ISSN 0042-9007 (Print) Volume 117 Number 2 FEBRUARY 2022

Wiley Editing
ISSN 1423-0410 (Online)

Services
There should be no barriers to getting your research
published, yet we know that manuscripts are often
returned for English language and formatting issues.

Let Wiley Editing Services provide you with expert help

to ensure your paper is ready for submission.
The International Journal of Transfusion Medicine
Don’t be rejected. Get published.

Volume 117 Number 2 February 2022

English Language Editing
Avoid English language errors
Get direct access to native English speakers who are experts in your
ȴHOGSURYLGLQJH[WHQVLYHHGLWVIRUODQJXDJHDQGVW\OH

Manuscript Formatting
6DYHYDOXDEOHWLPHIRUPDWWLQJWRDVSHFLȴFMRXUQDOVW\OH
Skilled formatters will ensure your manuscript and references are formatted
according to the style guide of your chosen journal.

Translation Service IN THIS ISSUE

Let your research do the talking 70 years of haemophilia care, a personal perspective

The International Journal of Transfusion Medicine

Receive accurate, high quality translation from Chinese, Spanish and
Perceptions on acceptability and reported
Portuguese by academics with advanced degrees.
consumption of marijuana by blood donors
prior to donation in the recreational use state
Figure Preparation of Colorado, USA
Improve the visual presentation of your research
Impacts of COVID-19 and elective surgery
:HJHQHUDWHSXEOLFDWLRQUHDG\ȴJXUHVWDEOHVDQGGLDJUDPVVHWWR\RXU
FKRVHQMRXUQDOȇVVSHFLȴHGVL]HUHVROXWLRQDQGOD\RXW cancellations on platelet supply and utilization
in the Canadian province of British Columbia
A first WHO reference reagent for the detection
Journal Recommendation
of anti-human platelet antigen
Make an informed decision
We will match your paper with 3-5 carefully selected journals from a range
of publishers that match your topic and publication goals.
18–465606

wileyeditingservices.com
 Vox Sanguinis
International Journal of Blood Transfusion
Official Journal of the International Society of Blood Transfusion
Founded 1956 by J. J. van Loghem, L. P. Holländer, J. Dausset, A. Hässig and J. Julliard (formerly Bulletin of the Central Laboratory of the Blood Transfusion Service of the Dutch Red
Cross, founded 1951)

Editor-in-Chief
Miquel Lozano, Barcelona, Spain

Section Editors
Blood Component Collection and Production International Forum
Denese C. Marks, Sydney, Australia Nancy M. Dunbar, Lebanon, NH, USA
Cellular Therapy Patient Blood Management
Zbigniew ‘Ziggy’ M. Szczepiorkowski, Lebanon, NH, USA Nelson Tsuno, Tokyo, Japan
Donors and Donations Reviews
Katja van den Hurk, Amsterdam, the Netherlands Zbigniew ‘Ziggy’ M. Szczepiorkowski, Lebanon, NH, USA
Haemovigilance Leo van de Watering, Amsterdam, the Netherlands
Claudia Cohn, Minneapolis, MN, USA Transfusion Medicine and New Therapies
Immunohaematology and Immunogenetics Pierre Tiberghien, Paris, France
Jill R. Storry, Lund, Sweden Transfusion-transmitted Disease and its Prevention
Clive Seed, Perth, Australia

Editorial Board
Arwa Al-Riyami, Muscat, Oman Wolfgang R. Mayr, Vienna, Austria
Claire Armour Barrett, Bloemfontein, South Africa Pieter van der Meer, Amsterdam, the Netherlands
Thierry Burnouf, Taipei, Taiwan Celina Montemayor, Toronto, Canada
Andreas Buser, Basel, Switzerland Shirley Owusu-Ofori, Kumasi, Ghana
Marcela Contreras, London, UK Luca Pierelli, Rome, Italy
Dana Devine, Vancouver, Canada France Pirenne, Créteil, France
Hendrik Feys, Mechelen, Belgium Sandra Ramirez, Ottawa, Canada
Ruchika Goel, Springfield, IL, USA Veera Sekaran Nadarajan, Kuala Lumpur, Malaysia
Salwa Hindawi, Jeddah, Saudi Arabia Rattiram Sharma, Chandigarh, India
Yanli Ji, Guangzhou, China Eilat Shinar, Ramat Gan, Israel
Micky Koh, London, UK Claude Tayou Tagny, Yaounde, Cameroon
Linda Larsson, Stockholm, Sweden Vip Viprakasit, Bangkok, Thailand
Bridon M’Baya, Blantyre, Malawi Silvano Wendel, São Paulo, Brazil

Scientific/Medical Illustrator Editorial Office

Alison Schroeer, Thompson, CT, USA Maria Davie, Edinburgh, UK
Technical Editor Production Editor
Doug Huestis, Tucson, AZ, USA Shirley Villeza, Manila, the Philippines

ISBT Standing Committee on Vox Sanguinis Past Editors-in-Chief

Gwen Clarke, Chairperson, Edmonton, Canada J. J. van Loghem, 1956–1960
Lin Fung, Brisbane, Australia W. H. Crosby, 1960–1963 (N. and S. America)
Eric Jansen, Amsterdam, the Netherlands L. P. Holländer, 1960–1970 (Europe)
Diana Teo, Singapore F. H. Allen, 1963–1977 (N. and S. America)
Miquel Lozano, Editor-in-Chief, Barcelona, Spain M. G. Davey, 1970–1980 (Africa, Asia and Australia)
N. R. Rose, 1977–1980 (N. and S. America)
Observers
C. P. Engelfriet, 1977–1996
Erica Wood, ISBT President, Melbourne, Australia
M. Contreras, 1996–2003
Jenny White, ISBT Executive Director, Amsterdam,
W. R. Mayr, 2003–2011
the Netherlands
D. Devine, 2011–2020
Claire Dowbekin, Publishing Manager, Wiley, Oxford, UK
 Vox Sanguinis
International Journal of Blood Transfusion
Aims and Scope
Vox Sanguinis reports on all issues related to transfusion medicine, from donor vein to recipient vein, including cellular therapies. Comments, reviews, original articles,
short reports and international fora are published, grouped into 10 main sections:
  1. Blood Component Collection and Production: Blood collection methods and devices (including apheresis); blood component preparation; inventory management;
collection and storage of tissues; quality management and good manufacturing practice; automation and information technology; plasma fractionation techniques
and plasma derivatives;
  2. Cellular Therapy: Cell-based therapies; CAR T-cell therapies; genetically modified cell therapies; cellular therapy (sources; products; processing and storage); stem
cells; cell-based regenerative medicine; cellular immunotherapy; molecular therapy;
  3. Donors and Donations: Donor recruitment and retention; donor selection; donor health
  4. Haemovigliance: Adverse events in blood and blood component donors and transfusion recipients; corrective and preventive measures of complications; near-misses
and errors in the transfusion chain; evaluation and outcomes of adverse events
  5. Immunohaematology and Immunogenetics: autoimmunity in haematology; alloimmunity of blood; pre-transfusion testing; complement in immunohaematology;
blood phenotyping and genotyping; genetic markers of blood cells and serum proteins: polymorphisms and function; parentage testing and forensic
immunohaematology;
  6. International Forum: Section in which topics related to any aspects of the transfusion chain (from technical to scientific, including socioeconomic and ethical facets)
are discussed by invited participants and summarized by guest editors
  7. Patient Blood Management: Bloodless surgery; preoperative anaemia; minimizing blood loss during surgery; alternatives to blood transfusion;
  8. Review: Comprehensive reviews or short comments of any aspect related to transfusion medicine;
  9. Transfusion Medicine: Transfusion indication, transfusion practice, thresholds and audits; transfusion efficacy assessment, clinical trials; therapeutic apheresis;
10. Transfusion-transmitted Disease and its Prevention: Identification and epidemiology of infectious pathogens transmissible by blood; donor testing for transfusion-
transmissible infectious pathogens; bacterial contamination of blood components; pathogen inactivation;
This comprehensive coverage has made the journal essential reading for a wide range of specialists interested in the present state of transfusion research and practice.
Journal Customer Services
For ordering information, claims and any enquiry concerning your journal subscription please go to https://hub.wiley.com/community/ support/onlinelibrary or contact
your nearest office. Americas: Email: [email protected]; Tel: +1 781 388 8598 or +1 800 835 6770 (toll free in the USA & Canada). Europe, Middle East and Africa:
Email: [email protected]; Tel: +44 (0) 1865 778315. Asia Pacific: Email: [email protected]; Tel: +65 6511 8000. Japan: For Japanese-speaking support, Email: cs-
[email protected]. Visit our Online Customer Help at https://hub.wiley.com/community/support/ onlinelibrary.
Information for Subscribers: Vox Sanguinis is published in 12 issues per year. Institutional subscription prices for 2022 are: Print & Online: US$2196 (US), US$2560 (Rest
of World), €1533 (Europe), £1189 (UK). Prices are exclusive of tax. Asia-Pacific GST, Canadian GST/HST and European VAT will be applied at the appropriate rates. For more
information on current tax rates, please go to www. wileyonlinelibrary.com/ tax-vat. The price includes online access to the current and all online backfiles to 1 January 2018,
where available. For other pricing options, including access information and terms and conditions, please visit www.wileyonlinelibrary.com/access.
Delivery Terms and Legal Title: Where the subscription price includes print issues and delivery is to the recipient’s address, delivery terms are Delivered at Place (DAP);
the recipient is responsible for paying any import duty or taxes. Title to all issues transfers FOB our shipping point, freight prepaid. We will endeavour to fulfil claims
for missing or damaged copies within six months of publication, within our reasonable discretion and subject to availability.
Back Issues: Single issues from current and recent volumes are available at the current single issue price from [email protected]. Earlier issues may be obtained from
Periodicals Service Company, 351 Fairview Avenue – Ste 300, Hudson, NY 12534, USA. Tel: +1 518 822-9300, Fax: +1 518 822-9305, Email: [email protected]
Abstracting and Indexing Services: The Journal is indexed by Abstracts in Anthropology (Sage); Abstracts on Hygiene & Communicable Diseases (CABI); Academic Search
(EBSCO Publishing); Academic Search Alumni Edition (EBSCO Publishing); Academic Search Premier (EBSCO Publishing); AGRICOLA Database (National Agricultural
Library); BIOBASE: Current Awareness in Biological Sciences (Elsevier); Biological Abstracts (Clarivate Analytics); BIOSIS Previews (Clarivate Analytics); CAB Abstracts®
(CABI); CABDirect (CABI); CAS: Chemical Abstracts Service (ACS); CSA Biological Sciences Database (ProQuest); CSA Environmental Sciences & Pollution Management
Database (ProQuest); CSA Virology & AIDS Abstracts (ProQuest); Current Contents: Life Sciences (Clarivate Analytics); Embase (Elsevier); Global Health (CABI); HEED:
Health Economic Evaluations Database (Wiley-Blackwell); Index Veterinarius (CABI); Journal Citation Reports/Science Edition (Clarivate Analytics); MEDLINE/PubMed
(NLM); Nutrition Abstracts & Reviews Series A: Human & Experimental (CABI); Pig News & Information (CABI); ProQuest Central (ProQuest); ProQuest Health & Medical
Complete (ProQuest); ProQuest Research Library (ProQuest); Protozoological Abstracts (CABI); PubMed Dietary Supplement Subset (NLM); Review of Medical &
Veterinary Entomology (CABI); Review of Medical & Veterinary Mycology (CABI); Rural Development Abstracts (CABI); Science Citation Index (Clarivate Analytics);
Science Citation Index Expanded (Clarivate Analytics); Tropical Diseases Bulletin (CABI); Veterinary Bulletin (CABI).
Copyright and Copying (in any format): Copyright © 2022 International Society of Blood Transfusion. All rights reserved. No part of this publication may be reproduced,
stored or transmitted in any form or by any means without the prior permission in writing from the copyright holder. Authorization to copy items for internal and personal
use is granted by the copyright holder for libraries and other users registered with their local Reproduction Rights Organisation (RRO), e.g. Copyright Clearance Center
(CCC), 222 Rosewood Drive, Danvers, MA 01923, USA (www.copyright.com), provided the appropriate fee is paid directly to the RRO. This consent does not extend to other
kinds of copying such as copying for general distribution, for advertising or promotional purposes, for republication, for creating new collective works or for resale.
Permissions for such reuse can be obtained using the RightsLink “Request Permissions” link on Wiley Online Library. Special requests should be addressed to: permissions@
wiley.com
Open Access: Vox Sanguinis accepts articles for Open Access publication. Please visit https://onlinelibrary.wiley.com/page/journal/14230410/homepage/fundedaccess.
html for further information about Open Access.
Copyright Policy: Papers accepted must be licensed for publication in Vox Sanguinis and a completed Copyright Transfer Agreement Form must accompany
every accepted paper. Authors will be required to license copyright in their paper to John Wiley & Sons Ltd. Upon acceptance of an article, corresponding authors
must log into Author Services to complete the licence agreement of their paper.
Wiley’s Corporate Citizenship initiative seeks to address the environmental, social, economic, and ethical challenges faced in our business and which are important to our
diverse stakeholder groups. Since launching the initiative, we have focused on sharing our content with those in need, enhancing community philanthropy, reducing our
carbon impact, creating global guidelines and best practices for paper use, establishing a vendor code of ethics, and engaging our colleagues and other stakeholders in our
efforts. Follow our progress at www.wiley.com/go/citizenship. Wiley is a founding member of the UN-backed HINARI, AGORA, and OARE initiatives. They are now
collectively known as Research4Life, making online scientific content available free or at nominal cost to researchers in developing countries. Please visit Wiley’s Content
Access – Corporate Citizenship site: http://www.wiley.com/WileyCDA/Section/id-390082.html. The journal to which you are submitting your manuscript employs a
plagiarism detection system. By submitting your manuscript to this journal you accept that your manuscript may be screened for plagiarism against previously published
works.
Disclaimer: The Publisher, International Society of Blood Transfusion and Editors cannot be held responsible for errors or any consequences arising from the use of
information contained in this journal; the views and opinions expressed do not necessarily reflect those of the Publisher or the International Society of Blood Transfusion
and Editors, neither does the publication of advertisements constitute any endorsement by the Publisher or the International Society of Blood Transfusion and Editors of
the products advertised.
VOX SANGUINIS (Online ISSN: 1423-0410 Print ISSN: 0042-9007) is published monthly. US mailing agent: Mercury Media Processing, LLC, 1850 Elizabeth Avenue, Suite
#C, Rahway, NJ 07065 USA. Periodical postage paid at Rahway, NJ. Postmaster: Send all address changes to VOX SANGUINIS, John Wiley & Sons Inc., C/O The Sheridan
Press, PO Box 465, Hanover, PA 17331, USA. For submission instructions, subscription and all other information visit: www.wileyonlinelibrary.com/journal/vox. Printed
in Singapore by C.O.S. Printers Pte Ltd.
 Vox Sanguinis (2022) 117, 153–296

117/2/2022 © 2022 International Society of Blood Transfusion

Contents

Review Transfusion Medicine and New Therapies

157 Recommendation for validation and quality 208 Estimation of the latent therapeutic demand for
assurance of non-invasive prenatal testing for immunoglobulin therapies in autoimmune
foetal blood groups and implications for IVD risk neuropathies in the United States A. Farrugia,
classification according to EU regulations M. Bansal & I. Marjanovic
F. B. Clausen, Åsa Hellberg, G. Bein, P. Bugert, 220 Therapeutic granulocyte infusion for patients with
D. Schwartz, T. D. Drnovsek, K. Finning, K. Guz, severe neutropaenia and neutrophilic dysfunction:
K. Haimila, C. Henny, H. O'Brien, A. Orzinska, New Zealand experience S. Chung,
K. Sørensen, S. Thorlacius, A. Wikman, O. Armstrong-Scott & R. Charlewood
G. A. Denomme, W. A. Flegel, C. Gassner, M. de Haas,
C. Hyland, Y. Ji, W. J. Lane, N. Nogués, M. L. Olsson, 227 Early packed red blood cell transfusion in major
T. Peyrard, C. E. van der Schoot, C. Weinstock trauma patients: Evaluation and comparison of
& T. Legler different prediction scores for massive transfusion
A. Cassignol, P. Bertein, P. Botti, A. Renard, J. Cotte &
Commentary J. Bordes

166 Seventy years of haemophilia care: A personal 235 The impact of COVID-19 outbreak on the
perspective C. Smit Transfusion Medicine Unit of a Northern Italy
Hospital and Cancer Centre D. Schiroli, L. Merolle,
G. Molinari, E. Di Bartolomeo, D. Seligardi, L. Canovi,
Original Articles
T. A. Pertinhez, P. Mancuso, P. Giorgi Rossi,
Donors and Donations R. Baricchi & C. Marraccini
169 Determinants of the intention to donate umbilical 243 Immediate intravenous iron administration
cord blood in pregnant women M. Fernandes, improves anaemia recovery following total knee
G. Alessandri, R. Abbad & C. Grano arthroplasty: A propensity-matched analysis
177 Perceptions on acceptability and reported H.-S. Park, S.-I. Bin, H.-J. Kim, J. Kim, H. Kim, Y. Ro &
consumption of marijuana by blood donors prior to W. U. Koh
donation in the recreational use state of Colorado, 251 Impacts of COVID-19 and elective surgery
USA K. Annen & M. G. DomBourian cancellations on platelet supply and utilization in
Blood Component Collection and Production the Canadian Province of British Columbia
K. E. Shopsowitz, C. Lim, A. W. Shih, N. Fishbane,
185 Qualifying coronavirus disease 2019 convalescent
B. R. Berry, M. Bigham, T. Petraszko, J. Trudeau,
plasma donors in Israel M. Izak, V. Gendelman,
M. Wyatt, M. T. S. Yan & D. Morrison
S. Bransburg-Zabary, E. Stoyanov, R. Gat, D. Cohen,
J. Chen, Y. Maor, A. Benov, B. Lev, O. Zimhony 259 Improvement of transfusion practice and reduction
& E. Shinar in red blood cell utilization in Belgian hospitals:
Results of a national survey and benchmarking
193 DEHT is a suitable plasticizer option for
J. Vanden Broeck, K. Beeckman, E. Van Gastel,
phthalate-free storage of irradiated red blood cells
L. De Keersmaecker, T. Devos, C. Gérard, L. Noens,
L. Larsson, S. Ohlsson, J. Derving, B. Diedrich,
D. Putzeys, K. Van Poucke, M. Haelterman,
P. Sandgren, S. Larsson & M. Uhlin
V. Deneys & R. Schots
Transfusion-transmitted Disease and its Prevention
Immunohaematology
201 HIV residual risk in Canada for apheresis source
268 A comparative study on perinatal outcomes of red
plasma donation without deferral for men who have
blood cell–alloimmunized pregnancies with
sex with men E. Aubé, A. Lewin, S. F. O'Brien,
anti-RhD in combination and anti-RhD alone in
Y. Grégoire, J. Pillonel, W. R. Steele, B. Custer,
China M. Yu, T. Tang, R. Zheng, M. Situ & J. Feng
K. L. Davison, M. Germain, C. R. Seed & F. Camirand
Lemyre, the Surveillance, Risk Assessment and 275 A first WHO reference reagent for the detection of
Policy Subgroup of the ISBT Transfusion anti-human platelet antigen-15b G. Sharp, A. Poles
Transmitted Infectious Diseases Working Party & L. Studholme
t
ea
n lin
a lo vo
x
ou
rn
n al/
i sj jo ur
th m/
This journal is available online at Wiley Online Library. w
co
V ie y.
Visit wileyonlinelibrary.com to search the articles r ar
lib
and register for table of contents e-mail alerts. li ne
on
ley
wi
Contents

282 Estimation of Lewis-negative alleles by 291 Risk of future haemolytic disease of the fetus
high-resolution melting analysis of three tag SNPs and newborn following the transfusion of
of FUT3 M. Soejima & Y. Koda Rh(D)-positive blood products to Rh(D)-negative
children M. H. Yazer, P. C. Spinella & J. N. Seheult
Letters to the Editor 293 Corrigendum
288 Using blood services platforms to facilitate 294 Diary of Events
COVID-19 vaccination programs E. Jaffe,
I. Abramovich, R. Sonkin & E. Shinar
289 The association of B blood group with SARS-CoV-
2-induced death in Babol, north of Iran S. Hassani,
N. Kalantari, S. F. Jalali, M. Shams, Z. Ahmadnia,
M. Aghajani Daroonkola, M. Bayani & H. Ghorbani
Received: 1 March 2021 Revised: 20 April 2021 Accepted: 4 June 2021

DOI: 10.1111/vox.13172

REVIEW ARTICLE

Recommendation for validation and quality assurance

of non-invasive prenatal testing for foetal blood groups
and implications for IVD risk classification according to EU
regulations

Frederik Banch Clausen1,2 | Åsa Hellberg2,3 | Gregor Bein4 | Peter Bugert5 |

Dieter Schwartz6 | Tadeja Dovc Drnovsek7 | Kirstin Finning8 | Katarzyna Guz9 |
Katri Haimila10 | Christine Henny11 | Helen O’Brien12 | Agnieszka Orzinska9 |
Kirsten Sørensen13 | Steinunn Thorlacius14 | Agneta Wikman15 |
2,16 2,17
Gregory Andrew Denomme | Willy Albert Flegel | Christoph Gassner2,18 |
Masja de Haas2,19,20 | Catherine Hyland2,12,21 | Yanli Ji2,22 |
2,23,24 2,25 2,3,26
William J. Lane | Núria Nogués | Martin L. Olsson |
Thierry Peyrard2,27 | C. Ellen van der Schoot2,28 | Christof Weinstock2,29 |
Tobias Legler30
1
Laboratory of Blood Genetics, Department of Clinical Immunology, Copenhagen University Hospital, Copenhagen, Denmark
2
cfDNA subgroup from the International Society of Blood Transfusion (ISBT) Working Party on Red Cell Immunogenetics and Blood Group Terminology (RCIBGT),
Amsterdam, The Netherlands
3
Department of Clinical Immunology and Transfusion Medicine, Office for Medical Services, Region Skåne, Sweden
4
Institute for Clinical Immunology and Transfusion Medicine, Justus-Liebig-University, Giessen, Germany
5
Institute of Transfusion Medicine and Immunology, Heidelberg University, Medical Faculty Mannheim, German Red Cross Blood Service Baden Württemberg –
Hessen, Mannheim, Germany
6
Department of Blood Group Serology and Transfusion Medicine, Medical University of Vienna, Vienna, Austria
7
Blood Transfusion Centre of Slovenia, Ljubljana, Slovenia
8
National Health Service Blood and Transplant, International Blood Group Reference Laboratory, UK
9
Institute of Hematology and Transfusion Medicine, Warsaw, Poland
10
Finnish Red Cross Blood Service, Helsinki, Finland
11
Interregional Blood Transfusion SRC, Berne, Switzerland
12
Clinical Services and Research, Australian Red Cross Lifeblood, Brisbane, Australia
13
Department of Immunology and Transfusion Medicine, Oslo University Hospital, Oslo, Norway
14
The Blood Bank, Landspitali University Hospital, Reykjavik, Iceland
15
Clinical Immunology and Transfusion Medicine Karolinska University Hospital and CLINTEC Karolinska Institutet, Stockholm, Sweden
16
Versiti Blood Research Institute and Diagnostic Laboratories, Milwaukee, Wisconsin, USA
17
Department of Transfusion Medicine, NIH Clinical Center, National Institutes of Health, Bethesda, Maryland, USA
18
Institute for Translational Medicine, Private University in the Principality of Liechtenstein, Triesen, Liechtenstein
19
Department of Immunohaematology Diagnostic Services, Sanquin Diagnostic Services and Sanquin Research, Amsterdam, The Netherlands
20
Department of Haematology, Leiden University Medical Center, Leiden, The Netherlands
21
School of Biomedical Sciences, Queensland University of Technology, Brisbane, Queensland, Australia

The content is solely the responsibility of the authors and does not necessarily represent the view of the authors’ employer, institute or national authorities.

Vox Sanguinis. 2022;117:157–165. wileyonlinelibrary.com/journal/vox © 2021 International Society of Blood Transfusion. 157
158 CLAUSEN ET AL.

22
Guangzhou Blood Center, Institute of Clinical Blood Transfusion, Guangzhou, China
23
Department of Pathology, Brigham and Women’s Hospital, Boston, Massachusetts, USA
24
Harvard Medical School, Boston, Massachusetts, USA
25
Immunohematology Laboratory, Blood and Tissue Bank, Barcelona, Spain
26
Department of Laboratory Medicine, Lund University, Lund, Sweden
27
Institut National de la Transfusion Sanguine, Centre National de Référence pour les Groupes Sanguins, Paris, France
28
Department of Experimental Immunohematology, Sanquin Research, Amsterdam, The Netherlands
29
Institute of Clinical Transfusion Medicine and Immunogenetics Ulm, German Red Cross Blood Service Baden-Württemberg-Hessen, and Institute of Transfusion
Medicine, University of Ulm, Ulm, Germany
30
Department of Transfusion Medicine, University Medical Center Göttingen, Göttingen, Germany

Correspondence
Frederik Banch Clausen, Laboratory of Blood Abstract
Genetics, Department of Clinical Immunology,
Background and Objectives: Non-invasive assays for predicting foetal blood group
Copenhagen University Hospital, Copenhagen,
Denmark. status in pregnancy serve as valuable clinical tools in the management of pregnan-
Email: [email protected]
cies at risk of detrimental consequences due to blood group antigen incompatibility.
Funding information To secure clinical applicability, assays for non-invasive prenatal testing of foetal
None blood groups need to follow strict rules for validation and quality assurance. Here,
we present a multi-national position paper with specific recommendations for vali-
dation and quality assurance for such assays and discuss their risk classification
according to EU regulations.
Materials and Methods: We reviewed the literature covering validation for in-vitro
diagnostic (IVD) assays in general and for non-invasive foetal RHD genotyping in
particular. Recommendations were based on the result of discussions between co-
authors.
Results: In relation to Annex VIII of the In-Vitro-Diagnostic Medical Device Regulation
2017/746 of the European Parliament and the Council, assays for non-invasive prena-
tal testing of foetal blood groups are risk class D devices. In our opinion, screening for
targeted anti-D prophylaxis for non-immunized RhD negative women should be placed
under risk class C. To ensure high quality of non-invasive foetal blood group assays
within and beyond the European Union, we present specific recommendations for vali-
dation and quality assurance in terms of analytical detection limit, range and linearity,
precision, robustness, pre-analytics and use of controls in routine testing. With respect
to immunized women, different requirements for validation and IVD risk classification
are discussed.
Conclusion: These recommendations should be followed to ensure appropriate assay
performance and applicability for clinical use of both commercial and in-house assays.

KEYWORDS
blood group, cell-free DNA, EU, foetal RHD genotyping, HDFN, quality assurance, validation

I N T R O D U CT I O N maintenance and quality assurance. Medical laboratories that offer

clinical testing are obliged to assess the information about the perfor-
In the care and management of pregnancies at risk due to blood group mance of an in-vitro diagnostic (IVD) assay provided by the manufac-
incompatibility, non-invasive foetal blood group genotyping based on turer or to validate their own in-house assay. In the European Union,
the analysis of cell-free foetal DNA from maternal plasma can be used IVD devices are classified on a scale of increasing risk to patients and
to predict the foetal blood group status [1]. For any diagnostic assay the public (class A, B, C or D) [2]. Blood group typing assays for trans-
used in medical laboratories, strict rules apply for validation, fusion, transplantation and cell administration are split between the
RECOMMENDATIONS FOR NON-INVASIVE TESTING OF FOETAL BLOOD GROUPS 159

highest risk class D (ABO, Rh, Kell, Kidd and Duffy systems) and were discussed among international experts at two conferences, first
class C (all other blood group systems) [2]. raised and discussed at the International Meeting of Cell-Free DNA, in
Non-invasive prenatal testing of foetal blood groups can have dif- Copenhagen 2019, and thereafter discussed by the cell-free DNA
ferent clinical objectives. Pregnant women can become immunized (cfDNA) subgroup from the International Society of Blood Transfusion
against a specific foetal blood group antigen, where the woman can (ISBT) Working Party on Red Cell Immunogenetics and Blood Group
produce IgG antibodies that, either in the current or next pregnancy, Terminology (RCIBGT), at the ISBT meeting in Basel in 2019. This
can be transported across the placenta and facilitate the destruction multi-national position paper was circulated among all co-authors,
of foetal red blood cells leading to haemolytic disease of the foetus whose comments were then addressed until this consensus was
and newborn (HDFN) [3]. Non-invasive prenatal testing of foetal reached. The recommendations of the manuscript are formulated and
blood groups can assist in the management of immunized women, endorsed by the cfDNA subgroup from the ISBT RCIBGT working
where the determination of foetal antigens helps clinicians to verify party.
the diagnosis and plan for timely treatment. A different objective
exists for non-immunized RhD negative women, where antenatal and
postnatal anti-D prophylaxis is administered to prevent the women RE SU LT S
from becoming immunized. In some countries, antenatal prophylaxis is
administered to all RhD negative women as universal prophylaxis, Legislation for the performance evaluation of IVD in
despite having no effect or purpose when a woman carries an RhD the European Union
negative foetus [4]. Rhesus immunoglobulin (RhIg) is regarded as com-
patible with foetal red blood cells; however, as a blood-derived prod- IVD assays for non-invasive prenatal testing of foetal blood groups
uct, it does possess a potential risk of carrying a contaminant or are medical devices. In May 2017, the In-Vitro-Diagnostic Medical
infectious agent and should therefore be used only when necessary Device Regulation (IVDR, 2017/746) [2] of the European Parliament
[5]. Moreover, there is worldwide shortage of RhIg [6]. In addition, and the Council about IVDs came into force. After transition periods,
RhIg donors require active (hyper) immunization, which, from an ethi- this regulation is repealing the previous Medical Devices Directive
cal perspective, further promotes the use of RhIG only when neces- 98/79/EC [9]. According to the new regulation, IVD is categorized
sary. A non-invasive prenatal test for foetal RhD (NIPT RhD) can into four risk classes A, B, C and D. In 2019, a corrigendum specified
identify those women who will benefit from prophylaxis, hereby guid- that devices used ‘to determine foeto-maternal blood group incom-
ing so-called targeted prophylaxis. As such, the prediction of the foe- patibility’ are included with devices ‘intended to be used for blood
tal RhD status helps by restricting the administration of prophylaxis, grouping,’ thus placing non-invasive prenatal testing of foetal
thus avoiding unnecessary treatment and unnecessary use of valuable blood groups under risk class D [10].
RhIg [7, 8]. Consequently, we find that the clinical objectives as well Assays that are used for blood group determination in order to
as the inherent risks associated with the clinical applications of non- guarantee immunocompatibility of blood, blood components, cells, tis-
invasive prenatal testing of foetal blood groups can be highly differ- sue or organs used for transfusion, transplantation or cellular therapy
ent, which should be reflected in the risk classification. need to fulfil the highest standards for risk class D IVD. In our opinion,
This report considers recommendations for the classification of non-invasive foetal RHD screening belongs to a different category in
IVD assays specifically for non-invasive prediction of foetal RHD terms of both severity and frequency of side effects due to unneces-
status, when applied as screening of non-immunized pregnant women sarily given or withheld administration of universal RhIg prophylaxis,
to guide targeted prophylaxis. In relation to Annex VIII of the In-Vitro- compared with side effects from incompatible transfusion due to
Diagnostic Medical Device Regulation 2017/746 of the European Par- incorrect blood group determination. It is therefore unnecessary
liament and the Council [2], we recommend that this IVD should be to categorize reagents for the non-invasive determination of the foe-
placed in risk class C. tal RhD status according to risk class D. In our opinion, these IVDs
In addition, we present specific requirements for assay validation belong to risk class C. Similarly, as already listed in the IVDR, such
and quality assurance. This set of recommendations should be taken reagents are equal to reagents with the intended use to screen for
into consideration by those responsible for the introduction non- genetic diseases in embryo or foetus, which are placed under risk
invasive prenatal testing of foetal blood groups in diagnostic class C.
laboratories. To establish recommendations for foeto-maternal blood
group testing, published recommendations for the validation of
molecular qualitative virus diagnostics tests can be used as refer-
METHODS ence points [11]. In addition, real-time polymerase chain reaction
(PCR) details have been published for non-invasive testing of
The authors, consisting of the leading experts in the field, reviewed >50,000 cases of RhD negative pregnant women using several
the literature about IVD assay validation in general and publications protocols [7, 12–25], demonstrating high assay performance, and
about performance studies dealing primarily with foetal RHD assays. from which the extensive experience can be applied to establish
Recommendations were formulated after discussions. Pertinent issues specific recommendations.
160 CLAUSEN ET AL.

Reagents, including calibrator and control reagents released for transfusion in an alloimmunized recipient can lead to severe
the determination of the blood groups belonging to the ABO, Rh haemolytic transfusion reactions. In this risk assessment, it should also
(C, c, D, E, e), and Kell systems, need to fulfil particular requirements. be noted that non-invasive foetal RHD genotyping has been reported
In 2002, common technical specifications (CTS) for IVD medical to identify more foetal RhD positive cases which were false negative
devices (CTS, Commission Decision 2002/364/EC) were laid down, by standard cord blood serology than causing false-negative results,
which include detailed requirements for lot release tests and perfor- effectively decreasing the overall risk of immunization [20].
mance evaluation of antibodies distributed for the determination of
blood group antigens [11]. In agreement with colleagues, Institut de
Biotechnologies Jacques Boy (Reims, France) [26], Deutsche Gesell- Estimate of the number of specimens required for
schaft für Transfusionsmedizin und Immunhämatologie (DGTI) [27] determining diagnostic accuracy
and the cfDNA subgroup of the ISBT RCIBGT working party, we find
that reagents with the intended use ‘Determination of the foetal RHD The total number to be investigated for the determination of diagnos-
status from maternal plasma for targeted anti-D prophylaxis’ are not tic accuracy is an important figure of any performance evaluation. For
comparable with antibodies used for the determination of the blood the validation of new anti-RH1 (anti-D) reagents, CTS requires a sam-
group antigen RhD (RH1), which precedes and aims to ensure an RhD ple of 3000 specimens for a new formulation; or if reagents have been
compatible transfusion of red blood cells. Manufacturers of non- clearly characterized, 1000 specimens still have to be investigated;
invasive foetal blood genotyping assays may consider CTS while plan- and if the application field or the mode of use changes, 500 specimens
ning a performance evaluation study; however, the requirements need need to be analysed. The CTS focusses on the safety of blood and
to be different due to sample specification (plasma instead of whole organ donations and thus requires extremely accurate results when
blood), target molecules (nucleic acids instead of red cells), reagent IVD is applied. For this reason, CTS confines to infectious disease
properties (e.g., oligonucleotides instead of antibodies), methodology tests for blood donation screening and to blood group serology
(nucleic acid amplification test instead of haemagglutination test) and assays. The requirements for the performance evaluation of anti-D
a different purpose of the test as well as considerably different immu- sera/reagents also include tests that show that RhD variants are
nization risks (as further discussed below). detected.
Based on the risk assessments presented above, a performance
evaluation study with 1000 specimens is considered sufficient when a
Non-invasive foetal RHD screening new technology for the determination of the foetal RHD status from
maternal blood is used (Table 1). With a new technology, assay perfor-
Risk assessment mance should be evaluated with diagnostic sensitivity and specificity.
For such evaluation, it is recommended that the lower limit of the
In the context of blood transfusion, high CTS demands are reasonable 95% confidence interval of the sensitivity should be ≥99% and that
because a false-negative RhD determination in blood donors or a the lower limit of the 95% confidence interval of the specificity should
false-positive determination in recipients of red cell concentrates may be ≥96%. Cord blood genotyping can be applied to resolve
result in anti-D immunization in 3%–70% of cases [28]. In contrast, a discrepancies.
false-negative result for the foetal RHD status, when applied as If a laboratory makes minor modifications to an existing published
screening to guide only targeted prophylaxis during pregnancy, entails protocol, fewer specimens are needed. Based on the extensive litera-
the omission of an antenatal anti-D prophylaxis only. If RhD is deter- ture about the application of real-time PCR for the determination of
mined in the newborn (e.g., from cord blood), this error is revealed, the foetal RHD status [7, 8, 12–25, 33–36], with real-time PCR details
and postnatal anti-D prophylaxis is administered accordingly. This published for >50,000 cases from several protocols [7, 12–25], the
approach is associated with an increase of the immunization risk from investigation of 100 specimens is sufficient for a performance evalua-
about 0.2%–0.4% to about 0.7%–1% [29, 30]. Thus, this risk is consid- tion of an assay using this well-validated and published real-time PCR
erably less than the immunization risk associated with a false-negative technology in pregnancies of at least 10 weeks of gestation. Oligonu-
result determined with serological blood group reagents in the context cleotides and probes for RHD exon 5 and 7 have been verified in a
of red blood cell transfusion. In some countries, a false-negative NIPT multicentre study [37]. The concentration of primer and probes has
RhD result will cause the omission of both antenatal and postnatal been previously published by Grootkerk-Tax et al. [12]. Other combi-
prophylaxis, and consequently, the risk of immunization is higher, nations of RHD exons have also been applied successfully [16–18,
around 4%–16% [4, 31, 32], elsewhere stated as a prevalence of 3.5% 21–23, 38–40]. It should be noted, however, that the majority of pub-
[1]. In these countries, cord blood serology has been terminated on lished studies have been done in predominantly Caucasian pregnan-
the basis of high performance of the foetal RHD genotyping assay, cies, and that the accuracy in non-Caucasian, especially in African
with diagnostic sensitivities around 99.9% [7]. Taken together, the populations, has not been established yet. In order to minimize manual
immunization risk in the context of screening non-immunized RhD errors, nucleic acid extraction should be performed with an automated
negative women is still much less than the anticipated risk in the con- procedure, which has already been validated and published with at
text of red blood cell transfusion, where also a mismatched least 1000 specimens at another site or in a side-by-side comparison
RECOMMENDATIONS FOR NON-INVASIVE TESTING OF FOETAL BLOOD GROUPS 161

T A B L E 1 Summary of requirements for the validation of assays RhD negative phenotype or a very weak RhD variant (e.g., DEL). In
with the intended use to determine the foetal RHD status from these cases, which are rare or of low frequency, it is still possible to
maternal blood
recommend universal antenatal anti-D prophylaxis, and the low rates
Diagnostic sensitivity 1000 specimens for a new technology; at of variants do not impact on the effectiveness of foetal RHD screening
and specificity least 100 specimens for a well-validated
programs [45]. However, the development and validation of assays
and published technology
with respect to these RhD variants are encouraged for non-Caucasian
Requirement for Plasma from D negative pregnant women,
populations. In general, laboratories and manufactures should always
specimen 10–29 weeks, including a general
validation documentation of collection tube type, describe the assay limitations related to variants.
transportation time and conditions in the
validation report
Analytical sensitivity Dilution series (1-in-2) WHO reference Analytical detection limit
material (NIBSC 07/222) in four
replicates, the dilution 1-in-2 must test
positive in four of four replicates For the determination of the analytical detection limit of a new NIPT
RhD assay, World Health Organization (WHO) reference material is
Measuring range and A dilution series (0.5 log) from 100 ng/ml
linearity RHD positive DNA spiked into RHD commercially available (Product Number 07/222, National Institute
negative plasma, tested in three for Biological Standards and Control [NIBSC], Potters Bar, Hertford-
replicates, the concentration ≤500 pg/ml shire, United Kingdom) [46]. After DNA extraction, a dilution series
plasma must test positive in three of
(1-in-2) of the DNA should be tested with four replicates of each dilu-
three replicates
tion. The dilution 1-in-2 must test positive in all four replicates; the
Intra-assay precision Eight replicates from a plasma pool, obtained
from RHD negative pregnant women results from higher dilutions are part of the validation dossier and fur-
with an RHD positive foetus, tested in ther characterize the assay. Additional calculations of detection limit
one run may be useful to assess a theoretical detection limit of the whole
Inter-assay precision Nine replicates from a plasma pool, obtained setup [47].
from RHD negative pregnant women
with an RHD positive foetus, tested in at
least three runs on three different days
The tests should be performed by different Measuring range and linearity
technicians
Robustness Three runs, 12 samples per run, six RHD The WHO reference material is not suitable for the determination of
positive and six RHD negative samples, measuring range and linearity because the concentration of RHD spe-
respectively, from patients or blood
cific nucleic acids is too low for this purpose. Suitable material is RHD
donors
positive DNA (preferably from leukocytes from a hemizygous donor)
spiked into RHD negative ethylenediaminetetraacetic acid (EDTA)
plasma at a concentration of 100 ng/ml. A semilogarithmic dilution
with such a validated, automated procedure with at least 100 speci- series is made with DNA in H2O and then spiked into RHD negative
mens [13, 14, 16–23, 25]. General pre-analytical issues should be plasma before DNA extraction. Each dilution point must be tested
evaluated with great care to optimize sensitivity [15, 37, 41, 42]. The with three PCR replicates. A concentration of ≤500 pg/ml plasma
pre-analytical conditions including the gestational week of the valida- must test positive in all three replicates.
tion specimens should resemble those of the clinical specimens that
are to be tested after the validation phase.
For Europe, it can be expected that in about 60%–65% RhD neg- Precision
ative pregnant women, a positive foetal RHD status is determined
and, in about 35%–40%, a negative foetal RHD status is obtained The intra-assay precision should be analysed using a plasma-pool from
[13, 18, 20, 22]. According to this distribution, it is acceptable that RHD negative pregnant women with RHD positive foetuses. The
diagnostic sensitivity and specificity are investigated in any given intra-assay precision can be determined in one run with at least eight
cohort of RhD negative pregnant women with sufficient accuracy aliquots of this plasma pool. For real-time PCR, the coefficient of vari-
without specifying detailed numbers for RHD positive and RHD nega- ation for intra-assay precision of cycle threshold (Ct)-values should
tive foetuses. Analytical sensitivity and specificity largely depend on be ≤5%.
the genomic target region. Assays that detect RHD exon 7 and at least In order to determine the inter-assay precision, a plasma pool as
one additional RHD exon allow for detection of most RHD variants described above is required. In three different runs on three different
[43, 44]. Other exon combinations have also been applied success- days, three aliquots of this plasma pool must be tested (in total nine
fully, and selection of exons may vary for the population. In aliquots). Different technicians should perform these tests. For
Europeans, it is not required to test for RHD variants, for example, if real-time PCR, the coefficient of variation for inter-assay precision of
the woman carries an aberrant RHD allele, which is associated with an Ct-values should be ≤5%.
162 CLAUSEN ET AL.

