transitioned from paid donation, then compulsory (but nonremunerated) donation, to voluntary nonremunerated donation. The volume of blood collection has steadily increased from 4,000,000 red blood cell (RBC; 200 mL/unit) units in 1998 to 22,000,000 RBC units in 2014. The percentage of human immunodeficiency virus (HIV) transfusion-transmitted infections in newly diagnosed HIV/AIDS has reduced from 29.6% in 2005 to less than 0.15% in 2013. The use of component therapy has increased from 18% in 1989 to 99% in 2014. CONCLUSIONS: Involvement of the government through enforcement of new statutes and regulations helped improve blood safety and clinical transfusion practice.

AABB validation study of the CDC's National Healthcare Safety Network Hemovigilance Module adverse events definitions protocol.

BACKGROUND: The utility of a hemovigilance system depends on appropriate, reproducible application of system definitions. This is even more important when submissions are not reviewed by an adjudicating body. We sought to determine how participants would code adverse reactions at institutions that had or had not received training on the application of definitions used in the CDC's National Healthcare Safety Network Hemovigilance Module (HVM). STUDY DESIGN AND METHODS: Facilities that were (11) or were not (11) submitting adverse reaction data to the HVM reviewed 36 hypothetical cases containing elements of 37 case definitions from 12 different diagnostic groups. Respondents were required to determine the type of adverse event, if any, and assign a case definition (diagnostic probability), severity, and imputability using the January 2013 HVM Surveillance Protocol definitions. Those submitting HVM data had access to an instructional slide set prepared by CDC using similar hypothetical cases. Concordance with expert analysis was determined for the two groups of respondents. RESULTS: The frequencies of agreement with the expert assessment were not different according to prior training exposure in any of the diagnostic groups, and results were totaled across both groups. Response accuracy varied by type of categorization (adverse event type, 72.1%; match with case definition, 76.5%; severity, 69.6%; imputablity, 64.4%) and by type of adverse event. CONCLUSION: Despite delineated definitions, considerable variability in responses was seen, and this was not reduced by the available training. This degree of inconsistency in application of the surveillance definitions could degrade the utility of comparative reports.

A centralized recipient database enhances the serologic safety of RBC transfusions for patients with sickle cell disease.

OBJECTIVES: To evaluate the utility of a centralized transfusion service model in preventing the transfusion of incompatible units in patients with sickle cell disease (SCD). METHODS: The serologic records of transfused patients with SCD were reviewed. The index hospital was where an alloantibody was initially detected. RESULTS: In total, 150 patients with SCD were evaluated; 66 (44.0%) of 150 were alloimmunized. In 42 (63.6%) of these patients, 1 or more antibodies evanesced. The median number of hospitals visited by patients with SCD for RBC transfusion with 1 or more evanesced antibodies was three (range, one to eight); the median number of nonindex hospitals was two (range, one to seven). Of the patients with evanesced antibodies, 28.6% received transfusions at various nonindex hospitals 20 or more times after the antibody evanesced. CONCLUSIONS: A centralized database can help identify patients with SCD who have evanesced alloantibodies and prevent issuing incompatible RBC units.

Routine use of a rapid test to detect bacteria at the time of issue for nonleukoreduced, whole blood-derived platelets.

BACKGROUND: The Pan Genera detection (PGD) test is used to screen platelet (PLT) products for bacterial contamination. We report the experience of using the PGD test on whole blood-derived PLTs (WBPs) at two large centralized transfusion services (CTS). STUDY DESIGN AND METHODS: Records of PGD test results were retrospectively reviewed. The PGD test was performed on individual WBP units or pools of WBPs ranging in size from 2 to 6 units at the time of issue. Bacterial culture was performed on PLT products with positive PGD tests, and at one CTS, the available cocomponents. RESULTS: A total of 70,561 WBP pools were screened with the PGD test. There were seven true-positive PGD tests and 242 false-positive tests (positive predictive value of PGD test, 2.81%). The overall contamination rate was 99 per 10(6) WBP pools (1:10,080; 95% confidence interval [CI], 40-204), and the false-positive rate was 3430 per 10(6) WBP pools (1:292; 95% CI, 3011-3890). All seven bacterial isolates were Gram positive. The median age of the individual WBP units in the seven contaminated pools was 5 days (range, 3-5 days) compared to 4 days (range, 1-5 days) in the false-positive pools (p=0.0012). The same bacteria isolated from a positive PLT pool also grew in one red blood cell cocomponent. CONCLUSION: After testing more than 70,000 WBP pools at two large CTSs, the rate of contaminated WBP pools detected by the PGD test was 99 per 10(6) pools (1:10,080).