Robustness inappropriate to use plasma from RhD positive individuals or plasma

from RhD negative individuals spiked with RHD positive leukocyte
There are different options to characterize the robustness of diagnos- DNA. The trend of the run control’s amplification signals can be
tic assays. Therefore, the laboratory or the manufacturer must decide assessed in order to recognize a decreasing analytical sensitivity. The
how this topic is addressed. However, experiments designed to evalu- extraction and amplification efficacy should be analysed in each sam-
ate the contamination risk are mandatory. Three subsequent runs with ple by detecting a known target sequence. Maternal, foetal or artificial
at least 12 samples each should be performed. Every run should com- (added to the specimen or the first extraction buffer) target sequences
prise at least six plasma samples from RHD positive and six plasma are appropriate. Recommended measures for quality assurance in rou-
samples from RHD negative patients or blood donors in a checker- tine operation are presented in Table 2.
board pattern. As a test of reproducibility, the laboratory should participate in an
external quality assurance (EQA) scheme. An international scheme has
been running as a workshop [49–51]. EQA is now offered from the
Quality assurance Danish EQA organizer DEKS. If no EQA is offered on a national level,
blinded samples can be exchanged with another laboratory at least
A diagnostic laboratory offering the determination of the foetal RHD once a year.
status from maternal blood requires a quality management
(QM) system. As part of a total QM system, each shipment of a new
lot of reagents must be tested with controls. Recommendations for Assays for non-invasive prediction of foetal blood
the storage of reagents are published [15]. group status in immunized women
If IVD products issued for testing the foetal RHD status are
assigned to Annex II of Directive 98/97/EG or risk class D of IVDR For immunized women, non-invasive prenatal genotyping of foetal
2017/746, lot-specific testing and release by a notified body is gene targets can predict antigen positivity, assisting in the manage-
required [2]. If placed under class C, lot-specific testing and release by ment of immunized pregnant women. The prediction of foetal anti-
a notified body is not required. gens helps clinicians to verify the diagnosis and plan for timely
treatment.
However, there is some degree of variation in how the knowledge
Pre-analytics in routine operation of the foetal blood group status is used in the management of the
immunized women. It may have a direct impact on indication of treat-
In general, pre-analytic issues need to be carefully assessed as pre- ment and how the pregnant woman is further monitored, or it may
analytical steps may affect the assay sensitivity. The interval of time have a less direct impact where knowledge of the foetal blood group
between venipuncture and separation of plasma from blood cells status is merely a contributing component to the diagnosis, combined
should be as short as possible when using EDTA tubes and should with other indications including antibody titres and ultrasound, and
preferably not exceed 5 days [41]. However, each laboratory further monitoring of the pregnant woman will continue unchanged
should perform their own evaluation of the allowed time duration
under the specific transportation conditions used by the individual
TABLE 2 Quality assurance in routine operation
laboratory. Each laboratory should also perform its own evaluation of
storage conditions as well as sample tubes used. For haemolytic speci- Transportation As short as possible, preferably ≤5 days from
time blood sampling until plasma separation
mens, a reduced sensitivity has been described; therefore, plasma
needs to be visually checked for haemolysis signs before nucleic acid Test for Visual inspection for haemolysis before nucleic
confounding acid extraction
extraction [41, 42].
factors
Positive run control One plasma pool from RHD negative pregnant
women with RHD positive foetuses
Controls in routine operation per runa
Negative run One plasma from RHD negative individual/
It is preferable to use a system for run control for monitoring the control plasma poolb
extraction efficiency as well as having a control system for amplifica- Extraction/ Either a human housekeeping gene or a
tion efficiency. Various solutions can be selected, and ultimately, each amplification heterologous DNA fragment is present
control during DNA extraction and amplification/
laboratory will have to decide upon a solution that is practical and
detection
efficient according to their setup.
a
For a run control, the recommendation is to collect the remaining The use of a plasma pool is recommended, not required. If not feasible, an
RHD positive DNA control is mandatory, preferably diluted to a
plasma from blood tubes after testing for foetal RHD and produce a
concentration of RHD positive DNA close to the detection limit.
plasma pool in order to use aliquots of this pool, preferably in every b
The use of a plasma pool is recommended, not required. If not feasible,
run [20]. Due to the smaller fragment size of cfDNA [48], it is an RHD negative DNA control is mandatory.
RECOMMENDATIONS FOR NON-INVASIVE TESTING OF FOETAL BLOOD GROUPS 163

or only slightly decreased. Therefore, depending on each country’s re-testing can be applied for cases with a negative result to ensure
management regime, an overall risk assessment is difficult. Still, a two independent negative results as basis for concluding a foetal
false-negative result may convey the risk that a pregnancy is not prop- blood group status and, thus, no risk of HDFN. In addition, the use
erly monitored, and a potentially affected foetus may not be discov- of foetal DNA controls to show the presence of foetal DNA in case
ered and treated in a timely manner. Hence, the risk associated with of a negative blood group genotyping result is strongly rec-
immunized women concerns the health of the foetus rather than the ommended [62].
risk of immunization of the woman as in the context of screening
non-immunized pregnant RhD negative women to guide targeted
prophylaxis. DI SCU SSION AND CO NCLUSIO N
The issue of risk classification for immunized women has been
discussed by the author group. There are different opinions as to Non-invasive foetal RHD genotyping has become a standard part of
whether testing of immunized women should be placed under risk antenatal care for RhD negative women in many countries. Foetal
class C or D according to EU regulations. This relates to different RHD genotyping screening programs allow RhD negative women
opinions as to how these women should be monitored. Overall, two carrying a compatible (RhD negative) foetus to avoid unnecessary
opinions are prevailing: One placing the testing under class C, where a administration of prophylactic RhIg and thus avoid exposure to a
high level of monitoring is continued despite predicting the foetus to blood-derived product as well as conserve the valuable RhIg for ratio-
be at no risk of HDFN, and one placing the testing under risk class D, nal use. NIPT RhD represents one of the first clinical applications of
where monitoring of the pregnant woman is substantially decreased cell-free foetal DNA and has been used as a clinical tool for almost
or even discontinued when the foetus is predicted to be at no risk. In 20 years [63]. High assay performance of real-time PCR assays has
addition, some favour risk class D because the risk concerns the foe- been extensively documented in the literature [7, 12–25], exemplify-
tus, whereas others favour risk class C. Outside of Europe, in ing the large experience with this assay. In contrast, assays for women
Australia, the risk classification is class 3, equivalent to the EU class C. immunized to blood group antigens other than RhD are reported in
In addition, several different techniques are used for testing of smaller numbers in the literature due to the rarity of cases [1].
immunized women, unlike the almost universally used real-time PCR The risk assessment of the immunization risk in the context of
for foetal RHD genotyping of non-immunized RhD negative women. targeted routine antenatal anti-D prophylaxis programs is found to be
These techniques include real-time PCR [52, 53], real-time PCR with in a different category when compared to the immunization risk in the
various modifications [54, 55], matrix-assisted laser desorption/ioniza- context of blood transfusion. In addition, it can be argued that non-
tion–time of flight (MALDI-TOF) mass spectrometry [56], targeted invasive foetal blood group genotyping serves different clinical pur-
DNA sequencing [57–60] and Droplet Digital PCR [61]. Furthermore, poses for immunized and non-immunized women, where the risk in
unlike foetal RHD genotyping of non-immunized women, the number immunized women concerns the foetus and the risk in non-immunized
of cases of immunized women are often much lower and quite rare women concerns foremost the women (and only later the foetus, if
for antigens other than RhD. Consequently, the number of published the woman becomes immunized and initiates a new pregnancy). Con-
cases and protocols is small compared to foetal RHD screening as cerning the potential risk associated with prophylaxis, demanding less
guide for targeted prophylaxis [1]. strict requirements for assays intended for screening non-immunized
Therefore, validation of these assays is more challenging, RhD negative women will help advance the implementation of screen-
although validation of assays for immunized women can be done ing programs and targeted prophylaxis, thereby avoiding unnecessary
using samples from non-immunized women when tested at equivalent treatment and thus overall decreasing the potential risk associated
gestational age as expected from clinical samples. with prophylaxis.
We recommend that the risk class decision for assays intended In relation to the EU regulation, we, the leading experts in the
for non-invasive prenatal testing of foetal blood groups in immunized field, do not agree that all assays for non-invasive prenatal testing of
women should be taken locally depending on the jurisdiction of each foetal blood groups should be placed under risk class D. We, thus, dis-
country, relating to each country’s regime for management of immu- agree specifically with the corrigendum of 2019 [10]. Alternatively,
nized women. The validation and quality assurance should generally we recommend at least placing non-invasive foetal RHD genotyping
follow the same principles as recommended for foetal RHD status. We under risk class C, when applied as a screening of RhD negative non-
recommended that manufacturers put specific claims on their NIPT immunized pregnant women to guide targeted prophylaxis.
products, with respect to the need of further monitoring of We have presented specific recommendations for assay valida-
alloimmunized women if the foetus has been tested antigen negative. tion in terms of performance evaluation, analytical detection limit,
For immunized women, it must be possible to convey a clinical range and linearity, precision, robustness, quality assurance, pre-
test result from an assay that has been tested only with a low number analytics and use of controls in routine testing. These recommenda-
of clinical cases. In a clinical setting, we recommend that clinicians tions are endorsed by expert colleagues and the cfDNA subgroup of
should be informed of this low number of cases, so that results are the ISBT RCIBGT working party. The resulting recommendations are
warranted by the number of samples tested and clinicians made aware designed to ensure appropriate assay performance and applicability
that caution should be taken for result interpretation. A strategy of for safe clinical use.
164 CLAUSEN ET AL.

ACKNOWLEDGEMEN TS decision making on Rh prophylaxis is feasible. Transfusion. 2008;48:

T.L. conceived the manuscript; T.L., Å.H. and F.B.C. wrote the draft. 2292–301.
14. Finning K, Martin P, Summers J, Massey E, Poole G, Daniels G. Effect of
All authors contributed with statement discussion and formulation. All
high throughput RHD typing of fetal DNA in maternal plasma on use of
authors critically revised and approved the manuscript. anti-RhD immunoglobulin in RhD negative pregnant women: prospec-
tive feasibility study. BMJ. 2008;336:816–8.
ORCID 15. Legler TJ, Heermann KH, Liu Z, Soussan AA, van der Schoot CE. Fetal
DNA: strategies for optimal recovery. Methods Mol Biol. 2008;444:
Frederik Banch Clausen https://orcid.org/0000-0003-2487-5373
209–18.
Åsa Hellberg https://orcid.org/0000-0002-7000-1899 16. Wikman AT, Tiblad E, Karlsson A, Olsson ML, Westgren M, Reilly M.
Katri Haimila https://orcid.org/0000-0001-8904-666X Noninvasive single-exon fetal RHD determination in a routine screening
Agneta Wikman https://orcid.org/0000-0002-8810-7974 program in early pregnancy. Obstet Gynecol. 2012;120:227–34.
17. Clausen FB, Christiansen M, Steffensen R, Jørgensen S, Nielsen C,
Gregory Andrew Denomme https://orcid.org/0000-0001-8727-
Jakobsen MA, et al. Report of the first nationally implemented clini-
1679
cal routine screening for fetal RHD in D- pregnant women to ascer-
Willy Albert Flegel https://orcid.org/0000-0002-1631-7198 tain the requirement for antenatal RhD prophylaxis. Transfusion.
Yanli Ji https://orcid.org/0000-0001-6792-5612 2012;52:752–8.
William J. Lane https://orcid.org/0000-0002-1097-5229 18. Clausen FB, Steffensen R, Christiansen M, Rudby M, Jakobsen MA,
Jakobsen TR, et al. Routine noninvasive prenatal screening for fetal
Martin L. Olsson https://orcid.org/0000-0003-1647-9610
RHD in plasma of RhD-negative pregnant women-2 years of screen-
ing experience from Denmark. Prenat Diagn. 2014;34:1000–5.
RE FE R ENC E S 19. Soothill P, Finning K, Latham T, Wreford-Bush T, Ford J, Daniels G.
1. van der Schoot CE, Winkelhorst D, Clausen FB. Non-invasive fetal Use of cffDNA to avoid administration of anti-D to pregnant women
blood group typing. In: Page-Christiaens L, Klein H-G, editors. Nonin- when the fetus is RhD-negative: implementation in the NHS. BJOG.
vasive prenatal testing (NIPT). Applied genomics in prenatal screen- 2015;122:1682–6.
ing and prenatal diagnosis. London: Academic Press; 2018. ISBN 20. de Haas M, Thurik FF, van der Ploeg CPB, Veldhuisen B,
978-0-12-814189-2. Hirschberg H, Soussan AA, et al. Sensitivity of fetal RHD screening
2. 2010/227/EU. Regulation (EU) 2017/746 of the European for safe guidance of targeted anti-D immunoglobulin prophylaxis:
Parliament and of the Council of 5 April 2017 on in vitro diagnostic prospective cohort study of a nationwide programme in The Nether-
medical devices and repealing Directive 98/79/EC and Commission lands. BMJ. 2016;355:i5789.
Decision. Off J. 2017;L 117:176–332. 21. Vivanti A, Benachi A, Huchet FX, Ville Y, Cohen H, Costa J-M. Diag-
3. de Haas M, Thurik FF, Koelewijn JM, van der Schoot CE. Haemolytic nostic accuracy of fetal rhesus D genotyping using cell-free fetal
disease of the fetus and newborn. Vox Sang. 2015;109:99–113. DNA during the first trimester of pregnancy. Am J Obstet Gynecol.
4. de Haas M, Finning K, Massey E, Roberts DJ. Anti-D prophylaxis: 2016;215:606.e1–5.
past, present and future. Transfus Med. 2014;24:1–7. 22. Haimila K, Sulin K, Kuosmanen M, Sareneva I, Korhonen A,
5. Kent J, Farrell AM, Soothill P. Routine administration of anti-D: the Natunen S, et al. Targeted antenatal anti-D prophylaxis program for
ethical case for offering pregnant women fetal RHD genotyping and RhD-negative pregnant women - outcome of the first two years of a
a review of policy and practice. BMC Pregnancy Childbirth. 2014; national program in Finland. Acta Obstet Gynecol Scand. 2017;96:
14:1–4. 1228–33.
6. Pegoraro V, Urbinati D, Visser GHA, di Renzo GC, Zipursky A, 23. Sørensen K, Kjeldsen-Kragh J, Husby H, Akkök ÇA. Determination of
Stotler BA, et al. Hemolytic disease of the fetus and newborn due to fetal RHD type in plasma of RhD negative pregnant women. Scand J
Rh(D) incompatibility: a preventable disease that still produces signif- Clin Lab Invest. 2018;78:411–6.
icant morbidity and mortality in children. PLoS One. 2020;15: 24. Saramago P, Yang H, Llewellyn A, Walker R, Harden M, Palmer S,
e0235807. et al. High-throughput non-invasive prenatal testing for fetal rhesus
7. van der Schoot CE, de Haas M, Clausen FB. Genotyping to prevent D status in RhD-negative women not known to be sensitised to the
Rh disease: has the time come? Curr Opin Hematol. 2017;24: RhD antigen: a systematic review and economic evaluation. Health
544–50. Technol Assess. 2018;22:1–172.
8. Clausen FB. Lessons learned from the implementation of non- 25. Yang H, Llewellyn A, Walker R, Harden M, Saramago P, Griffin S,
invasive fetal RHD screening. Expert Rev Mol Diagn. 2018;18: et al. High-throughput, non-invasive prenatal testing for fetal rhesus
423–31. D status in RhD-negative women: a systematic review and meta-
9. Directive 98/79/EC of the European Parliament and of the Council analysis. BMC Med. 2019;17:37.
of 27 October 1998 on in vitro diagnostic medical devices. Off J. 26. Rouillac-Le Sciellour C, Serazin V, Brossard Y, Oudin O, Le Van
1998;L 331:1–37. Kim C, Colin Y, et al. Noninvasive fetal RhD genotyping from mater-
10. Corrigendum to Regulation (EU) 2017/746 of the European Parlia- nal plasma. Use of a new developed free DNA fetal kit RhD. Transfus
ment and of the Council of 5 April 2017 on in vitro diagnostic medi- Clin Biol. 2007;14:572–7.
cal devices and repealing Directive 98/79/EC and Commission 27. Legler T, Bein G, Bugert P, Schwartz D. Validierung von
Decision 2010/227/EU. Off J. 2019;L 334:167. Testverfahren zur Bestimmung des fetalen RHD-Status aus dem Blut
11. 2002/364/EC. Commission decision of 7 May 2002 on common D-negativer Frauen in der Schwangerschaft. Transfusion. 2020;20:
technical specifications for in vitro-diagnostic medical devices. Off J. 50–4.
2002;L 131:0017–0030. 28. Williams LA 3rd, Sikora J, Aldrees R, Pham HP, Marques MB. Anti-Rh
12. Grootkerk-Tax MG, Soussan AA, de Haas M, Maaskant-van Wijk PA, alloimmunization after trauma resuscitation. Transfus Apher Sci.
van der Schoot CE. Evaluation of prenatal RhD typing strategies on cell- 2019;58:102652.
free fetal DNA from maternal plasma. Transfusion. 2006;46:2142–8. 29. Crowther CA, Middleton P, McBain RD. Anti-D administration in
13. Müller SP, Bartels I, Stein W, Emons G, Gutensohn K, Köhler M, et al. pregnancy for preventing rhesus alloimmunisation. Cochrane Data-
The determination of the fetal RhD status from maternal plasma for base Syst Rev. 2013;2:CD000020.
RECOMMENDATIONS FOR NON-INVASIVE TESTING OF FOETAL BLOOD GROUPS 165

30. Koelewijn JM, de Haas M, Vrijkotte TG, Bonsel GJ, van der 49. Clausen FB, Barrett AN, Krog GR, Finning K, Dziegiel MH. Non-
Schoot CE. One single dose of 200 microg of antenatal RhIG halves invasive foetal RhD genotyping to guide anti-D prophylaxis: an
the risk of anti-D immunization and hemolytic disease of the fetus external quality assurance workshop. Blood Transfus. 2018;16:
and newborn in the next pregnancy. Transfusion. 2008;48:1721–9. 359–62.
31. Liumbruno GM, D’Alessandro A, Rea F, Piccinini V, Catalano L, 50. Clausen FB, Barrett AN. The noninvasive fetal RHD genotyping
Calizzani G, et al. The role of antenatal immunoprophylaxis in the EQA2017 working group. Noninvasive fetal RHD genotyping to
prevention of maternal-foetal anti-Rh(D) alloimmunisation. Blood guide targeted anti-D prophylaxis – an external quality assessment
Transfus. 2010;8:8–16. workshop. Vox Sang. 2019;114:386–93.
32. Legler TJ. RhIg for the prevention Rh immunization and IVIg for the 51. Clausen FB, Hellberg Å. External quality assessment of noninvasive
treatment of affected neonates. Transfus Apher Sci. 2020;59: fetal RHD genotyping. Vox Sang. 2020;115:466–71.
102950. 52. Legler TJ, Lynen R, Maas JH, Pindur G, Kulenkampff D, Suren A,
33. Zhu YJ, Zheng YR, Li L, Zhou H, Liao X, Guo JX, et al. Diagnostic et al. Prediction of fetal Rh D and Rh CcEe phenotype from maternal
accuracy of non-invasive fetal RhD genotyping using cell-free fetal plasma with real-time polymerase chain reaction. Transfus Apher Sci.
DNA: a meta analysis. J Matern Fetal Neonatal Med. 2014;27: 2002;27:217–23.
1839–44. 53. Orzinska A, Guz K, Dębska M, Uhrynowska M, Celewicz Z, Wielgos M,
34. Londero D, Stampalija T, Bolzicco D, Castro Silva E, Candolini M, et al. 14 years of polish experience in non-invasive prenatal blood group
Cortivo C, et al. Fetal RHD detection from circulating cell-free fetal diagnosis. Transfus Med Hemother. 2015;42:361–4.
DNA in maternal plasma: validation of a diagnostic kit using auto- 54. Finning K, Martin P, Summers J, Daniels G. Fetal genotyping for the
matic extraction and frozen DNA. Transfus Med. 2019;29:408–14. K (Kell) and Rh C, c, and E blood groups on cell-free fetal DNA in
35. Runkel B, Bein G, Sieben W, Sow D, Polus S, Fleer D. Targeted ante- maternal plasma. Transfusion. 2007;47:2126–33.
natal anti-D prophylaxis for RhD-negative pregnant women: a sys- 55. Scheffer PG, van der Schoot CE, Page-Christiaens GC, de Haas M.
tematic review. BMC Pregnancy Childbirth. 2020;20:83. Noninvasive fetal blood group genotyping of rhesus D, c, E and of K
36. Legler TJ, Lührig S, Korschineck I, Schwartz D. Diagnostic perfor- in alloimmunised pregnant women: evaluation of a 7-year clinical
mance of the noninvasive prenatal FetoGnost RhD assay for the pre- experience. BJOG. 2011;118:1340–8.
diction of the fetal RhD blood group status. Arch Gynecol Obstet. 56. Li Y, Finning K, Daniels G, Hahn S, Zhong X, Holzgreve W. Noninva-
2021;1–6. https://doi.org/10.1007/s00404-021-06055-1 Epub sive genotyping fetal Kell blood group (KEL1) using cell-free fetal
ahead of print. DNA in maternal plasma by MALDI-TOF mass spectrometry. Prenat
37. Legler TJ, Liu Z, Mavrou A, Finning K, Hromadnikova I, Galbiati S, Diagn. 2008;28:203–8.
et al. Workshop report on the extraction of fetal DNA from maternal 57. Rieneck K, Bak M, Jønson L, Clausen FB, Krog GR, Tommerup N,
plasma. Prenat Diagn. 2007;27:824–9. et al. Next-generation sequencing: proof of concept for antenatal
38. Hyland CA, Gardener GJ, Davies H, Ahvenainen M, Flower RL, prediction of the fetal Kell blood group phenotype from cell-free
Irwin D, et al. Evaluation of non-invasive prenatal RHD genotyping of fetal DNA in maternal plasma. Transfusion. 2013;53:2892–8.
the fetus. Med J Aust. 2009;191:21–5. 58. Rieneck K, Egeberg Hother C, Clausen FB, Jakobsen MA, Bergholt T,
39. Hyland CA, Millard GM, O’Brien H, Schoeman EM, Lopez GH, Hellmuth E, et al. Next generation sequencing-based fetal ABO
McGowan EC, et al. Non-invasive fetal RHD genotyping for RhD blood group prediction by analysis of cell-free DNA from maternal
negative women stratified into RHD gene deletion or variant groups: plasma. Transfus Med Hemother. 2020;47:45–53.
comparative accuracy using two blood collection tube types. Pathol- 59. Orzinska A, Guz K, Mikula M, Kluska A, Balabas A, Ostrowski J, et al.
ogy. 2017;49:757–64. Prediction of fetal blood group and platelet antigens from maternal
40. Dovč-Drnovšek T, Klemenc P, Toplak N, Blejec T, Bricl I, Rožman P. plasma using next-generation sequencing. Transfusion. 2019;59:
Reliable determination of fetal RhD status by RHD genotyping from 1102–7.
maternal plasma. Transfus Med Hemother. 2013;40:37–43. 60. Wienzek-Lischka S, Bachmann S, Froehner V, Bein G. Potential of
41. Müller SP, Bartels I, Stein W, Emons G, Gutensohn K, Legler TJ. Cell- next-generation sequencing in noninvasive fetal molecular blood
free fetal DNA in specimen from pregnant women is stable up to group genotyping. Transfus Med Hemother. 2020;47:14–22.
5 days. Prenat Diagn. 2011;31:1300–4. 61. O’Brien H, Hyland C, Schoeman E, Flower R, Daly J, Gardener G.
42. Clausen FB, Jakobsen TR, Rieneck K, Krog GR, Nielsen LK, Tabor A, Non-invasive prenatal testing (NIPT) for fetal Kell, Duffy and Rh
et al. Pre-analytical conditions in non-invasive prenatal testing of blood group antigen prediction in alloimmunised pregnant women:
cell-free fetal RHD. PLoS One. 2013;8:e76990. power of droplet digital PCR. Br J Haematol. 2020;189:e90–e94.
43. Wagner FF, Frohmajer A, Flegel WA. RHD positive haplotypes in D 62. Scheffer PG, de Haas M, van der Schoot CE. The controversy about
negative Europeans. BMC Genet. 2001;2:10. controls for fetal blood group genotyping by cell-free fetal DNA in
44. Daniels G. Variants of RhD—current testing and clinical conse- maternal plasma. Curr Opin Hematol. 2011;18:467–73.
quences. Br J Haematol. 2013;161:461–70. 63. Finning KM, Martin PG, Soothill PW, Avent ND. Prediction of fetal D
45. Gordon LG, Hyland CA, Hyett JA, O’Brien H, Millard G, Flower RL, status from maternal plasma: introduction of a new noninvasive fetal
et al. Noninvasive fetal RHD genotyping of RhD negative pregnant RHD genotyping service. Transfusion. 2002;42:1079–85.
women for targeted anti-D therapy in Australia: a cost-effectiveness
analysis. Prenat Diagn. 2017;37:1245–53.
46. Metcalfe P, Rigsby P, Tait E, Urbaniak S. An international reference
reagent for the detection of RHD and SRY DNA in plasma. Vox Sang.
2012;102:243–9. How to cite this article: Clausen FB, Hellberg Å, Bein G,
47. Clausen FB, Urhammer E, Rieneck K, Krog GR, Nielsen LK, Bugert P, Schwartz D, Drnovsek TD, et al. Recommendation
Dziegiel MH. How to evaluate PCR assays for the detection of low- for validation and quality assurance of non-invasive prenatal
level DNA. APMIS. 2015;123:731–9.
testing for foetal blood groups and implications for IVD risk
48. Chan KC, Zhang J, Hui AB, Wong N, Lau TK, Leung TN, et al. Size
distributions of maternal and fetal DNA in maternal plasma. Clin classification according to EU regulations. Vox Sang. 2022;
Chem. 2004;50:88–92. 117:157–165.
Received: 7 June 2021 Accepted: 8 June 2021

DOI: 10.1111/vox.13173

COMMENTARY

Seventy years of haemophilia care: A personal perspective

I N T R O D U CT I O N that was impossible because of my thick throat. Luckily enough, a nurse

entered my room, she saw what was going on, and I guess she reduced the
Having lived for over 70 years with severe haemophilia A, I have infusion speed. It was an allergic reaction; I never again had one that serious.
taken up the challenge to portray not only my life with haemophilia
but also my career in haemophilia and patient advocacy. The result
has been a book named, Surviving Hemophilia, A Road Trip Through the P RO P H Y L A X I S A N D H O M E T R E A T M E N T
World of Healthcare. This commentary is an excerpt of the memories,
emotions and experiences contained in the book. Gradually, my physician thought that a regular dose of Factor VIII could
avoid most of my spontaneous bleeds, and prophylaxis entered my life.
It was soon to be followed by transfusions at home that solved my nee-
M Y E A R L Y HA E M O P H I L I A Y E A R S dle fear, one of my early anxieties. I got a normal, regular life and even
sports came within reach. In the mid-70s, the Italian doctors Pier Man-
Luckily enough, soon after my birth in 1951, I was quickly diagnosed with nucci and Zaverio Ruggeri introduced a winter sport holiday for
a spontaneous form of severe haemophilia A. At the same time, my par- haemophilia families in the Italian Dolomites. My Dutch haematologist
ents were told that I would not grow old, maybe I would reach the age Jan Wouter Ten Cate thought they were crazy, but they invited him to
of 20, maybe 30, as there was no effective treatment. Despite this per- take part in such a holiday. Jan Wouter Ten Cate asked me instead and
spective, my parents never patronized me and tried to give me a carefree since then, I enjoyed cross-country skiing together with families of the
youth. After a couple of life-threatening bleeds, my physician proposed Italian Haemophilia Society for almost 40 years. Winter sports had been
to treat me with growth hormones. I was 6 years old. This turned out to unimaginable for me, for my parents and my doctors. The confidence
be an experimental treatment with di-ethyl-stilbestrol (DES), introduced everyone with haemophilia gained from this experience reflects the free-
by the French doctor Raymond A. Turpin in the late 40s. At that time, dom and emancipation of the haemophilia community in those years.
DES was used as a medication for a variety of female reproductive prob-
lems and to inhibit growth in adolescent girls. DES caused dreadful
anomalies in women and their offsprings; as for me, DES inhibited my lin- THE WINTER OF D ES PAIR
ear growth at 1.45 m. This was one of the drastic experimental treat-
ments for haemophilia in the days when there was no treatment at all. I All feelings of optimism and freedom disappeared with the viral infec-
also took part in a more innocent experiment, the use of peanuts or pea- tions caused by hepatitis C virus (HCV) and human immunodeficiency
nut extract, a therapy developed by H. Bruce Boudreaux who himself virus (HIV) in the late 70s and 80s. In the end, these virus infections
had haemophilia. Both therapies did not have any positive outcome. caused a winter of despair in the international haemophilia community
with many thousands of HCV and HIV transmissions. Without an
effective treatment for HIV, almost all HIV-infected people with
A SP R I N G OF HO P E haemophilia died. Early HCV treatments were little effective; only
recently, there is a real therapy to treat HCV. For many people with
The real turnaround moment for haemophilia treatment came when Judith haemophilia, this treatment came into existence too late. They either
Pool described how to isolate blood plasma concentrates that contained died or lived with severe liver damage.
more Factor VIII on a volume basis than fresh plasma. In 1967, I received
my first infusion with Factor VIII concentrates. This had an unexpected
effect; I became stuffy and could hardly breathe. I thought I was going to C O M O R B I D I T Y I N OL D E R P E O P L E
die. I was lying in a single room with the door closed; I tried to scream, but WITH HAEMOPHILIA

Untreated bleeding episodes caused joint damage already in my early

Cees Smit (1951) is the author of ‘Surviving Hemophilia, A Road Trip Through the World of
Healthcare,’ Eburon Academic Publishers, Utrecht, 2020.
life, and experimental treatment limited my growth. I was infected

This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium,
provided the original work is properly cited.
© 2021 The Author. Vox Sanguinis published by John Wiley & Sons Ltd on behalf of International Society of Blood Transfusion.

166 wileyonlinelibrary.com/journal/vox Vox Sanguinis. 2022;117:166–168.

COMMENTARY 167

FIGURE 1 An example of healthcare contacts’ network of an ageing man with haemophilia

with HCV and HIV, and side-effects of HIV treatment resulted in seri- A F O RE S E E A B LE D I S A S T E R
ous renal complications. I am now a typical example of an ageing
man with haemophilia and complex comorbidity. This results in my The tainted blood history shows that somewhere on this road, people
extensive network of healthcare contacts (Figure 1). The combina- not only ignored an orange, but even a red light. Persons in charge
tion of the needle phobia from my childhood, together with comor- within the international haemophilia community focussed on the avail-
bidity and polypharmacy issues, creates for me the phenomenon of ability of Factor VIII to meet demands, while people with haemophilia
the ‘fear factor’, the lack of control when you are hospitalized and embraced the freedom they got from their medication. No one wanted
when you need medical treatment from physicians who are less to lose these acquired rights. In the 1970s, the plasma fractionation
familiar with haemophilia. An additional fear is related to my ageing industry knew the dangers of hepatitis viruses in the plasma supply.
process. I experience my health situation as fragile and easily Even some of their employees warned for these contaminations. This
unbalanced. makes it unacceptable to speak of a ‘tragic incident’ in haemophilia his-
tory. Tainted blood was a foreseeable disaster. Numerous accusations
have been made against national governments and are still being made
COMMERCIALISM OVER VOLUNTARISM in the Infected Blood Inquiry in the United Kingdom [2]. Until now, the
international plasma industry has escaped an investigation into the
There are some less beautiful sides of haemophilia care and plasma causes of their plasma procurement process and its inadequacies in
supply. These are the focus of the pharmaceutical industry on profits the 70s and 80s. People with haemophilia and their families still experi-
and the negligence to create a safe blood policy. It is less known that ence the burden of this tragedy. Many died and, those who survived
Judith Pool wrote a letter to the Nixon administration about the new live with ill health for more than 40 years.
US national blood policy. She wrote in 1974: ‘My concern […] is that
it in no way requires or even encourages the use of volunteer blood
for this purpose but assumes a continuation of the dangerous, expen- A NEW SPRING OF HOPE
sive, wasteful, and unethical purchase of plasma by pharmaceutical
houses to provide such therapeutic material’. Another memorable The voluntary blood and plasma sector never succeeded in getting con-
moment in time can be found in the famous book Journey, published trol over the blood and plasma supply. On the contrary, the haemophilia
by Suzanne and Robert Massie in 1976. They described the battle market is now a booming business with more than 60 companies active
between Hyland/Travenol (now Takeda) and the American Red Cross in a billion-dollar market. Plasma concentrates for haemophilia have been
over the production of Factor VIII concentrate, a battle that was won replaced by recombinant products, extended half-life products, factor-
by Hyland/Travenol. The Massies consider this as a great loss for the replacement therapies and gene therapies. For people with haemophilia,
system of solidarity and the voluntary non-remunerated blood and new therapies cause uncertainties, while existing products have its rela-
plasma donation. A third striking and neglected fact in the 70s is that tively unease like intravenous access. Unfortunately, people in three
the plasma industry ignored that their own factory workers fell ill with quarters of the world still have no access to haemophilia therapy because
hepatitis [1]. of the price of the haemophilia products and its relative scarcity.
168 COMMENTARY

G L O B A L I S M OV E R SE L F - R E L I A N C E FUNDING INF ORMATI ON

None.
For other plasma products, especially immunoglobulins, the world
is dependent on the large paid plasma donor pool in the United Cees Smit, Patient Expert
States. This plasma pool is no longer owned by US-based plasma
collection centres but by large international pharma companies. Department of epidemiology, Leiden University Medical Center (LUMC),
National blood transfusion services gradually lost their share on Hoofddorp, The Netherland
this market. European countries permitted the selling of their
plasma factories, BPL UK and Biotest in Germany, to China’s Correspondence
Creat. It shows that globalism has become a dominant factor in Cees Smit, Graan voor Visch 15715, 2132 EM Hoofddorp,
the blood and plasma industry and supersedes nation’s historical The Netherlands.
principle of self-sufficiency. At present, dealing with COVID-19, Email: [email protected]
we experience the impact of globalism on our health care. It is a
situation that was to be foreseen when we look at the history of Cees Smit (1951) is the author of ‘Surviving Hemophilia, A Road Trip
haemophilia treatment. History does not repeat itself, but it often Through the World of Healthcare,’ Eburon Academic Publishers,
rhymes. Utrecht, 2020.

ORCID
C O R O L LA R Y Cees Smit https://orcid.org/0000-0002-0735-7067

I am glad I have survived these 70 years living with haemophilia, and I RE FE RE NCE S
must acknowledge that I could have never come this far without sup- 1. Taylor JS, Shmunes E, Holmes AW. Hepatitis B in plasma fraction-
port. A large group of professionals, doctors, nurses, my friends, my ation workers. A seroepidemiologic study. JAMA. 1974;230:850–3.
2. Infected Blood Inquiry (2021). https://www.infectedbloodinquiry.
parents, my family, my partner and her children have always been at
org.uk. Assessed 7 June 2021
my side. I am also grateful to the many donors in the Netherlands
who voluntarily donate their blood plasma.
How to cite this article: Smit C. Seventy years of haemophilia
CONF LICT OF IN TE RE ST care: A personal perspective. Vox Sang. 2022;117:166–168.
The authors declare no conflicts of interest.
Received: 13 February 2021 Revised: 16 June 2021 Accepted: 22 June 2021

DOI: 10.1111/vox.13179

ORIGINAL ARTICLE

Determinants of the intention to donate umbilical cord blood

in pregnant women

Mariana Fernandes | Guido Alessandri | Rubi Abbad | Caterina Grano

Department of Psychology, Sapienza

University of Rome, Rome, Italy Abstract
Background and Objectives: Umbilical cord blood (UCB) donation is a behaviour pro-
Correspondence
Caterina Grano, Department of Psychology, moted by many countries’ health systems. However, UCB donation is not a wide-
Sapienza University of Rome, Via dei Marsi, spread behaviour among expectant mothers, and little is known about the reasons
78, 00185 Rome, Italy.
Email: [email protected] that may lead to it. The aim of the present study was to analyse the contribution of
Theory of Planned Behaviour (TPB) variables among both primiparous and multipa-
Funding information
rous women in predicting intention to donate UCB.
None
Materials and Methods: Three hundred seventy-six expectant mothers completed
questionnaires that captured sociodemographic data, parity, previous donation, atti-
tudes, subjective norms, perceived behavioural control (PBC) and intention to donate
UCB. Multigroup analysis structural equation modelling was conducted using Mplus
(version 8.02).
Results: Multigroup path analyses showed that intentions were strongly predicted by
subjective norms and moderately predicted by positive attitudes and PBC in both pri-
miparous and multiparous women. TPB constructs explained 71% of the variance in
intentions for both groups.
Conclusions: Future interventions to increase intention to donate among primiparous
and multiparous women could primarily consider the influence of partner and signifi-
cant others in determining positive intentions and secondarily target increasing posi-
tive attitudes and perceptions of control.

KEYWORDS
donor motivation, donors, embryonic stem cell

I N T R O D U CT I O N compatible donor [5]. Moreover, clinical studies have proved the plu-
ripotent nature of cord blood cells [6, 7], highlighting a wide range of
Umbilical cord blood (UCB) banking consists in the collection and stor- possible clinical applications in neonatology [6, 8], regenerative medi-
age of the UCB from the placenta and umbilical cord, soon after child- cine and immune modulation [9]. UCB can thus be used in the treat-
birth [1] to preserve potentially life-saving cells that are usually ment of various life-threatening diseases such as leukaemia, cerebral
considered medical waste [2, 3]. Cord blood transplantation is consid- palsy, neonatal hypoxic–ischaemic encephalopathy and diabetes.
ered a valid alternative treatment in haematologic diseases (malignant UCB can be stored either in a public bank for allogeneic use (avail-
and non-malignant), primary immunodeficiencies and metabolic disor- able to anyone who might be compatible) or in private banks for autolo-
ders [4, 5], especially in the case of scarce probability of having a gous or family use [1]. Several European and US professional

This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium,
provided the original work is properly cited.
© 2021 The Authors. Vox Sanguinis published by John Wiley & Sons Ltd on behalf of International Society of Blood Transfusion.

Vox Sanguinis. 2022;117:169–176. wileyonlinelibrary.com/journal/vox 169

170 FERNANDES ET AL.

organizations and institutions [10] support and promote the donation [24]. In this study, attitudes, normative beliefs, subjective norms,
of UCB to public institutions for altruistic and solidarity purposes awareness and behaviour control were all considered predictors of
(e.g., patients needing transplantation will be able to rapidly find a women’s intention to donate cord blood, explaining together 52%
matching donor), since the use of cord blood stored in private cord of the variance for intention to donate. However, it is not possible to
blood banks for autologous use rarely occurs [11]. Although in the pre- infer the contribution of each of these constructs in this study, as the
vious two decades, many public banks for allogeneic use have been regression coefficients were not reported by the authors and
established in Italy, actual donations remain very low, with only 2.2% of the results were only summarily described [24]. Other studies have
parents donating the cord blood to a public cord blood bank [12]. There considered only some of the TPB variables [19, 20]. For example, Kim
are several possible reasons for the current low cord blood donation and colleagues [19] examined the knowledge and attitudes of early
rates. For instance, although women and/or couples are aware of the post-partum women about storage, donation and disposal of their
possibility of donating UCB, they may either have little knowledge of cord blood, as well as sociodemographic factors influencing cord
donation possibilities or the procedures to be followed or even be blood donation, and found that knowledge, attitudes, income and
unaware of the uses of UCB [13, 14]. In the view of these premises, it source of information influenced cord blood donation. None of these
seems particularly relevant to understand the factors that influence studies has been conducted in the Italian context.
expectant parents’ decisions to donate cord blood, as this may be Based on these premises, the present study aims at extending
essential for encouraging donation. previous research on the predictors of the intention to donate UCB
Considering demographic factors, results of previous studies indi- considering primiparous and multiparous Italian women using the
cated that educational level and age are positively associated with well-established theoretical framework of the TPB.
UCB donation [5, 15, 16], while mixed results are reported on what
concerns household income [4, 16].
Although a growing number of studies highlighted the role of psy- METHODS
chosocial factors in explaining, predicting and promoting different
donation behaviours [17, 18], research on UCB donation is still very Participants
limited with only a few studies highlighting a positive association
between attitudes and UCB donation [19, 20]. Three hundred seventy-six pregnant women in the third trimester of
One of the most frequently applied models for predicting donation pregnancy participated in the study. To be included, women had to be
behaviours and social behaviours, in general, is the Theory of Planned in the last semester of pregnancy and be fluent in Italian. Eleven par-
Behaviour (TPB) [21]. In the TPB model, the most proximal determinant ticipants with pathologies precluding the donation of UCB were
of behaviour is considered intention to perform that behaviour [21]. In excluded from the study. A total of 365 participants remained.
turn, intentions are determined by attitudes (i.e., an evaluation or
appraisal of the behaviour as favourable or unfavourable), subjective
norms (i.e., perceived social pressure and expectations from significant Procedure
others to perform the behaviour) and perceived behavioural control
(PBC) (i.e., how much control individuals believe they have over the per- Expectant mothers were recruited through a variety of means
formance of the behaviour) [21, 22]. (e.g., information centres for maternity and birth, family associations,
Considering the TPB in the more general context of blood dona- cultural and sports associations). Participation in the study was anony-
tion, a systematic review reported that the model explained between mous and voluntary, and no incentives or payments were offered. The
31% and 72% of the variance in blood donation intentions and between survey was administrated as an online questionnaire. First, participants
54% and 56% in blood donation behaviour [17]. Across the studies, atti- read and signed a consent form and then were asked to provide their
tudes towards blood donation, subjective norms and PBC significantly email address. Afterwards, participants received an email containing a
predicted intention to donate blood, being the influence of subjective personal code and the web link to access the online questionnaire, to
norms on intentions less consistent [17]. Furthermore, PBC had the which they had access only through that same personal code. The study
strongest effect in predicting intentions to donate blood [17]. Similarly, was approved by the Ethical Committee of the Department of Psychol-
in a meta-analytic review [18] including 37 studies on blood donation ogy of ‘Sapienza’ University of Rome (Prot. 001066).
intentions and 24 predictive studies on blood donation behaviour,
intentions had the strongest associations with donation behaviour. In
addition, large positive effects were found for attitudes and PBC in Measures
predicting intentions, while a medium positive association was
observed for subjective norms [18]. Past donations were also positively The questionnaire, in addition to sociodemographic information (age,
associated with both intention to donate and donation behaviour [18]. parity, gestation month, current relationship, education level of the
Despite the successful use of the TPB in the context of blood expectant mother), included a question evaluating potential illnesses
donation [17, 23], to the best of our knowledge, only one study inves- precluding UCB donation (e.g., autoimmune diseases, infectious dis-
tigated the prediction of cord blood donation within this framework eases, etc.). Before starting to answer the questionnaire, women were
DETERMINANTS OF INTENTION TO DONATE CORD BLOOD 171

asked if they were aware of the possibility of donating or conserving How easy or difficult do you think it will be?’, ‘very difficult’ (1) to
UCB. All these questions were answered on a dichotomous scale ‘very easy’ (7). ‘Deciding to donate the cord blood of your baby is…’
(yes/no). ‘not at all up to me’ (1) to ‘completely up to me’ (7) and ‘How much
control do you feel you have over your decision to donate your baby
cord blood?’ ‘not at all under my control’ (1) to ‘completely under my
Theory of planned behaviour control’ (7). Items were scored on a 7-point Likert scale with higher
scores representing greater PBC over cord blood donation.
Regarding the formulation and scaling of the TPB variables, we Cronbach’s alpha resulted in 0.95.
followed the questionnaire construction recommendations of Fishbein Mothers’ intention to donate cord blood was measured through
and Ajzen (2010) [25]. The items used to measure the TPB constructs five items. An example of an item was ‘Do you intend to donate your
are listed in Appendix S1. baby cord blood?’ ‘I definitely do not’ (1) to ‘I definitely do’ (7). The
Attitudes towards UCB donation were assessed using a combination remaining items asked participants how strongly they wanted to
of outcome beliefs measured by 10 semantic differentials: ‘Donate the donate, how likely it was for them to though about donate and how
blood cord of my baby would be for me…’ ‘useless–useful,’ ‘difficult– much they would like to donate. Items were scored on a 7-point Likert
easy,’ ‘disappointing–rewarding,’ ‘senseless–sensible,’ ‘disadvantageous– scale with higher scores indicating stronger intentions to donate.
advantageous,’ ‘unsatisfactory–satisfactory,’ ‘wrong–right,’ Cronbach’s alpha resulted in 0.95.
‘stupid–wise,’ ‘unnerving–reassuring’ and ‘expensive–economical.’ Items Past donation/storage for primiparous/multiparous woman was
were scored on a 7-point Likert scale. Cronbach’s alpha resulted in 0.91. evaluated through a dichotomous question (yes/no).
Subjective norms regarding UCB donation were measured using a
combination of normative belief measures and motivation to comply
measures. Subjective norms were measured through five items rated on Data analysis
a 7-point Likert scale, ranging from completely disagree (1) to completely
agree (7). Example of items were as follows: ‘Most people who are Descriptive analysis was conducted with the statistical program
important to me think that I should donate the cord blood of my baby’ SPSS for Windows version 25.0 (IBM Corp, Armonk, NY). Data are
and ‘My partner would approve….’ Higher scores correspond to higher reported as mean  standard deviation (SD). Categorical data are
perceived social approval. Cronbach’s alpha resulted in 0.90. reported as counts and percentages. To estimate the hypothesized
PBC over UCB donation was measured using a combination of model, we used Mplus (version 8.02, Muthén & Muthén, Los
control belief measures and power items. PBC was assessed with Angeles, CA) [26] with the Satorra and Bentler [27] scaled
three items: ‘Suppose you decide to donate your baby cord blood. chi-square statistic (SBχ2) and robust standard errors, which takes

TABLE 1 Participants sociodemographic characteristics (N = 365)

Primiparous Multiparous
(n = 275) (n = 90)

N % N %
Months of gestation
Seventh 85 30.9 18 20.0
Eighth 101 36.7 42 46.7
Ninth 88 32.0 28 31.1
Tenth 1 0.4 2 2.2
Education level
Primary school — — 1 1.1
Middle school 30 10.9 17 18.9
Higher school 122 44.4 50 55.6
Degree 123 44.7 20 22.1
Missing — — 0.5 2.2
Current relationship
Married 142 51.6 59 65.6
Cohabitant 119 43.3 30 33.3
Have a partner (not cohabitant) 14 5.1 1 1.1
Mean age (SD) 31.22 5.24 33.09 5.63

Abbreviation: SD, standard deviation.

172 FERNANDES ET AL.

into account the non-normal distribution of the data (Mplus RE SU LT S

estimator = MLM). According to a multifaceted approach to the
assessment of model fit [28], the following criteria were employed The final sample was composed of 365 participants, with a mean age
to evaluate the goodness of tested models: the Satorra–Bentler χ 2
of 31.67 (SD = 5.39; age range = 18–44). Primiparous constituted
likelihood ratio statistic supplemented by the comparative fit index 75.3% (n = 275) of the sample. Sociodemographic characteristics for
(CFI) and the root mean square error of approximation (RMSEA), the entire sample are reported in Table 1. The vast majority of respon-
with associated confidence intervals. The significance value of χ 2 is dents reported being aware of the possibility of donating or conserv-
sensitive to large sample sizes and easily produces a statistically ing UCB (97.5%). Considering past UCB donation, only 2.7% had
significant result [29]. Values of RMSEA < .08 and CFI > .90 were previously donated UCB (n = 10).
considered acceptable [30].
Multiple-group structural equation modelling was used to investi-
gate measurement invariance across the two groups (primiparous and Correlations
multiparous) [26]. In our approach, the equivalence across the two
groups was tested as follows: after evaluating the configural model, Table 2 presents the descriptive statistics and the zero-order correla-
each level of invariance was evaluated separately for groups and then tions among the mean scores on the overall measures of attitudes,
simultaneously. The complete sequence of models is shown in subjective norm, PBC, past behaviour and behavioural intention. All
Table 3. Overall, configural, metric, scalar and strong invariance [31] correlations resulted significant and in the expected direction, except
were evaluated. The plausibility of the equality constraints implied by for those involving past behaviour in the multiparous group. Indeed,
the different models was examined with the Satorra–Bentler SB⊿χ 2 past behaviour resulted uncorrelated with all of the variables included
difference test between nested models (i.e., the sex constrained in the model. It was, thus, decided to remove this variable from all
model vs. the unconstrained model [32]). subsequent structural equation models.

TABLE 2 Zero-order correlations among study variables

Multiparous ATT NS PBC INT PB Mean SD

ATT 1 6.20 1.42
NS 0.56** 1 4.52 2.28
PBC 0.19* 0.30** 1 5.57 1.99
INT 0.52** 0.66** 0.44** 1 4.62 2.18
PB 0.05 0.05 0.03 0.04 1 0.87 0.38
Primiparous ATT NS PBC INT PB Mean SD
ATT 1 6.07 1.66
NS 0.66** 1 4.68 2.32
PBC 0.39** 0.45** 1 5.43 1.99
INT 0.69** 0.87** 0.52** 1 4.95 2.32
PB — — — — — —

Note: **p < 0.01; *p < 0.05.

Abbreviations: ATT, attitudes; INT, behavioural intention; NS, subjective norms; PB, past behaviour; PBC, perceived behavioural control; SD, standard
deviation.

TABLE 3 Model fit and comparisons

Title SBχ2 scr df CFI RMSEA (95%CI)

M1. Unconstrained model 862.416 1.20 448 0.926 0.074 (0.067, 0.082)
M2. Equal loadings 881.717 1.21 467 0.926 0.073 (0.065, 0.080)
M4. Equal structural paths 885.133 1.21 470 0.926 0.073 (0.065, 0.080)
SB⊿χ2 ⊿df p
M2 versus M1 22.12 19 0.28
M3 versus M2 3.42 3 0.31

Abbreviations: CFI, comparative fit index; CI, confidence intervals; RMSEA, root mean square error of approximation.
DETERMINANTS OF INTENTION TO DONATE CORD BLOOD 173

F I G U R E 1 The hypothesized model with completely standardized estimates for (a) multiparous and (b) primiparous
Note: ATT, attitudes; PBC, perceived behavioural control; INT, behavioural intention; SN, subjective norms

Structural equation analysis no restriction imposed. As reported in Table 3, this model fitted the
data very well. Following standard procedures, first, across-group
Our theoretical model was tested as a multiple-group structural equa- invariance was imposed on factor loadings and then on structural
tion model simultaneously on primiparous and multiparous groups. In regression paths. Overall, these constraints did not change the fit of
this model, attitudes, subjective norms, PBC and behavioural intention the model (See Table 3, bottom part). The final model with all restric-
were specified as latent variables saturated, respectively, by 10, 5, 3 tion included is represented in Figure 1, panels (a) and (b). Overall,
and 5 items. Parameters were all freely estimated across groups with behavioural intention was strongly predicted by subjective norms,
174 FERNANDES ET AL.

moderately predicted by attitudes and PBC. Subjective norms and professionals) as their opinion may influence expectant mothers inten-
attitudes were strongly correlated, whereas there were moderate cor- tions. Interventions should also consider increasing positive attitudes
relations between subjective norms and PBC and between attitudes towards cord blood donation and enhance women’s perceptions of con-
and PBC. Explained variance was high (R = 0.71 in both groups).
2
trol over this behaviour. Targeting these constructs may be particularly
relevant, given that cognition-based interventions have been shown to
be effective in other contexts, such as blood donation [35]. Although the
DISCUSSION present study did not consider whether expectant mothers were coun-
selled by health care providers to donate their cord blood, future studies
Stemming from the TPB, the aim of the present study was to explore could examine whether perceptions of control and social pressure
the extent to which attitudes, subjective norms and PBC influence the change as a result of the counselling sessions received.
intention to donate UCB in Italian primiparous and multiparous preg- For example, educational interventions and motivational inter-
nant women [21]. views to decrease ambivalence and resistance towards UCB donation
For decades, TPB has been applied to effectively explain, predict may be considered as means to increase intentions to donate.
and promote blood donation behaviour [17, 18], but, surprisingly, only The present study has some limitations that need to be acknowl-
one study used this framework to predict blood cord donation [24]. edged. First, despite the fact that assessing UCB donation behaviour
The present study findings extend previous research in the Italian would have been optimal, it is important to note that only 2.2% of
context supporting the robustness of the TPB model in explaining women donates UCB in Italy donate UCB [12], similar to other coun-
blood cord donation intentions. The fit indices across the cross- tries [36, 37]. Therefore, a much larger initial sample would have been
sectional analyses confirmed the model structure both in primiparous necessary to estimate the intention–behaviour pathway. Neverthe-
and multiparous women and the invariance across the two groups. less, we know from previous studies and meta-analyses on TPB that
Together, the variables included in the TPB model explained 71% of intentions generally range from 13.8% to 29% for different health
the variance in intentions to donate UCB in both samples, with sub- behaviours (e.g., physical activity, diet, breastfeeding, safe sex, absti-
jective norms being the strongest determinant of intentions and atti- nence from drugs) [38, 39] and from 19% to 38% for blood donation
tudes and PBC significantly contributing to explaining intentions both behaviour [18]. Future population studies are needed to evaluate how
in primiparous and multiparous women. This amount of explained var- behavioural intentions will determine actual donations in the specific
iance is consistent with the results reported in other studies on blood context of UCB donation. Secondly, the study considered a conve-
donation [17]. Unfortunately, regarding cord blood donation in spe- nience sample. Participants were expectant mothers who volunteered
cific, further comparisons are limited, since Natan’s study [24] suffers to be part of the study, and they may differ from mothers who were
from a few methodological issues, namely because items were ill- not available (i.e., a general tendency to be altruistic). Therefore, cau-
defined and regression results were only partially reported. tion is needed in generalizing. Thirdly, the present research focussed
Differently from previous studies on blood donation [17, 18], it on testing the prediction of intentions within one model, the TPB.
seems that subjective norms play a stronger role in determining inten- Thus, the structural paths examined were those specified when fol-
tions in the present study. The direction of the path from subjective lowing this framework. Evaluating the contribution of other models,
norms to intention was positive, implying that higher perceived social as well as integrating the TPB model with other constructs (self-effi-
approval for cord blood donation results in higher intention for UCB cacy, self-identity, etc.) could hypothetically generate distinct results.
donation. This finding is not surprising considering that blood cord Finally, although previous studies in the blood donation context
donation is generally a shared decision within a couple and that sub- reported that past behaviour was predictive of intentions, this variable
jective norms also encompassed a partners’ approval measure. In fact, could not be used in the present study because only 2.7% of partici-
previous studies showed that pregnant women consider their hus- pants had donated UCB in the past.
band/spouse to have a key role in the decision about cord blood In conclusion, the results of the present study demonstrate the
donation [4, 33]. Based on these findings, it would be desirable that usefulness of TPB in predicting intentions to donate UCB and high-
future studies should also include the perspective of the partner and light the importance of partner and significant others in forming posi-
partner–actor interactions in determining blood cord donation inten- tive intentions to donate UCB. Unlike other studies on blood and
tions. Other explanations for this finding are also possible, such as the organ donation [17, 18, 40], it appears that subjective norms play a
fact that although cord blood donation is well known, the women stronger role in determining intentions in the context of UCB. This is
underline the need of knowing more clear procedures on cord blood consistent with other research on bone marrow donation [41]. These
banking processes [34] and turn to relevant others to form their inten- aspects, along with positive attitudes and perceived control over cord
tions. Moreover, consistently with other studies on blood donation blood donation, should be considered when planning future interven-
[17, 18], attitudes and PBC were also found to have a predictive role tions to increase donation intentions.
on intended behaviour donation.
Some practical implications can be drawn based on the present find- AC KNOW LEDG EME NT S
ings. Interventions directed towards raising UCB donation intentions M.F. wrote the manuscript and revised the data analyses;
may need to target relevant others (partner, families and health G.A. conducted the data analyses and reviewed the manuscript;
DETERMINANTS OF INTENTION TO DONATE CORD BLOOD 175

R.A. worked on the study implementation, collected research data and among patients attending Nnamdi Azikiwe university teaching hospital,
contributed to the introduction section of the manuscript; Nnewi, south East Nigeria. Asian J Transfus Sci. 2017;11:162.
17. Masser BM, White KM, Hyde MK, Terry DJ. The psychology of
C.G. designed the research study, supervised the collection of the
blood donation: current research and future directions. Transfus
data and helped writing and editing the manuscript. The authors wish Med Rev. 2008;22:215–33.
to thank all the expectant mothers who participated in the study. 18. Bednall TC, Bove LL, Cheetham A, Murray AL. A systematic review
and meta-analysis of antecedents of blood donation behavior and
intentions. Soc Sci Med. 2013;96:86–94.
CONF LICT OF IN TE RE ST
19. Kim M, Han S, Shin M. Influencing factors on the cord-blood dona-
The authors declare that they have no conflict of interest. tion of post-partum women. Nurs Health Sci. 2015;17:269–75.
20. Fernandez CV, Gordon K, Van Den Hof M, Taweel S, Baylis F.
ORCID Knowledge and attitudes of pregnant women with regard to collec-
tion, testing and banking of cord blood stem cells. CMAJ. 2003;168:
Mariana Fernandes https://orcid.org/0000-0002-7005-2805
695–8.
Guido Alessandri https://orcid.org/0000-0003-4731-5248 21. Ajzen I. The theory of planned behavior. Organ Behav Hum Decis
Caterina Grano https://orcid.org/0000-0002-1899-0773 Process. 1991;50:179–211.
22. Fishbein M, Ajzen I. Belief, attitude, intention, and behavior: an introduc-
tion to theory and research. Reading, MA: Addison-Wesley; 1975.
RE FE R ENC E S
23. France JL, Kowalsky JM, France CR, McGlone ST, Himawan LK,
1. Peberdy L, Young J, Massey DL, Kearney L. Parents’ knowledge,
Kessler DA, et al. Development of common metrics for donation atti-
awareness and attitudes of cord blood donation and banking
tude, subjective norm, perceived behavioral control, and intention
options: an integrative review. BMC Pregnancy Childbirth. 2018;18:
for the blood donation context. Transfusion. 2014;54:839–47.
395–395.
24. Ben NM, Grinberg K, Galula S, Biton M. Factors affecting Israeli
2. Hass R, Kasper C, Böhm S, Jacobs R. Different populations and sources
womenʼs decision whether to donate cord blood. MCN Am J Matern
of human mesenchymal stem cells (MSC): a comparison of adult and
Child Nurs. 2014;39:96–101.
neonatal tissue-derived MSC. Cell Commun Signal. 2011;9:12.
25. Fishbein M, Ajzen I. Predicting and changing behavior: the reaso-
3. Broder SM, Ponsaran RS, Goldenberg AJ. US public cord blood bank-
ned action approach. New York, NY: Psychology Press; 2010 xix,
ing practices: recruitment, donation, and the timing of consent.
518–xix, 518.
Transfusion. 2013;53:679–87.
26. Muthén LK, Muthén BO. Mplus user’s guide. 8th ed. Los Angeles,
4. Katz G, Mills A, Garcia J, Hooper K, McGuckin C, Platz A, et al. Banking
CA: Muthén & Muthén; 2018.
cord blood stem cells: attitude and knowledge of pregnant women in
27. Satorra A, Bentler P. In: University of California Department of Psychol-
five European countries. Transfusion. 2011;51:578–86.
ogy, editor. Scaling corrections for statistics in covariance structure analy-
5. Bhandari R, Lindley A, Bhatla D, Babic A, Mueckl K, Rao R, et al.
sis. Los Angeles, CA: California Ditigal Library. UCLA statistics series 2;
Awareness of cord blood collection and the impact on banking.
1988 p. 0–13.
Pediatr Blood Cancer. 2017;64:1–9.
28. Tanaka JS. Multifaceted conceptions of fit in structural equation
6. Harris DT. Non-haematological uses of cord blood stem cells. Br J
models. In: Bollen RA, Long JS, editors. Testing structural equation
Haematol. 2009;147:177–84.
models. Newbury Park, CA: Sage Publications; 1993. p. 10–39.
7. Harris DT. Cord blood stem cells: a review of potential neurological
29. Kline RB. Principles and practice of structural equation modeling.
applications. Stem Cell Rev. 2008;4:269–74.
New York, NY: Guilford Press; 2016.
8. Bianchi M, Papacci P, Valentini CG, Barbagallo O, Vento G, Teofili L.
30. Hu L, Bentler PM. Cutoff criteria for fit indexes in covariance struc-
Umbilical cord blood as a source for red-blood-cell transfusion in
ture analysis: conventional criteria versus new alternatives. Struct
neonatology: a systematic review. Vox Sang. 2018;113:713–25.
Equ Modeling A Multidiscip J. 1999;6:1–55.
9. Biazar E. Use of umbilical cord and cord blood-derived stem cells for tis-
sue repair and regeneration. Expert Opin Biol Ther. 2014;14:301–10. 31. Meredith W. Measurement invariance, factor analysis and factorial
10. Petrini C. Ethical issues in umbilical cord blood banking: a compara- invariance. Psychometrika. 1993;58:525–43.
tive analysis of documents from national and international institu- 32. Bollen KA. A new incremental fit index for general structural equa-
tions. Transfusion. 2013;53:902–10. tion models. Sociol Methods Res. 1989;17:303–16.
11. Shearer WT, Lubin BH, Cairo MS, Notarangelo LD. Cord blood 33. Dinç H, Şahin NH. Pregnant women’s knowledge and attitudes about
banking for potential future transplantation. Pediatrics. 2017;140: stem cells and cord blood banking. Int Nurs Rev. 2009;56:250–6.
e20172695. 34. Salvaterra E, Casati S, Bottardi S, Brizzolara A, Calistri D, Cofano R,
12. Istituto Superiore Di Sanità.: Report 2020: Banche di sangue di et al. An analysis of decision making in cord blood donation through
cordone ombelicale [Internet]. Vol 31. 2020. https://www. a participatory approach. Transfus Apher Sci. 2010;42:299–305.
centronazionalesangue.it/rapporti-di-attivita-delle-banche-sco/ 35. Godin G, Vézina-Im LA, Bélanger-Gravel A, Amireault S. Efficacy of
13. Grano C, Scafa V, Zucaro E, Abad R, Lombardo C, Violani C. interventions promoting blood donation: a systematic review.
Knowledge and sources of information on umbilical cord blood dona- Transfus Med Rev. 2012;26:224–237.e6.
tion in pregnant women. Cell Tissue Bank. 2020;21:279–87. 36. World Marrow Donor Association.: D2.1 Report on 2019 unrelated stem cell
14. Karagiorgou LZ, Pantazopoulou MNP, Mainas NC, Beloukas AI, transplants. 2020. https://wmda.info/wp-content/uploads/2021/02/2020_
Kriebardis AG. Knowledge about umbilical cord blood banking among D2.1-Report-on-2019-unrelated-stem-cell-transplants-MJ.pdf
Greek citizens. Blood Transfus. 2014;12:24–7. 37. Cell Trials Data.: Percentage of births banking cord blood. 2021.
15. Jordens CFC, Kerridge IH, Stewart CL, O’Brien TA, Samuel G, https://celltrials.org/maps-cell-and-gene-therapy/percentage-of-
Porter M, et al. Knowledge, beliefs, and decisions of pregnant births-banking-cord-blood
Australian women concerning donation and storage of umbilical cord 38. Guo JLL, Wang TFF, Liao JYY, Huang CMM. Efficacy of the the-
blood: a population-based survey. Birth. 2014;41:360–6. ory of planned behavior in predicting breastfeeding: meta-
16. Okocha C, Ezeama N, Aneke J, Onubogu C, Okafor C, Egbunike C. analysis and structural equation modeling. Appl Nurs Res. 2016;
Knowledge and attitude of donating and using cord blood for transfusion 29:37–42.
176 FERNANDES ET AL.

39. McEachan RRC, Conner M, Taylor NJ, Lawton RJ. Prospective pre- SUPPORTING INFORMATION
diction of health-related behaviours with the theory of planned Additional supporting information may be found online in the
behaviour: a meta-analysis. Health Psychol Rev. 2011;5:97–144.
Supporting Information section at the end of this article.
40. Hyde MK, Knowles SR, White KM. Donating blood and organs:
using an extended theory of planned behavior perspective to
identify similarities and differences in individual motivations to
How to cite this article: Fernandes M, Alessandri G, Abbad R,
donate. Health Educ Res. 2013;28:1092–104.
Grano C. Determinants of the intention to donate umbilical
41. Hyde MK, White KM. Testing an extended theory of planned behav-
ior to predict young people’s intentions to join a bone marrow donor cord blood in pregnant women. Vox Sang. 2022;117:169–176.
registry. J Appl Soc Psychol. 2013;43:2462–7.
Received: 5 April 2021 Revised: 24 June 2021 Accepted: 7 July 2021

DOI: 10.1111/vox.13183

ORIGINAL ARTICLE

Perceptions on acceptability and reported consumption

of marijuana by blood donors prior to donation in the
recreational use state of Colorado, USA

Kyle Annen1,2 | Melkon G. DomBourian1,2

1
Department of Pathology, University of
Colorado School of Medicine, Aurora, Abstract
Colorado, USA
Background and Objectives: Blood donor opinions and behaviours regarding mari-
2
Department of Pathology and Laboratory
Medicine, Children’s Hospital Colorado,
juana use are not well known nor is the potential impact to the blood supply. We
Aurora, Colorado, USA sought to assess opinions and frequency of marijuana use in proximity to blood
donation via a survey of blood donors at a hospital-based blood collection site in a
Correspondence
Kyle Annen, Department of Pathology and state where recreational marijuana use has been legal since 2012.
Laboratory Medicine, Children’s Hospital
Materials and Methods: Blood donors at least 18 years of age who donated between
Colorado, 13123 East 16th Avenue, Aurora,
CO 80045, USA. 2014 and 2019 were surveyed electronically, with all responses kept anonymous to
Email: [email protected]
encourage engagement and accurate reporting.
Results: Overall response rate was 8.03% (12,186 surveys sent with 979 responses).
Of responding donors, 23.5% indicated that they felt that consuming various forms
of marijuana was acceptable prior to blood donation. Marijuana use <72 h prior to
blood donation was reported in all demographic groups surveyed except age
18–24 years. Of donors who reported daily marijuana use, 47.4% indicated >20
donations and 52.6% indicated apheresis platelet donation.
Conclusion: Nearly one quarter of responding blood donors feel that marijuana use is
acceptable prior to blood donation, and nearly every demographic group surveyed
indicated use of marijuana <72 h prior to donation. These results suggest the need
for additional research to determine if marijuana-related metabolites in collected
blood products negatively impact recipients, particularly vulnerable populations such
as children and pregnant women. These results may inform whether changes in
donor screening or testing for marijuana use are warranted.

KEYWORDS
blood donors, cannabis, marijuana, THC

I N T R O D U CT I O N Colorado and Washington, became the first to legalize recreational

marijuana use [1, 2]. Since then, marijuana legalization for
It is currently unknown whether blood donors in the United States recreational purposes has expanded to 15 states in total, with many
use marijuana prior to blood donation and if so, in what frequency. It other states having passed related legislation [3]. In addition, legisla-
is also unknown if the legalization of marijuana for medical or recrea- tion to decriminalize marijuana at the federal level was passed by The
tional use has impacted the perception by blood donors that mari- US House of Representatives in 2020 and awaits review in the Senate
juana use is safe in proximity to blood donation. In 2012, two states, as of this writing [4]. In a household survey conducted in the states of

Vox Sanguinis. 2022;117:177–184. wileyonlinelibrary.com/journal/vox © 2021 International Society of Blood Transfusion. 177
178 ANNEN AND DOMBOURIAN

Washington, Colorado, Oregon and New Mexico in 2017, 41% of blood products on a recipient is currently unknown. THC has been
respondents endorsed having used marijuana in their lifetime and 9% found in plasma products [13, 14], and approximately, 10% is bound
endorsed use in the past month [5]. In a cross-sectional US survey to red cells [15, 16]. While the large volume of distribution of THC
from 2002 to 2014, the perception that marijuana presented no risk would seem to make significant transfusion-associated marijuana
to the user increased from 5.6% to 15.1% [6]. As this trend of exposure seem less likely in adults, the potential impact of THC on
expanded legalization continues, the prevalence of marijuana use is vulnerable populations through a blood transfusion, such as pregnant
expected to increase as public opinion is trending towards acceptance women, premature infants, or the foetus during intrauterine transfu-
[7]. This raises the question of whether blood donors who reside in sion, has not been established. THC rapidly crosses the placenta, and
states or countries where marijuana is legal for recreational use may maternal marijuana use has been shown to negatively impact foetal
perceive that the drug is harmless and feel that it is acceptable to brain development and lead to pregnancy complications such as spon-
donate after recent marijuana consumption. taneous pre-term birth, although these studies are based on more
The US Food and Drug Administration (FDA) requires screening than a single THC exposure [17–19].
questions for specific medications with a set deferral period, such as Children’s Hospital Colorado Blood Donor Center is a hospital-
those that are known to cause teratogenicity (vitamin A derivatives, based collection centre in the US state of Colorado, which collects
5-alpha-reductase inhibitors); that impact the effectiveness of the approximately 8000 units of whole blood and 5000 units of apheresis
product (anticoagulants) or indicate a possible risk of bacterial con- platelets annually for hospital use in neonatal, paediatric and high-risk
tamination (antibiotics). Additionally, the FDA requires questions that pregnancy care. Our blood donor centre supports a wide range of ser-
screen for intravenous drug use (IVDU) but does not screen for pre- vices including intrauterine transfusion; heart, liver, and kidney trans-
scription opioids or recreational drugs that may be consumed orally or plant; as well as haematology and oncology including bone marrow
by inhalation [8]. In addition, the Universal Donor Health Question- transplant. Given both the changes in societal perspectives on mari-
naire (DHQ) does not ask donors about marijuana use and commonly juana use, and unknown impact of possible THC transmission by
used legal drugs such as tobacco or alcohol are also not specifically blood transfusion into the vulnerable populations that our institution
addressed [8, 9]. This is despite nicotine and its metabolites, as well as regularly serves, we wanted to better understand donor attitudes and
prescription drugs such as opiates, benzodiazepines, stimulants, or behaviour related to marijuana use. Here, we report a survey of blood
barbiturates having been proven to be present in whole blood units donors at our collection centre to define and quantify both the per-
[9, 10]. In 2019, the FDA held a public hearing to obtain stakeholder spectives regarding, and actual usage of, marijuana in proximity to
views on marijuana and marijuana-containing products [11]. This hear- blood product donation.
ing did not extend to blood donation issues, and currently, the FDA
does not require the testing of blood donors or blood products for
marijuana or any illicit drugs. M A T E R I A L S A N D M ET H O D S
Despite recreational marijuana consumption being legal in several
countries around the world, there is a paucity of international litera- Study design
ture on marijuana use specifically among blood donors. International
guidance documents regarding blood and blood components such as A cross-sectional study was conducted from August 2019 through
the European Parliament Commission Directive 2004/33/EC mirror December 2019 via an 18-question survey delivered via email
guidance in the United States and does not specifically address mari- (surveymonkey.com, San Mateo, CA). Responses were anonymous
juana use, but does mention non-prescribed drugs in the context of to encourage truthful reporting. The survey questions addressed
infectious disease risk from intravenous/intramuscular (IV/IM) route perceptions of marijuana use in relation to blood donation and
[12]. Through formal correspondence with colleagues within the Inter- transfusion, personal use of marijuana products, demographic infor-
national Society of Blood Transfusion (ISBT) Haemovigilance Working mation and donation history. An optional comments section was
Group, we were able to identify policies in countries where medical available for most non-demographic questions. The option to skip
marijuana use has been legalized and recreational use has been any question was available, thus a respondent could indicate mari-
decriminalized or legalized including: the Netherlands, Denmark, Nor- juana use but decline to answer gender, and so forth. Institutional
way, Finland and Canada. Deferral periods for the European countries review board approval was obtained from the University of Colo-
listed ranged from 24 h post single use, 4 weeks for habitual use and rado. Participants who had donated between 1 January 2014 and
up to 12 months for any use. Canadian marijuana deferral policy 9 October 2019 were sent the survey. Responses were collected for
focusses on whether a potential donor is intoxicated and would be 3 months.
unable to understand screening questions and therefore be unable to
provide informed consent, but does not specify a specific period of
time for deferral (personal correspondence, ISBT Haemovigilance Study participants
Working Group, 16 June 2021).
The potential impact of Δ9-tetrahydrocannabinol (THC), the psy- Eligibility criteria included Children’s Hospital Colorado blood donors
choactive component of marijuana, or its metabolites in donated 18 years of age and older who qualified for blood donation under the
BLOOD DONOR MARIJUANA USE 179

regulatory requirements set by the FDA. Donors are qualified based T A B L E 1 Demographic distribution of survey respondents with
on a brief physical exam with established acceptable vital signs, mini- proportion who indicated use of marijuana since legalization in
Colorado
mum haematocrit level and successful completion of the DHQ, which
includes screening questions regarding medications (e.g., antibiotics or Indicated marijuana
use <72 h prior to blood
medications with known teratogenicity), IVDU or other blood expo-
donation (% of total
sure, travel history (e.g., malaria or variant Creutzfeldt Jakob exposure) Demographics Response (%) respondents)
and high-risk sexual practices (e.g., sex with a prostitute) [8, 20]. All
Gender
donors had donated at least once during the survey time period.
Male 336 (34.4%) 16 (4.8%)
Donors who did not supply an email address or provided incomplete
Female 640 (65.4%) 20 (3.1%)
responses were excluded from the survey. Participants were informed
Other 1 (<1%) 1 (100%)
that the survey was anonymous to encourage truthful reporting, thus
responses were not tracked for follow-up. Age (years)
18–24 32 (3.27%) 0
25–34 200 (20.4%) 14 (7%)
Analysis 35–44 246 (25.2%) 11 (4.5%)
45–54 219 (22.4%) 3 (1.4%)
Results were analysed based on the proportion of respondents who 55–64 165 (16.9%) 3 (1.8%)
answered the question, using SurveyMonkey (San Mateo, CA) and 65+ 116 (11.9%) 6 (5.1%)
Microsoft Excel (Redmond, WA). Results are reported using descrip- Donation site (multiple option)
tive statistics.
Mobile drive 460 (47%) 21 (4.6%)
Donor centre 654 (66.8%) 23 (3.5%)
Frequency of donation in units
RESULTS
0 5 (0.5%) 0 (0%)
1–5 319 (32.6%) 10 (3.1%)
Demographics and donation history
6–10 207 (21.2%) 6 (2.9%)

A total of 12,186 donors met inclusion criteria and were sent surveys, 11–15 88 (9%) 5 (5.7%)

from which 979 individual donors took part, for a response rate of 16–20 59 (6.1%) 2 (3.4%)
8.03%. The majority of respondents identified as female (65.4%) and Greater than 20 295 (30.1%) 14 (4.7%)
one respondent identified as ‘other’. Age ranges were grouped in Type of products donated (multiple option)
10-year increments up to age 65 years (see Table 1), with the majority Whole blood 887 (90.6%) 30 (3.4%)
of respondents in the 35- to 44-year-old category (25%). This is gen- Apheresis platelets 262 (26.8%) 15 (5.7%)
erally comparable to the demographic distribution at our centre, with Apheresis plasma 114 (11.6%) 9 (7.9%)
a female preponderance. For comparison, of 935 blood donors at our
Apheresis red cells 81 (8.3%) 9 (11.1%)
center between 2019 and 2020, 56% were female. Majority age
group was 35–44 (24%), and donations represented 72% whole blood
and 22% apheresis platelets. The most common donation setting in
our survey responses was the main hospital donor centre (66.8%) with reported frequency of using marijuana (edibles, smoke and/or vape)
the remaining donors presenting to mobile drives. Product donations as less than once a month (40.3%) with a minority reporting using it
were primarily whole blood units (90.6%) with the highest reported daily (6.8%) or a few times a week (5.8%). For donors who reported
number of individual donations was 1–5 units (32.6%), followed by marijuana use, 43 of 312 respondents (13.8%) stated they have used
greater than 20 units (30.1%). Donors were able to indicate that they marijuana within 72 h of blood donation. Within this user subgroup of
had donated more than one type of product (e.g., select both ‘whole respondents, 32.5% reported donating more than 20 units of blood
blood’ and ‘platelets on machine’). products, 34.8% were platelet donors and 16.2% indicated donating
both. When looking at those who reported daily use, 47.4% (9/19)
had donated >20 units; 52.6% (10/19) were platelet donors and
Marijuana use reported by blood donors 26.3% (5/19) indicated donating both (see Figure 1).

When questioning if a donor had used marijuana (not including can-

nabidiold (CBD) products) since legalization, 32.6% of respondents Concerns about reporting use
stated they had tried it at least once. Of these, 2.9% of the donors
reported using marijuana prescribed by a physician, with migraines, Several questions were designed to directly address marijuana use by
anxiety and pain being common reasons for treatment. Most donors donors (see Table 2). Donors were queried ‘If you were asked, at the
180 ANNEN AND DOMBOURIAN

T A B L E 2 Survey responses: Self-reported marijuana use and

frequency of marijuana by blood donors

Answer
Survey question Response choices (%)
Remember this survey is Yes, edibles 256 (26.1)
anonymous: Have you Yes, smoke or 200 (20.4)
personally used marijuana vape
(not including CBD oil) at any
time since it became legal in No 272 (62.6)
Colorado (medical or
recreational)? N = 728
You indicated that have used Just tried it once 101 (32.6)
marijuana (edibles, smoke or Less than once a 125 (40.3)
F I G U R E 1 Blood donors who used marijuana frequently reported vape), how often do you use month
a proportionally higher rate of repeated donations and apheresis it? N = 310
A few times a 45 (14.5)
platelet donations than non-users
month
A few times a 18 (5.8)
week
time of donation, if you had used a marijuana product recently
Daily 21 (6.8)
(<72 h), would you feel comfortable answering honestly?’ Most
Have you used marijuana prior to Yes, edibles 19 (6.1)
respondents indicated ‘Yes’. Some reasons cited included that they
blood donation (72 h or less)?
do not use it (67.8%), and because it is legal (13.1%) or because they
N = 312
do not believe it affects the blood donation (3.5%). Responses for
Remember this survey is Yes, smoke or 24 (7.7)
‘No’ to this question included being concerned about that informa- anonymous: Have you vape
tion being part of the donor record (1.4%); feeling it should not mat- personally used marijuana No 272 (87.2)
ter (1.9%) and that the question was too personal (0.7%). Comments (not including CBD oil) at any
time since it became legal in
for this question highlighted concerns for potential legal or financial
Colorado (medical or
risk, such as one respondent who stated ‘I would be concerned recreational)? N = 728
about it being on record, especially not knowing the confidentiality
standards for donor records off the top of my head. It is legal in Col-
orado, but not nationally, and many employers have different stan-
dards, and I don’t know what my insurance company’s reaction use of marijuana. When asked ‘how would you feel if you knew that
would be.’ Another respondent pointed out that workplace drives at you or someone you cared about could receive a unit of blood from
locations where the employer does not allow marijuana use when someone who had recently consumed a marijuana product?’ the
off-duty could be a motivation to lie about recent use: ‘My donation majority (43.5%) of respondents answered that they would ‘be con-
was mobile and made at my place of work. I would be concerned cerned but accept the transfusion if it was really needed’ with 22.9%
that this information could be shared or accidentally seen.’ Several of respondents saying ‘it wouldn’t matter’ and 5% saying they would
comments also raised concerns about social judgement, such as ‘refuse the blood product even if it could negatively affect their
‘I wouldn’t think my use has any bearing on my ability to give blood health.’ Comments generally reflected that the impact was unknown
at all and therefore might be less likely to answer honestly. Also, the and the respondent would trust their medical professional, although
stigma.’ some had concerns. One respondent commented ‘This would be a
grand travesty and complete breach of confidence in the medical
world. We come to hospitals and clinics to receive care that is SAFE,
Perceptions on marijuana use in blood donation EFFECTIVE, and most importantly PROVEN. Marijuana is very far
from being proven. It has no place when it comes to emergency
To help identify the changes in cultural perception on acceptability medicine.’ Several others were concerned about allergic reactions,
of marijuana use in relation to blood transfusion, we designed sev- hypersensitivity to THC or the impact on their job if it were to cause
eral questions to address this issue (see Table 3). Most respondents a positive test.
felt that it was unacceptable to use marijuana prior to blood dona- Questioning the timing of marijuana use prior to donation showed
tion; however, 23.5% of respondents indicated it was acceptable. Of 13.2% respondents thought timing did not matter and no deferral was
note, three donors who indicated that they have used marijuana needed. The majority (30.3%) thought donors should wait at least
<72 h prior to a blood donation responded that they felt it was not 72 h. Other time periods included at least 1 day (14.1%), more than a
acceptable to use marijuana prior to blood donation; two indicated week (14.4%) and more than a month (12.4%). Some respondents
they would not feel comfortable answering the question honestly if (3.8%) thought no length of time would be sufficient and that people
they were asked in screening and one of these donors indicated daily who use marijuana should be permanently deferred.
BLOOD DONOR MARIJUANA USE 181

TABLE 3 Survey responses: Donor perceptions of marijuana use in proximity to blood donations

Answer Uses marijuana <72

Survey question Response choices (%) h? (N = 37)
Do you think it is acceptable for a blood donor to Yes 230 (23.5) 31
consume or smoke marijuana prior to a blood No 684 (69.9) 3
donation?
If you were asked, at the time of donation, if you Yes, because I don’t use it 664 (67.8) 0 (0)
had used a marijuana product recently (<72 h), Yes, because it is legal in Colorado 128 (13.1) 11
would you feel comfortable answering
honestly? Yes, because I don’t believe it affects the blood 34 (3.5) 11
donation.
No, because that question is too personal 7 (0.7) 1
No, because I’d be concerned about that being in 14 (1.4) 5
my donor record
No, because it isn’t a question that should 19 (1.9) 7
matter for blood donation
Yes, other 56 (5.7) 2
No, other 8 (0.8) 0
If it were considered acceptable for someone to Anytime, I don’t think it matters 129 (13.2) 21
consume marijuana prior to blood donation, They should wait at least a day 138 (14.1) 8
how long should a person wait to donate after
consuming? They should wait at least 72 h 297 (30.3) 7
They should wait more than a week 141 (14.4) 0
They should wait more than a month 124 (12.4) 0
People who use marijuana should be 37 (3.8) 0
permanently deferred (never allowed to
donate)
How would you feel if you knew that you or It wouldn’t matter 225 (22.9) 30
someone you cared about could receive a unit I would be concerned but accept the transfusion 426 (43.5) 6
of blood from someone who had recently if it was really needed
consumed a marijuana product?
I don’t know 204 (20.8) 1
I would refuse the blood product even if it could 49 (5) 0
negatively affect my health

Note: Donors had the option to skip questions.

DISCUSSION Honesty, when answering the DHQ is critical to the safety of the
blood supply. Although many highly personal questions are asked
The objective of this study was to determine the opinions and behav- about sexual activity, it is known that blood donors are not always
iours of blood donors related to marijuana use both in general and in honest or comfortable with providing responses. A German study rel-
proximity to blood donation in a state with legalized recreational use. evant to blood donor truthfulness in reporting recreational drug use
Through our anonymous survey, we found that among donors respon- found that of 186 donors surveyed, all denied using recreational
dents almost a quarter found marijuana use acceptable prior to blood drugs. However, hair and urine samples collected at the time of dona-
donation. We also saw that among donors that use marijuana daily, tion from these same surveyed donors showed that 10 donors tested
almost half had donated greater than 20 times, and many of them positive for recreational drugs with six positives for cannabinoids [23].
were apheresis platelet donors. In our survey, while the majority of donors indicated they would be
Although 15 states have legalized recreational marijuana [3], as of honest in answering a question about marijuana use, some did state
this writing, it is still illegal at the federal level in the United States, and they would not be honest, feeling that the question should not be in
many employers consider a positive drug test grounds for termination. the donor record (1.4%), is too personal (0.7%) or that it should not
This is true in the state of Colorado currently; although a House Bill was matter for blood donation (1.9%).
introduced in January of 2020 that would prohibit firing an employee The legal and potential employment implications for a recipient
for using marijuana when off-duty, it was postponed indefinitely [21]. must also be considered. Although confidentiality measures are in
Countries with nationally legalized recreational marijuana use such as place to enable blood donors to securely and honestly answer the
Canada are also addressing this dilemma [22]. This issue is reflected in DHQ, privacy breaches may occur. Lawsuits have been brought alleg-
the small proportion of blood donors (1.4%) who stated they would be ing negligence both in screening donors and failure to inform recipi-
concerned about marijuana use being in the blood donor record. ents of risks. Cases related to confidentiality breaches regarding
182 ANNEN AND DOMBOURIAN

human immunodeficiency virus in the blood supply provide guidance blood donors were sent a routine Donor Experience Survey during
for informed consent and record-keeping policies that would be appli- the same time period as our survey, with a response rate of 15.7%.
cable to donor marijuana use status [24]. We may also have unintentionally introduced a bias with the
Although transfusion services and blood collection centres have question by choosing ‘72 h’ as our cutoff for several questions.
generally not been found liable for harm if they were following stan- The timeframe of 72 h was chosen for the survey as there is evidence
dard practices, some lawsuit cases ruled that the blood supplier or that that is the approximate amount of time in which THC will
trade group should have enacted screening or provided recommenda- become undetectable in urine [28]. One question: ‘If it were consid-
tions when the industry became aware of the risks to the recipient ered acceptable for someone to consume marijuana prior to blood
[25, 26]. Transmission of small amounts of drugs – from prescriptions donation, how long should a person wait to donate after consuming?’
or recreational use – is possible but is not generally disclosed as part had the majority (30.3%) of responses indicate the 72 h was the ideal
of the blood transfusion consent [10]. Because we do not know the window. Due to our use of ‘72 h’ as a timeframe for marijuana use in
transmissibility of marijuana in blood products, it has not been proven this and several other questions, we may have inadvertently signalled
that there would be no impact should a recipient receive a blood or that this was a preferred response. This question did receive a wide
platelet transfusion on an outpatient basis and then have workplace range of responses, with 14.1% choosing ‘they should wait at least a
or school drug testing, a traffic accident or other incident that could day’ and 3.8% stating that ‘people who use marijuana should be per-
incorrectly implicate them as a marijuana user. The legal risk is further manently deferred (never allowed to donate).’
complicated by the interstate transfer of blood products, where a unit One of the reasons that little attention has been paid to the
could be collected in a legal state and transfused in one where mari- amount of marijuana that may be transfusion-transmitted is that it is
juana is still illegal. Although the timing of a positive drug test after a believed to be an insignificant amount for an adult. Whether mari-
receiving a THC-positive unit may seem unlikely, a chronic marijuana juana metabolites can be transmitted by blood transfusion in a quan-
user may have a higher baseline plasma level of THC due to high lipo- tity significant enough to cause harm to a vulnerable patient such as a
philic binding, with subsequent redistribution, resulting in a potentially neonate is unknown. Studies on marijuana use during pregnancy have
higher dose of THC in the donated unit [16]. Knowing the transmissi- reported increased likelihood of low birth weight and require care in
bility of THC in the blood supply may be helpful for patients in this the neonatal intensive care unit [29–31]. In addition, there is no ‘mini-
unfortunate predicament and may also support deferring for mari- mum safe dose’ established for marijuana use in pregnancy. If it is
juana usage. determined that any significant amount of THC is present in donated
There were several weaknesses of the study design. A survey units, a deferral may be needed until the safety limit of marijuana can
sent prior to legalization was not conducted; therefore, it is unclear if be established. The FDA seeks to protect vulnerable populations from
use of marijuana, when illegal, significantly varied when compared to blood transfusion risks – users of teratogenic mediations such as
use after legalization. As this was an electronically administered sur- finasteride are deferred for 30 days, even though the ingested dose is
vey, there is possible bias in the response group. In an effort to minimally transmitted through a blood transfusion after only a 3-day
encourage truthful reporting, the survey was not linked to respon- deferral [32]. Comparatively, although it has not been proven or stud-
dent’s email addresses, thus, we could not send reminders to partici- ied, there is a perception by the blood industry that a single exposure
pants who did not respond and could not follow-up with respondents of marijuana, diluted into a recipient, would be essentially harmless, as
to verify their responses. We chose not to re-send the survey to all discussed by CBS-Hema-Quebec Donor Selection Criteria Working
donors who met inclusion criteria to avoid receiving repeat responses. Group, an advisory group that includes medical, scientific, quality and
We also do not know if our sample was diluted by surveying a 5-year regulatory affairs, and operational experts from both organizations, as
range of donors, as many may no longer be active donors and thus well as a pharmacist and a donor representative (personal correspon-
not responsive to communication from our centre. This likely means dence, Mindy Goldman, 4 June 2021).
that our respondents were those who had a high likelihood to respond The findings in this study shows that, among our population of
to surveys or were highly interested in the topic and possibly recent blood donors, marijuana use in proximity to donation does occur.
donors. This may also account for the high percentage of donors who Marijuana use less than 72 h prior to blood donation was reported in
reported both donating more than 20 units and frequent marijuana all demographic groups surveyed except for the 18–24 age group.
use, skewing our results towards a higher percentage than is repre- Overall, of donors who indicated they had used marijuana less than
sentative of the overall donor population. The converse is also possi- 72 h prior to blood transfusion, 32.5% indicated that they had
ble: the topic of marijuana use may have dissuaded some donors from donated more than 20 times, and although the raw numbers for this
responding. We also note that donor respondents were a female group are very small, it is of interest that frequent marijuana users
majority; a review of the literature shows that sex differences in can- appear to be frequent donors as noted in Figure 1: within those who
nabis use do exist. Males are found to use marijuana more frequently reported daily use, 52.6% indicated they were apheresis platelet
and in higher doses when compared to females, which may imply that donors and 47.4% indicated they had donated greater than 20 prod-
our results underestimate marijuana use in our donor population [27]. ucts, with 26.3% indicating both. It is possible that this number repre-
In addition, the non-incentivized participation rate of 8.03% was sents a small group of marijuana users who are highly motivated
lower than anticipated. For comparison, Children’s Hospital Colorado donors and also motivated survey responders; as this is one of the
BLOOD DONOR MARIJUANA USE 183