The immunohematologic and patient safety benefits of a centralized transfusion database.

BACKGROUND: The transfusion medical record is an important tool for providing safe and appropriate blood. However, many patients seek care at more than one hospital and this record is usually not portable. We posited that a centralized transfusion service database (CTS-D) offers benefits through tracking blood types, transfusion requirements, and detecting wrong blood in tube (WBIT). STUDY DESIGN AND METHODS: Records held in the CTS-D from 1997 to 2010 were queried to enumerate those seen at more than one hospital versus one hospital only. Transfusion-related attributes were collected including red blood cell (RBC) antibodies, transfusion requirements, and reactions. WBITs detected due to historical ABO typing were tallied. A review of blood orders that required alteration based on requirements held in the CTS-D was completed. RESULTS: There were 724,584 records; 10.9% of patients had been tested or received blood transfusion at more than one hospital. Of the 63,973 records with RBC alloantibodies, a greater proportion of patients were seen at more than one hospital versus one hospital only (7.11% vs. 3.97%, p < 0.005). Of the 97,687 patient records that required special processing, patients seen at one hospital had a lower rate than those at more than one hospital (12.13% vs. 24.59%, p < 0.005). There were 77 WBITs (0.18 WBITs per 1000 patients). An in-depth review of WBITs found an additional 26.3% (5 of 19) were detected because the current and historical ABO types were from two different hospitals within the CTS. CONCLUSIONS: The CTS-D provides a universal transfusion record that improves patient safety. As health care systems are enlarged, centralization of the transfusion component of the medical record should be considered.

Evolution in a centralized transfusion service.

The metropolitan Seattle area has utilized a centralized transfusion service model throughout the modern era of blood banking. This approach has used four laboratories to serve over 20 hospitals and clinics, providing greater capabilities for all at a lower consumption of resources than if each depended on its own laboratory and staff for these functions. In addition, this centralized model has facilitated wider use of the medical capabilities of the blood center's physicians, and a county-wide network of transfusion safety officers is now being developed to increase the impact of the blood center's transfusion expertise at the patient's bedside. Medical expectations and traffic have led the blood center to evolve the centralized model to include on-site laboratories at facilities with complex transfusion requirements (e.g., a children's hospital) and to implement in all the others a system of remote allocation. This new capability places a refrigerator stocked with uncrossmatched units in the hospital but retains control over the dispensing of these through the blood center's computer system; the correct unit can be electronically cross-matched and released on demand, obviating the need for transportation to the hospital and thus speeding transfusion. This centralized transfusion model has withstood the test of time and continues to evolve to meet new situations and ensure optimal patient care.

Evaluation of the properties of components prepared and stored after holding of whole blood units for 8 and 24 hours at ambient temperature.

BACKGROUND: The capability of holding whole blood (WB) units at ambient temperature, overnight, should help in platelet (PLT) concentrate preparation logistics. We summarize the results of a study conducted in the early 1990s that compared, in particular, PLT and red blood cell (RBC) in vivo viability properties following storage after preparation after 8- and 24-hour WB hold periods. STUDY DESIGN AND METHODS: Individuals donated units of WB on two occasions. Centrifugation at 20 to 24°C to separate PLTs and additive system RBC placement at 1 to 6°C was completed 8 hours after phlebotomy or after 24 hours in randomized order. Components were not leukoreduced. Studies including in vitro biochemical and hematologic analyses and autologous in vivo RBC and PLT evaluations were conducted at two sites. RESULTS: RBC 24-hour in vivo (mean ± SD) recoveries (single-label approach), after 35 days of storage, were 79.2 ± 4.3 and 79.4 ± 3.9% (n = 9; p > 0.05), with WB holding periods of 8 and 24 hours, respectively. With 42 days of storage, recovery after a 24-hour hold was slightly less than with an 8-hour hold (72.9 ± 6.5% vs. 76.0 ± 5.4%; n = 17; p < 0.05). RBC 2,3-diphosphoglycerate acid levels were substantially less after the 24-hour hold compared to after the 8-hour hold (n = 18; p < 0.05). PLT recovery after 5 days of storage with 8- and 24-hour hold periods were similar, 51.1 ± 14.9 and 50.6 ± 17.7%, respectively (n = 18; p > 0.05). The PLT survival variable and in vitro properties reflecting storage quality also showed no significant difference. CONCLUSION: RBC and PLT in vivo variables, and most in vitro variables, were not significantly different after storage with WB holding times of 8 and 24 hours except for a slight diminution of RBC recovery with the 24-hour hold after 42 days of storage.