most desirable populations for blood donation, this represents a con- 2. Washington state initiative 502. https://www.sos.wa.gov/_assets/
cern for the blood supply, should these individuals ever need to be elections/initiatives/i502.pdf (2012). Accessed 15 Mar 2021.
3. Hansen, C., Alas, H. States where marijuana is legal: A guide
deferred from donating apheresis platelets.
to marijuana legalization. U.S. news and world report:
Platelets, as a plasma-based product, may have more potential for Best states. 2020. https://www.usnews.com/news/best-states/
transmission of THC to vulnerable populations [15]. If this is the case, articles/where-is-marijuana-legal-a-guide-to-marijuana-
then a question on the DHQ with a related deferral period may need legalization. Accessed 15 Mar 2021.
4. Colorado House of Representatives: 116th congress. HR-3884.
to be established. In 2014, Booth and Gehrie proposed that a 2-week
https://www.congress.gov/116/bills/hr3884/BILLS-116hr3884rh.
deferral would be appropriate to allow for complete metabolism of pdf (2019). Accessed 24 Feb 2021.
any marijuana use [33]. As we found a dedicated population of repeat 5. Pacula R, Jacobson M, Maksabedian EJ. In weeds: a baseline view of
platelet donors who use marijuana in our survey, these results, if cannabis use among legalizing states and their neighbours. Addiction.
2016;111:973–80.
applicable to the other states or countries that allow recreational mar-
6. Compton WM, Han B, Jones CM, Blanco C, Hughes A. Marijuana use
ijuana use, suggest that a 2-week deferral would have a significant and use disorders in adults in the USA, 2002-14: analysis of annual
impact on the donor blood supply. If blood product testing for THC cross-sectional surveys. Lancet Psychiatry. 2016;3:954–64.
and its related metabolites was used in place of screening questions, 7. Carliner H, Brown QL, Sarvet AL, Hasin DS. Cannabis use,
attitudes and legal status in the U.S. a review. Prev Med. 2017;104:
this would create additional delay and cost as detection in whole
13–23.
blood would utilize the high complexity methodology of liquid chro-
8. Federal Drug Administration: Guidance for industry: Implementation of
matography tandem mass spectrometry [34]. acceptable full-length and abbreviated donor history questionnaires
Colorado was one of the first states to legalize recreational mari- and accompanying materials for use in screening donors of blood and
juana; the results we have reported here may be a harbinger for the blood components. https://www.f.gov/regulatory-information/search-
fda-guidance-documents/implementation-acceptable-full-length-and-
trends of use by blood donors as recreational marijuana use becomes
abbreviated-donor-history-questionnaires-and-accompanying (2020).
more socially acceptable both nationally and internationally. Based Accessed 13 Jan 2021.
on our results, with a proportion of nearly every demographic group 9. Wiencek JR, Gehrie EA, Keiser AM, Szklarski PC, Johnson-Davis KL,
in our survey indicating they had used marijuana less than 72 h prior Booth GS. Detection of nicotine and nicotine metabolites in units of
banked blood. Am J Clin Pathol. 2019;151:516–21.
to blood donation, there is no doubt that some blood donors are
10. Gehrie EA, Keiser A, Haglock-Adler CJ, Strathmann F, Booth GS.
using marijuana. Now it is up to the stewards of the blood supply to Detecting pharmaceuticals in the red blood cell inventory of a hospi-
determine whether or not there is enough THC transmitted via tal blood bank. J Pediatr. 2017;189:227–231.e1.
blood transfusion to have a negative impact on the recipient, 11. FDA: Question and answer summary on cannabis use. https://www.
fda.gov/news-events/public-health-focus/fda-regulation-cannabis-
whether it be physiological or present legal or financial ramifications.
and-cannabis-derived-products-questions-and-answers (2019).
Further studies on the impact of THC transmitted by blood transfu- Accessed 24 Feb 2021.
sion, especially platelet products, on the recipient are warranted to 12. Commission directive 2004/33/EC of 22 March 2004: implementing
determine if additional action such as screening or deferral is certain technical requirements for blood and blood components. Off
J Eur Union 2004:L 91/25–39. https://www.ema.europa.eu/en/
needed.
documents/regulatory-procedural-guideline/commission-directive-
2004/33/ec-22-march-2004-implementing-directive-2002/98/ec-
ACKNOWLEDGEMEN TS european-parliament-council-regards-certain-technical-requirements-
K.A. designed the study, performed data analysis, created the tables blood-blood-components_en.pdf.
13. Kraemer M, Madea B, Hess C. Detectability of various cannabinoids
and image, and edited the manuscript. M.G.D. performed data analysis
in plasma samples of cannabis users: indicators of recent cannabis
and prepared the manuscript.
use? Drug Test Anal. 2019;11:1498–506.
The authors would like to thank Dr. Mindy Goldman of the CBS- 14. Peters FT, Maurer HH, Hellstern P. Prevalence of illicit drug use in
Hema-Quebec Donor Selection Criteria Working Group and the mem- plasmapheresis donors. Vox Sang. 2003;84:91–5.
bers of the ISBT Hemovigilance Working Party for their insight and 15. Widman M, Agurell S, Ehrnebo M, Jones G. Binding of (+)- and
(minus)-delta-1-tetrahydrocannabinols and (minus)-7-hydroxy-delta-
assistance in identifiying international policies on this topic.
1-tetrahydrocannabinol to blood cells and plasma proteins in man.
J Pharm Pharmacol. 1974;26:914–6.
CONF LICT S OF INTE R ES T 16. Huestis MA. Human cannabinoid pharmacokinetics. Chem Biodivers.
The authors declare no conflicts of interest. 2007;4:1770–804.
17. Lee CC, Chiang CN. Maternal-fetal transfer of abused substances:
pharmacokinetic and pharmacodynamic data. NIDA Res Monogr.
ORCID 1985;60:110–47.
18. Wang X, Dow-Edwards D, Anderson V, Minkoff H, Hurd YL. Discrete
Kyle Annen https://orcid.org/0000-0002-7192-5057
opioid gene expression impairment in the human fetal brain associ-
ated with maternal marijuana use. Pharmacogenomics J. 2006;6:
RE FE R ENC E S 255–64.
1. Colorado Joint Select Committee Legislation: Amendment 64 19. Leemaqz SY, Dekker GA, McCowan LM, Kenny LC, Myers JE,
Implementation. https://www.colorado.gov/pacific/sites/default/files/ Simpson NAB, et al. Maternal marijuana use has independent effects
13%20Amendment%2064%20LEGIS.pdf (2013). Accessed 15 Mar on risk for spontaneous preterm birth but not other common late
2021. pregnancy complications. Reprod Toxicol. 2016;62:77–86.
184 ANNEN AND DOMBOURIAN

20. Transfusion Services Standards Committee. AABB standards for between maternal characteristics, breastfeeding patterns, and neona-
blood banks and transfusion services. 32nd ed. Bethesda, MD: AABB tal outcomes. J Pediatr. 2018;197:90–6.
Press; 2020.p. 64–70. 30. El Marroun H, Brown QL, Lund IO. An epidemiological, developmen-
21. Colorado 73rd General Assembly: Employee protection lawful off- tal and clinical overview of cannabis use during pregnancy. Prev
duty activities. HB20-189. https://leg.colorado.gov/bills/hb20-1089. Med. 2018;116:1–5.
Accessed 25 Feb 2021. 31. Gunn JK, Rosales CB, Center KE, Nuñez A, Gibson SJ, Christ C, et al.
22. Stam L. Employment and human rights law in Canada. https://www. Prenatal exposure to cannabis and maternal and child health outcomes:
canadaemploymenthumanrightslaw.com/2018/10/legal-pot-drug- a systematic review and meta-analysis. BMJ Open. 2016;6:e009986.
testing-off-duty-substance-use/. Accessed 15 Mar 2021. 32. Becker CD, Stichtenoth DO, Wichmann MG, Schaefer C, Szinicz L.
23. Mahl MA, Hirsch M, Sugg U. Verification of the drug history given by Blood donors on medication - an approach to minimize drug burden
potential blood donors: results of drug screening that combines hair for recipients of blood products and to limit deferral of donors.
and urine analysis. Transfusion. 2000;40:637–41. Transfus Med Hemother. 2009;36:107–13.
24. Bierig JR. Legal issues for blood banks. Arch Pathol Lab Med. 1994; 33. Booth GS, Gehrie EA. Implications of legalized recreational marijuana
118:442–53. on the United States blood supply. Transfusion. 2014;54:1902–3.
25. Kelly C, Barber JP. Legal issues in transfusion medicine. Is blood 34. Hubbard JA, Smith BE, Sobolesky PM, Kim S, Hoffman MA, Stone J,
banking a medical profession? Clin Lab Med. 1992;12:819–33. et al. Validation of a liquid chromatography tandem mass spectrome-
26. Blood bank trade group held liable for HIV-tainted transfusion. AIDS try (LC-MS/MS) method to detect cannabinoids in whole blood and
Policy Law. 1996;11:1–11. breath. Clin Chem Lab Med. 2020;58:673–81.
27. Cuttler C, Mischley LK, Sexton M. Sex differences in cannabis use
and effects: a cross-sectional survey of cannabis users. Cannabis
Cannabinoid Res. 2016;1:166–75.
28. Huestis MA, Mitchell JM, Cone EJ. Urinary excretion profiles of How to cite this article: Annen K, DomBourian MG.
11-nor-9-carboxy- 9-tetrahydrocannabinol in humans after single Perceptions on acceptability and reported consumption of
smoked doses of marijuana. J Anal Toxicol. 1996;20:441–52.
marijuana by blood donors prior to donation in the
29. Crume TL, Juhl AL, Brooks-Russell A, Hall KE, Wymore E,
recreational use state of Colorado, USA. Vox Sang. 2022;117:
Borgelt LM. Cannabis use during the perinatal period in a state with
legalized recreational and medical marijuana: the association 177–84.
Received: 2 March 2021 Revised: 18 May 2021 Accepted: 18 May 2021

DOI: 10.1111/vox.13162

ORIGINAL ARTICLE

Qualifying coronavirus disease 2019 convalescent plasma

donors in Israel

Marina Izak1 | Veronica Gendelman1 | Sharron Bransburg-Zabary1 |

Evgeniy Stoyanov1 | Roni Gat2,3 | Daniel Cohen4 | Jacob Chen5,6,7 |
Yasmin Maor7,8 | Avi Benov5,9 | Boaz Lev10 | Oren Zimhony11 | Eilat Shinar1,2

1
National Blood Services, Magen David Adom,
Ramat Gan, Israel Abstract
2
Faculty of Health Sciences, Ben-Gurion Background and Objectives: Passive immunization using investigational COVID-19
University of the Negev, Beer Sheva, Israel
3
convalescent plasma (CCP) is a promising therapeutic strategy and could improve
Clinical Research Center, Soroka University
Medical Center, Beer Sheva, Israel outcome if transfused early and contain high levels of anti-SARS-CoV-2 antibodies.
4
School of Public Health, Tel Aviv University, We report the management of a national CCP collection and distribution program in
Tel Aviv, Israel
Israel.
5
Trauma and Combat Medicine Branch, Israel
Defense Forces Medical Corps, Ramat Gan, Materials and Methods: From 1 April 2020 to 15 January 2021, 4020 volunteer
Israel donors donated 5221 CCP units and 837 (20.8%) donors donated more than once.
6
Hospital Management, Meir Medical Center,
Anti-nucleocapsid IgG antibodies were determined using chemiluminescent immuno-
KfarSaba, Israel
7
Sackler Faculty of Medicine, Tel Aviv
assay method (Abbott). A statistical model based on repeated IgG tests in sequential
University, Tel Aviv, Israel donations was created to predict the time of antibody decline below sample/cut-off
8
Infectious Disease Unit, Wolfson Medical (S/CO) level of 4.0.
Center, Holon, Israel
9 Results: Ninety-six percent of CCP donors suffered a mild disease or were asymp-
The Azrieli Faculty of Medicine, Bar-Ilan
University, Safed, Israel tomatic. Older donors had higher antibody levels. Higher antibody levels (S/CO ≥4)
10
Epidemic Unit, Ministry of Health, were detected in 35.2% of the donors. Low positive (S/CO ≥1.4–3.99) were found in
Jerusalem, Israel
11
37%, and 27.8% had undetectable antibodies (S/CO ≤1.4). The model predicted
Infectious Disease Unit, Kaplan Medical
Center, Rehovot, Israel decrease antibody thresholds of 0.55%/day since the first CCP donation, providing
guidance for the effective timing of future collections from donors with high anti-
Correspondence
Marina Izak, Faculty of Health Sciences, body levels.
Magen David Adom, Ramat Gan, Israel. Conclusions: An efficient CCP collection and distribution program was achieved,
Email: [email protected]
based on performing initial and repeated plasma collections, preferably from donors
with higher antibody levels, and only antibody-rich units were supplied for therapeu-
tic use. The inventory met the quantity and quality standards of the authorities,
enabled to respond to the growing demand of the medical system and provide a
product that may contribute to improve prognosis in patients with COVID-19.

KEYWORDS
antibodies, convalescent plasma, donors

Vox Sanguinis. 2022;117:185–192. wileyonlinelibrary.com/journal/vox © 2021 International Society of Blood Transfusion. 185
186 IZAK ET AL.

I N T R O D U CT I O N COVID-19 as potential CCP donors and conduct laboratory tests to

qualify and supply CCP units for treatment. CCP collections were initi-
Coronavirus disease 2019 (COVID-19), caused by severe respiratory ated on 1 April 2020, according to the first FDA protocol [20], with
syndrome coronavirus 2 (SARS-CoV-2), is one of the biggest global local modification to comply with the MOH regulations and the IRB
health threats of the last century. protocol [21], similar to programs established simultaneously around
At the time of this writing, a year into the pandemic, specific the world [22].
treatment remains elusive [1]. Although the available vaccines may Potential CCP donors required evidence of COVID-19 by molecu-
become a principal game changer in the prevention of new infection, lar tests and two consecutive negative test results after symptomatic
passive immunization by transfusion of COVID-19 convalescent recovery, thereafter 14-day deferral needed before plasma collection.
plasma (CCP) is still used widely. This strategy is based on century-old Recovered COVID-19 patients were referred to the MDANBS by
reports that describe the efficacy of treating patients during the 1918 various sources, including MOH’s database, the Israeli Defense
influenza A pandemic by transfusions of CCP [2–4] and from small Forces, cohorts in closed ethnic communities and social media.
reports, showing encouraging clinical benefit of CCP in patients with Donors gave their consent for transfer of personal data to MDANBS.
severe COVID-19 [5–7]. All CCP donors were non-remunerated volunteers whose health his-
Based on these reports, the Israeli Ministry of Health (MOH) tories complied with MOH and MDANBS criteria for blood donations.
requested Magen David Adom National Blood Services in Israel Only males or nulliparous females were recruited to mitigate the risk of
(MDANBS) to establish an investigational CCP program as a part of a transfusion-associated acute lung injury (TRALI). A Donor Recruitment
national COVID-19 treatment protocol. Call Center was established and operated by trained MDANBS person-
As of today, data accumulated worldwide suggest that transfusion nel, who conducted telephone interviews with potential donors to
of CCP is safe and effective [8, 9]. Recent data from matched con- assure compliance with requirements. Pre-donation screening included
trolled studies [10, 11], from randomized clinical trial [12] and from evaluation of potential donors’ records in the MDANBS computer data-
retrospective analysis [13] showed benefit of CCP in patients treated base (Progesa, MAK-system) to identify prior disqualifying deferral.
early with CCP containing high-titre antibodies (Ab), while others did
not show decrease in mortality [14, 15]. Based on these data,
U.S. Food and Drug Administration (FDA) issued on 4 February 2021 Plasma collection
a revision of the Emergency Use Authorization (EUA) for CCP and lim-
ited the authorization to the use of high-titre CCP only [16]. Several CCP collections were initially performed at the main MDANBS plas-
trials are ongoing, investigating clinical benefit of CCP [17] and stan- mapheresis donation centre that routinely performs apheresis plasma
dardization of serological and neutralization assays [18]. collections and had been involved in similar projects previously [23].
In Israel, transfusing CCP is currently an integral component of To respond to the rapidly growing demands for CCP, four additional
the early treatment of COVID-19, as a part of a national investiga- donation sites were opened, additional mobile apheresis equipment
tional program. All aspects of CCP collection, processing, testing and (MCS+, Haemonetics, Covina, CA) was purchased and extra apheresis
distribution to hospitals nationwide are centrally performed by Magen operators were trained.
David Adom National Blood Services (MDANBS), to assure standardi- In addition to the standard MDANBS Donor Health Question-
zation, quality and impartiality. The treatment protocol was based on naire, every donor signed an informed consent for the CCP collection.
transfusion of two CCP units (200 ml each) 24 h apart, to patients Apheresis collections of 600 ml of CCP were obtained, divided into
approved by the MOH research committee. The results of treating three 200-ml units, frozen at 30 C within 22 h and labelled as Aphere-
the first group of COVID-19 patients have been previously reported sis Convalescent Plasma according to ISBT-128 standards (ISBT code
[19], and the correlation of clinical benefit with higher anti-SARS- E9743). As anti-SARS-CoV-2Ab testing was not available during the first
CoV-2 Ab in transfused CCP was shown. 2 weeks of CCP collections, we saved archive samples for further studies.
A key question for every CCP collection and distribution centre is
how to select the right plasma donors. In this article, we report our
experience accumulated since 1 April 2020, in recruiting CCP donors Donors’ testing
and in inventory management, as our aim is to qualify and supply for
transfusion CCP units with highest anti-SARS-CoV-2 antibody levels. Blood samples from each donation were tested according to MOH
and MDANBS standards and the IRB protocol, including ABO/Rh,
Treponema pallidum haemagglutinin assay (PK7300 Beckman Coulter,
MATERIALS AND METHODS Brea, CA), red blood cells (RBC) antibody screening (Erythra, Grifols,
Spain), serological tests for human immunodeficiency virus I/II (HIV-
Donor population I/II), hepatitis B virus (HBV), hepatitis C virus (HCV), human
T-lymphotropic virus I/II (HTLV-I/II) (Alinity S, Abbott, Green Oaks, IL)
The Ethics Committee of the MOH approved an Institutional Review and individual donor nucleic acid testing (ID-NAT) for HIV-I/II, HCV,
Board (IRB) protocol to recruit individuals who recovered from HBV and West Nile virus (WNV) (Panther, Grifols, Spain).
CCP COLLECTION, QUALIFICATION AND DISTRIBUTION IN ISRAEL 187

Convalescent patients
Applications for contacted
CP
by phone call n=51,116
N=56
Did not meet the criteria for CCP
donation or refused to donate
Scheduled for donation n=45,202
n=5,914 Missed the appointment or
scheduled for later
n=1,658
Arrived for donation
n=4,256 Differed due to health issues (low
Hb, etc.)
n=192
Donated
Units discarded after donation
n=4,020 (abnormal blood tests)
n=44

Donated once Donated more than once

n=3,183 (79.2%) n=837 (20.8%)

Donated Donated Donated

twice three times four times
n= 561 n=212 or more
(14%) (5.3%) n=64
(1.6%)

FIGURE 1 Recruitment of convalescent plasma donors (1 April–15 January, 2021). CCP, COVID-19 convalescent plasma;Hb, haemoglobin

Anti-SARS-CoV-2 antibodies Positive result was defined as S/CO≥1.4 [29, 30]. Having accumu-
lated a sufficient CCP inventory (since 1 October 2020), we quali-
Commercially available assays for anti-SARS-CoV-2 Ab differ by the Ab fied for transfusion CCP units by S/CO: one unit had an Ab level
subclass (IgM, IgA, IgG or total antibody), the targeted antigen (subunit of S/CO ≥7.0 and another– S/CO ≥4.0, thus an average S/CO≥4.5
1[S1] of the spike protein, nucleocapsid protein [N] or the receptor- was provided, in line with the later decision of FDA, issued on
binding domain [RBD]) and by assay method, that is, lateral flow assay 4 February 2021 [16].
(LFA) [24, 25], neutralizing Ab assay (nAb) [26, 27], enzyme-linked 3. Viral neutralization assay
immunosorbent assay (ELISA) [28] and chemiluminescent immunoassay As initial reports indicated a positive correlation between anti-S and
(CLIA) [29, 30]. For this project, we used multiple laboratory methods anti-N IgG values and nAb activity [22, 26], we compared our results
to test the presence of different anti-SARS-CoV-2 Ab. of anti-S by ELISA (EUROIMMUN) and anti-S by CLIA (Abbott) with
results of neutralization studies for the first 53 CCP units. The Israeli
1. Anti-S (S1 subunit) SARS-CoV-2 Ab Institute for Biological Research team performed the test, using a
Serum samples were tested for anti-S IgG and IgA, using ELISA modified plaque reduction neutralization test (PRNT) with Vero E6
(EUROIMMUN AG, Germany), performed in the Research Labora- cells (ATCC® CRL-1586™), as described previously [19].
tories of the School of Public Health, Tel Aviv University during 4. Pre-donation anti-S SARS-CoV-2 rapid point of care (POC) assay
the first month of the project (April, 2020). A positive result was We evaluated anti-S BELTEST-IT COV-2 Rapid Test (PharmAct AG,
defined as a sample to calibrator absorbance (S/CO) ratio Germany) LFA [31], as a part of the pre-donation screening on a cap-
≥ 1.1 [28]. illary blood sample, to avoid collections of plasma from donors with
2. Anti-N (nucleocapsid protein) SARS-CoV-2 Ab undetectable Ab. Results were obtained within 15 min. Only poten-
Starting 1 May 2020, all CCP collections were tested for anti-N by tial donors with positive IgG by POC results proceeded to the aphe-
CLIA, performed on the Architect i2000 SR (Abbott, Green Oaks, resis collection. Venous blood samples were obtained as well for all
IL) automated immunoassay analyser [29]. Testing also included potential donors (including first and subsequent donations) and
samples retained from the first month’s apheresis collections. tested for anti-N by CLIA.
188 IZAK ET AL.

4,020 donors

Anti-SARS-CoV-2 IgG
undetectable
n=1,117 (27.8%)

Anti-SARS-CoV-2 IgG Anti-SARS-CoV-2 IgG Anti-SARS-CoV-2 IgG >7

1.4-3.99 S/CO 4-7.0 S/CO S/CO
n=1,487 (37%) n=952 (23.7%) n=464 (11.5%)

Donors will be invited for further

CCP collections
according to the initial antibody
level (35.2%)

F I G U R E 2 Identification of convalescent plasma donors suitable for further donations, based on anti-nucleocapsid antibody IgG results. CCP,
COVID-19 convalescent plasma;S/CO, sample/cutoff

F I G U R E 3 Correlation between neutralizing antibody (nAb) activity (x-axis) and two serological assays (y-axis): (a) anti-S, EUROIMMUN and
(b) anti-nucleocapsid (anti-N), CLIA; Abbott. Vertical dotted lines represent the cutoff for nAb positivity at the indicated titre. The dashed
horizontal lines represent the cutoff for serological assays positivity. R is calculated by Spearman correlation test

Statistical analysis The last detectable Ab is the dependent variable, a function of: (1)
time between first donation to subsequent donation and (2) initial
Descriptive statistics are presented as mean  SD for continuous vari- antibody level. Data were analysed using R software (Version 3.5.1).
ables and compared using independent t-test or Mann–Whitney test.
Categorical variables presented as number of observations and per-
centage and compared using Pearson χ2 test. To assess correlations RE SU LT S
between Ab levels measured by EUROIMMUN and CLIA tests, we
used Spearman’s correlation test. Donors’ demographics and eligibility
A statistical prediction model of decline of anti-N Ab in subse-
quent CCP donations was created by generalized linear mixed models From 1 April 2020, until 15 January 2021, the CCP Call Center per-
(GLMM) with a random intercept for each participant. GLMM was formed 51,116 telephone interviews (217/day), resulting in 5914
used to account for clustering with link function fitted to distributions. (11.6%) scheduled appointments. Most of the donors referred were
CCP COLLECTION, QUALIFICATION AND DISTRIBUTION IN ISRAEL 189

F I G U R E 4 Gradual decrease of anti-nucleocapsid (anti-N) antibody levels (%) in convalescent plasma donors through 150 days since first
donation. Anti-N anti-SARS-CoV-2 IgG antibodies detected by chemiluminescent microparticle immunoassay (CLIA; Abbott, Green Oaks, IL).
Donors with sample/cut-off ratio (S/CO ≥1.4) were analysed. Ab level in the first donation for each donor was defined as 100% and, in
subsequent donations, was calculated (%) relative to his/her first donation. Light grey lines represent donors’ anti-N levels (in %) through
subsequent donations; trend-line represents the mixed model regression analysis using R software. Ab, antibodies

ineligible for donation due to health reasons or unwilling to donate Anti-SARS-CoV-2 antibodies
plasma. Only 72% (4256/5914) of the scheduled donors arrived to
the Apheresis collection centres, of whom 4064 donated CCP and First 230 CCP collections were tested by EUROIMMUN. Anti-S1 IgG
192 (4.5%) were further deferred on site due to health reasons Ab were undetectable in 17% (39/230). As the results of the test were
(i.e., low haemoglobin, abnormal blood pressure or other health con- not available at the time of CCP release, 19 of these CCP units were
ditions). The percentage of deferrals on site for CCP donors was transfused. Two-weeks’ follow-up was available for 15 patients that
lower than for regular blood donors: 11.3% blood donors deferred in received at least one unit of CCP with undetectable anti-S1 IgGAb
2019 and 10.5% in 2020 due to health reasons. Disqualifying labora- levels, and as discussed previously [19], the lower mean antibody level
tory test results for transfusion-transmitted diseases, abnormal in transfused plasma was a predictor of worse outcome in the group
blood count or presence of clinically significant Ab to blood group of 49 patients analysed in the study [19].
antigens were found in blood samples of 1.08% CCP donors (44/ All 5221 donations from the 4020 donors were tested for anti-N
4064), comparing to 0.53% among blood donors in 2019 and 0.58% by CLIA; 1117/4020 (27.8%) had an anti-N S/CO <1.4 (negative);
in 2020. Final analyses were performed on 4020 CCP donors 1487/4020 (37.0%) had S/CO of 1.4–3.99 (low positive); 952/4020
(Table S1, Figure 1), their mean age was 32.6  12.9 years and (23.7%) had S/CO 4.0–7.0 (positive) and 464/4020 (11.5%) had S/CO
736/4020 (18%) were female and 36% were first-time donors. >7.0 (high positive) (Figure2). Antibody levels were higher in older
About 96% of the CCP donors (3859/4020) were asymptomatic or donors:96 individuals older than 60 years had a mean  SD S/CO of
had a mild COVID-19; 161/4020 (4%) had moderate disease. Analy- 5.74  2.65, while the youngest 506 donors (age 17–20 years) had a
sis of the first 726 CCP donors’ self-reports revealed that the mean mean  SD S/CO of 3.74  1.79 (p < 0.001) (Table S2). In stratifica-
time from the onset of symptoms to the first CCP collection was tion to age strata, there was no difference in antibody levels between
45.6  14.5 days. genders.
190 IZAK ET AL.

Results of 199 rapid point-of-care test (POC) were compared to by a multi-assay laboratory protocol and supplying it to COVID-19
the anti-N results: 171/199 (85.9%) showed concordant results, with patients in Israel. Over 2300 COVID-19 patients treated until
143 positive and 28 negative by both assays. Disagreement of POC 15 January 2021, as a part of an investigational, multi-institutional
with anti-N CLIA was observed in 28/199 samples. Of them, 27/199 national program.
were positive by the POC assay and negative by the anti-N assay and Patients recovered from COVID-19 were referred by the MOH
1/199 negative by the POC assay and positive by the anti-N assay. or responded to calls in the social media to become CCP donors. To
The positive predictive value (PPV) for POC was 0.84 and the negative facilitate recruitment of eligible individuals, personnel of CCP Call
predictive value (NPV) was 0.97, with sensitivity of 99.3% and specific- Center conducted health-screening interviews. Although only 11.6%
ity of 50.9%. Comparison of anti-S by EUROIMMUN and anti-N by of the calls yielded appointments, of which 72% of the donors showed
CLIA was performed on 139 samples; PPV of EUROIMMUN was 0.96 up, remarkable low percentage of donors (4.5%) were deferred on
and NPV was 0.63 (sensitivity of 92.4% and specificity of 79.2%). site, compared to a deferral rate of 11% in our regular blood drives.
NAb activity was determined in 53 CCP units from 29 donors, as The low percentage of deferrals on site was probably a result of pre-
described previously [19]. The median nAb titre was 1:160 (inter- donation telephone interviews with potential donors to assure compli-
quartile range [IQR] 1:160–1:640, range 1:20–1:2560). NAb titre was ance with requirements, the tactics that were not accepted for regular
<1:160 in eight CCP units (15.1%) and ≥ 1:160 in 84.9%. Taken as a blood donations. The relatively high percentage of CCP units dis-
continuous variable, the IgG anti-S by EUROIMMUN yielded positive carded due to abnormal blood tests (1.08% for CCP vs. 0.42% for
correlation r = 0.69 (p < 0.001) with nAb after logarithmic transforma- blood donations) could be a result of high percentage of first-time
tion of both variables. IgG anti-N by CLIA showed a positive correla- donors among CCP donors comparing to regular blood donors
tion of r = 0.66 (p < 0.001) with nAb, both by Spearman’s rank (36% vs. 20%, respectively).
correlation (Figure3). Almost all our CCP donors were asymptomatic or had mild
COVID-19 disease, as most of potential donors with moderate or
severe disease were not qualified for CCP collection, usually due to
Gradual decrease of antibody level over time persistent symptoms. The addition of pre-donation screening method
enabled to collect CCP only from donors who showed the presence of
Of 4020 apheresis donors, 837 (20.8%) had more than one CCP col- antibodies, thus, saving time and resources of both the donors and
lection; 561 donated twice, 212 donated three times, 44 donated four the blood services.
times, 16 donated five donations and 4 had six donations (Table S1). Based on published data on efficacy of antibody-rich CCP and
As we aimed to deliver only antibody-rich CCP to COVID-19 patients, according to recently approved FDA policy [16], we used a statistical
only donors with initial anti-N IgG level of S/CO ≥ 4.0 were invited prediction model to optimize our CCP inventory of units with higher
for subsequent donations, and two CCP units were delivered to IgG antibody levels. Since the model predicted that anti-N Ab
patients: one unit with IgG S/CO ≥ 7.0 and one with S/CO ≥ 4.0, pro- decreased at 0.55%/day from the time of the first donation (Tables S3
viding an average S/CO ≥ 4.5. and S4 and Figure 4), only donors with higher antibody levels were re-
The timeframe of Ab decrease was predicted by statistical model scheduled for further plasma donations, keeping short periods
as 0.55%/day (Table S3, Figure4) and could be calculated relatively to between collections (2 weeks). All donated units were retested for
the level at the first collection, for example, if initial S/CO was 10.0, anti-N antibodies in subsequent donations.
the decrease to S/CO of 4.0 will take 92.65 days (Table S4). Our study has few limitations. One is the fact that 88.4% of indi-
viduals who were referred to the MDANBS were found to be non-
eligible due to health issues, parity in women or refused to donate.
Blood groups in CCP donors Consistently low compliance rate of COVID-19 convalescents to
donate plasma was recently described by our colleagues [34]. We
The prevalence of ABO/Rh blood groups in CCP donors was similar to need to further study this phenomenon. Secondly, we used anti-N Ab
that of blood donors in the general Israeli population, according to the as a surrogate marker of anti-viral neutralizing activity; however,
MDANBS database from 2019. Higher prevalence of blood group A follow-up of repeated nAb or anti-S IgG was not available during the
and a lower prevalence of group O among COVID-19 patients study period. Thirdly, no anti-N IgGAb were found in high proportion
reported previously [32, 33] was not seen in CCP donors in Israel; of our CCP donors, in agreement with published data on lower Ab
however, higher percentage of AB group in convalescents (9%) com- levels anticipated in cohort of asymptomatic individuals [35] and in
paring to blood donors (8%) was statistically significant (p = 0.002). patients with mild symptoms [36, 37], with rapid decline of anti-N and
nAb[38]. Anti-S antibodies were undetectable in 17% of our donors
by EUROIMMUN test and in 14% by the rapid POC lateral flow test.
DISCUSSION We are awaiting for follow-up results of nAb tests in a larger group of
CCP donors that will be performed by the Israeli Institute for Biologi-
This report describes the steps taken to rapidly establish a program cal Research; they will be helpful to understand better the relationship
for the recruitment of volunteer CCP donors, qualifying their plasma between anti-N and nAb.
CCP COLLECTION, QUALIFICATION AND DISTRIBUTION IN ISRAEL 191

Another limitation was unknown period between recovery and 3. Stoll HF. Value of convalescent blood and serum in treatment of
donation in many donors. It is clear that early CCP donation soon after influenza pneumonia. JAMA. 1919;73:478–82.
4. Luke TC, Kilbane EM, Jackson JL, Hoffman SL. Meta-analysis: conva-
recovery is associated with higher antibody levels, but unfortunately,
lescent blood products for Spanish influenza pneumonia: a future
the donors’ information provided from the sites to MDABS was H5N1 treatment? Ann Intern Med. 2006;145:599–609.
uneven and sometimes incomplete. 5. Shen C, Wang Z, Zhao F, Yang Y, Li J, Yuan J, et al. Treatment of
Currently, all expectations are concentrated on the results of vacci- 5 critically ill patients with COVID19 with convalescent plasma.
JAMA. 2020;323:1582–9.
nation programs. Although over 3 million individuals were already vacci-
6. Zhang B, Liu S, Tan T, Huang W, Dong Y, Chen L, et al. Treatment
nated in Israel by the Pfizer-BioNTech vaccine [39], urgent requests for with convalescent plasma for critically ill patients with SARS-CoV-2
CCP units for new COVID-19 patients are being added daily. Gaps in infection. Chest. 2020;158:e9-e13.
knowledge still exist for the timeframe for SARS-CoV-2 antibody forma- 7. Duan K, Liu B, Li C, Zhang H, Yu T, Qu J, et al. Effectiveness of con-
valescent plasma therapy in severe COVID19 patients. Proc Natl
tion and decline, the relationship between antibody levels and the sever-
Acad Sci U S A. 2020;117:9490–6.
ity of COVID-19 disease and the protective effect of antibodies against 8. Salazar E, Perez KK, Ashraf M, Chen J, Castillo B, Christensen PA,
re-infection with the SARS-CoV-2 wild type or variants [40, 41]. et al. Treatment of COVID19 patients with convalescent plasma.
In this ongoing project, we focussed on rapid creation of sufficient Am J Pathol. 2020;190:1680–90.
9. Joyner MJ, Senefeld JW, Klassen SA, Mills JR, Johnson PW,
CCP inventory, by collection of CCP with higher antibody levels, as we
Theel ES, et al. Effect of Convalescent Plasma on Mortality among
believe that a better outcome for COVID-19 patients can be achieved by
Hospitalized Patients with COVID19: Initial Three-Month Experience
providing CCP transfusion early during the course of the disease. This (published online ahead of print on 12 August 2020). medRxiv 2020.
challenging task was achievable and maybe less complicated in Israel, https://doi.org/10.1101/2020.08.12.20169359.
where the country’s blood collection processing and supply is concen- 10. Liu STH, Lin HM, Baine I, Wajnberg A, Gumprecht JP,
Rahman F, et al. Convalescent plasma treatment of severe
trated in a centralized Blood Services Establishment. Building a CCP inven-
COVID-19: a propensity score-matched control study. Nat Med.
tory was achieved by being part of an integral, multi-disciplinary program 2020;26:1708–13.
involving community stakeholders (hospitals), governmental regulators 11. Salazar E, Christensen PA, Graviss EA, Nguyen DT, Castillo B, Chen J,
(MOH) and research laboratories, all supporting the national program for et al. Significantly decreased mortality in a large cohort of coronavi-
rus disease 2019 (COVID-19) patients transfused early with conva-
donor recruitment and laboratory qualification of CCP. We encourage
lescent plasma containing high-titer anti-severe acute respiratory
additional programs to share their experiences to support a timely deter- syndrome coronavirus 2 (SARS-CoV-2) spike protein IgG. Am J
mination of best practices for CCP programs during the current pandemic. Pathol. 2021;191:90–107.
12. Libster R, Pérez Marc G, Wappner S, Coviello S, Bianchi A, Braem V,
et al. Early high-titer plasma therapy to prevent severe covid-19 in
ACKNOWLEDGEMEN TS
older adults. N Engl J Med. 2021;384:610–8.
We thank all the volunteer donors and MDANBS employees for con- 13. Joyner MJ, Carter RE, Senefeld JW, Klassen SA, Mills JR,
tributionto the CCP collection and distribution. We thank S. Gerald Johnson PW, et al. Convalescent plasma antibody levels and the risk
Sandler for review of the manuscript. We thank Nir Paran and Tomer of death from Covid-19 (published online ahead of print on
13 January 2021). N Engl J Med. 2021;384:1015–27.
Israely for performance of neutralizing antibody assay. Special thanks
14. Simonovich VA, Burgos Pratx LD, Scibona P, Beruto MV,
to Lena Novack for statistical analysis and to the medical staff of hos- Vallone MG, Vazquez C, et al. A randomized trial of convalescent
pitals in Israel for treating thousands of COVID-19 patients. plasma in COVID-19 severe pneumonia. N Engl J Med. 2021;384:
M.I. wrote the first draft of manuscript, S.B.-Z. analysed the data, 619–29.
15. Agarwal A, Mukherjee A, Kumar G, Chatterjee P, Bhatnagar T,
V.G., E.S. and D.C. coordinated the testing, R.G. performed statistical
Malhotra P. Convalescent plasma in the management of moderate
analysis, J.C., Y.M., A.B., B.L. and O.Z. contributed to the design of
covid-19 in adults in India: open-label phase II multicenter random-
project and reviewed the final manuscript, E.S. devised the project, ized controlled trial (PLACID trial). BMJ. 2020;371:m3939.
planned experiments and wrote the final manuscript. 16. FDA: revision of the Emergency Use Authorization (EUA) for
COVID-19 convalescent plasma. https://www.fda.gov/news-events/
fda-brief/fda-brief-fda-updates-emergency-use-authorization-covid-
CONF LICT OF IN TE RE ST
19-convalescent-plasma-reflect-new-data. Accessed 4 February
No conflict of interest to disclose. 2021.
17. Eckhardt CM, Cummings MJ, Rajagopalan KN, Borden S, Bitan ZC,
Wolf A, et al. Evaluating the efficacy and safety of human anti-SARS-
ORCID CoV-2 convalescent plasma in severely ill adults with COVID-19: a
Marina Izak https://orcid.org/0000-0002-4297-6373 structured summary of a study protocol for a randomized controlled
trial. Trials. 2020;21:499.
18. Wouters E, Steenhuis M, Schrezenmeier H, Tiberghien P, Harvala H,
Feys HB, et al. Evaluation of SARS-CoV-2 antibody titers and
RE FE R ENC E S potency for convalescent plasma donation: a brief commentary (pub-
1. Infectious Disease Society of America Guidelines on the Treatment lished online ahead of print on 23 December 2020). Vox Sang. 2020;
and Management of Patients with COVID19. 2020. https://www. 116(5):493–6.
idsociety.org/COVID19guidelines. Accessed 20 November 2020. 19. Maor Y, Cohen D, Paran N, Israely T, Ezra V, Axelrod O, et al. Com-
2. Ross CW, Hund EJ. Treatment of pneumonic disturbance complicat- passionate use of convalescent plasma for treatment of moderate
ing influenza. JAMA. 1919;72:640–5. and severe pneumonia in COVID19 patients and association with
192 IZAK ET AL.

IgG antibody levels in donated plasma. EClinicalMedicine. 2020;26: 33. Wu BB, Gu DZ, Yu JN, Yang J, Shen WQ. Association between ABO
100525. blood groups and COVID19 infection, severity and demise: a system-
20. FDA: Recommendations for investigational COVID-19 convalescent atic review and meta-analysis. Infect Genet Evol. 2020;84:104485.
plasma. https://www.fda.gov/media/136798. Accessed 15 January 34. Wang HE, Ostrosky-Zeicher L, Katz J, Wanger A, Bai Y, Sridhar S,
2021. et al. Screening donors for COVID-19 convalescent plasma (publi-
21. MOH. https://www.health.gov.il/Subjects/disease/corona/Documents/ shed online ahead of print on 24 December 2020). Transfusion.
bz-259818720.pdf. Accessed 9 July 2020. 2020;61:1047–52.
22. Wendel S, Kutner JM, Machado R, Font~ao-Wendel R, Bub C, 35. Long QX, Tang XJ, Shi QL, Li Q, Deng HJ, Yuan J, et al. Clinical and
Fachini R, et al. Screening for SARS-CoV-2 antibodies in convales- immunological assessment of asymptomatic SARS-CoV-2 infections.
cent plasma in Brazil: preliminary lessons from a voluntary Nat Med. 2020;26:1200–4.
convalescent donor program. Transfusion. 2020;60:2938–51. 36. Röltgen K, Wirz OF, Stevens BA, Powell AE, Hogan CA, Najeeb J,
23. Shimoni Z, Niven MJ, Pitlick S, Bulvik S. Treatment of West Nile virus et al. SARS-CoV-2 antibody responses correlate with resolution of
encephalitis with intravenous immunoglobulin. Emerg Infect Dis. RNAemia but are short-lived in patients with mild illness (published
2001;7:759. online ahead of print on 8 August 2020). medRxiv. 2020. https://doi.
24. Flower B, Brown JC, Simmons B, Moshe M, Frise R, Penn R, et al. org/10.1101/2020.08.15.20175794.
Clinical and laboratory evaluation of SARS-CoV-2 lateral flow assays 37. Lynch KL, Whitman JD, Lacanienta NP, Beckerdite EW, Kastner SA,
for use in a national COVID-19 seroprevalence survey. Thorax. Shy BR, et al. Magnitude and kinetics of anti-SARS-CoV-2 antibody
2020;75:1082–8. responses and their relationship to disease severity (published online
25. Andrey DO, Cohen P, Meyer B, Torriani G, Yerly S, Mazza L, et al. ahead of print on 14 July 2020). Clin Infect Dis. 2020;72:301–8.
Head-to-head accuracy comparison of three commercial COVID19 38. Brochot E, Demey B, Touzé A, Belouzard S, Dubuisson J, Schmit JL,
IgM/IgG serology rapid tests. J Clin Med. 2020;9:2369. et al. Anti-spike, anti-nucleocapsid and neutralizing antibodies in
26. Luchsinger LL, Ransegnola BP, Jin DK, Muecksch F, Weisblum Y, SARS-CoV-2 inpatients and asymptomatic individuals. Front
Bao W, et al. Serological assays estimate highly variable SARS-CoV-2 Microbiol. 2020;11:584251.
neutralizing antibody activity in recovered COVID-19 patients. J Clin 39. Polack FP, Thomas SJ, Kitchin N, Absalon J, Gurtman A, Lockhart S,
Microbiol. 2020;58:e02005-20. et al. Safety and efficacy of the BNT162b2 mRNA Covid-19 vaccine.
27. Wajnberg A, Amanat F, Firpo A, Altman DR, Bailey MJ, Mansour M, N Engl J Med. 2020;383:2603–15.
et al. Robust neutralizing antibodies to SARS-CoV-2 infection persist 40. Mo H, Zeng G, Ren X, Li H, Ke C, Tan Y, et al. Longitudinal profile of
for months. Science. 2020;370:1227–30. antibodies against SARS-coronavirus in SARS patients and their clini-
28. Theel ES, Harring J, Hilgart H, Granger D. Performance characteristics cal significance. Respirology. 2006;11:49–53.
of four high-throughput immunoassays for detection of IgG antibodies 41. Kellam P, Barclay W. The dynamics of humoral immune responses
against SARS-CoV-2. J Clin Microbiol. 2020;58:e01243-20. following SARS-CoV-2 infection and the potential for re-infection.
29. Manalac J, Yee J, Calayag K, Nguyen L, Patel PM, Zhou D, et al. J Gen Virol. 2020;101:791–7.
Evaluation of Abbott anti-SARS-CoV-2 CMIA IgG and Euroimmun
ELISA IgG/IgA assays in a clinical lab. Clin Chim Acta. 2020;510:
SUPPORTING INFORMATION
687–90.
30. Meschi S, Colavita F, Bordi L, Matusali G, Lapa D, Amendola A, et al. Additional supporting information may be found online in the
Performance evaluation of Abbott ARCHITECT SARS-CoV-2 IgG Supporting Information section at the end of this article.
immunoassay in comparison with indirect immunofluorescence and
virus microneutralization test. J Clin Virol. 2020;129:104539.
31. BELTEST-IT COV-2 Rapid Test (PharmAct) https://pharmact.de/en/ How to cite this article: Izak M, Gendelman V, Bransburg-
beltest-it-cov-2. Accessed 10 June 2020. Zabary S, et al. Qualifying coronavirus disease 2019
32. Ellinghaus D, Degenhardt F, Bujanda L, Buti M, Albillos A, Invernizzi P,
convalescent plasma donors in Israel. Vox Sang. 2022;117:
et al. Genomewide association study of severe COVID19 with respira-
tory failure. N Engl J Med. 2020;383:1522–34. 185–192.
Received: 8 March 2021 Revised: 20 June 2021 Accepted: 20 June 2021

DOI: 10.1111/vox.13177

ORIGINAL ARTICLE

DEHT is a suitable plasticizer option for phthalate-free storage

of irradiated red blood cells

Linda Larsson1,2 | Sara Ohlsson2 | Julia Derving2 | Beatrice Diedrich1,2 |

Per Sandgren1,2 | Stella Larsson2 | Michael Uhlin1,2

1
Department of Clinical Science, Intervention
and Technology (CLINTEC), Karolinska Abstract
Institutet, Stockholm, Sweden
Background and Objectives: Due to increasing concerns about possible endocrine-
2
Department of Clinical Immunology and
Transfusion Medicine, Karolinska University
disrupting properties, the use of the plasticizer di(2-ethylhexyl) phthalate (DEHP) will
Hospital, Stockholm, Sweden be banned in future blood storage. Di(2-ethylhexyl) terephthalate (DEHT) provides
sufficient red blood cell (RBC) quality during conventional blood bank storage. It is
Correspondence
Linda Larsson, Department of Clinical Science, important that a new plasticizer also maintains acceptable quality during exposure to
Intervention and Technology (CLINTEC), ANA
high cell stress, such as irradiation, which is commonly used to prevent graft-versus-
Futura 8, Karolinska Institutet, SE-171
77 Stockholm, Sweden. host disease.
Email: [email protected]
Materials and Methods: A total of 59 RBC units were collected and processed in
polyvinyl chloride (PVC)-DEHT or PVC-DEHP blood bags combined with either
saline-adenine-glucose-mannitol (SAGM) or phosphate-adenine-glucose-guanosine-
saline-mannitol (PAGGSM) additive solution. All units were X-ray irradiated on day 2
post-collection. Sampling for assessment of parameters of storage lesion was per-
formed on day 2 pre-irradiation and day 14 and 28 post-irradiation.
Results: Though irradiation increased cell stress, DEHT/PAGGSM and current com-
mon European preference DEHP/SAGM were equally affected up to 14 days post-
irradiation for all measured parameters. At day 28, haemolysis and microvesicle count
were slightly increased in DEHT, whereas extracellular potassium ions, glucose, lac-
tate, pH, mean corpuscular volume and microvesicle phosphatidylserine remained
unaffected by plasticizer choice throughout storage. No individual unit exceeded
0.8% haemolysis, not even in DEHT/SAGM, the combination overall most affected
by irradiation. Of the four combinations, membrane stability was least impacted in
DEHP/PAGGSM.
Conclusion: We demonstrate that DEHT is a suitable plasticizer for storage of RBCs
after X-ray irradiation cell stress. This strengthens the option of DEHT as a viable
non-phthalate substitute for DEHP.

KEYWORDS
DEHP, DEHT, irradiation, phthalate, plasticizer, red blood cells

This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium,
provided the original work is properly cited.
© 2021 The Authors. Vox Sanguinis published by John Wiley & Sons Ltd on behalf of International Society of Blood Transfusion.

Vox Sanguinis. 2022;117:193–200. wileyonlinelibrary.com/journal/vox 193

194 LARSSON ET AL.

I N T R O D U CT I O N It is important that a new plasticizer is versatile enough to main-

tain acceptable RBC quality not only at ideal storage conditions, but
The normal lifespan of a red blood cell (RBC) is around 120 days in cir- also during exposure to common blood bank stressors. Therefore, we
culation, but maximum storage time of a refrigerated RBC concentrate wanted to subject RBCs stored in PVC-DEHT bags, combined with
(RCC) is usually limited to five to seven weeks due to a degenerative SAGM or PAGGSM, respectively, to X-ray irradiation and compare
process called RBC storage lesion. This is a descriptive name for the the results to corresponding DEHP storage. This study sheds light on
large number of metabolic, oxidative and morphological changes that the characteristics of DEHT for RBC storage during the condition of
RBCs undergo during blood bank storage conditions, which ultimately irradiation cell stress.
leads to cell rupture: haemolysis [1]. When RBCs lyse, haemoglobin is
released. Haemoglobin is a strong inducer of reactive oxygen species
(ROS) that, just like RBC microvesicles (RMV), are linked to pro- M A T E R I A L S A N D M ET H O D S
inflammatory and pro-thrombotic processes [1, 2]. Therefore, it is crucial
to keep the number of haemolysed cells in an RCC as low as possible up Blood collection and component processing
until the point of transfusion. The upper limit for haemolysis has been
set to 0.8% of red cell mass in the Council of Europe member states [3]. A total of 59 RCCs were produced through collection and separation
The haemolysis generation is affected by exposure to different of whole blood (WB; 450 ml  10% in 63 ml citrate-phosphate-dex-
manufacturing and storage stressors [4–6]. One of the most extreme trose) from regular, voluntary, non-remunerated blood donors, follow-
stressors is irradiation. Irradiation, usually through X-ray or gamma ing Karolinska standard operating procedures [23, 25]. Briefly, these
methods [7], is an important measure to inhibit proliferation of resid- include separation with Macospin (3130  g, 11 min) and MacoPress
ual white blood cells (WBCs) and thereby prevent transfusion- Smart Revo (both Macopharma, Mouvaux, France) on the collection
associated graft-versus-host disease [3, 8, 9]. However, irradiation day (no overnight WB hold). Blood bag systems corresponding to the
stimulates the creation of ROS which, in turn, damages the RBC mem- commercially available bottom-and-top NPT reference (Macopharma),
brane through induction of lipid peroxidation and oxidation of including identical needle and leucoreduction filters, were man-
+ +
membrane-bound proteins, including inhibition of Na /K ATPase ufactured completely of PVC-DEHT or PVC-DEHP, respectively. These
[10–12]. This premature loss of membrane integrity increases the rate bag systems contained either 100 ml SAGM or PAGGSM, respectively,
of both haemolysis and RMV generation [6, 10–14] as well as the for RBC resuspension. This set-up allowed four study arms to be cre-
accumulation rate of extracellular potassium ions (K+), associated with ated: DEHT/SAGM, DEHT/PAGGSM, DEHP/SAGM and DEHP/
hyperkalaemia-linked cardiac arrest [10, 15, 16]. Consequently, the PAGGSM (blood type A; DEHP/SAGM n = 14, all other arms n = 15).
shelf-life of irradiated RCCs is frequently reduced between 24 h and The donors were chosen randomly for each arm; no pool-and-split
2 weeks. Among the more stringent practises is the ‘14 + 14 days’ strategy was applied. The RCCs were cold stored (2–6 C) within 8 h of
rule (maximum 14 days old RCCs at time of irradiation, maximum donation. Tube sealers from the Qseal range (Conroy Medical AB,
14 days storage post-irradiation), which is advocated in for instance Upplands Väsby, Sweden) and sterile docking device TSCDII
Sweden and the UK. [3, 17, 18]. (TerumoBCT, Lakewood, CO) were used for both the DEHT and DEHP
The use of polyvinyl chloride (PVC) storage containers plasticized material.
with di(2-ethylhexyl) phthalate (DEHP) has, in combination with suit- On day (d) 2 post-collection, the RCCs were X-ray irradiated (can-
able additive solution (AS), long been a cornerstone to maintain low ister center dose 37.85 Gy, maximum 38.53 Gy, minimum 26.42 Gy;
haemolysis levels throughout storage. DEHP leaches from the PVC Raycell Mk2, Best Theratronics, Ottawa, Canada) following guideline
matrix and incorporates into the RBC membrane phospholipid bilayer, specifications [3]. The RCCs were sampled on d2 pre-irradiation for
thereby stabilizing it and prolonging the time to lysis [19, 20]. Due to baseline values, then on d14 and d28 post-irradiation. Sampling was
increasing concerns about possible endocrine-disrupting properties, performed through sterile connection of 40 ml sampling bags (also
the regulatory restrictions for medical devices have recently been made of PVC-DEHT or PVC-DEHP, respectively) to the RCCs, very
updated with the aim to abolish the use of DEHP in blood storage gentle mixing during sampling, followed by immediate analysis.
[21]. Therefore, it is urgent to remove DEHP without compromising The RBCs were exposed to the same plasticizer (DEHT or DEHP,
the blood component quality [22], a conundrum that has delayed the respectively) at all time points, from collection until study end. Fur-
replacement process for a long time. thermore, the RCCs of different plasticizers were handled and stored
Although no other plasticizer has, so far, shown identical RBC separately to avoid accidental cross-contamination, as were the bag
storage quality to DEHP, there are a few promising candidates. In a systems during the manufacturing process.
previous study [23], we demonstrated that a structural isomer to
DEHP, di(2-ethylhexyl) terephthalate (DEHT), provides adequate RBC
quality during seven weeks of storage in both saline-adenine-glucose- Assessment of RBC quality
mannitol (SAGM; primary AS in Europe) and phosphate-adenine-glu-
cose-guanosine-saline-mannitol (PAGGSM). Satisfying quality of The impact of X-ray irradiation on the different plasticizer/AS combi-
DEHT combined with AS-1 has also been confirmed [24]. nations was determined through collective assessment of membrane
STORAGE OF IRRADIATED RBCs IN PLASTICIZER DEHT 195

effects and metabolic effects during storage. The RCCs were analysed irradiation in all study arms, independent of plasticizer/AS combina-
for haemoglobin (HbRCC; g/l), haematocrit (Hct) and mean corpuscular tion. DEHT/PAGGSM stayed similar to both DEHP arms throughout
volume (MCV; fl) through Swelab Alfa Plus Basic hematology analyzer storage, whereas storage in DEHT/SAGM resulted in slightly higher
(Boule Diagnostics AB, Spånga, Sweden). After centrifugation of concentration than both PAGGSM arms (d14: p < 0.001, d28:
the samples twice (1450  g, 10 min, 20 C), free supernatant p < 0.05; Figure 1b, Table 1). The two SAGM arms (DEHT/SAGM
haemoglobin (Hbsupernatant; g/l) was measured with HemoCue plasma/ vs. DEHP/SAGM) remained similar throughout storage. The difference
low haemoglobin photometer (Radiometer Medical ApS). Haemolysis in concentration between the study arms stayed within 2.4 mmol/l
(%) was calculated as (100-Hct)  Hbsupernatant (g/l) / HbRCC (g/l). Extra- (min-max range of the d28 means: 57.2–59.6 mmol/l).
cellular K+ (mmol/l), pH, glucose (mmol/l) and lactate (mmol/l) were Increased RMV count signifies loss of membrane integrity. The
assessed with ABL 800 Flex blood gas analyser (Radiometer Medical RMV count was similar for all study arms at d14 post-irradiation.
ApS, Brønshøj, Denmark). Measurements of RMV count and RMV However, at d28, DEHT/SAGM had the highest observed count,
phosphatidylserine externalization (Annexin V positive RMVs) were followed by DEHT/PAGGSM (Figure 1c, Table 1). Although the overall
performed by flow cytometry (CytoFLEX, Beckman Coulter, Brea, CA), percentage of RMVs exposing the apoptosis marker phos-
protocols identical to previous study [23]. phatidylserine increased after irradiation (p < 0.05), there was no dif-
In addition, residual WBCs were counted with cell counter ference between the study arms at the separate time points (Table 1).
ADAM-rWBC (NanoEnTek, Seoul, South Korea) at storage start. All Glucose, lactate and pH are markers of metabolic capacity. The
donors were screened for required serological markers [3], and glucose concentration and pH were both initially higher in SAGM
all RCCs were bacteriologically tested at storage end (Karolinska Uni- storage than in PAGGSM (glucose: p < 0.05, pH: p < 0.001), but the
versity Laboratory, Clinical Chemistry and Clinical Microbiology differences ceased with storage time. Correspondingly, the lactate
departments). generation was faster and indicated an overall higher metabolism dur-
ing SAGM storage. MCV is sensitive to osmolality. A distinct increase
in MCV was detectable from d14 onwards in both SAGM arms,
Statistical analysis whereas in PAGGSM, MCV decreased compared to d2 (p < 0.001;
Figures 2a–c, Table 1). No distinct differences could be related to
After verifying a Gaussian distribution (D’Agostino-Pearson normality plasticizer for glucose, lactate, pH or MCV.
test), mean  standard deviation (SD) was computed (GraphPad Prism
v.8.2 for Windows; GraphPad Software Inc., La Jolla, CA). To test sta-
tistical significance between the study arms, two-way analysis of vari- DI SCU SSION
ance (ANOVA) was applied with Holm-Sidak’s correction as post-hoc
test, while differences between time points within a single study arm This study demonstrates that DEHT is a suitable plasticizer for storage
were compared using repeated measures one-way ANOVA. of RBCs after applying severe cell stress in the form of X-ray
irradiation.
The observed haemolysis in irradiated RBCs stored in DEHT in
Ethics statement combination with either SAGM or PAGGSM (Figure 1(a)) was in accor-
dance with the previously shown pattern of non-irradiated RBC units
As all used material was fully anonymised, the Stockholm Regional stored in DEHT [23], however, expectedly amplified. The non-
Ethical Review Board did not consider an ethical application irradiated d49 haemolysis values were exceeded somewhere between
necessary. d14 and d28 post-irradiation. This similarity suggests that irradiation
itself does not interfere with the DEHP removal and/or DEHT addi-
tion in other ways than the accelerated speed in haemolysis genera-
RESULTS tion observed also after irradiation of conventional DEHP-stored
RBCs [6, 11, 13, 14]. Even with DEHP, the possibility of individual
All RBC units fulfilled European and national quality criteria in HbRCC, units exceeding 0.8% haemolysis during post-irradiation storage
Hct and WBC count [3, 17]. Screenings for bacteria and infectious remains a health risk, in particular to the more fragile patient catego-
markers were negative. ries such as, for instance, paediatric patients. The literature for irradia-
Haemolysis was similar for DEHT/PAGGSM and DEHP/SAGM at tion of non-DEHP storage containers is scarce, likely because removal
d14 post-irradiation. However, during the second half of storage, it of DEHP has been linked to critical haemolysis levels in non-irradiated
increased faster in DEHT, especially in combination with SAGM, and units [26]. This study indicates that PVC-DEHT containers provide
at d28, all four study arms differed (p < 0.001; Figure 1a, Table 1). The sufficient RBC storage environment with existing commercial ASs for
highest individual unit (DEHT/SAGM) reached 0.65% haemolysis, at least 14 days, but even up to 28 days post-irradiation, when irradia-
meaning all units remained safely below 0.8% at storage end. tion is carried out early after donation.
+
Elevated extracellular K signifies membrane leakage and/or The extracellular K+ results were also encouraging. The minor
+
suppressed metabolism. Extracellular K increased sharply after concentration differences between the four plasticizer/AS
196 LARSSON ET AL.

F I G U R E 1 (a) RBCs stored in DEHT had acceptable haemolysis throughout storage, especially when combined with PAGGSM additive
solution. No units exceeded 0.8%. The bars show mean with a scatter dot plot representing the individual values. (b) Irradiation implicated sharp
increases in extracellular K+ concentration in all study arms. DEHT/SAGM was marginally higher than both PAGGSM arms post-irradiation but
similar (ns) to DEHP/SAGM. (c) RBC microvesicle count was similar in both plasticizers at d14 post irradiation but higher in DEHT at d28.
PAGGSM detained the microvesicle generation compared to SAGM. Values in (b) and (c) are displayed as mean  standard deviation. Significance
levels: *p < 0.05, no significance (ns). Radiation symbols represent post-irradiation. DEHT/SAGM, DEHT/PAGGSM, DEHP/PAGGSM: n = 15,
DEHP/SAGM: n = 14. DEHT, di(2-ethylhexyl) terephthalate; DEHP, di(2-ethylhexyl) phthalate; SAGM, saline-adenine-glucose-mannitol;
PAGGSM, phosphate-adenine-glucose-guanosine-saline-mannitol; RBC, red blood cell

combinations were likely not of clinical significance and suggest that relationship was demonstrated between RMV count and both plas-
the irradiation itself had substantially more impact than the possible ticizer and AS, the latter likely related to the ATP dependency of
theoretical influence of plasticizer or AS (Figure 1b). the RMV generation; the lower RMV count in PAGGSM-stored
RMVs are shed from the RBC membrane as part of the stor- units suggests a slower reduction of ATP. Though ATP was not
age lesion process, as ROS increases and antioxidants and ATP are analysed in this study, the associated markers pH, lactate and glu-
depleted [2]. In this study, the generation of RMVs had similar but cose all trended towards a higher metabolic rate in SAGM storage
earlier onset than in non-irradiated units [23]. This is likely attrib- (Figure 2a,b).
uted to increased ROS generation caused by irradiation, both Phosphatidylserine is a phospholipid normally located in the inner
directly and through the positive feedback loop provided by the leaflet of the RBC bilayer membrane. Membrane proteins of ATPase
haemolysis-induced increase of free haemoglobin [11, 12]. A character control the asymmetry between the inner and outer leaflet,
TABLE 1 Red blood cell analysis results pre-irradiation (day 2) and post-irradiation (day 14 and 28).
Day 2 (pre irradiation) Day 14 (post irradiation) Day 28 (post irradiation)
STORAGE OF IRRADIATED RBCs IN PLASTICIZER DEHT

Analysis parameter DEHT/SAGM DEHT/PAGGSM DEHP/SAGM DEHP/PAGGSM DEHT/SAGM DEHT/PAGGSM DEHP/SAGM DEHP/PAGGSM DEHT/SAGM DEHT/PAGGSM DEHP/SAGM DEHP/PAGGSM

Haemolysis (%) 0.01  0.01 0.03  0.02 <0.01 0.01  0.01 0.21  0.04a,b,c 0.15  0.03a,e 0.12  0.03b,f 0.07  0.01c,e,f 0.49  0.09a,b,c 0.35  0.07a,d,e 0.28  0.04b,d,f 0.18  0.03c,e,f
RMV, count (103/μl 0.05  0.03 0.03  0.02 0.04  0.02 0.04  0.01 0.19  0.09 0.17  0.07 0.13  0.09 0.17  0.07 0.60  0.37a,b,c 0.40  0.24a,d,e 0.28  0.15b,d 0.21  0.08c,e
supernatant)
RMV, externalized 57.0  18.8c 51.3  25.1e 48.3  22.4f 32.4  14.7c,e,f 56.7  11.8 47.0  10.9 53.8  13.8 47.1  10.9 71.7  7.6 63.1  10.7 66.4  12.7 58.3  9.7
phosphatidylserine (%)
Extracellular K+ (mmol/l) 6.6  1.1 5.6  0.5 6.0  1.0 5.9  0.5 52.6  2.8a,c 49.1  3.1a 50.6  2.1 49.1  3.0c 59.6  2.5a,c 57.2  2.6a 58.2  2.0 57.4  2.5c
pH 37 C 6.950  0.017a,b,c 6.811  0.032a,d 6.916  0.028b,d,f 6.804  0.031c,f 6.606  0.019a,c 6.569  0.028a 6.580  0.032 6.573  0.030c 6.432  0.031 6.403  0.044 6.414  0.041 6.415  0.034
Glucose (mmol/l) 29.1  1.5a,c 27.7  1.7a,d 29.0  0.8d,f 27.0  1.3c,f 22.0  0.8 21.3  1.2 21.8  1.5 21.3  1.1 18.0  1.5 17.1  1.9d 18.7  1.9d 17.7  1.4
Lactate (mmol/l) 5.2  0.8 4.3  0.4 5.3  0.7 4.5  0.5 19.5  1.9c 17.4  1.6 18.7  1.9 16.8  1.7c 26.4  3.3c 24.5  3.9 24.9  3.9 22.9  3.0c
MCV (fl) 93.3  3.0 92.6  5.5 94.0  3.3 93.4  4.2 100.0  3.6a,c 89.7  5.0a,d 99.3  3.3d,f 89.9  3.7c,f 105.0  3.8a,c 90.9  4.8a,d 103.0  3.5d,f 90.4  3.5c,f

Note: Data are presented as mean  standard deviation. DEHT/SAGM, DEHT/PAGGSM, DEHP/PAGGSM: n = 15, DEHP/SAGM: n = 14. Significant differences (p < 0.05) are shown by letters a-f.
Abbreviations: DEHT, di(2-ethylhexyl) terephthalate; DEHP, di(2-ethylhexyl) phthalate; SAGM, saline-adenine-glucose-mannitol; PAGGSM, phosphate-adenine-glucose-guanosine-saline-mannitol; RMV, red blood cell microvesicles; MCV, mean corpuscular volume.
a
DEHT/SAGM versus DEHT/PAGGSM.
b
DEHT/SAGM versus DEHP/SAGM.
c
DEHT/SAGM versus DEHP/PAGGSM.
d
DEHT/PAGGSM versus DEHP/SAGM.
e
DEHT/PAGGSM versus DEHP/PAGGSM.
f
DEHP/SAGM versus DEHP/PAGGSM.
197
198 LARSSON ET AL.

F I G U R E 2 (a) pH, (b) lactate and (c) MCV were unaffected by plasticizers. Discernible differences could be linked to additive solution
composition, which influence (a) decreased or (b, c) increased with time. Values are displayed as mean  standard deviation. Significance levels:
*p < 0.05, no significance (ns). Radiation symbols represent post-irradiation. DEHT/SAGM, DEHT/PAGGSM, DEHP/PAGGSM: n = 15, DEHP/
SAGM: n = 14. DEHT, di(2-ethylhexyl) terephthalate; DEHP, di(2-ethylhexyl) phthalate; SAGM, saline-adenine-glucose-mannitol; PAGGSM,
phosphate-adenine-glucose-guanosine-saline-mannitol; MCV, mean corpuscular volume

and when ATP levels decrease, there is a progressive translocation of Furthermore, it was not possible to directly measure ATP. Instead,
phosphatidylserine to the outer leaflet [27], where it plays a signalling indirect markers such as pH, glucose, lactate and K+ had to be used
role for phagocytosis [28]. The literature is ambiguous as to whether for conclusions about the metabolism. Yet another limitation is the
the percentage of phosphatidylserine positive RMVs from non-irradi- impact of donor variability as, for logistical reasons, no pool-and-split
ated RBCs increases during storage or not; several studies, our own strategy was applied during processing. This is a weakness when com-
previous study included, seem to suggest that it does not [23, 29, 30]. paring different study arms; however, it is simultaneously an impor-
However, as irradiation induces increased oxidizing damage to tant asset for demonstrating that despite donor variation, no
membrane-bound proteins [11, 12], it may be hypothesized that the individual units exceeded the limit for haemolysis, independent of the
observed increase in the frequency of RMVs exposing phos- plasticizer or AS they were stored in (Figure 1a).
phatidylserine may be a consequence of irradiation damage. Possibly, Numerous processing conditions affect the RBC storage lesion,
irradiation targets ATPase the same way as for extracellular K+. Phos- which complicates inter-blood establishment comparisons [31, 32]. A
phatidylserine externalization could not be related to AS the way it strength of this study is that the irradiated RCCs were processed in par-
could for non-irradiated units. The irradiation damage to the ATPase allel with the non-irradiated RCCs of our previous study about DEHT
may have outweighed the potential later effects of AS- and storage storage, using identical processing specifications, disposable materials,
time-dependent metabolic depletion, especially as the RBCs were irra- equipment, sampling procedure, laboratory staff and facilities. This opti-
diated as early as d2 post-collection. mizes the use of that study for comparison and reference [23].
Irradiation appears to primarily impact the storage lesion parame- This study demonstrates an encouraging quality of irradiated RBCs
ters connected to the stability of the RBC membrane. Expectedly, the stored in PVC-DEHT blood bags combined with the European primary
same parameters were affected when the membrane-stabilizing plasti- AS choice SAGM. The membrane integrity loss caused by DEHP
cizer DEHP was exchanged for DEHT. However, there seems to be removal can, however, be counteracted to some extent by instead
no negative physical impact by irradiation on the DEHT plasticizer choosing PAGGSM. DEHT/PAGGSM differed very little, if at all, to the
itself, as parameters that are unaffected by choice of plasticizer in current DEHP/SAGM storage. It is well established that SAGM is a sub-
non-irradiated RBCs remained unaffected also in this study. The optimal storage solution for RBCs and that, quality-wise, more
AS-dependent differences for glucose, lactate, pH and MCV were favourable options exist [26, 33–35]. However, it does remain the first-
well in accordance with previous findings for non-irradiated units [23] hand choice in most European countries, likely due to its long-standing
without any detectable dissimilarities that could be ascribed to poten- regulatory approval, cost, and the extensive evaluation efforts a transi-
tial irradiation damage to the plasticizer itself. Furthermore, sealing, tion would require. Perhaps, when the new medical device regulation is
sterile docking and general bag handling were performed without taken into effect and a new plasticizer will be mandatory, it may be a
remark throughout the study. good opportunity to simultaneously consider a change of AS.
This study has both strengths and limitations. One limitation is It could also be a good time to revise the guidelines for irradiation,
that irradiation was performed on d2, while it is allowed up until d28 as has been previously suggested [13], and opt for a more conserva-
post-collection in many countries [3], and there is evidence con- tive post-irradiation timeframe. An important observation of this
necting irradiation of older RCCs to higher haemolysis rates [6, 13]. study was that neither the haemolysis levels nor the RMV count of
This study gives a good first indication, but a similar follow-up study RBCs stored in DEHT/PAGGSM differed statistically to DEHP/SAGM
with older units would be a necessary complement for further conclu- at 14 days post-irradiation. A maximum of 14 days post-irradiation
sions. Moreover, the irradiation process was performed by X-ray tech- storage time is already routine practice in Sweden and the UK [17,
nique. Although proven to be equal to gamma technique for RBCs 18]. Such a revision would further facilitate the transition to DEHP-
stored in DEHP [7], gamma irradiation of DEHT should be verified. free materials.
STORAGE OF IRRADIATED RBCs IN PLASTICIZER DEHT 199

In summary, this work investigates the additional stress of irradia- 10. Moreira OC, Oliveira VH, Benedicto LB, Nogueira CM, Mignaco JA,
tion on red blood cells stored in PVC bags plasticized with DEHT. A Fontes CFL, et al. Effects of gamma-irradiation on the membrane
ATPases of human erythrocytes from transfusional blood concen-
satisfactory post-irradiation storage capacity of DEHT with either
trates. Ann Hematol. 2008;87:113–9.
SAGM or PAGGSM additive solution is demonstrated. This adds fur- 11. Antosik A, Czubak K, Gajek A, Marczak A, Glowacki R, Borowczyk K,
ther weight to the suitability of DEHT as a non-phthalate replacement et al. Influence of pre-storage irradiation on the oxidative stress
option for DEHP. markers, membrane integrity, size and shape of the cold stored red
blood cells. Transfus Med Hemother. 2015;42:140–8.
12. Anand AJ, Dzik WH, Imam A, Sadrzadeh SM. Radiation-induced red
ACKNOWLEDGEMEN TS cell damage: role of reactive oxygen species. Transfusion. 1997;37:
The authors want to acknowledge Dr Hans Gulliksson for eminent 160–5.
mentorship; Blenda Westeson, Felicia Daggert, Tove Friis- 13. Serrano K, Chen D, Hansen AL, Levin E, Turner TR, Kurach JDR,
et al. The effect of timing of gamma-irradiation on hemolysis and
Christensen, Jenny Eriksson, Alexandra Sabo, Erik Dahl, Andreas
potassium release in leukoreduced red cell concentrates stored in
Eriksson, Sandra Valencia and Hanna-Stina Ahlzén for valuable contri- SAGM. Vox Sang. 2014;106:379–81.
butions at blood collection and component production; Macopharma 14. Qadri SM, Chen D, Schubert P, Devine DV, Sheffield WP. Early
(especially Stefan Reichenberg, Naïma Frizi and former employees γ-irradiation and subsequent storage of red cells in SAG-M additive
solution potentiate energy imbalance, microvesiculation and suscep-
Sonia Chatellier and Åse Tolf) for supporting the research idea and
tibility to stress-induced apoptotic cell death. Vox Sang. 2017;112:
providing material; and the volunteer blood donors whose participa-
480–3.
tion made this study possible. 15. Zhang Z, Shen Y, Li B, Li N. Cardiac arrest triggered by hyperkalemia
L.L. provided the research idea, study design, laboratory work, during red blood cell transfusion: a case report and literature review.
data analysis and manuscript composition; S.O. and J.D. assisted on Transfus Med. 2017;27:384–6.
16. Hall TL, Barnes A, Miller JR, Bethencourt DM, Nestor L. Neonatal
laborations; B.D. managed the blood collection; P.S. and
mortality following transfusion of red cells with high plasma potas-
S.L. provided technical and medical expertise; M.U. supervised the sium levels. Transfusion. 1993;33:606–9.
study. All authors revised the manuscript and contributed with valu- 17. Svensk Förening för Transfusionsmedicin.: Handbok för
able input. Blodcentraler. KITM, 12 Mar 2015. 2015. https://transfusion.se/app/
uploads/2020/04/blodkomponenter-framstallning-och-anvandning.
pdf. Accessed 16 Feb 2021.
CONF LICT OF IN TE RE ST 18. Foukaneli T, Kerr P, Bolton-Maggs PHB, Cardigan R, Coles A,
The authors declare no conflict of interest. Gennery A, et al. Guidelines on the use of irradiated blood compo-
nents. Br J Haematol. 2020;191:704–24.
19. Rock G, Tocchi M, Ganz PR, Tackaberry ES. Incorporation of
ORCID
plasticizer into red cells during storage. Transfusion. 1984;24:
Linda Larsson https://orcid.org/0000-0002-0196-1773 493–8.
Michael Uhlin https://orcid.org/0000-0002-2177-6727 20. Horowitz B, Stryker MH, Waldman AA, Woods KR, Gass JD,
Drago J. Stabilization of red blood cells by the plasticizer,
diethylhexylphthalate. Vox Sang. 1985;48:150–5.
RE FE R ENC E S
21. European Union. Regulation (EU) 2017/745 of the European Parlia-
1. Hess JR. Red cell changes during storage. Transfus Apher Sci. 2010; ment and of the Council of 5 April 2017 on medical devices. Official
43:51–9. J. 2017;L117:96.
2. Rubin O, Canellini G, Delobel J, Lion N, Tissot JD. Red blood cell 22. van der Meer PF, Reesink HW, Panzer S, Wong J, Ismay S, Keller A,
microparticles: clinical relevance. Transfus Med Hemother. 2012;39: et al. Should DEHP be eliminated in blood bags? Vox Sang. 2014;
342–7. 106:176–95.
3. Council of Europe. Guide to the preparation, use and quality assur- 23. Larsson L, Sandgren P, Ohlsson S, Derving J, Friis-Christensen T,
ance of blood components. 20th ed. Strasbourg, France: Council of Daggert F, et al. Non-phthalate plasticizer DEHT preserves adequate
Europe Publishing; 2020. blood component quality during storage in PVC blood bags. Vox
4. Sowemimo-Coker SO. Red blood cell hemolysis during processing. Sang. 2021;116:60–70.
Transfus Med Rev. 2002;16:46–60. 24. Graminske S, Puca K, Schmidt A, Brooks S, Boerner A, Heldke S,
5. Hansen AL, Kurach JD, Turner TR, Jenkins C, Busch MP, Norris PJ, et al. In vitro evaluation of di(2-ethylhexyl)terephthalate-plasticized
et al. The effect of processing method on the in vitro characteristics polyvinyl chloride blood bags for red blood cell storage in AS-1 and
of red blood cell products. Vox Sang. 2015;108:350–8. PAGGSM additive solutions. Transfusion. 2018;58:1100–7.
6. de Korte D, Thibault L, Handke W, Harm SK, Morrison A, 25. Larsson L, Larsson S, Derving J, Watz E, Uhlin M. A novel protocol
Fitzpatrick A, et al. Timing of gamma irradiation and blood donor sex for cryopreservation of paediatric red blood cell units allows
influences in vitro characteristics of red blood cells. Transfusion. increased availability of rare blood types. Vox Sang. 2019;114:
2018;58:917–26. 711–20.
7. Bashir S, Naik F, Cardigan R, Thomas S. Effect of X-irradiation on the 26. Gulliksson H, Meinke S, Ravizza A, Larsson L, Hoglund P. Storage of
quality of red cell concentrates. Vox Sang. 2011;101:200–7. red blood cells in a novel polyolefin blood container: a pilot in vitro
8. Kopolovic I, Ostro J, Tsubota H, Lin Y, Cserti-Gazdewich CM, study. Vox Sang. 2017;112:33–9.
Messner HA, et al. A systematic review of transfusion-associated 27. Daleke DL. Regulation of transbilayer plasma membrane phospho-
graft-versus-host disease. Blood. 2015;126:406–14. lipid asymmetry. J Lipid Res. 2003;44:233–42.
9. Moroff G, Luban NL. Prevention of transfusion-associated graft-ver- 28. Willekens FL, Werre JM, Kruijt JK, Roerdinkholder-Stoelwinder B,
sus-host disease. Transfusion. 1992;32:102–3. Groenen-Döpp YAM, van den Bos AG, et al. Liver Kupffer cells
200 LARSSON ET AL.

rapidly remove red blood cell-derived vesicles from the circulation by 33. Hess JR. An update on solutions for red cell storage. Vox Sang.
scavenger receptors. Blood. 2005;105:2141–5. 2006;91:13–9.
29. Almizraq R, Tchir JD, Holovati JL, Acker JP. Storage of red blood cells 34. Lagerberg JW, Korsten H, Van Der Meer PF, De Korte D. Prevention
affects membrane composition, microvesiculation, and in vitro qual- of red cell storage lesion: a comparison of five different additive solu-
ity. Transfusion. 2013;53:2258–67. tions. Blood Transfus. 2017;15:456–62.
30. Almizraq RJ, Holovati JL, Acker JP. Characteristics of extracellular 35. Zehnder L, Schulzki T, Goede JS, Hayes J, Reinhart WH.
vesicles in red blood concentrates change with storage time and Erythrocyte storage in hypertonic (SAGM) or isotonic (PAGGSM)
blood manufacturing method. Transfus Med Hemother. 2018;45: conservation medium: influence on cell properties. Vox Sang. 2008;
185–93. 95:280–7.
31. D’Alessandro A, Culp-Hill R, Reisz JA, Anderson M, Fu X, Nemkov T,
et al. Heterogeneity of blood processing and storage additives in dif-
ferent centers impacts stored red blood cell metabolism as much as
storage time: lessons from REDS-III-omics. Transfusion. 2019;59: How to cite this article: Larsson L, Ohlsson S, Derving J,
89–100. Diedrich B, Sandgren P, Larsson S, et al. DEHT is a suitable
32. Shih AW, Apelseth TO, Cardigan R, Marks DC, Bégué S, Greinacher
plasticizer option for phthalate-free storage of irradiated red
A, et al. Not all red cell concentrate units are equivalent: international
survey of processing and in vitro quality data. Vox Sang. 2019;114: blood cells. Vox Sang. 2022;117:193–200.
783–94.
Received: 10 March 2021 Revised: 16 June 2021 Accepted: 16 June 2021

DOI: 10.1111/vox.13176

ORIGINAL ARTICLE

HIV residual risk in Canada for apheresis source plasma

donation without deferral for men who have sex with men

Eliana Aubé1,2 | Antoine Lewin2,3 | Sheila F. O’Brien4 | Yves Grégoire5 |

6 7
Josiane Pillonel | Whitney R. Steele | Brian Custer8,9 | Katy L. Davison10 |
Marc Germain5 | Clive R. Seed11 | Félix Camirand Lemyre1,12 | the Surveillance,
Risk Assessment and Policy Subgroup of the ISBT Transfusion Transmitted Infectious Diseases
Working Party

1
Department of Mathematics, Université de
Sherbrooke, Sherbrooke, Quebec, Canada Abstract
2
Medical Affairs and Innovation, Héma- Background and Objectives: In Canada, men having sex with men (MSM) are
Québec, Montreal, Quebec, Canada
3
deferred for 3 months from last sexual contact to reduce human immunodeficiency
Faculty of Medicine and Health Science,
Université de Sherbrooke, Sherbrooke, virus (HIV) risk to recipients. The aim of this paper was to model the Canadian resid-
Quebec, Canada ual risk of HIV-positive source plasma incorporating pathogen inactivation (PI) under
4
Epidemiology & Surveillance, Canadian Blood
no MSM deferral scenarios for apheresis plasma donations.
Services, Ottawa, Ontario, Canada
5
Medical Affairs and Innovation, Héma- Materials and Methods: A combined Bayesian network (BN) and Monte Carlo
Québec, Quebec City, Quebec, Canada approach were implemented to estimate the HIV residual risk under 3-month defer-
6
Direction des Maladies Infectieuses, Santé
ral compared with no deferral without quarantine scenarios for MSM donors. Models
Publique France, St. Maurice, France
7
Scientific Affairs, American Red Cross,
involve the stochastic generation of donation and its infection status based on its
Rockville, Maryland, USA corresponding simulated donor profile. Viral load reduction conferred by PI used by
8
Epidemiology and Health Policy Science, source plasma fractionators was simulated. Model parameters were derived from
Vitalant Research Institute, San Francisco,
California, USA Héma-Québec and Canadian Blood Services data, viral loads in a large sample of
9
Department of Laboratory Medicine, UCSF, HIV-positive US blood donors, CSL Behring documentation and from published data.
San Francisco, California, USA
Results: In the most likely scenario for the 3-month deferral model, there were 2.71
10
Immunisation, Hepatitis & Blood Safety,
Health Protection Agency, London, UK positive donations per 1,000,000 donations (95% confidence interval [CI] 2.63–
11
Donor and Product Safety (DAPS) Policy 2.78). For the no-deferral model, there were 3.01 positive donations per 1,000,000
Unit, Australian Red Cross Lifeblood, Perth,
donations (95% CI 2.94–3.09). For both scenarios, the risk of having an infectious
Western Australia, Australia
12
Centre de recherche, Centre Hospitalier
pool was 0 in 300,000 pools (95% CI 0–0.0000123) after consideration of PI.
Universitaire de Sherbrooke, Sherbrooke, Conclusion: Based on simulation results, there would be a negligible HIV residual risk
Quebec, Canada
associated with the removal of a time-based MSM deferral without quarantine for
Correspondence source plasma incorporating PI.
Félix Camirand Lemyre, Faculté des sciences,
Université de Sherbrooke, 2500 boulevard de KEYWORDS
l’Université, Sherbrooke, c1/QC J1K 2R1,
Canada, deferral, HIV, MSM, pathogen inactivation technology, residual risk
Canada.
Email: [email protected]

Funding information
Canadian Blood Services; Fonds de Recherche
du Québec - Nature et Technologies; Natural
Sciences and Engineering Research Council of
Canada

Vox Sanguinis. 2022;117:201–207. wileyonlinelibrary.com/journal/vox © 2021 International Society of Blood Transfusion. 201
202  ET AL.
AUBE

I N T R O D U CT I O N conservatively assumed to be one HIV RNA copy per pool. A pool

designates the combination of a number of individual plasma dona-
Since human immunodeficiency virus (HIV) emerged in the early 1980s, tions required to reach a predefined volume of plasma. Individual
major efforts have been made to ensure blood products’ safety and qual- donations are assumed to originate from donors who are considered
ity in order to avoid transmission of infectious agents to recipients [1]. safe to donate based on their answers to numerous risk questions,
One approach has been to exclude from donation those at higher risk for including those regarding sexual risks (see Data S1) and whose dona-
HIV. In North America, beginning in the early 1980s, men having sex tion has tested negative for HIV-1/2 antibody and HIV-1 NAT. This
with men (MSM) was identified as a very-high-risk group for acquired allows an estimate of what the residual risk would be in a model
immunodeficiency syndrome (AIDS). Although little was known about where the MSM deferral for source plasma is 3 months (current) ver-
HIV as the causative agent of AIDS at the time, MSM were permanently sus no MSM deferral (proposed).
deferred from blood donation if they had sex with another man even
once, since 1977. Since then, scientific knowledge concerning HIV infec-
tion, diagnostics and transmission modes have considerably expanded. Model
Sensitive assays have reduced the window period (WP) of infection and
computer systems have all but eliminated the chance of releasing a prod- A two-stage Monte Carlo procedure was employed to estimate the
uct in error. Internationally, MSM deferral from blood donation was ini- rate of HIV-positive pools in N = 300,000 pools of fractionated
tially permanent, but numerous countries have changed the deferral plasma treated with one or more PI steps.
from a lifetime to progressively shorter periods [2]. Based on develop-
ment of highly accurate testing methods and introduction of pathogen
inactivation (PI), completely lifting the ban on donation from sexually First stage
active MSM has been suggested [3]. In particular, it has been proposed
that MSM could safely donate plasma intended for fractionation, given A BN is a probabilistic graphical model that uses a directed acyclic
that each donation is screened by serology and nucleic acid testing graph (DAG) to specify conditional relationships among a collection of
(NAT), and that the fractionation process includes very effective PI steps. variables. In the BN DAG, nodes represent variables, and edges reflect
However, it is necessary to carefully assess risks associated with this conditional dependencies. A joint probability distribution can then be
change since this group of potential donors remains at higher risk of HIV constructed from the graphical structure by elicitation of the set of
infection in Western countries. For example, at the end of 2016, the conditional probabilities of each node given its parents in the graph [7].
Public Health Agency of Canada estimated that 49.1% of HIV-infected Figure 1 shows the BN DAG considered for our simulation models. It
persons in Canada were MSM [4]. includes donor variables known to be associated with risk of infection:
In Canada, the lifetime exclusion for MSM blood donors was age at the time of donation, sex, donor status (new/repeat), MSM
abolished in 2013 and replaced by a 5-year deferral. In 2016, deferral (yes/no), compliance with MSM deferral (yes/no, in the 3-month sce-
changed from 5 years to 12 months, and, in 2019, Canada authorized nario only), level of risk (high/low, in the no-deferral scenario only),
MSM blood donations after 3 months since last sexual contact with HIV infectious state (not infected/infected within the WP/infected
another man [5]. These policy changes were supported by mathemati- and not in WP) and NAT (+/
). According to the BN graph in
cal models simulating various scenarios which allowed estimating the Figure 1, the joint probability over these variables factors as:
risk of releasing potentially infectious donations [5, 6]. Based on these
estimations, simulated HIV residual risk remains extremely low, given Pðage, sex, MSM, status, compliance, HIV state NATÞ
that sensitivity of NAT has improved to the extent that the HIV WP ¼ PðageÞPðsexageÞPðMSMsexÞPðstatusage, sex, MSMÞPðcomplianceMSMÞ
 PðHIV statesex, MSM, status, complianceÞPðNATHIV stateÞ:
for donation testing in Canada is 8–10 days [5].
One approach now being considered in Canada is the possible
implementation of less restrictive MSM deferral for plasma intended The Elicitation section contains a description of how bodies of
for fractionation. In this context, we developed a stochastic BN model knowledge were used to model each factor above.
with Monte Carlo simulations to estimate the risk of HIV contamina-
tion of plasma-derived products if sexually active MSM were allowed
to donate. Second stage

The Monte Carlo algorithm performed simulations of each donation

MATERIALS AND METHODS individually and independently from the joint distribution specified by
the BN. To form plasma pools, simulated plasma donations (750 ml)
Context were iteratively combined until a pooled volume ranging randomly
from 4000 L to 6000 L was reached (personal communication with
We define HIV residual risk as the probability of having an infectious Rima Khalil, Héma-Quebec Stable products director). The HIV viral
pool of plasma after using PI processes. The infectious threshold is load of a pool was calculated as the sum of the viral load of each
HIV RESIDUAL RISK AND APHERESIS SOURCE PLASMA 203

seen from the data that the proportion of new donors (which needed
to be simulated in our model) varied significantly among these three
groups. Distribution of sex within each age group (PðsexjageÞ) was
determined similarly.

MSM donor proportion

PðMSM ¼ yesjsex ¼ femaleÞ was set to 0, and PðMSMjsex ¼ maleÞ was
adjusted based on the MSM male donor proportion for the 12-month
deferral model described in O’Brien et al. [8], see Data S1.

Donor status
The distribution of new and repeat donors (Pðstatusjage,sex,MSMÞ)
was elicitated by matching sample proportions from Héma-Québec
and CBS internal data.

MSM donor non-compliance with MSM deferral criteria

The rate of MSM failing to comply with deferral criteria was deter-
mined based on male non-compliance rate retrieved from the most
likely scenario of O’Brien et al. [5]. Non-compliance in MSM
F I G U R E 1 Directed acyclic graph (DAG) of human (Pðcompliance ¼ nojMSM ¼ yesÞ) was estimated using the non-
immunodeficiency virus (HIV) risk estimation for source plasma
compliance rate in men and MSM proportion:

Pðnon-compliance in menÞ
Pðcompliance ¼ nojMSM ¼ yesÞ ¼
PðMSM ¼ yesÞ
positive donation adjusted for the volume of each plasma unit and the
total volume of the pool. The effect of the PI process was simulated
based on the viral load of each pool. Further, Pðcompliance ¼ nojMSM ¼ noÞ was set to 0.

HIV state
Elicitation For the conditional distribution of HIV state, parameters associated to
HIV incidence/prevalence and WP probability were considered.
Simulation process flowchart HIV prevalence/incidence: Here, prevalence is defined as the pro-
portion of positive donations in a donation pool, whereas the inci-
The BN DAG in Figure 1 was used as a process flowchart for donor dence rate is defined as the proportion of new HIV cases in a
profile simulations. First, age was generated, second, sex conditional donation pool over a year. Given the limited number of HIV-positive
on age, third, MSM status conditional on sex and so on. In the no- donations in our datasets (three positive donations in 5 years), preva-
deferral model, rate of non-compliance was used as a proxy for the lence and incidence rate estimations were not categorized by age
percentage of recently active high-risk MSM potential donors (calcula- group and were determined for all groups of donors G ¼
tions are shown in the Scenarios section). Although not explicitly Gðsex,MSM,status,complianceÞ corresponding to a plausible choice for donor
shown in Figure 1, infectious state was also set to depend on preva- profile variables.
lence and incidence rates. These were made specific to sex, donor sta- To determine group-wise HIV prevalence and incidence rates,
tus, MSM status and compliance variables. For newly positive donations in Canada (CBS and Héma-Québec) from 1 July 2014 to
donations, an extra simulation step was carried to consider WP risk. 30 June 2019 were considered. For groups matching status ¼ repeat,
prevalence was elicitated according to the formula

Parameters used in the first stage number of HIV infected donations in Gstatus¼repeat
Prevalence in Gstatus¼repeat ¼
number of donations tested in Gstatus¼repeat

Donor age and sex

Distribution of donors by age group (PðageÞ) was set to match sample Prevalence in groups with status ¼ new was considered 10 times
proportions calculated based on plasma donation data from Canadian higher than the one of their matched analogue with status ¼ repeat (i.
Blood Services (CBS) and Héma-Québec collected in 2018 (January e., same characteristics except for status) obtained from the above
1–December 31). The minimum age for donation is 17 at CBS and Equation [5].
18 at Héma-Québec. Three age groups were considered: 17/18–24, To determine group-wise HIV incidence, an annual mean of 2.15
25–44 and above 44. This stratification was made since it could be donations per donor was used to estimate the number of annual
204  ET AL.
AUBE

donations. Group-specific incident cases rate was determined from HIV- HIV serological and NAT
positive donations made by repeat donors who had made an HIV-negative Performance, limit of detection and system failure parameters were
donation in the 12 months preceding their positive donation. For groups elicitated based on routinely used procedures in Canadian blood oper-
associated with status ¼ repeat, incidence rate was calculated as: ators involving Cobas multiplex and Procleix Ultrio Plus package
inserts [10, 11], see Data S1 for details.
number of incident cases in Gstatus¼repeat
Incidence rate in Gstatus¼repeat ¼ :
number of donations tested in Gstatus¼repeat
PI process
Donor screening and testing of donations and fractionation pools are
Incidence rate in groups with status ¼ new was elicitated follow- the first and second steps in product safety, while virus reduction
ing the approach of Davison et al. and set equal to the one of their using PI is the third and last step applied to pooled plasma destined
matched analogues with status ¼ repeat multiplied by a truncated-at- for plasma-derived medical products [12] (see also Data S1). The log
zero normal random variable with mean of 1.65 (SD 1.0334). This reduction factor (LRF) is used to quantify this reduction, for instance,
adjustment factor was put forward following the analysis of first time one LRF corresponds to inactivating 90% of the pathogen with the
and repeat English donor data during 2001–2007 [6]. viral load being reduced by a factor of 10 (i.e., 1 log). HIV LRF for
For cases where MSM ¼ yes, prevalence and incidence propor- PDMD ranges from 15.3 (Privigen, CSL Behring, Bern, Switzerland) to
tions were set equal to the number of HIV-positive donations from 9.6 (RiaSTAP, CSL Behring, Marburg, Germany) (Personal communica-
non-compliant MSM donors (current criteria), and the estimated num- tion with CSL Behring). For modelling purposes, the lowest mean LRF
ber of non-compliant MSM donors was based on post-donation inter- for HIV was used.
views and compliance survey. Model parameter values related to NAT performance and PI log
WP donations: WP is the stage from infection to time at which reduction are detailed in Data S1.
pathogen becomes detectable by a given diagnostic procedure. The
probability of having an incident donation being made during the WP
was considered independent of the donor’s profile given their ‘newly Scenarios
infected’ status, and set equal to:
Simulations were conducted according to a 3-month MSM deferral
PðWP donationjincident case ¼ yesÞ model and to a model with no MSM deferral without plasma unit
WP length
¼ : quarantine. Both were simulated under four scenarios: most likely,
Median interdonation interval length
optimistic and pessimistic i and ii. Values for the parameters are
shown in Tables 1–3.
The median interdonation interval length was calculated based on 1. Most likely scenario: HIV incidence, prevalence, MSM donor
2018 CBS and Héma-Québec data for plasma donation. proportion and non-compliance rates were taken as described in the
Elicitation section. For the no-deferral model, proportion of males
with high-risk MSM behaviour was determined using the proportion
Parameters used in the second stage of non-compliant male donors as proxy for high-risk MSM behaviour
non-compliant male donor proportion ð3-monthÞ2
as follows: 1þnon-compliant male donor proportion ð3-monthÞ.
Viral load The 3-month non-compliant male donor proportion was multi-
For WP donations, Weusten et al. model [9] was used to specify the plied by 2 for the no-deferral model since MSM donors who were
viral load distribution. The maximum WP viral load was set to 14.5 non-compliant before lifting of the deferral were considered to remain
copies/ml. Viral loads post WP were simulated based on U.S. HIV- blood donors after the lifting, and the number of high-risk newly eligi-
positive blood donation database (1156 positive blood donations from ble MSM donors was considered to be proportional to that number.
the period of 2010 to 2018). Complete modelling details regarding 2. Optimistic scenario: HIV incidence, prevalence and MSM
viral load distribution are given in Data S1. donor proportion were set as in the most likely scenario. Non-

TABLE 1 MSM parameter values according to deferral model

Three-month deferral (95% CI) No deferral (95% CI)

Proportion of MSM among donor population 0.0162 (0.0099–0.0254) 0.0172 (0.0110–0.0264)
Proportion of non-compliant MSM 0.0010 (0.0003–0.0020) –
Proportion of men at high risk of HIV – 0.0020 (0.0006–0.0040)
Proportion of MSM among new donors by age group 17/18–24 0.0522 (0.0348–0.0843) 0.0562 (0.0398–0.0864)
25–44 0.0279 (0.0225–0.0324) 0.0332 (0.0282–0.0375)
>44 0.0158 (0.0135–0.0186) 0.0218 (0.0197–0.0245)

Abbreviations: HIV, human immunodeficiency syndrome; MSM, men having sex with men.
HIV RESIDUAL RISK AND APHERESIS SOURCE PLASMA 205

T A B L E 2 Human immunodeficiency virus (HIV) prevalence and

2.81  10
13

1.63  10
6

1.63  10
8
incidence according to deferral models

Worst-case

0.02872

0.01041
HIV prevalence in the donor population

8617
57.4
HIV prevalence per

–
Sex Donor type 100,000 donations
Men First time 0.724

1.97  10
14

14
Repeat 0.117

Pessimistic ii

7

1.72  10
8

8

1.99  10
6

6
9.73 10
8
2.01  10

1.00  10

1.64  10

2.28  10
Women First time 0.613

0.01045
0.01331
3136
3994
Repeat 0.306

5.68
6.32
Probability that donation is an incident case for HIV in donor
population

HIV incident donation rate

1.95  10
13

14

9.75  10
8

8

1.65  10
8

8

2.05  10
6

6
Pessimistic i
Sex per 100,000 donations

1.95  10

9.70  10

1.59  10

2.00  10
0.00962
0.01240
Men 0.039

2885
3719
5.10
5.96
Women 0.102

MSM HIV prevalence

HIV prevalence

1.82  10
14

14
Donor type Compliance per 100,000 donations

9.17  10
8

8

1.55  10
8

8

1.15  10
6

6
Optimistic

1.66  10

8.16  10

1.49  10

1.26  10
First time Non-compliant or high risk 155.40

0.00420
0.00445
1259
1335
2.71
2.86
MSM donor population 65.28
Repeat Non-compliant or high risk 15.54
MSM donor population 6.53

MSM HIV incident donation probability

1.93  10
14

14

9.76  10
8

8

1.75  10
8

8

1.77  10
6

6
Most likely

1.95  10

9.79  10

1.47  10

2.20  10
HIV incident rate

0.00441
0.00494
Donor type Compliance per 100,000 donations
1323
1483
2.79
3.01

Repeat Non-compliant or high risk 15.54

MSM donor population 2.7

Abbreviation: MSM, men having sex with men.

3-month deferral

3-month deferral

3-month deferral

3-month deferral

3-month deferral

3-month deferral

3-month deferral
Deferral model

No deferral

No deferral

No deferral

No deferral

No deferral

No deferral

No deferral
TABLE 3 Parameter values for window period and interdonation
interval

Parameter Value Reference

Window period (mean [SD])a 9 (0.6) days O’Brien et al.
Human immunodeficiency virus (HIV) risk estimate

Abbreviations: NAT, nucleic acid testing; PI, pathogen inactivation.

[5]
Median copies per pool after NAT and PI (RNA copies/pool)

Interdonation interval (median 14 days [7– HQ and CBS

Mean copies per pool after NAT and PI (RNA copies/pool)
Mean viral concentration per pool after NAT and PI (RNA

Maximum copies per pool after NAT and PI (RNA copies/

[IQR; Q1, Q3]) 34]

Number of pools with a viral load (in 300,000 pools)

a
Variability modelled by a normal distribution.
HIV positive donations per 1,000,000 donations

Abbreviations: CBS, Canadian Blood Services; HQ, Héma-Québec; IQR,

Probability of getting a pool with a viral load

interquartile range; SD, standard deviation.

compliance in male donors for 3-month deferral and high-risk MSM

proportion for no-deferral model were reduced by half.
3. Pessimistic scenarios:

i. Compared to the most likely scenario, incidence, prevalence,

non-compliance and/or high-risk MSM proportion were doubled,
copies/ml)

and MSM donor proportion was increased by a factor of 1.5.

TABLE 4

pool)

ii. Parameters were set as in pessimistic i, except for MSM donor

proportion. For the latter, the estimation of MSM proportion
within the general population in Canada of 2.9% (95% confidence
206  ET AL.
AUBE

interval [CI] 1.8–3.9) was used, combined with the donor rate cal- profile of plasma-manufactured products, with respect to HIV trans-
culated for the general population [13]. mission, would likely remain unaffected. Of note, our assumed thresh-
old is arguably conservative, see, for example, Zanetti et al. [16],
In addition, a worst-case scenario was simulated for the no-deferral which states, regarding plasma for transfusion, that one human infec-
model using MSM donor proportion and MSM HIV incidence and tious dose corresponds to approximately 500–1000 HIV RNA copies.
prevalence rates based on the MSM general population (i.e., 443 per See also the study of HIV acquisition in Kleinman et al. [17].
100,000 MSM persons and 9.60%, respectively) [14, 15]. In the worst-case scenario, only three infectious donations made out-
side the WP ended up in a pool, none of these, due to an NAT failure. This
corresponds to three procedural errors in every 10,000 years in Canada.
RESULTS Each source of data employed to estimate the model parameters
has its limitations, and some parameters were more challenging to
In the most likely scenario for the 3-month deferral model, the probabil- estimate. For instance, a response bias might have been present in
ity of having a pool with HIV RNA was 0.00441. An average number of surveys used to estimate the non-compliance rate among men for the
HIV copies per pool after use of PI of 9.7606  10
8 (95% CI MSM deferral criteria [18]. Non-compliance in Canada is low and may
9.7598  10
8–9.7615  10
8) was obtained, with a median of be different in other countries [19]. Our estimation of these rates was

8
6
1.7536  10 copies and a maximum of 1.7672  10 copies. For based on the ones presented by O’Brien et al. [8], obtained from a
the no-deferral scenario, probability of having a pool with HIV RNA Canadian survey, to reflect Canadian donor behaviour with greater
was 0.00494. The number of HIV copies per pool was 9.7856  10
8 accuracy. Another limitation is that this simulation study did not con-
(95% CI 9.7847  10
8–9.7865  10
8), along with a median number sider the pre-exposure prophylaxis (PrEP) medication used to reduce

8
6
of 1.4700  10 and a maximum number of 2.2000  10 copies the burden of HIV. PrEP use could interfere with HIV screening
per pool. The mean number of HIV copies per billion litres ranges from among blood donors and may result in undetectable RNA by nucleic
0.019334 to 0.019511 corresponding to a difference of 0.000177 cop- acid/viral load assays for blood component qualification [20]. PrEP
ies per billion litres between the averages of the two deferral models. medication use is a deferral criterion for blood donors in Canada, and
For the worst-case scenario, the probability of having a pool with hence, its inclusion in the model should arguably not significantly
HIV RNA was 0.02872. The average number of HIV copies per pool increase the proportion of donations with HIV viral loads below the
after use of PI was of 1.6294  10
6 (95% CI 1.6244  10
6– NAT detection limit under a realistic compliance rate assumption.
1.6346  10
6), with a median of 1.6300  10
8 copies and a maxi- Despite the need to generate evidence-based data supporting
mum of 0.010409. Results are presented in Table 4. There, absolute policy changes regarding MSM ensuring the safety of recipients, few
differences between mean viral concentrations are smaller than 1.59 studies have evaluated intention to donate plasma among MSM. A
 10
15 copies/ml and are thus not clinically significant. qualitative study performed by Grace et al. indicated that time-based
Finally, under the one HIV RNA copy per pool assumed threshold, deferral would maintain the key issues regarding equity and fairness
none of the 300,000 pools simulation rounds, each round representing related to blood screening practices for MSM in Canada [21].
around two billion donations, would have been considered infectious According to participants, policies regarding MSM blood donations
(i.e., no manufactured product would have contained one virion of HIV should be more aligned with sound and up-to-date scientific evidence.
after the use of PI process). A sensitivity analysis was conducted to Our results strongly suggest that allowing sexually active MSM to
highlight variables that are likely to affect risk estimates (Data S2). donate plasma for fractionation would be just as safe as the current
approach of time-limited deferral, without having to impose a poten-
tially inefficient and costly quarantine strategy that would be per-
DISCUSSION ceived as unduly burdensome and discriminatory.
Demand for plasma-derived medicinal products in Canada has
Our model allowed to simulate HIV contamination risk in pools of increased over the past 10 years [22]. Our study supports the safety
plasma with a change from 3 months to no deferral without quaran- of expanding source plasma donation to include donors not subject to
tine for MSM, considering the use of PI steps in the fractionation pro- deferral for MSM. Based on the number of source plasma donations
cess. The results suggest that the latter risk of HIV transmission in in Canada in 2018 (106,654 donations), no deferral model would lead
plasma-derived products is negligible. According to our model, all- to an increase of about 1119 donations (1.05%) if we added newly eli-
owing sexually active MSM to donate would not significantly increase gible donors according to pessimistic ii scenario.
the risk of occurrence of an HIV RNA-positive pool. Our results also Based on our modelling, there would be a negligible HIV contami-
suggest that, while the vast majority of pools would not contain HIV nation risk associated with a change from 3-month MSM deferral to
at all, the average viral load within the HIV RNA-positive pools would no MSM deferral for source plasma donors in Canada. Given the aim
only be slightly higher under the no-deferral model compared to the of implementing less restrictive MSM deferral policies in Canada, our
one under the 3-month deferral model and would be well under results support that it would be a safe next step to permit MSM to
the one-per-pool HIV RNA copy threshold. Therefore, the safety donate source plasma without a time-based deferral.
HIV RESIDUAL RISK AND APHERESIS SOURCE PLASMA 207

FUND ING INFORMATION 9. Weusten J, Vermeulen M, van Drimmelen H, Lelie N. Refinement of

This research received funding support from Canadian Blood Services a viral transmission risk model for blood donations in seroconversion
window phase screened by nucleic acid testing in different pool sizes
MSM Research Grant program, funded by the federal government
and repeat test algorithms. Transfusion. 2011;51:203–15.
(Health Canada) and the provincial and territorial ministries of health, 10. Roche Molecular Systems, Inc. cobas® MPX Multiplex HIV, HCV &
from the National Sciences and Engineering Research Council of Canada HBV Nucleic Acid Test for Use on the cobas® 6800/8800 Systems -
and the Fond de recherche du Québec – Nature et Technologies. Package Insert [Internet]. Roche Molecular Systems, Inc. 2020 [cited
2021 May 26]. https://www.fda.gov/media/115031/download
Australian governments fund Australian Red Cross Lifeblood to provide
11. Grifols Diagnostic Solutions Inc. Procleix® UIltrio Plus® Assay - Pack-
blood, blood products and services to the Australian community. age Insert [Internet]. Grifols Diagnostic Solutions Inc. 2020 [cited
2021 May 26]. https://www.fda.gov/media/73892/download
12. Klamroth R, Gröner A, Simon TL. Pathogen inactivation and removal
methods for plasma-derived clotting factor concentrates. Transfu-
AUTHOR CONTRIBUTIONS
sion. 2014;54:1406–17.
All authors contributed to the design, reviewed model outputs and 13. Pillonel J, Pelat C, Tiberghien P, Sauvage C, Danic B, Martinaud C,
contributed to interpretation. E.A. performed the analysis and drafted et al. The evolving blood donor deferral policy for men who have sex
the manuscript, and all authors revised it critically. All authors with men: impact on the risk of HIV transmission by transfusion in
France. Transfusion. 2020;60:525–34.
approved of the submitted version.
14. Rich AJ, Lachowsky NJ, Sereda P, Cui Z, Wong J, Wong S, et al. Esti-
mating the size of the MSM population in metro Vancouver, Canada,
CONF LICT OF IN TE RE ST using multiple methods and diverse data sources. J Urban Health.
The authors declare that there are no conflicts of interest identified. 2018;95:188–95.
15. Challacombe L. The epidemiology of HIV in gay men and other men
who have sex with men [Internet]. CATIE. 2015 [cited 2021
ORCID
May 26]. http://www.catie.ca/sites/default/files/epi-msm%20EN%
Antoine Lewin https://orcid.org/0000-0003-1748-4198 202015%2002%2025b.pdf
Sheila F. O’Brien https://orcid.org/0000-0002-5332-2789 16. Zanetti AR, Bodini U, Clerici M, Romanò L, Paolini E, Biasin M, et al.
Yves Grégoire https://orcid.org/0000-0002-1096-698X Transfusion of red blood cells from an HIV-RNA-positive/anti-HIV-
negative donor without HIV infection in the recipient. Transfusion.
Whitney R. Steele https://orcid.org/0000-0001-7955-3115
2007;47:1328–9.
Katy L. Davison https://orcid.org/0000-0002-6337-892X 17. Kleinman SH, Lelie N, Busch MP. Infectivity of human immunodefi-
Clive R. Seed https://orcid.org/0000-0002-0234-4507 ciency virus-1, hepatitis C virus, and hepatitis B virus and risk of
Félix Camirand Lemyre https://orcid.org/0000-0003-3277-2729 transmission by transfusion. Transfusion. 2009;49:2454–89.
18. Seed CR, Lucky TT, Waller D, Wand H, Lee JF, Wroth S, et al. Com-
pliance with the current 12-month deferral for male-to-male sex in
RE FE R ENC E S Australia. Vox Sang. 2014;106:14–22.
1. Perkins HA, Busch MP. Transfusion-associated infections: 50 years 19. Sauvage C, Spinardi R, Pelat C, Pouget T, Danic B, Woimant G, et al.
of relentless challenges and remarkable progress. Transfusion. 2010; Noncompliance with blood donor selection criteria - Complidon
50:2080–99. 2017, France. Transfusion. 2020;60:73–83.
2. Goldman M, W-Y Shih A, O’Brien SF, Devine D. Donor deferral poli- 20. Custer B, Quiner C, Haaland R, Martin A, Stone M, Reik R, et al. HIV
cies for men who have sex with men: past, present and future. Vox antiretroviral therapy and prevention use in US blood donors: a new
Sang. 2018;113:95–103. blood safety concern. Blood. 2020;136:1351–8.
3. Pathogen Reduction of Platelet Donations as an Alternative Proce- 21. Grace D, Gaspar M, Lessard D, Klassen B, Brennan DJ, Adam BD,
dure to MSM Donor Deferral. U.S. Food & Drug Administration. et al. Gay and bisexual men’s views on reforming blood donation
2019. p. 12. https://www.fda.gov/media/127506/download policy in Canada: a qualitative study. BMC Public Health. 2019;
4. Public Health Agency of Canada. Summary: Estimaties of HIV Inci- 19:772.
dence, Prevalence and Canada’s Progress on Meeting the 90–90-90 22. Hotchko M, Robert P. Recent market status and trends of fraction-
HIV Targets, 2016 [Internet]. Ottawa: Public Health Agency of ated plasma products. Ann Blood [Internet]. 2018 [cited 2021 May
Canada; 2018 [cited 2021 May 25]. 15 p. http://publications.gc.ca/ 26];3:19 https://aob.amegroups.com/article/view/4330
collections/collection_2018/aspc-phac/HP40-216-2018-eng.pdf
5. O’Brien SF, Grégoire Y, Pillonel J, Steele WR, Custer B, Davison KL,
SUPPORTING INFORMATION
et al. HIV residual risk in Canada under a three-month deferral for
men who have sex with men. Vox Sang. 2020;115:133–9. Additional supporting information may be found online in the
6. Davison KL, Gregoire Y, Germain M, Custer B, O’Brien SF, Supporting Information section at the end of this article.
Steele WR, et al. Changing the deferral for men who have sex with
men - an improved model to estimate HIV residual risk. Vox Sang.
2019;114:666–74. How to cite this article: Aubé E, Lewin A, O’Brien SF,
7. Koski T, Noble JM. Bayesian networks: an introduction. Chichester,
Grégoire Y, Pillonel J, Steele WR, et al. HIV residual risk in
West Sussex, London: Wiley; 2009.
8. O’Brien SF, Osmond L, Fan W, Yi Q-L, Goldman M. Compliance with Canada for apheresis source plasma donation without deferral
time-based deferrals for men who have sex with men. Transfusion. for men who have sex with men. Vox Sang. 2022;117:
2019;59:916–20. 201–207.
Received: 6 January 2021 Revised: 3 May 2021 Accepted: 4 May 2021

DOI: 10.1111/vox.13134

ORIGINAL ARTICLE

Estimation of the latent therapeutic demand for

immunoglobulin therapies in autoimmune neuropathies
in the United States

Albert Farrugia1 | Megha Bansal2 | Ivan Marjanovic3

1
Faculty of Medicine and Health Sciences,
University of Western Australia, Perth, Abstract
Western Australia, Australia
Background and Objectives: The use of immunoglobulin (IG) solutions as an immu-
2
Private Consultant, Health Economics and
Outcomes Research
nomodulatory therapy in certain neurological conditions has become an established
3
Private Consultant in Neurology, Milan, Italy modality and represents a significant proportion of total IG use. The estimation of
the evidence-based potential demand designated as the latent therapeutic demand
Correspondence
Albert Farrugia, Faculty of Medicine and
(LTD) for IG in these diseases is required for adequate planning of the plasma supply
Health Sciences, University of Western required to manufacture the product.
Australia, Perth, Western Australia 6009,
Australia.
Materials and Methods: The diseases studied included chronic inflammatory demye-
Email: [email protected] linating polyneuropathy (CIDP), Guillain–Barré syndrome (GBS) and multifocal motor
neuropathy (MMN). The LTD for IG was assessed using a decision analysis model,
using Microsoft Excel. The model analysed the epidemiological and clinical factors
contributing to IG usage. One-way sensitivity analysis and probabilistic sensitivity
analysis derived the LTD in grams per 1000 inhabitants. The key variables included
the treatment schedule and the prevalence of the disease.
Results: The model estimates that an average annual IG demand and standard devia-
tion for CIDP, GBS and MMN in the United States is 83.05
24.5, 6.1
3.2 and
36.1
25.5 g/1000 inhabitants, respectively.
Conclusion: Together with previous work on the LTD for IG in immunodeficiencies,
these results indicate that current IG usage reflects the estimated LTD for the main
indications for IG in the United States The wide range of LTD found in all these stud-
ies emphasizes the need for more precise assessment of the underlying variables,
particularly disease prevalence and dosage. Further studies on other indications such
as secondary immunodeficiencies will augment these results and will assist in guiding
demand planning for IG use and plasma collection in the United States and inform
blood policy in other countries.

KEYWORDS
immunoglobulins, IVIG, plasma derivatives

[Correction added on 24 June 2021 after first online publication: The affiliation of the 2nd
author, Megha Bansal, was corrected in this version.]

208 © 2021 International Society of Blood Transfusion. wileyonlinelibrary.com/journal/vox Vox Sang. 2022;117:208–219.
ESTIMATION OF THE LATENT THERAPEUTIC DEMAND FOR IMMUNOGLOBULIN THERAPIES 209
IN AUTOIMMUNE NEUROPATHIES IN THE UNITED STATES

I N T R O D U CT I O N in the hands (commonly), with differences from one side of the body
to the other in the specific muscles involved. MMN is a chronic dis-
The use of immunoglobulin (IG) solutions purified from human plasma ease which responds well and uniquely to IVIG treatment [17] and is
has become increasingly established as a therapy in various neurologi- indicated on the basis of guidelines (EFNS 2006); nevertheless, only
cal diseases in the past 20 years [1]. This group of indications has one IVIG product is licensed for this indication in the United States. In
overtaken primary immune deficiency (PID) as the largest users of IG Europe, it is approved for all brands.
[2] and thus represents an important driver for the collection of Table 1 from the U.S. Food and Drug Administration (FDA) sum-
plasma for fractionation. The United States is one of the few countries marizes the status of the current immunoglobulin products licensed
in the world which collects more than enough plasma for its use and by indication in the United States [11]. It is noted that several of these
exports a considerable volume of plasma products outside the coun- products do not carry the label for all the neurological indications
try. Despite this, shortages of IG for different diseases have become which are the subject of this paper. While usage of these products for
increasingly visible in the past few years [3, 4]. Plasma for fraction- indications which have not been approved constitutes off-label use, it
ation is a finite resource, and, in many countries, growth in its collec- is known that clinicians and insurers support the prescribing of these
tion has lagged behind the demand for IG [5]. Estimating the potential products for the whole spectrum of neurological indications, basing
demand for IG indications is important for adequate planning of their decisions on which indications are supported through the rele-
plasma collection and management of the limited supply. In previous vant medical guidelines. Although regulatory authorities do not sub-
work, the concept of latent therapeutic demand (LTD) has been used scribe to the concept that immunoglobulins are generic products,
to estimate the requirements for IG in PID [6, 7] in the hypothetical clinical practice indicates that the safety and efficacy profiles of all
scenario when usage is unconstrained by supply and reimbursement. licensed products are similar.
Given that the usage of IG in neurological conditions now constitutes Despite the relatively small number of products specifically
the single largest group of indications, an estimate of the LTD for IG licensed for the above diseases, IG consumption in the United States
arising from these diseases is relevant. for these conditions is very high. In this study, the LTD for IG for
Chronic inflammatory demyelinating polyneuropathy (CIDP) is a these three indications has been estimated using a decision analysis
chronic neurological disorder characterized by progressive weakness model previously described for estimating LTD of Factor VIII in
and impaired sensory function in the legs and arms. Treatment for haemophilia [18] and IG in PID [6].
CIDP includes corticosteroids such as prednisone, which may be pre-
scribed alone or in combination with immunosuppressant drugs. Plas-
mapheresis (plasma exchange) and intravenous immunoglobulin (IVIG) METHODS
therapy are effective [8], but plasma exchange can cause a rebound in
the condition [9] and causes substantial inconvenience to the patients Model structure
. Following the first approval of IG for use in CIDP by the US Food
and Drug Administration (FDA) 2008 [10], IG use for this indication For modelling the LTD for IG in the three neuropathies described in
has expanded greatly, despite the lack of specific licensure for most of the Introduction, we used the decision analysis methodology similar
the products on the U.S. market [11]. This reflects the acceptance by to the methods developed by Stonebraker et al [6, 7] for modelling
medical specialists that the efficacy of the product transcends particu- the LTD of IVIG in PID. The variables used in the model are shown in
lar brands, a situation which mirrors the regulatory oversight of IG Tables 1 and 2. The model is based on the relationships of the epide-
product in Europe, where the CIDP indication is embedded in the miological and clinical factors, integrated as shown in the flow charts
summary of product characteristics (SmPC) for intravenous immuno- in Figure 1. Each disease follows a different treatment schedule and
globulin (IVIG) products, allowing all brands to be prescribed for CIDP depending on the incidence/prevalence of the disease the demand in
if they have shown efficacy and safety according to the relevant terms of grams per 1000 inhabitants is calculated. For each disease,
Guideline of the European Medicines Agency (EMA) [12, 13]. variables included prevalence, dosage and variations in treatment pat-
Guillain–Barré syndrome (GBS) is an acute neurological disorder terns. For CIDP, a treatment schedule based on an initial treatment of
characterized by presence of different autoantibodies against the 24 weeks followed by a maintenance period of 24 weeks at a lower
peripheral nervous system. Treatments include plasmapheresis and dose and frequency [19]. Since the America Association of Neurology
IVIg therapy, modalities which have similar efficacy rates [14, 15]. The (AAN) does not specify dosage levels for IVIG [16], the dosage was
benefit of IVIG in children with GBS is poorly supported by evidence, varied over a wide range in the analysis to reflect dosages reported by
but the modality forms part of clinical practice by specialists in the dis- a range of studies and organizations [19, 20]. In GBS, a dose for 3–
ease [16]. Despite IVIG being indicated on the basis of clinical associa- 6 days was followed by a second complete dose in case of relapse
tion guidelines [16], no IVIG product in the United States is [15, 16]. In MMN, a loading dose was followed by a maintenance dose
specifically licensed for GBS. In Europe, it is approved for use in all every 1–6 weeks depending on the clinical status of the patient
brands. [17, 21, 22].
Multifocal motor neuropathy (MMN) is an asymmetric, progres- The model was built in Microsoft Excel. The analysis was con-
sive neuromuscular motor disorder characterized by muscle weakness ducted on one-way sensitivity analysis and probabilistic sensitivity
210 FARRUGIA ET AL.

T A B L E 1 Immunoglobulin products licensed for use in the United TABLE 2 : Base-case values for population variables in the model
States. From Reference [11]
Parameter Value Source
Product Manufacturer Indications US population 317,297,938 Census 2014
Gammagard liquid Baxter Healthcare • Primary Humoral Prevalence
Corporation Immunodeficiency
CIDP 7.7/100,000 Laughlin 2009 [27]
• Multifocal Motor
Neuropathy GBS 2.3/100,000 NIH 2011 Genetics Home
reference31
Gammagard S/D Baxter Healthcare • Primary Humoral
Corporation Immunodeficiency MMN 1.65/100,000 Lawson 2014 [32]
• B-cell Chronic Male:female
Lymphocytic
CIDP 56.6:43.4% Laughlin 2009 [27]
Leukaemia
• Immune GBS 54.6:45.4% HCUP 2012 [41]
Thrombocytopenic MMN 67.5:32.5% Berg 2014 [40]
Purpura
Age groups (years)
• Kawasaki
syndrome CIDP
Gammaplex Bio Products • Primary Humoral 18–44 18.8% Laughlin 2009 [27]
Laboratory Immunodeficiency 45–65 40.9%
• Immune
65+ 40.9%
Thrombocytopenic
Purpura GBS
Bivigam Biotest • Primary Humoral 1–18 9.11% HCUP 2012 [41]
Pharmaceuticals Immunodeficiency 18–44 27.58%
Corporation
45–64 35.57%
Carimune NF CSL Behring AG • Primary Humoral
65+ 27.70%
Immunodeficiency
• Immune MMN
Thrombocytopenic 18–44 60.50% Azulay 1997 [42], Berg-Vos 2002
Purpura [17], Harbo 2009 [43]
45–64 36.80%
Privigen CSL Behring AG • Primary Humoral
65+ 2.60%
Immunodeficiency
• Immune Weight (kg) –
Thrombocytopenic male
Purpura 1–17 44.5 CDC 2010 [44]
Gamunex-C Grifols Therapeutics, • Primary Humoral 18–44 83.5
Gammaked Inc. Immunodeficiency
45–65 88.7
(distributed by: • Immune
Kedrion Thrombocytopenic 65+ 83.8
Biopharma) Purpura Weight (kg) –
• Chronic female
Inflammatory
1–17 39.4
Demyelinating
Polyneuropathy 18–44 67.8
Flebogamma DIF Instituto Grifols, SA • Primary Humoral 45–65 73.7
5% and 10% Immunodeficiency 65+ 68.3
Octagam Octapharma • Primary Humoral
Abbreviations: CIDP, chronic inflammatory demyelinating polyneuropathy;
Pharmazeutika Immunodeficiency
GBS, Guillain–Barre syndrome; MMN, multifocal motor neuropathy.
Produktionsges.
m.b.H

For the Normal Distribution : NORM:INVðRANDð Þ, mean, std deviationÞ

analysis. We have used probability distributions around variables as
appropriate. These distributions were generated using Monte Carlo Tornado diagrams were used to describe the sensitivity analyses
simulations with Excel, as has been described [23]. Normal and uni- (Figure 2), while the probability distributions were used to generate
form distributions were used for some model parameters. The follow- histograms (Figure 3). Excel has the capability to generate both these
ing formulas were used within Excel: products [24, 25].
The respective models for the three conditions have been made
For the Uniform Distribution : min þ ðmax  minÞ*RANDðÞ available [26] as a resource for researchers and for comment.
ESTIMATION OF THE LATENT THERAPEUTIC DEMAND FOR IMMUNOGLOBULIN THERAPIES 211
IN AUTOIMMUNE NEUROPATHIES IN THE UNITED STATES

varied range of 1.9–7.7 per 100,000 worldwide [28–30]. Since the

study done in the United States showed higher prevalence compared
to the worldwide estimation, we assumed 7.7 per 100,000 as the base
case and conducted one-way sensitivity analysis around prevalence of
1.9–8.9 per 100,000.
According to National Institutes of Health, the prevalence of GBS
is quite varied, 0.6–4 cases per 10,000 [31]. We have taken an aver-
age of 2.3 cases per 100,000. The one-way sensitivity analysis utilizes
0.6 and 4 cases per 100,000 people as input values.
The prevalence of MMN is estimated from various sources to be
0.3-3 cases per 100,000 people [32]. The average of 1.65 cases per
100,000 people is used as the base case.

Patients treated with IVIG

CIDP is a rare disease and the review of medical records in the United
States identified 23 cases treated for CIDP in the Rochester Epidemi-
ology Project [27]. The prevalence computed in this project was 7.7
per 100,000 with a yearly incidence of 1.6 cases per 100,000, in the
studied population of Olmstead County in Minnesota. The population
of Olmstead County was 144,921 in the year in which the project was
conducted. Among the cases identified with CIDP, 59% were treated
with IVIG. Another retrospective study done in Massachusetts Gen-
eral Hospital identified 28.9% of patients being treated with IVIG in
CIDP [33]. We used a uniform distribution to vary the patients treated
in the initial period.
In the maintenance or extension period, the CIDP patients that
responded well to IVIG were infused with lower dosage of IVIG (usu-
ally 1 g/kg) and at lower frequency (usually every 3–4 weeks) to avoid
a relapse. Multiple studies and trials indicated that 50–70% patients
responded well to IVIG [8, 34]. A retrospective study done in Spanish
hospitals showed 45.3% of the patients responded well to the treat-
ment and were assigned to the group with low frequency of infusions.
This group was further divided among patients receiving drug at 6–8
week interval, 8–12 week and >12 weeks [35] .
A retrospective chart review study shows 86.9% of the patients
suffering from GBS receive IVIG either alone or in combination with
plasmapheresis [14]. Out of these, patients who relapse are provided
with a second complete dose of IVIg. The evidence shows 8–16% of
F I G U R E 1 Flow charts showing the contribution of the relevant the patients relapse [15, 36].
variables used to calculate latent therapeutic demand (LTD) for use of A review study [22] for treatment of MMN reports 80% of the
IVIG in chronic inflammatory demyelinating polyneuropathy (CIDP) patients respond to IVIG. We used this figure (80%) as the number of
(a), Guillain–Barre syndrome (GBS) (b) and multiple motor neuron
people treated with IVIG in the United States due to lack of exact
neuropathy (MMN) (c)
data. Since treatment of MMN is an ongoing treatment, we assumed
100% of these patients are treated with IVIG.

Data sources
Dose of IVIG
Prevalence
We noted the administration of IVIG in CIDP for 1 year with
A review of medical records in the United States estimated the preva- 24 weeks on the initial period and 24 weeks in the maintenance
lence of CIDP as 8.9 per 100,000 [27]. Other reports provided a period. Most of the trials and studies mentioned a loading dose of
212 FARRUGIA ET AL.

TABLE 3 Base-case values for treatment patterns for CIDP, GBS and MMN

Parameter Value Source

Treatment of CIDP
Patients treated with IVIG in the initial 59% Laughlin 2009 (Table 1) [27]
period
Patients treated with IVIG in the 45.3% Querol 2013 [35]
extension period
Time period of IVIG administration 24 weeks – initial period; 24 weeks – Hughes 2008 ICE study [19]
extension period
Volume of IVIG used per person in the 2 g/kg over 2–4 days and then 1 g/kg over Hughes 2008 ICE study [19]
initial period 1–2 days
Volume of IVIG used per person in the 1 g/kg over 1–2 days Hughes 2008 ICE study [19]
extension period
Frequency of dosage in the initial Every 3 weeks Hughes 2008 ICE study [19]
period
Frequency of dosage in the initial Every 6–8 weeks or 8–12 weeks Querol 2013 [35]
period or >12 weeks
Patients on low frequency of dosage in
the extension period
6–8 weeks 33.3% Querol 2013 [35]
8–12 weeks 35.9%
>12 weeks 30.8%
Treatment of GBS
% Treated with IVIG 86.9% Oczko-Walker 2010 [14]
% Treated with IVIG for second dose 12% va Doorn 2010, NBA 2012 Australian
(patients who relapse) guidelines [39]
IVIG administration days 3.5 days Patwa 2012 AAN guidelines [16]
Volume of IVIG used per person 1.2 g/kg Patwa 2012 AAN guidelines [16]; NBA
2012 Australian guidelines [39]
Frequency of dosage may be repeated in case of relapse or no van Doorn 2010 [15]
response
Treatment of MMN
% Treated with IVIG 80.0% Jinka 2014 [22]; 80% respond to IVIG
treatment
% Treated with IVIG maintenance 100.0% Aassumption
Interval between IVIG dose Every 3.5 weeks Berg-vos 2002 [17]
Volume of loading IVIG 2 g/kg Jinka 2014 [22]
Volume of IVIG for maintenance 1.1 g/kg Berg-vos 2002 [17]; Berg 2014 [40], NBA
Australia 2012 guidelines [39]

Abbreviations: CIDP, chronic inflammatory demyelinating polyneuropathy; IVIG, intravenous immunoglobulin; GBS, Guillain–Barre syndrome; MMN,
multifocal motor neuropathy.

2 g/kg of IVIg in the initial period over 2–4 days. For the remaining 6–24 weeks. The patients were divided in the groups of 6–8 weeks,
period, a range of 0.4–2 g/kg of IVIG is given every 3 weeks [19, 8–12 weeks and 12–24 weeks [35].
20, 37]. Most of the dosages recommended by guidelines such as the A randomized controlled trial for IVIG in GBS used 0.4 g/kg over
EFNS/PNS [38] and the Australian Guidelines [36] reflect the dosages 3–6 days [16]. The Australian guidelines recommended dosage to be
used in the seminal clinical trial of Hughes et al [19] which we have 2 g/kg over 2–5 days [39]. We used an average of 1.2 g/kg over
used in our analysis. The American Academy of Neurology suggests 4 days as the base case and varied the dose from 0.4–2 g/kg over 3–
the dosage, frequency and duration of treatment based on clinical 6 days in one-way sensitivity analysis.
assessment [16]. We varied the dosage using a uniform distribution. The evidence shows an induction dose of 2 g/kg of IVIG is admin-
In the extension period, the same range of 0.4–2 g/kg of IVIG was istered in MMN [22]. The maintenance dose varies between 0.1 and
used, but the frequency of infusion was reduced from 3 weeks to 2 g/kg every 1–6 weeks annually [17, 39, 40].
ESTIMATION OF THE LATENT THERAPEUTIC DEMAND FOR IMMUNOGLOBULIN THERAPIES 213
IN AUTOIMMUNE NEUROPATHIES IN THE UNITED STATES

TABLE 4 Variables used in the sensitivity analysis

Parameter Values for one-way SA Probability distribution Source

Treatment in CIDP
Prevalence of CIDP 1.9–8.9/100,000 Laughlin 2009 [27]
Patients treated with IVIG in initial 28.9–59% Uniform () Darabi 2006 [33], Laughlin 2009 [27]
period
Patients treated with IVIG in 35–70% Uniform () Gorson 2012, Sarti 2009 [34]
extension period
Volume of IVIG used per person for 0.4–2.0 Uniform () Australian 2012 guidelines [20] and
maintenance (gm/kg) Eftimov 2013 [37]
Frequency of dosage in the extension
period
1st group 6–8 weeks Uniform () Querol 2013 [35]
2nd group 8–12 weeks Uniform ()
3rd group >12 weeks Uniform ()
Treatment in GBS
Prevalence of GBS 6–40/1000,000 Uniform () NIH 2011 Genetics Home reference31
% Treated with IVIG for second dose 8–16% Uniform () Van Doorn 2010 pg S77 [15], NBA 2012
(patients who relapse) Australian guidelines [39]
IVIG administration days 3–6 days Uniform () Patwa 2012 AAN guidelines [16]
volume of IVIG used per person 0.4–2 g/kg Uniform () Patwa 2012 AAN guidelines [16]; NBA
2012 Australian guidelines [39]
Treatment in MMN
Prevalence of MMN 0.3–3/100,000 Uniform () Lawson 2014 [32]
Interval between IVIG dose every 1–6 weeks Uniform () Berg-vos 2002 [17]
volume of IVIG for maintenance 0.1–2.0 g/kg Uniform () Berg-vos 2002 [17]; Berg 2014 [40], NBA
Australia 2012 guidelines [39]
Weight (kgs – 10th, 50th and 90th Male Uniform ()
percentiles)
1–17 27.95–44.5–74.10 Extended Swanson–Megill CDC 2010 [44]
18–44 62.60–83.5–112.75 approximation

45–64 69.10–88.7–113.1
65+ 65.40–83.8–104.5
Female
1–17 27.45–39.35–65.25
18–44 52.5–67.8–102.6
45–64 54.7–73.7–106.3
65+ 49.8–68.3–91.2

Abbreviations: CIDP, chronic inflammatory demyelinating polyneuropathy; IVIG, intravenous immunoglobulin; GBS, Guillain–Barre syndrome; MMN,
multifocal motor neuropathy.

Age, gender and weight of patients GBS inflicts both adults and children so we included another age
group of 1–18 years of age to GBS. The proportions of patients in
In the cases where dose is administered per kg of the weight of each age group were gathered from Healthcare Cost and Utilization
patients, it is necessary to distinguish age and gender groups. The data Project (HCUP) of the U.S. Agency for Healthcare Research and Qual-
for gender and age for CIDP were obtained from a retrospective study ity [41]. Several follow-up studies were used to populate the propor-
conducted in the United States. [27] The age groups are classified as tions in each age group in MMN [17, 42, 43].
18–44, 45–64 and 65+ years of age. There was only one case of a The weight distribution was extracted from a report by the
4-year-old in this study, which we excluded so that the population is Centers for Disease Control [44], specified weight by age and
aligned with the adult population in the clinical trial which we gender. We used extended the Swanson–Megill (ESM) approxima-
used [19] . tion [45], as described by Stonebraker et al [7] to obtain 10th,
214 FARRUGIA ET AL.

TABLE 5 LTD for immunoglobulin in primary immunodeficiencies and neurological diseases. From Stonebraker et al [7] and this study

Mean LTD immunoglobulin

Condition g/103 population References
Common variable immune deficiency (CVID) 65.4
73.6 Stonebraker et al [7]
X-linked agammaglobulinemia (XLA) 25.5
27.6 Stonebraker et al [7]
Severe combined immune deficiency (SCID) 13.4
13.5 Stonebraker et al [7]
Wiskott–Aldrich syndrome (WAS) 0.5
0.4 Stonebraker et al [7]
hyper IGM syndrome (HIGM) 0.3
0.3 Stonebraker et al [7]
Chronic inflammatory demyelinating polyneuropathy (CIDP) 83.05
24.5 This study
Guillain–Barré syndrome (GBS) 6.1
3.2 This study
Multifocal motor neuropathy (MMN) 36.1
25.5 This study
Total mean immunoglobulin consumption 230.35

Abbreviation: LTD, latent therapeutic demand.

50th and 90th percentiles for use in the probabilistic sensitivity immunoglobulin for all the three disease studied, but the one-way
analysis. sensitivity analysis also allowed generation of the range of values con-
tributing to the final estimate.
This study has several limitations. Although we sought to assess
RESULTS the LTD of the neurological indications for which immunoglobulin is
prescribed in the United States, these three conditions are not an
Using the variables shown in Tables 3 and 4 for one-way sensitivity exhaustive list of these indications in the medical literature [46]. Inclu-
analysis, Tornado diagrams were generated which showed the dosage sion of these other rare conditions may be expected to augment the
of immunoglobulin and the disease prevalence were the most impor- total LTD estimated in our study, and immunoglobulin has been used
tant parameters influencing the decision analysis model in all the three to treat exacerbations myasthenia gravis for many years [47],
conditions (Figure 2). Extension and second doses for CIDP and GBS although the evidence base is poor [48], and new therapies are emerg-
respectively were less important in influencing demand. In MMN, the ing for the disease [49]. While one product is approved for this indica-
interval between doses was highly influential. tion in Europe, none are approved in the United States, and we were
The model allowed the generation of a probabilistic distribution unable to access any data regarding its usage, possibly because of a
for LTD for each condition (Figure 3) which was skewed for all the lack of insurance for this condition.
three diseases. From this distribution, an average demand and stan- Any study addressing the uncertainties underpinning LTD suffers
dard deviation for CIDP, GBS and MMN of 83.05
24.5, 6.1
3.2 from the limitations imposed by these uncertainties. As is shown by
and 36.1
25.5 g/1000 inhabitants, respectively, was estimated. The the sensitivity analysis, disease prevalence influences LTD greatly and
median extracted from the histograms of the probability distribution is the subject of debate and controversy, particularly for CIDP, the
were 111, 9.5 and 65 g/1000 inhabitants, respectively. indication associated with the highest LTD. In a recent meta-analysis,
Broers et al [50] found a pooled prevalence of 2.81 per 100,000 over
11 studies included. Of note was this meta-analyses’ exclusion of the
DISCUSSION widely cited study conducted by the Mayo Clinic for Olmstead
County in the United States [27], which was used in our study to
This study augments previous work by Stonebraker and colleagues in specify prevalence in the United States. The Mayo study was included
which decision analysis methodology is used to estimate the in the studies considered by Broers et al, but data were not incorpo-
unconstrained demand for plasma protein therapies [6, 7, 16]. The rated in the final analysis because Broers et al considered that the
level of uncertainty and professional variation underpinning many of crude rates for the data were not visible. Broers et al comment that
the key elements contributing to demand enable this methodology, the inclusion of the Mayo Clinic study would have increased the
through the sensitivity analyses which are a feature of the model, to pooled prevalence, and this study, which is the most widely cited for
synthesize the various values offered through clinical research and the prevalence of CIDP in the United States in the literature, reports a
professional into an estimate for LTD. The relative contributions of prevalence similar to a later study extracting prevalence data for the
the different variables into the final estimate are shown in the Tor- whole of the USA extracted from an insurance claims data base [51],
nado diagram generated by the one-way sensitivity analysis, allowing increasing our confidence in the prevalence recorded by the Mayo
a rapid pictorial assessment of the ranking of these variables in the Clinic .
LTD (Figure 2). Unsurprisingly, the model identified dosage and dis- Earlier studies in other geographies, for example, Japan [52] and
ease prevalence as the highest contributors to the LTD for Australia [29], and a more recent study from a single centre in Chile
ESTIMATION OF THE LATENT THERAPEUTIC DEMAND FOR IMMUNOGLOBULIN THERAPIES 215
IN AUTOIMMUNE NEUROPATHIES IN THE UNITED STATES

F I G U R E 2 One-way sensitivity analysis (tornado diagram) – For variables included in the decision analysis model for CIDP (a), GBS (b) and
MMN (c). CIDP, chronic inflammatory demyelinating polyneuropathy; GBS, Guillain–Barre syndrome; MMN, multifocal motor neuropathy
216 FARRUGIA ET AL.

F I G U R E 3 Distribution of IVIG use in the US, extracted from the model and estimated for patients with CIDP (a), GBS (b) and MMN (c).
CIDP, chronic inflammatory demyelinating polyneuropathy; GBS, Guillain–Barre syndrome; IVIG, intravenous immunoglobulin; MMN, multifocal
motor neuropathy
ESTIMATION OF THE LATENT THERAPEUTIC DEMAND FOR IMMUNOGLOBULIN THERAPIES 217
IN AUTOIMMUNE NEUROPATHIES IN THE UNITED STATES

[53] have reported lower prevalence rates, while more recent studies scenario assumes that the underlying factors shaping demand are sim-
in Australia [54] report prevalence rates of 10 per 100,000. In this lat- ilar to those in high-usage countries. These factors assume, for exam-
ter study, a difference was noted between prevalence levels esti- ple, dosage regimens such as those we have drawn from in
mated using the diagnostic criteria of the European Federation of influencing demand. Dosage in immunoglobulin therapies is a contro-
Neurological Societies/Peripheral Nerve Society (EFNS/PNS) [38], versial topic [62], and the guideline-driven dosages underpinning
considered to be the most widely accepted criteria for the diagnosis much of current practice in the developed economies do not reflect
of CIDP [50], and the patient numbers listed in a data base including possible individual variations in the required trough levels in PID
patients prescribed immunoglobulin for CIDP. patients [63], which, if used in determining dosages, may well influ-
Reported differences in disease prevalence can arise from a num- ence the demand. Similarly, scrutiny of the doses used in the neuropa-
ber of causes. In clinical studies, diagnostic alignment to different sets thies may also result in altered regimens, as many of these doses were
of criteria may result in these differences. In CIDP, the criteria of the derived empirically from the initial clinical observations on the use of
EFNS/PNS are considered to have a higher diagnostic sensitivity than IVIG in idiopathic thrombocytopenic purpura (ITP) with minimal rele-
those previously published by the American Academy of Neurology vance to the different pathologies involved in the neuropathies. The
(AAN) [55], which may underestimate prevalence and incidence of the global supply of IG is already under substantial stress [64], and
disease [56]. A study in the UK reported a twofold difference between increased effort is required to collect the volumes of plasma needed
the prevalence of CIDP defined by the EFNS/PNS and that defined by to generate these amounts of IG. These pressures should heighten
the AAN [56], and this difference was also found in the meta-analysis efforts in understanding better the use of immunoglobulin in all the
of Broers et al. [50] However, the uncertainty generated by these dif- disease where it is present as a modality. They should also accelerate
ferent criteria is compounded by the fact that in more than 47% of the search for alternatives to IG therapy, with some therapies already
cases, CIDP is misdiagnosed as other diseases and clinical adherence showing promise as candidates for treating the autoimmune neurolog-
to diagnostic criteria is poor [57]. Differences in geographies and ical diseases assessed in the present study [65]. In the interim, we pro-
demographics may also contribute to these variations. pose that decision analyses for estimating latent demand for therapies
Overall, uncertainty in this key variable emphasizes the role of sensi- should be considered as useful tools in assisting policy makers and
tivity analysis in defining and quantifying the extent to which findings funders in planning access to medical interventions.
such as those in the present study may be influenced by further research
in the clinical and epidemiological definition of the disease. While meta- AC KNOW LEDG EME NT S
analyses such as those of Broers et al [50] reflect the uncertainty in prev- A.F. designed the study, wrote the paper and provided critical com-
alence through the prediction interval, this role is taken up by the Tor- ment regarding the provision of immunoglobulin. M.B. designed and
nado Diagrams which are included in the present study. executed the decision analysis model in Microsoft Excel. I.M. provided
In this study, we considered the sole administration route for input in the clinical features and evidence based use of immunoglobu-
immunoglobulin in treating these diseases to be intravenous. Although lin in neurological diseases.
one product has been approved for subcutaneous administration in
treating CIDP, there are still no data available on the extent to which CONFLIC T OF INT ER E ST
this route has been adapted for CIDP [58]; furthermore, a European All three authors declare past provision of services to companies
meta-analysis of studies conducted in Europe suggests that current manufacturing immunoglobulin therapies. None of these conflicts was
practice involves similar doses with IVIG and SCIG in most cases of current at the time of writing this paper.
patients with neuropathies [59]. The US FDA’s recommendation to
correct the dosage of immunoglobulin when transitioning PID patients
ORCID
from IVIG to SCIG is not reflected in the case of CIPD [60].
Albert Farrugia https://orcid.org/0000-0003-0804-7463
Estimation of LTD for other groups of indications, notably condi-
tions such as idiopathic thrombocytopenic purpura and immunodefi-
RE FE RE NCE S
ciency secondary to haematological malignancies, will increase insight
1. Dalakas MC. Use of intravenous immunoglobulin in neurology. In:
in the factors influencing demand for these conditions. In the interim, Imbach P, editor. Antibody therapy: Substitution – Immuno-
our study suggests that the underlying demand for IG in the main neu- modulation – Monoclonal immunotherapy. Cham: Springer Interna-
rological diseases approaches 125 g/1000 inhabitants. In Table 5, we tional Publishing; 2018. p. 101–9. https://doi.org/10.1007/978-3-
list the LTD for immunoglobulin consumption for the indications 319-68038-5_8.
2. Grifols. Investor and Analyst Meeting 2019 [Internet]; 2019. https://
where this has been studied [7] (this study). This suggests that approx-
www.grifols.com/documents/51507592/90066043/IAD+2019
imately 75% of the amount of IG used currently in the countries using +Grifols.pdf/e2505992-1b9c-4168-9f08-b68a75170c4c, Accessed
high amounts IG, including the United States, Canada and Australia February 12, 2020
[61], represents the consumption expected from decision analysis 3. Center for Biologics Evaluation and Research. Information About
Immune Globulin (Human) Product Shortage. FDA [Internet]; 2019.
modelling for these two groups of indications. As diagnosis and treat-
http://www.fda.gov/vaccines-blood-biologics/safety-availability-
ment access improve in other, lower consuming countries, the actual biologics/information-about-immune-globulin-human-product-shortage,
usage is expected to reflect the levels estimated in these models. This Accessed February 2, 2020
218 FARRUGIA ET AL.

4. Loftus P. Drug shortage leaves patients without immune-disorder polyradiculoneuropathy (ICE study): A randomised placebo-
treatment. Wall Street Journal; 2019. https://www.wsj.com/articles/ controlled trial. Lancet Neurol. 2008;7:136–44.
drug-shortage-leaves-patients-without-immune-disorder-treatment- 20. Australian National Blood Authority. Chronic inflammatory demyelin-
11565343023, Accessed December 2, 2020 ating polyneuropathy - dosage of immunoglobulin; 2014 Available
5. Farrugia A, Grazzini G, Quinti I, Candura F, Profili S, Liumbruno GM. from https://www.blood.gov.au/pubs/ivig/conditions-for-which-
The growing importance of achieving national self-sufficiency in IVIg-has-an-established-therapeutic-role.html#cdn-03, Accessed
immunoglobulin in Italy. The emergence of a national imperative. February 12, 2020
Blood Transfus. 2019;17:449–58. 21. Australian National Blood Authority. Multifocal motor neuropathy –
6. Stonebraker JS, Farrugia A, Gathmann B, ESID Registry Work- Dosage of immunoglobulin; 2014.
ing Party, Orange JS. Modeling primary immunodeficiency dis- 22. Jinka M, Chaudhry V. Treatment of multifocal motor neuropathy.
ease epidemiology and its treatment to estimate latent Curr Treatment Opt Neurol. 2014;16. https://doi.org/10.1007/
therapeutic demand for immunoglobulin. J Clin Immunol. 2014; s11940-013-0269-y.
34:233–44. 23. Winston W. Microsoft Excel data analysis and business modeling.
7. Stonebraker JS, Hajjar J, Orange JS. Latent therapeutic demand 6th ed. Redmond, Washington: Pearson; 2019.p. 717–26 https://
model for the immunoglobulin replacement therapy of primary support.microsoft.com/en-us/office/introduction-to-monte-carlo-
immune deficiency disorders in the USA. Vox Sang. 2018;113: simulation-in-excel-64c0ba99-752a-4fa8-bbd3-4450d8db16f1.
430–40. 24. Thakur M. Tornado Chart in Excel j Step by Step Examples to Create
8. Gorson KC. An update on the management of chronic inflammatory Tornado Chart [Internet]. WallStreetMojo; 2019. https://www.
demyelinating polyneuropathy. Ther Adv Neurol Disord. 2012;5: wallstreetmojo.com/tornado-chart-in-excel/, Accessed April 14,
359–73. 2021
9. Hahn AF, Bolton CF, Pillay N, Chalk C, Benstead T, Bril V, et al. 25. Cynthia Counter. Probability histogram with Excel simple [Internet];
Plasma-exchange therapy in chronic inflammatory demyelinating 2015. https://www.youtube.com/watch?v=LOA3kKOnUH0, Accessed
polyneuropathy. A double-blind, sham-controlled, cross-over study. April 14, 2021
Brain. 1996;119(Pt 4):1055–66. 26. Farrugia A, Bansal M. Model construction - Latent Therapeutic
10. Stump E. FDA approves immune globulin product for CIDP. Neurol- Demand for Immunoglobulin in neuropathies [Internet]. 2021.
ogy Today. 2008;8:1. https://www.researchgate.net/project/Latent-Therapeutic-Demand-
11. Center for Biologics Evaluation and Research. Immune Globulins. for-Immunoglobulin-in-neuropathies, Accessed April 18, 2021
FDA; 2020. https://www.fda.gov/vaccines-blood-biologics/ 27. Laughlin RS, Dyck PJ, Melton LJ, Leibson C, Ransom J, Dyck PJB.
approved-blood-products/immune-globulins, Accessed December Incidence and prevalence of CIDP and the association of diabetes
2, 2020 mellitus. Neurology. 2009;73:39–45.
12. European Medicines Agency. Guideline on the clinical investigation 28. Lunn MPT, Manji H, Choudhary PP, Hughes RAC, Thomas PK.
of human normal immunoglobulin for intravenous administration Chronic inflammatory demyelinating polyradiculoneuropathy: A
(IVIg) EMA/CHMP/BPWP/94033/2007 rev. 4; 2020. https://www. prevalence study in south East England. J Neurol Neurosurg Psychia-
ema.europa.eu/en/documents/scientific-guideline/draft-guideline- try. 1999;66:677–80.
clinical-investigation-human-normal-immunoglobulin-intravenous- 29. Mcleod JG, Pollard JD, Macaskill P, Mohamed A, Spring P,
administration-ivig_en-1.pdf Khurana V. Prevalence of chronic inflammatory demyelinating poly-
13. European Medicines Agency. Guideline on core SmPC for human neuropathy in New South Wales, Australia. Ann Neurol. 1999;46:
normal immunoglobulin for intravenous administration (IVIg); 2018. 910–13.
https://www.ema.europa.eu/en/documents/scientific-guideline/ 30. Mygland A, Monstad P. Chronic polyneuropathies in Vest-Agder,
guideline-core-smpc-human-normal-immunoglobulin-intravenous- Norway [Internet]. Vol. 8, European journal of neurology. Eur J Neu-
administration-ivig-rev-5_en.pdf rol. 2001;8(2):157–65.
14. Oczko-Walker M, Manousakis G, Wang S, Malter JS, 31. US National Library of Medicine. Guilliann Barre syndrome.
Waclawik AJ. Plasma exchange after initial intravenous immuno- Bethesda, MD: National Institutes of Health; 2011 http://ghr.nlm.
globulin treatment in Guillain-Barré syndrome: Critical nih.gov/condition/guillain-barre-syndrome.
reassessment of effectiveness and cost-efficiency. J Clin 32. Lawson VH, Arnold, WD. Multifocal motor neuropathy: A review of
Neuromuscul Dis. 2010;12:55–61. pathogenesis, diagnosis, and treatment. Neuropsychiatr Dis Treat.
15. van Doorn PA, Kuitwaard K, Walgaard C, van Koningsveld R, Ruts L, 2014;10:567–76.
Jacobs BC. IVIG treatment and prognosis in Guillain-Barré syndrome. 33. Darabi K, Abdel-Wahab O, Dzik WH. Current usage of intravenous
J Clin Immunol. 2010;30(suppl 1):S74–8. immune globulin and the rationale behind it: The Massachusetts
16. Patwa HS, Chaudhry V, Katzberg H, Rae-Grant AD, So YT. Evidence- General Hospital data and a review of the literature. Transfusion.
based guideline: Intravenous immunoglobulin in the treatment of 2006;46:741–53.
neuromuscular disorders: Report of the therapeutics and technology 34. Sarti L, Falai T, Pinto F, Tendi E, Matà S. Intravenous immune globulin
assessment Subcommittee of the American Academy of neurology. usage for neurological and neuromuscular disorders: An academic
Neurology. 2012;78:1009–15. centre, 4 years experience. Neurol Sci. 2009;30:213–8.
17. Van den Berg-Vos R, Franssen H, Wokke J, Van Den Berg L. 35. Querol L, Rojas-Garcia R, Casasnovas C, Sedano MJ, Muñoz-
Multifocal motor neuropathy: Long term clinical and electrophysio- Blanco JL, Alberti MA, et al. Long-term outcome in chronic inflamma-
logical assessment of intravenous immunoglobulin maintenance tory demyelinating polyneuropathy patients treated with intravenous
treatment. Brain. 2002;125:1875–86. immunoglobulin: A retrospective study. Muscle Nerve. 2013;48:
18. Stonebraker JS, Amand RE, Bauman MV, Nagle AJ, Larson PJ. Model- 870–6.
ling haemophilia epidemiology and treatment modalities to estimate 36. National Blood Authority of Australia. Criteria for clinical use of
the unconstrained factor VIII demand. Haemophilia. 2004;10:18–26. immunoglobulin in Australia; 2012 https://www.criteria.blood.gov.
19. Hughes RAC, Donofrio P, Bril V, Dalakas MC, Deng C, Hanna K, au/CheckEligibility, Accessed October 7, 2019
et al. Intravenous immune globulin (10% caprylate-chromatography 37. Eftimov F, Winer JB, Vermeulen M, de Haan R, van Schaik IN. Intra-
purified) for the treatment of chronic inflammatory demyelinating venous immunoglobulin for chronic inflammatory demyelinating
ESTIMATION OF THE LATENT THERAPEUTIC DEMAND FOR IMMUNOGLOBULIN THERAPIES 219
IN AUTOIMMUNE NEUROPATHIES IN THE UNITED STATES

polyradiculoneuropathy. Cochrane Database Syst Rev. 2013;12: 53. Cea G, Idiaquez JF, Salinas R, Matamala JM, Villagra R, Stuardo A.
CD001797. Epidemiology of chronic inflammatory demyelinating polyneuropathy
38. den Bergh PYKV, Hadden RDM, Bouche P, Cornblath DR, Hahn A, in the south-eastern area of Santiago. Chile J Clin Neurosci. 2020;
Illa I, et al. European Federation of Neurological Societies/Peripheral 74:271–3.
Nerve Society Guideline on management of chronic inflammatory 54. Australian Medical Services Advisory Committe C of
demyelinating polyradiculoneuropathy: Report of a joint task force A. Immunoglobulin for chronic Inflammatory Demyelinating Poly-
of the European Federation of Neurological Societies and the Periph- neuropathy (CIDP). MSAC Application 1564, Assessment Report.
eral Nerve Society – First Revision. Eur J Neurol. 2010;17:356–63. [Internet]. 2019 Nov. Report No.: MSAC application no. 1564.
39. National Blood Authority (Australia). Criteria for the clinical use of http://www.msac.gov.au/internet/msac/publishing.nsf/Content/
intravenous immunoglobulin in Australia; 2012. 2B740EDAE80B75DFCA25837E00096D86/$File/1564_MSAC_
40. Van den Berg LH. The Long Term Treatment of Multifocal Motor Assessment%20Report_Redacted.pdf
Neuropathy with Intravenous Immunoglobulin; 2014. (European 55. American Academy of Neurology. Research criteria for diagnosis of
Neurological Review). chronic inflammatory demyelinating polyneuropathy (CIDP). Report
41. Agency for Health care Research and Quality. HCUP-US NIS over- from an ad hoc Subcommittee of the American Academy of neurol-
view; 2020. https://www.hcup-us.ahrq.gov/nisoverview.jsp, Accessed ogy AIDS task force. Neurology. 1991;41:617–8.
December 8, 2020 56. Rajabally YA, Simpson BS, Beri S, Bankart J, Gosalakkal JA. Epidemio-
42. Azulay JP, Rihet P, Pouget J, Cador F, Blin O, Boucraut J, et al. Long logic variability of chronic inflammatory demyelinating poly-
term follow up of multifocal motor neuropathy with conduction neuropathy with different diagnostic criteria: Study of a UK population.
block under treatment. J Neurol Neurosurg Psychiatry. 1997;62: Muscle Nerve. 2009;39:432–8.
391–4. 57. Allen JA, Lewis RA. CIDP diagnostic pitfalls and perception of treat-
43. Harbo T, Andersen H, Hess A, Hansen K, Sindrup SH, Jakobsen J. ment benefit. Neurology. 2015;85:498–504.
Subcutaneous versus intravenous immunoglobulin in multifocal 58. Allen JA, Gelinas DF, Freimer M, Runken MC, Wolfe GI. Immuno-
motor neuropathy: A randomized, single-blinded cross-over trial. Eur globulin administration for the treatment of CIDP: IVIG or SCIG?
J Neurol. 2009;16:631–8. J Neurol Sci. 2020;408:116497.
44. Fryar GO, Gu, Q, Ogden, CL. Anthropometric reference data for chil- 59. Racosta JM, Sposato LA, Kimpinski K. Subcutaneous versus intrave-
dren and adults: United States, 2007-1010. Vital Health Stat. 2012; nous immunoglobulin for chronic autoimmune neuropathies: A meta-
11:1–48. analysis. Muscle Nerve. 2017;55:802–9.
45. Woodruff J, Dimitrov NB. Optimal discretization for decision analy- 60. Behring CSL. Immune Globulin Subcutaneous (Human) (IGSC), 20%
sis. Oper Res Persp. 2018;5:288–305. Liquid, Hizentra - Package insert; 2018
46. Živkovic S. Intravenous immunoglobulin in the treatment of neuro- 61. Government of Canada PS and PC. Protecting access to immune
logic disorders. Acta Neurol Scand. 2016;133:84–96. globulins for Canadians: Final report of the expert panel on immune
47. Karelis G, Balasa R, De Bleecker JL, Stuchevskaya T, Villa A, Van globulin product supply and related impacts in Canada; 2018. http://
Damme P, et al. A phase 3 multicenter, prospective, open-label effi- publications.gc.ca/site/eng/9.854785/publication.html. Accessed
cacy and safety study of immune globulin (human) 10% Cap- October 8, 2019
rylate/chromatography purified in patients with myasthenia gravis 62. Kerr J, Quinti I, Eibl M, Chapel H, Späth PJ, Sewell WAC, et al. Is dos-
exacerbations. ENE. 2019;81:223–30. ing of therapeutic immunoglobulins optimal? A review of a three-
48. Gajdos P, Chevret S, Toyka KV. Intravenous immunoglobulin for decade long debate in Europe. Front Immunol. 2014;5:629.
myasthenia gravis. Cochrane Database of Systematic Reviews; 2012. 63. Bonagura VR. Illustrative cases on individualizing immunoglobulin
https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858. therapy in primary immunodeficiency disease. Ann Allergy Asthma
CD002277.pub4/full, Accessed April 16, 2021 Immunol. 2013;111:S10–3.
49. Guptill JT, Soni M, Meriggioli MN. Current treatment, emerging 64. Prevot J, Jolles S. Global immunoglobulin supply: Steaming
translational therapies, and new therapeutic targets for autoimmune towards the iceberg? Curr Opin allergy Clin Immunol. 2020;20(6):
myasthenia gravis. Neurotherapeutics. 2016;13:118–31. 557–64.
50. Broers MC, Bunschoten C, Nieboer D, Lingsma HF, Jacobs BC. Inci- 65. Patel DD. Neonatal Fc receptor in human immunity: Function and
dence and prevalence of chronic inflammatory demyelinating poly- role in therapeutic intervention. J Allergy Clin Immunol. 2020;
radiculoneuropathy: A systematic review and meta-analysis. NED. 146:12.
2019;52:161–72.
51. Bril V, Blanchette CM, Noone JM, Runken MC, Gelinas D,
Russell JW. The dilemma of diabetes in chronic inflammatory
demyelinating polyneuropathy. J Diabetes Complications. 2016; How to cite this article: Farrugia A, Bansal M, Marjanovic I.
30:1401–7. Estimation of the latent therapeutic demand for
52. Iijima M, Koike H, Hattori N, Tamakoshi A, Katsuno M, Tanaka F, immunoglobulin therapies in autoimmune neuropathies in the
et al. Prevalence and incidence rates of chronic inflammatory demye-
United States. Vox Sang. 2022;117:208–219.
linating polyneuropathy in the Japanese population. J Neurol Neuro-
surg Psychiatry. 2008;79:1040–3.
Received: 15 January 2021 Revised: 2 June 2021 Accepted: 3 June 2021

DOI: 10.1111/vox.13170

ORIGINAL ARTICLE

Therapeutic granulocyte infusion for patients with severe

neutropaenia and neutrophilic dysfunction: New Zealand
experience

Shanee Chung1 | Oliver Armstrong-Scott2 | Richard Charlewood1

1
New Zealand Blood Service, Auckland,
New Zealand Abstract
2
Department of General Medicine, Auckland Background and Objectives: Studies have shown granulocyte transfusions (GTXs)
District Health Board, Auckland, New Zealand
may be beneficial in neutropaenic patients with severe systemic infections.
Correspondence New Zealand Blood Service has a policy for provision of granulocytes to
Shanee Chung, The Leukemia/BMT Program
New Zealand’s District Health Boards. We set out to explore utilization of therapeu-
of B.C., 2775 Laurel Street, Floor
10, Vancouver, BC V5Z 1G1, Canada. tic granulocyte infusions in New Zealand.
Email: [email protected]
Materials and Methods: Patients who received GTXs in the 16-year period between
2000 and 2016 were identified by the New Zealand electronic blood management
system, eProgesa. Information pertaining to recipient demographics, disease-related
factors, methods of granulocyte collection and clinical outcomes was obtained by the
review of electronic transfusion and clinical records.
Results: Forty-five septic patients received granulocyte support for a total of
263 days. The median age of the recipients was 16 (range 0–74) years. Seventy-nine
percent of the recipients had an underlying haematological malignancy with 50%
having acute leukaemia. The median neutrophil count on the last day of GTX was
0.02  109/L (range 0–16.32). Sixty-three percent (27/43 patients with available
data) had persisting severe neutropaenia when the GTXs were stopped. The median
duration of support was 3 (range 1–32) days. Forty-six percent of granulocyte collec-
tions were performed via apheresis. Of the 44 patients, for whom survival outcome
was available, 18 (41%) survived the acute illness.
Conclusion: GTXs were infrequently used, most commonly in the setting of an
underlying haematological malignancy. This may be explained by the current weak
evidence base supporting this therapeutic modality. Procuring a sufficiently large
dose of granulocytes for infusion remains an issue for adult recipients.

KEYWORDS
apheresis – donation, blood components, granulocyte concentrate

I N T R O D U CT I O N prolonged period of severe neutropaenia, which predisposes them to

invasive bacterial and fungal infections that are associated with high
Cancer patients receiving intensive chemotherapy or those undergo- morbidity and mortality [1]. The strongest predictor of survival from
ing haematopoietic stem cell transplantation (SCT) often experience a these infections is improvement of the neutrophil count [2].

220 © 2021 International Society of Blood Transfusion. wileyonlinelibrary.com/journal/vox Vox Sanguinis. 2022;117:220–226.
THERAPEUTIC GRANULOCYTE INFUSION IN SEVERE NEUTROPAENIA/NEUTROPHILIC DYSFUNCTION 221

Granulocyte transfusion (GTX) has been around for many New Zealand’s policy and procedures for GTX
decades with the first documented GTX in 1934 [3]. Significant
enthusiasm developed for GTX in the 1970s, but controlled trials in NZBS provides a ‘vein-to-vein’ transfusion service for the nation, and
the 1970s and 1980s yielded mixed results [4], which were partially it is responsible for provision of all blood products. The national policy
attributed to differences in qualities and doses of the granulocytes for provision of therapeutic granulocyte products states that GTX
transfused. There was resurgence of interest in clinical application of may be considered in patients with
GTXs in the 1990s with the advent of leukapheresis and clinical utili-
zation of granulocyte colony stimulating factor (G-CSF). Granulocyte 1. persistent neutropaenia of less than 0.2  109/L, which is
collection without stimulation of the donor may yield 0.1–1  10 10
expected to persist for longer than 5 days,
granulocytes, but this yield can increase to 4–8  1010 cells after 2. either septicaemia or life-threatening local infection that is not
stimulation with G-CSF and steroids [5]. A study has found that responding to 72 h of appropriate antimicrobial therapy or proven
granulocytes obtained from donors stimulated by G-CSF retained or probable fungal or yeast infection that is refractory to appropri-
immunological functions such as chemotaxis, respiratory burst, adhe- ate antifungal therapy,
sion and bactericidal and antifungal activity, which remained 3. good long-term prognosis from the underlying disorder and
unchanged even after storage of the granulocyte product for 4. a suitable donor.
24 hours [6].
The few published multicentre randomized controlled trials for Contraindications include poor long-term prognosis of the
GTX suffered from poor recruitment, and they were not able to underlying disorder, known as human leukocyte antigen (HLA)
show clinical benefit in the intervention arm [7, 8]. However, the alloimmunization (relative), severe respiratory compromise and
RING study did show the subjects who received a higher dose of requirement for ventilatory support.
granulocytes (mean dose greater than 0.6  109/kg per transfusion) The granulocyte products can be provisioned within 24 h of a
had superior outcomes in a subgroup analysis [8]. The latest request through the on-call Transfusion Medicine Specialist on 6 days
Cochrane review recommends regarding the use of GTX as investi- of the week, from Monday to Saturday. A donor search is made on a list
gational, given the lack of sufficient evidence from randomized con- of ABO- and RhD-compatible NZBS apheresis donors. Family members
trolled trials to support or refute its use in patients with of the patient, who meet requirements of the NZBS Collection Stan-
neutropaenia and severe infection to reduce mortality [9]. In the dards as evaluated by a NZBS Medical Officer, may become donors,
recent years, several single centres published retrospective analyses but directed family donation is relatively contraindicated if allogeneic
of their real-life experience with GTX therapy. In one such study, SCT is being considered due to risk of HLA alloimmunization. Given the
providing GTX to severe neutropaenic patients with treatment- limited size of the donor pool, granulocytes are only matched for ABO
refractory, life-threatening infections resulted in 15/27 (56%) and RhD types, unless the recipient is known to have HLA antibodies.
patients surviving to hospital discharge [10]. In another, 18/22 Apheresis units from single donors expose the recipients to a limited
(81.8%) patients with neutropaenia due to haematological malig- number of HLA antigens compared to buffy coat units from several
nancy survived severe refractory abdominal infection when treated donors. Collection of an apheresis unit, however, requires priming of the
with GTX until recovery of absolute neutrophil count (ANC >1 donor with recombinant G-CSF (5–10 μg/kg) with or without dexametha-
 109/L) or significant clinical improvement. The patients achieving sone (8 mg) in the evening prior to the collection date, whereas buffy coat
control of the infection within 7 days of the first GTX had signifi- units are routinely set aside for platelet pooling at NZBS collection sites.
cantly better overall survival (p < 0.001) [11]. Nguyen et al. described Consequently, buffy coat units are used for GTX if an apheresis donation
use of GTX as a bridge to an urgent allogeneic SCT in 19 severe neu- cannot be arranged in a timely manner. For procurement of buffy coat
tropaenic patients with severe uncontrolled infection, where 90% of units, whole blood donations, collected into ‘top-and-bottom’ bags, are
the GTX recipients were able to proceed to SCT with 80% continu- centrifuged at 3616g (using a Heraeus 6000i centrifuged at 3300 rpm for
ing on GTX support until neutrophil engraftment. Following the SCT, 11 min) at room temperature, and the bottom red cell layer and the top
10% of the patients eventually succumbed to the initial infection, for plasma layer are removed using a Macopress automated blood component
which they received GTX. They showed an association between separator. The remaining buffy coat is transferred to a bag that is suitable
delay in provision of GTX and delay in proceeding to HCT for transfusion. For apheresis collection, NZBS currently uses Spectra
(p < 0.0001), suggesting a potential role for GTX in facilitating an Optia®, which utilizes continuous-flow centrifugation and optical detection
urgent SCT [12]. In the current multicentre retrospective observa- technology. Other machines that had been in use over the preceding years
tional study, we report on New Zealand’s real-life experience in use include Haemonetics MCS+® and Fresenius Kabi’s COM.TEC®.
of GTX for treating paediatric and adult patients with severe infec- All granulocyte products are irradiated prior to infusion to prevent
tion in the setting of neutropaenia or neutrophil dysfunction associ- transfusion-associated graft-versus-host disease. The requirements
ated with wide ranging haematological and non-haematological for an adequate buffy coat unit are volume of 35–65 ml and granulo-
disorders over 16 years. This project was carried out as a quality cyte and platelet contents of ≥1  109 and ≥5  1010 per unit,
improvement audit for New Zealand Blood Service (NZBS), and an respectively. The volume of an apheresis unit is defined locally, and it
ethics application was not required as per the local guidelines. is between 200 and 500 ml in practice. Each apheresis unit should
222 CHUNG ET AL.

contain ≥1  1010 of granulocytes. For quality control, each donation RE SU LT S

of granulocyte products is tested for volume and granulocyte content,
and these data are entered into a statistical process control system, Between 2000 and 2016, 45 patients received GTX for a total of
NWA, to identify trends and outliers. The collected granulocytes must 263 days in New Zealand. Twenty-eight (62%) patients were females.
be transfused as soon as possible but can be stored for up to 24 h The median age of the recipients was 16 with a range between zero
from collection if they are kept at 20–24 C without agitation. and 74 years (Figure 1). Fifty-six percent of the recipients had under-
lying acute leukaemia, and another 22% had other haematological
malignancies including large granulocytic leukaemia (two patients),
METHODS chronic lymphocytic leukaemia (one patient), classic Hodgkin lym-
phoma (one patient), Burkitt lymphoma (one patient), chronic myeloid
An electronic search was performed on eProgesa, the electronic blood leukaemia (one patient), Juvenile myelomonocytic leukaemia (one
management system used by NZBS since year 2000, to identify all patient) and T-cell lymphoproliferative disorder, NOS (one patient).
recipients of buffy coat and/or apheresis granulocyte units during the Four patients had underlying aplastic anaemia. Four recipients did not
16-year period between 2000 and 2016. We collected data related to have a primary haematological diagnosis. These included two paediat-
the recipient’s demographic features, the underlying disease status ric patients with chronic granulomatous disease leading to granulocyte
(primary diagnosis, indication for GTX), the intervention (method of dysfunction and two patients with solid organ malignancy (Figure 2).
procurement of granulocyte products, number of infusions given per All patients had documented bacterial or fungal infections that were
patient, dose of granulocytes given per patient) and the clinical out- refractory to conventional therapy. None received GTX as a second-
come (neutrophil count post-GTX, survival). The Transfusion Nurse ary prophylaxis to prevent progression or recurrence of chronic
Specialists working for the eight largest District Health Boards (DHBs) infection.
around New Zealand reviewed the electronic records of the GTX Eleven out of the 20 DHBs in New Zealand have a dedicated
recipients belonging to their respective DHBs and other smaller DHBs Haematology Department, and the remaining DHBs are served by
supported by their DHBs and collected the pre-specified set of data. neighbouring DHBs with a Haematology Service. The majority of the
The authors analysed the data and prepared the manuscript. GTXs (21 patients, 47%) were carried out by Starship Hospital,
the largest tertiary paediatric service in New Zealand. Auckland City
Hospital, the adult equivalent, was the second biggest user of GTX
STATISTICAL ANALYSIS during the study period, but their usage was much less than that of
Starship Hospital, with only six cases comprising 13% of the total
Pearson’s chi-squared test with Yates’ continuity correction was used number. This was slightly more than that seen in the rest of the coun-
for the correlation between increment in the neutrophil count during try. Other haematology centres in both North and South Islands of
treatment and survival to the time of discharge from hospital. Kaplan– New Zealand had one to four cases of GTX therapy in the 16-year
Meier analysis was used to estimate survival. The statistical software period. There did not appear to be any apparent regional variation in
used was R (R Foundation for Statistical Computing, Vienna, 2018). the clinical practice; for example, Wellington Hospital Haematology

FIGURE 1 Age distribution of recipients. This graph illustrates the number of recipients of therapeutic granulocyte infusion in each age
bracket
THERAPEUTIC GRANULOCYTE INFUSION IN SEVERE NEUTROPAENIA/NEUTROPHILIC DYSFUNCTION 223

F I G U R E 2 Underlying primary diagnoses of granulocyte transfusion (GTX) recipients. This graph illustrates the underlying primary diagnoses
of the patients who received therapeutic granulocyte infusion for neutropaenia or neutrophil dysfunction in this series

Centre, serving Capital and Coast, Hutt Valley and Wairarapa DHBs in neutropaenia and invasive pulmonary aspergillosis. Despite receiving
the lower end of the North Island (a population of 525,910 people by 23 therapeutic doses of buffy coats and 14 apheresis units of GTX,
the Ministry of Health 2020/21 estimate), used GTX on four occa- she eventually succumbed to the acute infection. Her WBC at the
sions, whereas Dunedin Hospital Haematology Department, serving commencement of GTX was 0.14  109/L, and the peak WBC during
the lower end of the South Island (a population of 344,900 people by the course of GTX was 0.44  109/L (no differentials available).
the same estimate) used it for three patients. The survival outcome data were available for all but one recipient.
Thirty-four out of the 40 recipients with available data (85%) had Twenty-six (58%) recipients were alive at the time of discharge from
severe neutropaenia (defined as a neutrophil count of less than 0.5  hospital. Eighteen (40%) died from severe infection.
9
10 /L) at baseline and three (7.5%) had moderate neutropaenia The information on the granulocyte dose per kilogram body
(defined as a neutrophil count between 0.5 and 1  109/L). Four of weight was available for a subset of 30 patients. Fifteen patients
the five patients with missing data had only the total white blood cell received 0–100  108/kg of granulocytes, and 40% of them survived
(WBC) count without differentials reported on the day. Three of the the acute infection. Of the 11 patients who received 100–200  108/
44 patients with available data (6.8%) had a neutrophil count above 1 kg of granulocytes, 55% survived. All four patients who received
 109/L. Only one patient had a missing data for this latter category greater than 200  108/kg of granulocytes were alive at the time of
because all of the patients who had a WBC reported without differen- discharge from hospital (p = 0.1014).
tials had a WBC of less than 1. The patients with a starting neutrophil Based on the available data, the post-GTX neutrophil count could
count of 1  10 /L or greater included a patient with an underlying
9
be determined for a subset of 42 patients. During the time interval
chronic granulomatous disease (ANC 16.32  109/L), a paediatric between the first GTX and the day after the last GXT, 27 (64%)
acute leukaemia patient with typhlitis and abdominal wall cellulitis patients had a rise in their neutrophil count of 0.2  109/L or greater,
(ANC 3.29  10 /L) and an adult patient with a bone marrow failure
9
and 14 (52%) of these patients survived to hospital discharge. Out of
syndrome and Escherichia coli bacteraemia (ANC 1.1  10 /L). The 9
the 15 patients who had less than 0.2  109/L increase in their neu-
median neutrophil count at baseline was 0.02  10 /L with a range
9
trophil count, only five (33%) survived (p = 0.4055). All but six
between 0 and 16.32  109/L. patients had an increment in their neutrophil count at one or more
One hundred thirty-nine (139/303, 45.9%) therapeutic doses of measurements while receiving GTX. Of the six patients who failed to
granulocytes were collected via apheresis, and 164 (164/303, 54.1%) have any rise in their neutrophil count during the treatment course,
doses were derived from buffy coats (10–12 buffy coats per one ther- only one patient (17%) was alive at the time of discharge from hospi-
apeutic dose). The median number of GTXs given per patient was four tal, in comparison to 18 out of the 36 patients (50%) who had at least
with a range between one and 37 doses (Figure 3). The patient who a transient rise in their neutrophil count (p = 0.282).
received the highest cumulative dose of GTX was a 7-year-old female Eight GTX incidents were associated with a reported transfusion
with a brain stem glioma, who had chemotherapy-induced reaction according to the NZBS haemovigilance record. The
224 CHUNG ET AL.

F I G U R E 3 Number of therapeutic granulocyte doses given per patient. This graph illustrates distribution of the number of therapeutic doses
of granulocytes given per patient. Each therapeutic dose of granulocytes consists of either one apheresis unit or 10–12 buffy coats

commonest transfusion reaction was mild to moderate allergic reac- patients receiving GTX in each. The number of cases varied between
tion (four patients; two patients with buffy coat units and two one and four in other years, except for in 2002, when GTX was not given
patients with apheresis units). Two patients experienced non- at all. This is not surprising as there has not been any groundbreaking
haemolytic febrile transfusion reaction (one with buffy coat unit and improvement in the process of granulocyte collection or a major publica-
one with apheresis unit) and one patient developed transfusion- tion providing stronger evidence for benefit of GTX over this time. The
associated circulatory overload after receiving an apheresis unit. majority of the recipients were paediatric patients located at Starship
There was one reported case of transfusion-associated polycythaemia Hospital, the country’s largest tertiary children’s hospital. This may be
in a paediatric patient who received a cumulative sum of 23 buffy coat partially because GTX is perceived to be more efficacious in paediatric
units and 14 apheresis units over the treatment course. patients, who are able to receive a higher number of granulocytes per
kilogram body weight due to their smaller size. It may also be because
doctors and the parents, who often are the surrogate decision-makers
DISCUSSION for paediatric patients, are more likely treat the very young patients
aggressively and/or explore alternative therapeutic options when con-
In our retrospective study, we reviewed the GTX usage in ventional measures are failing. Cost may also be an issue favouring pae-
New Zealand over a 16-year period between 2000 and 2016 and diatric recipients. Although all blood components are free of charge to
explored the associated clinical practice and patient outcomes. the patient, NZBS operates on a cost recovery model, with the
Although the NZBS protocol suggests GTX be given to patients with a government-funded DHBs paying for each component. A single aphere-
neutrophil count of less than 0.2  109/L, 11 (24.4%) recipients had sis granulocyte component costs approximately €500, while a buffy coat
a neutrophil count of 0.2  109/L or greater at the time of treatment. from a single whole blood donation costs approximately €145.
This is because NZBS allows the requesting clinicians to make the New Zealand is a small country, hence, therapeutic granulocytes
decision to access GTX support for a septic patient after careful con- can generally be transported from a collection site to where a patient
sideration of the patient’s underlying disease, neutrophil function, is situated by land or air in a timely manner. Provision of all blood
anticipated trajectory of the neutrophil count and severity of the products in New Zealand is coordinated by a single crown-owned,
acute infection, in addition to the absolute neutrophil count. non-profit entity, NZBS, so there are insignificant variations in the
There was little fluctuation in the demand for GTX over the years. cost or logistic complexities related to GTX provision around the
The maximum number of cases was seen in 2008 and 2012, with five country. Consequently, there does not appear to be any regional
THERAPEUTIC GRANULOCYTE INFUSION IN SEVERE NEUTROPAENIA/NEUTROPHILIC DYSFUNCTION 225

variation in the usage of GTX. The overall number of cases has been pressure [8]. In the same study, 20% of the recipients and less than 5% of
low from all parts of the country. This is likely reflective of the clini- the transfusions were associated with grade 3–4 reactions, such as hyp-
cians’ uncertainty or scepticism towards the therapeutic benefit of oxia, tachycardia, hypotension and allergic reaction. There were no deaths
GTX in treating severe infections in immunocompromised patients, attributable to GTX and no significant association between the granulo-
given the lack of clear evidence based on large well-conducted cyte dose administered and the occurrence of a transfusion reaction. In
randomized controlled trials. New Zealand, reporting of transfusion reactions is voluntary, which may
It is interesting to note that at 42 days (the RING trial’s endpoint), lead to under-reporting. Our observed rate of transfusion reactions for
our survival was over 70%, which is in keeping with the RING trial’s fresh components is 2% from audits of over 1000 transfusion episodes,
expected indicator of success [8]. We have failed to demonstrate whereas our reported rate via the haemovigilance program is 0.3%. Trans-
improved survival in patients who received a higher dose of fusion reactions associated with GTX are, however, less likely to have
granulocytes in the subgroup analysis, but it could be attributable to been unreported compared with those associated with other fresh com-
the small sample size and the incomplete data set. We were not able ponents as provision of therapeutic granulocyte products demands close
to determine the granulocyte dose per kilogram body weight for communication among the treating clinicians, the transfusion medicine
patients who received GTX prior to 2008 due to missing data on the specialists and the NZBS staff during the treatment course.
granulocyte content in the infusion products in our current electronic There are potential ethical and safety concerns pertaining to
quality assurance database. We also did not find statistically signifi- exposing healthy volunteer donors to medications, such as G-CSF and
cant difference in survival in patients who had a rise in their neutro- dexamethasone, for mobilization of granulocytes. Quillen et al.
phil count of at least 0.2  109/L compared to those who did not followed 83 out of 92 apheresis granulocyte donors who received
(p = 0.4055), but in order to have a 95% chance of detecting a statisti- three or more doses of G-CSF at 5 mcg/kg and 8 mg of oral dexa-
cally significant difference, around 144 patients were needed in each methasone between 1994 and 2002 for a median follow-up period of
group. Likewise, there was no statistically significant difference in sur- 10 years. They compared the health outcomes of these granulocyte
vival between the patients who had any rise in the neutrophil count donors with matched control platelet donors and found there was no
during the course of GTX and those who did not (p = 0.282), but we difference in the incidence of malignancies, coronary artery disease
would have needed 24 patients in each arm to detect a statistically and thrombosis [13]. To our knowledge, there is no definite evidence
significant difference (50% vs. 16.7%) with 80% certainty. in the current literature that a short-term use of G-CSF and dexa-
GTX has a number of challenges. It is mandatory to have special- methasone leads to any significant long-term morbidity or mortality.
ized expertise and facility, such as leukapheresis and HLA matching, These medications, however, still have potential side effects, such as
to produce the granulocyte products and the manufactured granulo- bone pain, tenderness at the injection site, transient hyperglycaemia
cyte products need to reach the intended recipients within 24 hours. and so forth. Therefore, it is important to counsel the potential donors
This may be particularly challenging in peripheral centres where the carefully before obtaining their informed consent.
manufacturing expertise and facility are not available. The NZBS pro- Improvements in diagnostic strategies, antimicrobial therapy and
tocol considers a minimum therapeutic dose of granulocytes to be supportive care in modern medicine have reduced the perceived need
1  1010, although >2  1010 is desirable. Each bag of buffy coat unit for GTX, and there is lack of strong evidence for GTX in treating neu-
should contain at least 1  109 of granulocytes, and at least 10 bags tropaenic patients with severe infection; a limited number of publi-
of buffy coat are given per infusion to achieve the therapeutic dose. shed randomized trials were underpowered to demonstrate potential
Currently, there is no formal requirement to adjust the number of benefit of GTX owing to small sample sizes and slow recruitment. As
buffy coat units to be given based on the patient’s weight although the latest 2015 Cochrane review has concluded, there is also not
this does happen in practice. The recommended granulocyte dose enough evidence to refute the benefit of GTX, and there are case
from the 2009 Cochrane review is 3  108/kg, which equates to 2.1 series and anecdotal evidence that show potential benefit of GTX;
 10 10
per infusion for a 70 kg individual. As the average weight of hence, it seems premature and scientifically unjustified to dismiss
the population increases in the current obesity pandemic, provision of therapeutic utility of GTX entirely [9, 14]. We believe GTX should be
a sufficient dose of therapeutic granulocytes will likely become more considered in critically unwell patients with severe neutropaenia or
and more challenging. neutrophil dysfunction who fail to respond to conventional antimicro-
There are concerns over treatment-related adverse effects, particu- bial or antifungal therapy and surgical measures to control an infective
larly pulmonary events, CMV transmission and HLA alloimmunization. source but otherwise have good long-term prognosis from their
GTX appears safe in our study. During the 16-year period where underlying diagnoses. Without firm evidence directing the clinicians
45 patients received a total of 303 therapeutic doses, only eight transfu- and patients in one way or another, full discussion should be held
sion incidents were reported to have caused a transfusion reaction. The with regards to the potential benefits and risks of GTX, and the con-
commonest transfusion reaction was an allergic reaction, and none was troversial nature of the treatment before a consensus decision is
life-threatening. This is significantly less than what was reported in the reached for each patient. It is essential that such a decision is made
RING study, where 41% of the 114 participants and 28% of the transfu- promptly so that the search for suitable donors can begin without
sions were associated with grade 1–2 transfusion reactions, the com- delay and GTX can be delivered in a timely manner at an appropriate
monest of which were fever, chills and/or modest changes in the blood dose, before the patient is in extremis.
226 CHUNG ET AL.

Our study has a number of limitations. It is a retrospective data 2. Gerson SL, Talot GH, Hurwitz S, Strom BL, Lusk EJ, Cassileth PA.
analysis, and the data quality was dependent on the availability and Prolonged granulocytopenia: the major risk factor for invasive pul-
monary aspergillosis in patients with acute leukaemia. Ann Intern
completeness of historic documentation. The number of subjects
Med. 1984;100:345–51.
included in this study is small although we were able to capture all 3. Strumia MM. The effect of leukocytic cream injections in the treat-
patients who received GTX during the study period by virtue of the ment of the neutropenias. Am J Med Sci. 1934;187:527–44.
presence of a single national electronic platform for patient blood 4. Strauss RG. Therapeutic granulocyte transfusions in 1993. Blood.
1993;81:1675–8.
management. There is no comparison group because we anticipated it
5. Price TH. Granulocyte transfusion in the G-CSF era. Int J Hematol.
would be difficult to define a matched historic control group as the 2002;76:77–80.
GTX recipients included in this analysis were a heterogeneous group 6. Drewniak A, Boelens JJ, Vrielink H, Tool ATJ, Bruin MCA, van den
with varying underlying diagnoses, different types of infections and a Heuvel-Eibrink M, et al. Granulocyte concentrates: prolonged func-
tional capacity during storage in the presence of phenotypic changes.
wide range of baseline neutrophil counts.
Haematologica. 2008;93:1058–67.
In order to prove or refute the clinical benefit of GTX in septic 7. Seidel MG, Peters C, Wacker A, Northoff H, Moog R, Boehme A,
patients with severe neutropaenia or neutrophil dysfunction, a random- et al. Randomized phase III study of granulocyte transfusions in neu-
ized controlled trial with an adequate power is needed. Such efforts to tropenic patients. Bone Marrow Transplant. 2008;42:679–84.
8. Price TH, Boeckh M, Harrison RW, McCullough J, Ness PM,
date have been hindered by failure to enrol enough participants to
Strauss RG, et al. Efficacy of transfusion with granulocytes from
achieve an adequate power. The next best approach would be to set
G-CSF/dexamethasone-treated donors in neutropenic patients with
up an international prospective registry of therapeutic GTX, so the out- infection. Blood. 2015;126:2153–61.
come data can be collected in a much larger scale. The Biomedical 9. Estcourt LJ, Stanworth S, Doree C, Blanco P, Hopewell S, &
Excellence for Safer Transfusion (BEST) Collective has set up a web- Trivella, M. Granulocyte transfusions for treating infections in
patients with neutropenia or neutrophil dysfunction. Cochrane Data-
based international collaborative registry of GTX [15]. NZBS has been
base Syst Rev. 2015;6:CD005341.
working with this international registry to establish participation since 10. Ediriwickrema A, Virk M, Andrews J. Granulocyte transfusions in a
2018. It is anticipated that as the Registry data matures over time, cohort of neutropenic patients with life-threatening infections and
more information will become available to shed more light on real-life hematologic diseases. Blood. 2019;134:3696.
11. Richard-Carpentier G, Kantarjian HM, Ravandi F, Kadia TM,
experience in therapeutic use of GTX and trend in clinical outcome.
Borthakur G, Garcia-Manero G, et al. Safety and efficacy of non-
irradiated granulocyte transfusions (GTX) in neutropenic patients
ACKNOWLEDGEMEN TS with severe or refractory abdominal infections: a single center retro-
The first and last authors designed the study protocol, analyzed the spective analysis of 119 transfusions in 22 patients. Blood. 2018;
132:3815.
data and wrote the manuscript. The second author contributed in data
12. Nguyen TM, Scholl K, Idler I, Wais V, Körper S, Lotfi R, et al. Granulo-
collection. We are grateful to our Transfusion Nurse Specialists, Liz cyte transfusions – bridging to allogeneic hematopoietic stem cell
Thrift, Rachel Donegan, Fiona King, Christopher Corkery, Jo Lilley, transplantation. Leuk Lymphoma. 2020;6:481–4.
Jomyn Diedericks, Suzi Rishworth and Graeme Sykes for their assis- 13. Quillen K, Byrne P, Yau YY, Leitman SF. Ten-year follow-up of
unrelated volunteer granulocyte donors who have received multiple
tance in data collection.
cycles of granulocyte-colony-stimulating factor and dexamethasone.
Transfusion. 2009;49:513–8.
CONF LICT OF IN TE RE ST 14. O’Donghaile D, Childs RW, Leitman SF. Blood consult: granulocyte
There is no conflict of interest to declare. This reserach received no transfusions to treat invasive aspergillosis in a patient with severe
aplastic anemia awaiting mismatched hematopoietic progenitor cell
specific grant from any funding agency in the public, commercial, or
transplantation. Blood. 2012;119:1353–5.
not-for-profic sectors.
15. Pagano MB, Morton S, Cohn CS, Gross S, Kutner J, Lewin A, et al. An
international registry of granulocyte transfusions. Transfus Med
ORCID Hemother. 2018;45:318–22.
Shanee Chung https://orcid.org/0000-0002-0375-4858
Richard Charlewood https://orcid.org/0000-0002-1798-1189

How to cite this article: Chung S, Armstrong-Scott O,

RE FE R ENC E S
Charlewood R. Therapeutic granulocyte infusion for patients
1. Bodey GP, Buckley M, Sathe YS, Freireich EJ. Quantitative relation-
with severe neutropaenia and neutrophilic dysfunction: New
ships between circulating leukocytes and infection in patients with
acute leukaemia. Ann Intern Med. 1966;64:328–40. Zealand experience. Vox Sang. 2022;117:220–226.
Received: 4 February 2021 Revised: 29 May 2021 Accepted: 3 June 2021

DOI: 10.1111/vox.13171

ORIGINAL ARTICLE

Early packed red blood cell transfusion in major trauma

patients: Evaluation and comparison of different prediction
scores for massive transfusion

Arnaud Cassignol1 | Pauline Bertein1 | Paul Botti2 | Aurélien Renard3 |

Jean Cotte4 | Julien Bordes4

1
SMUR Department, Sainte-Musse Public
Hospital, Toulon, France Abstract
2
Radiology Department, University Hospital of Background and Objectives: Our study sought to evaluate and compare different
Geneva, Genève, Switzerland
3
prediction scores for massive transfusion in-hospital packed red blood cell (PRBC)
Emergency Department, Sainte-Anne Military
Hospital, Toulon, France transfusions.
4
Anesthesia and Intensive Care Department, Materials and Methods: Between January 2013 and December 2018, 1843 trauma
Sainte-Anne Military Hospital, Toulon, France
patients were enrolled in the registry of a level-1 trauma centre. All prehospital and
Correspondence in-hospital variables needed to calculate the Shock Index and RED FLAG, Assessment
Arnaud Cassignol, 80 rue du Corail, Résidence of Blood Consumption (ABC) and Trauma Associated Severe Hemorrhage (TASH)
Eden Park, Bâtiment Kalmia, 83220, Le
Pradet, France. scores were prospectively collected in the registry. The primary endpoint was the ini-
Email: [email protected] tiation of transfusion within the first hour of the patient’s arrival at the hospital.
Results: A total of 1767 patients were included for analysis with a mean age of 43 years
(19) and a mean Injury Severity Score of 15 (14). The in-hospital TASH score had the
highest predictive performance overall (area under the curve [AUC] = 0.925, 95% confi-
dence interval [CI] [0.904–0.946]), while the RED FLAG score (AUC = 0.881, 95% CI
[0.854–0.908]) had the greatest prehospital predictive performance compared to the
ABC score (AUC = 0.798, 95% CI [0.759–0.837]) and Shock Index (AUC = 0.795, 95%
CI [0.752–0.837]). Using their standard thresholds, the RED FLAG score was the most
efficient in predicting early transfusion (sensitivity: 87%, specificity: 76%, positive predic-
tive value: 25%, negative predictive value: 99%, Youden index: 0.63).
Conclusion: The RED FLAG score appears to outperform both the ABC score and the
Shock Index in predicting early in-hospital transfusion in trauma patients managed by
pre-hospital teams. If adopted, this score could be used to give advance warning to
trauma centres or even to initiate early transfusion during pre-hospital care.

KEYWORDS
early transfusion, major trauma, massive transfusion, pre-hospital score, RED FLAG

I N T R O D U CT I O N hospital care, while 27% occur within the first 48 h in hospital [2].
Between 30% and 40% of these deaths are attributable to massive
Trauma is the primary cause of death among people under 45 years of haemorrhage [3]. In recent years, various scores have been developed
age worldwide [1], and 66% of trauma deaths occur during pre- to quickly predict the need for massive transfusion. There are

Vox Sanguinis. 2022;117:227–234. wileyonlinelibrary.com/journal/vox © 2021 International Society of Blood Transfusion. 227
228 CASSIGNOL ET AL.

14 prediction scores and algorithms for massive transfusion in major ≤13 g/dl, pre-hospital orotracheal intubation and unstable pelvic frac-
trauma patients [4]. Of these, the Assessment of Blood Consumption ture upon clinical examination. Each variable counts as 1 point. A RED
(ABC) score and the Shock Index are most frequently used in pre- FLAG score ≥ 2 is considered positive.
hospital care [5, 6], while the Trauma Associated Severe Hemorrhage The ABC score (pre-hospital gold standard) is calculated using four
(TASH) score remains the gold standard in hospitals [7]. The objective variables: heart rate ≥ 120 bpm, systolic blood pressure ≤ 90 mmHg,
of these scores is to identify major trauma patients requiring transfu- the penetrating effect of the trauma and presence of free intraperito-
sion at an early stage in order to anticipate the amount of labile blood neal fluid (e.g., by Focused Assessment with Sonography for Trauma
products needing to be ordered by the receiving trauma centre [8]. [FAST]). Each variable counts as 1 point. An ABC score ≥ 2 is consid-
Although the capability for pre-hospital blood transfusion is being ered positive.
implemented in some areas, it is not yet widely accepted as a standard The Shock Index is calculated using two variables: maximum heart
practice [9]. Effective pre-hospital management and triage of these rate and minimum systolic arterial blood pressure. A Shock Index ≥1 is
major trauma patients to dedicated trauma centres reduce their mor- considered positive.
tality by 25% [10]. In 2018, a French trauma team developed the RED
FLAG prediction score for massive transfusion, which is applicable
from pre-hospital care [11]. The main objective of our study was to In-hospital score
evaluate the RED FLAG score in predicting early packed red blood cell
(PRBC) transfusion, defined as the initiation of blood transfusion The TASH score (in-hospital gold standard) is calculated using eight
within the first hour of hospital arrival of major trauma patients in a variables (Table 1): patient gender, haemoglobin level, base excess,
level-1 trauma centre, as well as comparing this score to other trans- systolic blood pressure, heart rate, presence of free intraperitoneal
fusion scores currently in use in pre-hospital (ABC score and Shock fluid (e.g., by FAST), unstable pelvic fracture and open femur fracture.
Index) and in-hospital (TASH score) care. A simplified point score can be directly correlated to the probability of
massive transfusion.

MATERIALS AND METHODS

Epidemiological analysis
Study type
Data collected during pre-hospital care include age, gender, type of
We conducted a monocentric retrospective study in the Sainte Anne trauma, mechanism of trauma, Glasgow Coma Scale (GCS), heart rate,
Military Hospital of Toulon (Southeast of France), a level-1 trauma systolic and diastolic blood pressure, respiratory rate, pulse oximetry,
centre with every necessary medical and surgical resources for the pre-hospital resuscitation (orotracheal intubation, catecholamines and
treatment of any type of traumatic injury. It receives an average of fluids administered), capillary haemoglobin and FAST ultrasound
300 trauma patients per year. A registry of the hospital’s trauma results. These variables were used to calculate the Shock Index and
patients was started in 2013 to collect data prospectively in accor- RED FLAG and ABC scores. In-hospital data included clinical examina-
dance with the Utstein-style guidelines [12]. Raw data were collected tion (GCS, heart rate, systolic and diastolic blood pressure, respiratory
prospectively on paper by physicians and then entered into an elec- rate, pulse oximetry), biological analysis (standard blood, coagulation
tronic database. A clinical research assistant regularly verifies the and arterial blood gas tests) and medical imaging (chest and pelvis x-
integrity and completeness of the data and collects patient outcomes rays, FAST ultrasound and whole-body scan). The primary endpoint
upon discharge. All patients in the registry from January 2013 to was early transfusion, defined as the transfusion of PRBCs within the
December 2018 were included in the study so long as major trauma first hour after the trauma patient’s arrival at the hospital.
was suspected in pre-hospital care. The calculation of prediction
scores for transfusion in each trauma patient (RED FLAG, ABC score,
Shock Index and TASH score) was performed retrospectively by a sin- Statistical analysis
gle operating physician using the registry data. Both the Establishment
Ethics Committee and the Data Protection Committee have agreed to We first performed a descriptive analysis of our study population.
the study (DR-2016-234, request for authorization no. 911461v2). Descriptive statistics included frequencies and percentages for cate-
gorical variables and mean values (standard deviation [SD]) for contin-
uous variables. We then compared early transfused patients with
Scores included in the study non-transfused patients. The Mann–Whitney U-test was used to com-
pare quantitative variables, while the chi-squared test was used for
Pre-hospital scores qualitative variables. Comparison of each score’s diagnostic perfor-
mance to predict early in-hospital PRBC transfusion was evaluated via
The RED FLAG score is calculated using five variables: Shock Index analysis of the receiver operating characteristic curve (ROC curve)
≥1, mean arterial blood pressure ≤ 70 mmHg, capillary haemoglobin using the same method as DeLong et al. [13]. Sensitivity, specificity,
BLOOD TRANSFUSION SCORES IN MAJOR TRAUMA PATIENTS 229

TABLE 1 Trauma associated severe hemorrhage (TASH) score and probability of massive transfusion (MT)

Variables Values Points TASH score Probability of MT

Haemoglobin (g/dl) <7 8 1–8 <5%
<9 6 9 6%
<10 4 10 8%
<11 3 11 11%
<12 2 12 14%
Base excess (mEq/L) < 10 4 13 18%
< 6 3 14 23%
< 2 1 15 29%
Systolic blood pressure (mmHg) <100 4 16 35%
<120 1 17 43%
Heart rate (bpm) >120 2 18 50%
Free intraperitoneal fluid (via FAST) 3 19 57%
Clinically unstable pelvic fracture 6 20 65%
Open or displaced femur fracture 3 21 71%
Male sex 1 22 77%
Total points = TASH score 23 82%
24+ >85%

Abbreviation: FAST, focused assessment with sonography for trauma.

positive predictive values and negative predictive values were calcu- RED FLAG score was the best performing pre-hospital score
lated with their standard thresholds in order to examine each score’s (p < 0.05). Table 5 presents the predictive performance of each
predictive capability regarding early PRBC transfusion (RED FLAG score with their standard thresholds, with the RED FLAG score dem-
score ≥ 2, ABC score ≥ 2, Shock Index ≥1, TASH score ≥ 16). onstrating the best predictive performance for early PRBC transfu-
The statistical software XLSTAT pro v.22.1.3 was used for data sion. However, the TASH score’s predictive performance improved
acquisition and analysis, as well as for the comparison of ROC curves. (sensitivity [Se] = 87%, specificity [Sp] = 86%, positive predictive
Statistical significance was set at p < 0.05. value [PPV] = 37%, negative predictive value [NPV] = 99%, Youden
index = 0.73) after the positivity threshold was lowered (TASH ≥9).

RESULTS
DI SCU SSION
Population
Our study aimed to evaluate and compare prediction scores for
Of the 1843 trauma patients included in the registry during the study massive transfusion, especifically, in the case of early PRBC trans-
period, 1767 patients (96%) were included in the analysis. A fusion defined as the initiation of transfusion within the first hour
study population flowchart and reasons for exclusion can be seen in of arrival at our level-1 trauma centre. The RED FLAG score
Figure 1. Study population characteristics are reported in Table 2. appeared to be the highest performing pre-hospital scores studied
in predicting the need for early PRBC transfusion with
AUC = 0.88, 95% CI [0.85–0.91].
Aims The RED FLAG score was recently developed by a French trauma
team and includes the Shock Index and additional clinical variables
After admission to the level-1 trauma centre, 151 patients were trans- that denote a patient’s severity of trauma [11]. As such, it can be con-
fused within the first hour. Table 3 presents the clinical, paraclinical sidered as a slightly more adapted score than the Shock Index in the
and severity differences between trauma patients who received early early detection of trauma-induced haemorrhage. To our knowledge, it
PRBC transfusion and those who did not. The early transfused popu- has never been compared to other in-hospital and pre-hospital scores
lation was more often in critical condition than the non-transfused that predict the need for massive transfusion. Our study is also origi-
population. The ROC curves and AUC differences for each score stud- nal in introducing a notion of temporality with the concept of early
ied are included in Figure 2 and Table 4. The in-hospital TASH score transfusion in the comparison of these scores. Although we were
was better than all pre-hospital scores studied (p < 0.05), while the unable to verify whether patients who received an early PRBC
230 CASSIGNOL ET AL.

FIGURE 1 Study population flowchart. ABC, assessment of blood consumption; TASH, trauma associated severe hemorrhage

transfusion required a massive transfusion thereafter, the results of The Shock Index has also been relevant in geriatric and paediatric
our study suggest an interest in the RED FLAG score as it seemed to populations [21, 22].
perform better than both the ABC score and the Shock Index in In the hospital, the TASH score is the gold standard to predict
predicting the need for early PRBC transfusion. transfusion needs [23], which acts as a simplified scoring system
The ABC score is the standard reference in pre-hospital care [5]. where increasing points can be directly correlated to an increasing
In the hospital, it has demonstrated greater diagnostic performance probability of massive transfusion without having a clear positivity
than the TASH score in predicting the need for transfusion in trauma threshold. We used the threshold ≥16 in our study as it has been com-
patients [14]. In our study, the ABC score appeared to perform less monly used in the literature [24], though different thresholds for the
well than the TASH score in predicting early PRBC transfusion. This TASH score have been suggested [25, 26], and a threshold of ≥8.5
could be explained by the use of pre-hospital variables in its calcula- was found to have greater predictive performance (Se = 84.4% and
tion versus in-hospital variables for the TASH score. It also appeared Sp = 78.4%) [27]. In our study, by decreasing the positivity threshold
to be less effective than the RED FLAG score and comparable to the to ≤9, the predictive performance of the score was increased regard-
Shock Index. In addition to systolic blood pressure and heart rate (also ing early PRBC transfusion. However, having to wait for pathological
Shock Index components), the ABC score considers free intraperito- and/or imaging results at the hospital may delay both calculating the
neal fluid via FAST and penetrating trauma as well. These last two var- score and starting transfusion. The relevance of the RED FLAG score
iables did not prove to be significant in predicting transfusion in a in initiating early in-hospital PRBC transfusion is interesting but
recent study [15], which may explain the comparability of the two should be the subject of a study specifically dedicated to this
scores. objective.
The Shock Index is the oldest score included in our study. It was In the majority of studies comparing massive transfusion predic-
first used as an indicator of shock severity [16], then, as a prediction tion scores, the TASH score appears to perform the best [24, 26, 28].
score for trauma mortality [17] and, finally, as a predictive test for However, there are limitations to using the TASH score in pre-hospital
massive transfusion [6]. The Shock Index has been demonstrated to care, such as the large number of variables required for its calculation
be an early indicator of hypovolaemic shock and blood transfusion (eight variables) and the absence of a real threshold for initiating
upon hospital arrival [18]. It is a measure of the severity of hypo- transfusion. The Shock Index (two variables), the ABC score (four vari-
volaemia, particularly, in patients in the compensatory stage of shock ables) and the RED FLAG score (five variables) with their fixed thresh-
who have a heart rate and/or systolic blood pressure within the toler- olds (≥1, ≥2 and ≥2, respectively) are more appropriate in pre-hospital
ance limits, despite significant blood loss [19]. By selecting the thresh- emergency care for starting PRBC transfusion on scene and/or to
old of SI ≥1, its calculation can be simplified for pre-hospital care [20]. alert the trauma centre.
BLOOD TRANSFUSION SCORES IN MAJOR TRAUMA PATIENTS 231

TABLE 2 Study population characteristics (n = 1767 patients) accessible to the pre-hospital care team. It permits the rapid identifi-
Characteristics Value cation of trauma patients who require the mobilization of significant
human and material resources to control bleeding.
Sex: Male, n (%) 1359 (77)
It is important to use the appropriate algorithms and standard
Age, mean (SD) 42.8 (19.2)
procedures in pre-hospital care in order to anticipate the transfu-
Mechanism of injury, n (%)
sion requirements of major trauma patients and to alert the
• Blunt 1616 (91.5)
trauma centre. Without advance warning from pre-hospital teams,
• Penetrating 151 (8.5)
trauma leaders are often restricted to awaiting the patient’s arrival
Type of trauma, n (%)
at the hospital to activate a massive transfusion protocol. In this
• Motorcycle 709 (40) context, the scores are of less interest as patients are already in
• Car crash 397 (22.5) the hospital, leading to a loss of preparedness for ordering labile
• Fall 322 (18.4) blood products. Thus, advance warning of early PRBC transfusion
• Pedestrian 107 (6) by pre-hospital teams is necessary for increasing preparedness by
• Gunshot 82 (4.6) anticipating the orders of labile blood products in the hospital and
• Stab wound 62 (3.5) readying operating rooms before the arrival of the major trauma

• Other 88 (5) patient.

Our study has several limitations. Although the data collected
Transport to hospital, n (%)
were prospective and mostly complete, the study design was ret-
• EMS 1079 (61)
rospective, monocentric and observational. Scores were calculated
• HEMS 622 (35)
retrospectively, and it was impossible to determine whether their
• Private transportation 66 (4)
use during pre-hospital care would have affected in-hospital care.
Severity scores
The retrospective nature also prevented us from analysing the
MGAP score, n (%)
temporal component of score use, a fundamental element of all
• 3–17a 196 (11.1) pre-hospital care. It was not possible for us to confirm whether
• 18–22b 311 (17.6) patients who had received an early PRBC transfusion received a
• 23–29 c 1260 (71.3) massive transfusion afterward, as our study focussed on early
Injury severity score, mean (SD) 15 (14) PRBC transfusion upon arrival in hospital. Neither can we confirm
Prediction scores for MT, n (%) whether patients who were not transfused in the first hour did not
• TASH score ≥ 16 82 (4.6) later require a blood transfusion. RED FLAG had been developed

• ABC score ≥ 2 149 (8.4) primarily in a blunt trauma population and reflects distinct local
procedures (medicalization of pre-hospital emergency care), such
• Shock Index ≥1 310 (18)
as orotracheal intubation in the field or measuring haemoglobin
• RED FLAG ≥2 527 (30)
levels by capillary test. Consequently, RED FLAG might not be
In-hospital evolution
applicable when either pre-hospital haemoglobin assessment is
• Blood transfusion within the first hour, n (%) 151 (8.5)
unavailable or pre-hospital intubation is not commonly performed,
• Length of stay, day, mean (SD) 11 (16)
or in the context of penetrating trauma, particularly concerning
• Mortality, n (%) 129 (7.3)
military combat where high-energy penetrating trauma is more fre-
Abbreviations: ABC, assessment of blood consumption; EMS, emergency quent. Penetrating trauma accounted for only 8.5% of our cohort,
medical service; HEMS, helicopter emergency medical service; MGAP and no subgroup analysis was performed. Therefore, we cannot
score, mechanism, glasgow coma scale, age, systolic blood pressure; MT,
determine the relevance of the scores evaluated in this population.
massive transfusion; TASH, trauma associated severe hemorrhage.
a
High risk of mortality. A multicentric prospective study is necessary to both validate our
b
Intermediate risk of mortality. results and allow us to standardize our pre-hospital care protocols
c
Low risk of mortality. accordingly.
In summary, it remains difficult to detect active bleeding and
anticipate the need for transfusion in major trauma patients. The RED
FLAG score appears to outperform both the ABC score and the Shock
Many of the existing scores that attempt to predict massive trans- Index in predicting the necessity of early PRBC transfusion in major
fusion are not derived from pre-hospital variables [4]. Our study was trauma patients during pre-hospital care. Its use in current practice
based solely on pre-hospital variables used in the design of the RED could provide two significant advantages: (i) providing advance warn-
FLAG score, ABC score and Shock Index. Only the TASH score was ing to receiving trauma centres of the need to initiate an early transfu-
designed using in-hospital variables. As Hamada et al. suggest, the sion and (ii) beginning blood transfusion during pre-hospital care for
RED FLAG score has the advantages of simplicity and pragmatism due teams equipped with labile blood products in their mobile emergency
to its use of variables corresponding to criteria that are directly units (EMS and/or HEMS).
232 CASSIGNOL ET AL.

TABLE 3 Comparison of patient transfused or not within first hour in hospital

Blood transfusion within first hour

YES NO

Characteristics n = 151 n = 1616 p-value

Sex male, n (%) 105 (70) 1254 (78) 0.025
Age, mean (SD) 46 (21) 43 (19) 0.099
Prehospital physiological variables, mean (SD)
• GCS 9 (5) 13 (4) <0.001
• Heart rate (bpm) 103 (31) 89 (21) <0.001
• Systolic blood pressure (mmHg) 94 (37) 123 (24) <0.001
• Diastolic blood pressure (mmHg) 53 (28) 72 (18) <0.001
• Mean blood pressure (mmHg) 66 (31) 88 (19) <0.001
• Peripheral oxygen saturation (%) 89 (18) 97 (6) <0.001
• Respiratory rate (cpm) 20 (7) 19 (5) 0.082
• Capillary haemoglobin (g/dl) 11 (3) 13 (2) <0.001
Prehospital resuscitation
• Mechanical ventilation, n (%) 88 (58) 245 (15) <0.001
• Catecholamine administration, n (%) 90 (60) 103 (6) <0.001
• Tranexamic acid administration, n (%) 102 (68) 240 (15) <0.001
• Osmotherapy, n (%) 18 (12) 56 (4) <0.001
• Fluid administration, ml, mean (SD) 944 (786) 378 (399) <0.001
In-hospital physiological variables, mean (SD)
• GCS 8 (6) 12 (5) <0.001
• Heart rate (bpm) 102 (34) 87 (21) <0.001
• Systolic blood pressure (mmHg) 86 (34) 122 (24) <0.001
• Diastolic blood pressure (mmHg) 46 (24) 72 (17) <0.001
• Mean blood pressure (mmHg) 58 (27) 88 (18) <0.001
• Peripheral oxygen saturation (%) 91 (17) 99 (23) <0.001
• Respiratory rate (cpm) 19 (6) 19 (5) 0.99
Biological parameters, mean (SD)
• Haemoglobin (g/dl) 9.9 (2.4) 13.6 (4.3) <0.001
• Blood platelets (G/L) 185 (76) 229 (66) <0.001
• Prothrombin 50 (20) 83 (15) <0.001
• Fibrinogen 1.6 (1) 2.7 (1) <0.001
Blood gas parameters, mean (SD)
• pH 7.18 (0.18) 7.37 (0.19) <0.001
• PO2 206 (128) 169 (104) 0.002
• PCO2 46 (17) 39 (10) <0.001
• HCO3 16.7 (4.6) 22.3 (3.3) <0.001
• Base excess 10.6 (6.4) 2.6 (3.7) <0.001
• Lactate 6.1 (4.8) 2 (1.5) <0.001
Medical imagery, n (%)
• Pneumothorax 17 (11) 57 (3.5) <0.001
• Haemothorax 37 (25) 41 (2.5) <0.001
• Unstable pelvic fracture 38 (25) 99 (6.1) <0.001
• Free intraperitoneal fluid (via FAST) 40 (26) 81 (5) <0.001
(Continues)
BLOOD TRANSFUSION SCORES IN MAJOR TRAUMA PATIENTS 233

TABLE 3 (Continued)

Blood transfusion within first hour

YES NO

Characteristics n = 151 n = 1616 p-value

Severity, n (%)
• Emergency haemostatic surgery 51 (34) 52 (3.2) <0.001
• ISS > 15 132 (87) 622 (39) <0.001
• In-hospital death 55 (36) 74 (4.6) <0.001

Abbreviations: FAST, focused assessment with sonography for trauma; GCS, glasgow coma scale; ISS, injury severity score.

F I G U R E 2 Receiving operating curves of four scores (Shock Index and RED FLAG, ABC and TASH scores) for early packed red blood cell
transfusion. ABC, assessment of blood consumption; TASH, trauma associated severe hemorrhage

TABLE 4 Comparison of the predictive performance of the Shock Index and TASH, RED FLAG and ABC scores in the cohort (n = 1767
patients)

Scores AUC difference 95% CI p-value

RED FLAG versus Shock Index 0.086 0.08–0.10 <0.001
RED FLAG versus ABC score 0.083 0.07–0.09 <0.001
ABC score versus Shock Index 0.003 0.01–0.10 0.539
TASH score versus RED FLAG 0.044 0.04–0.05 <0.001
TASH score versus ABC score 0.127 0.12–0.14 <0.001
TASH score versus Shock Index 0.130 0.12–0.14 <0.001

Abbreviations: ABC, assessment of blood consumption; AUC, area under the curve; TASH, trauma associated severe hemorrhage.

TABLE 5 Diagnostic properties of each score (Shock Index and TASH, RED FLAG and ABC scores) with standard thresholds

Scores Thresholds Sensitivity Specificity PPV NPV Youden index

RED FLAG ≥2 87% 76% 25% 99% 0.63
Shock Index ≥1 62% 87% 30% 96% 0.48
ABC score ≥2 44% 95% 45% 95% 0.39
TASH score ≥16 36% 98% 67% 94% 0.35

Abbreviations: ABC, assessment of blood consumption; NPV, negative predictive value; PPV, positive predictive value; TASH, trauma associated severe
hemorrhage.
234 CASSIGNOL ET AL.

ACKNOWLEDGEMEN TS major trauma–the Utstein style” (as of July 17, 1999). An interna-
The authors thank the doctors of the emergency room and intensive tional trauma Anaesthesia and critical care society (ITACCS). Acta
Anaesthesiol Belg. 2000;51:18–38.
care unit of the military hospital of Saint Anne for their work on the
13. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas
registry for major trauma patients. under two or more correlated receiver operating characteristic cur-
A.C. and P.B. contributed to study design, data analysis and inter- ves: a nonparametric approach. Biometrics. 1988;44:837–45.
pretation, literature search, and writing. A.R., J.C. and J. B. contributed 14. Krumrei NJ, Park MS, Cotton BA, Zielinski MD. Comparison of mas-
sive blood transfusion predictive models in the rural setting.
to data collection and critical revision. P.B. contributed to data analy-
J Trauma Acute Care Surg. 2012;72:211–5.
sis and interpretation. The final version of the manuscript has been 15. Schroll R, Swift D, Tatum D, Couch S, Heaney JB, Llado-Farrulla M,
validated by all of the authors. et al. Accuracy of shock index versus ABC score to predict need for
massive transfusion in trauma patients. Injury. 2018;49:15–9.
16. Allgöwer M, Burri C. Shock index. Dtsch Med Wochenschr 1946.
CONF LICT OF IN TE RE ST
1967;92:1947–50.
The authors report no conflicts of interest. 17. Cannon CM, Braxton CC, Kling-Smith M, Mahnken JD, Carlton E,
Moncure M. Utility of the shock index in predicting mortality in trau-
ORCID matically injured patients. J Trauma. 2009;67:1426–30.
18. Mutschler M, Nienaber U, Münzberg M, Wölfl C, Schoechl H,
Arnaud Cassignol https://orcid.org/0000-0002-8403-0076
Paffrath T, et al. The shock index revisited - a fast guide to transfu-
sion requirement? A retrospective analysis on 21,853 patients
RE FE R ENC E S derived from the TraumaRegister DGU. Crit Care. 2013;17:R172.
1. Murray CJ, Lopez AD. Alternative projections of mortality and dis- 19. Birkhahn RH, Gaeta TJ, Terry D, Bove JJ, Tloczkowski J. Shock index in
ability by cause 1990-2020: global burden of disease study. Lancet. diagnosing early acute hypovolemia. Am J Emerg Med. 2005;23:323–6.
1997;349:1498–504. 20. Moffat B, Vogt KN, Inaba K. The shock index: is it ready for
2. Evans JA, van Wessem KJP, McDougall D, Lee KA, Lyons T, primetime? Crit Care. 2013;17:196.
Balogh ZJ. Epidemiology of traumatic deaths: comprehensive 21. Ohmori T, Kitamura T, Ishihara J, Onishi H, Nojima T, Yamamoto K,
population-based assessment. World J Surg. 2010;34:158–63. et al. Early predictors for massive transfusion in older adult severe
3. Kauvar DS, Lefering R, Wade CE. Impact of hemorrhage on trauma trauma patients. Injury. 2017;48:1006–12.
outcome: an overview of epidemiology, clinical presentations, and 22. Strutt J, Flood A, Kharbanda AB. Shock index as a predictor of mor-
therapeutic considerations. J Trauma. 2006;60:S3–11. bidity and mortality in pediatric trauma patients. Pediatr Emerg Care.
4. Tonglet ML. Early prediction of ongoing hemorrhage in severe 2019;35:132–7.
trauma: presentation of the existing scoring systems. Arch Trauma 23. Yücel N, Lefering R, Maegele M, Vorweg M, Tjardes T, Ruchholtz S,
Res. 2016;20:5. et al. Trauma associated severe hemorrhage (TASH)-score: probability
5. Nunez TC, Voskresensky IV, Dossett LA, Shinall R, Dutton WD, of mass transfusion as surrogate for life threatening hemorrhage after
Cotton BA. Early prediction of massive transfusion in trauma: simple multiple trauma. J Trauma Inj Infect Crit Care. 2006;60:1228–37.
as ABC (assessment of blood consumption)? J Trauma. 2009;66: 24. Subramanian A. Assessing the efficiency of scoring system for
346–52. predicting the probability of massive transfusion in trauma patients.
6. Vandromme MJ, Griffin RL, Kerby JD, McGwin G, Rue LW, J Hematol Thromboembolic Dis. 2014;02:161.
Weinberg JA. Identifying risk for massive transfusion in the relatively 25. Borgman MA, Spinella PC, Holcomb JB, Blackbourne LH, Wade CE,
normotensive patient: utility of the prehospital shock index. Lefering R, et al. The effect of FFP:RBC ratio on morbidity and mor-
J Trauma. 2011;70:384–8. tality in trauma patients based on transfusion prediction score. Vox
7. Maegele M, Lefering R, Wafaisade A, Theodorou P, Wutzler S, Sang. 2011;101:44–54.
Fischer P, et al. Revalidation and update of the TASH-score: a scoring 26. Poon K, Lui C, Tsui K. Comparison of the accuracy of local and inter-
system to predict the probability for massive transfusion as a surro- national prediction models for massive transfusion in major trauma
gate for life-threatening haemorrhage after severe injury. Vox Sang. patients. Hong Kong J Emerg Med. 2012;19:189–97.
2011;100:231–8. 27. Brockamp T, Nienaber U, Mutschler M, Wafaisade A, Peiniger S,
8. Yeguiayan J-M, Garrigue D, Binquet C, Jacquot C, Duranteau J, Lefering R, et al. Predicting on-going hemorrhage and transfusion
Martin C, et al. Medical pre-hospital management reduces mortality requirement after severe trauma: a validation of six scoring systems
in severe blunt trauma: a prospective epidemiological study. Crit and algorithms on the TraumaRegister DGU. Crit Care. 2012;16:R129.
Care. 2011;15:R34. 28. Chico-Fern andez M, García-Fuentes C, Alonso-Fern andez MA, Toral-
9. Cassignol A, Marmin J, Mattei P, Goffinet L, Pons S, Renard A, et al. Vazquez D, Bermejo-Aznarez S, Alted-Lo pez E. Massive transfusion
Civilian prehospital transfusion - experiences from a French region. predictive scores in trauma. Experience of a transfusion registry.
Vox Sang. 2020;115:745–55. Med Intensiva. 2011;35:546–51.
10. MacKenzie EJ, Rivara FP, Jurkovich GJ, Nathens AB, Frey KP,
Egleston BL, et al. A national evaluation of the effect of trauma-
center care on mortality. N Engl J Med. 2006;354:366–78.
How to cite this article: Cassignol A, Bertein P, Botti P,
11. Hamada SR, Rosa A, Gauss T, Desclefs J-P, Raux M, Harrois A, et al.
Development and validation of a pre-hospital “Red Flag” alert for Renard A, Cotte J, Bordes J. Early packed red blood cell
activation of intra-hospital haemorrhage control response in blunt transfusion in major trauma patients: Evaluation and
trauma. Crit Care. 2018;22:113. https://doi.org/10.1186/s13054- comparison of different prediction scores for massive
018-2026-9.
transfusion. Vox Sang. 2022;117:227–234.
12. Dick WF, Baskett PJ, Grande C, Delooz H, Kloeck W, Lackner C,
et al. “Recommendations for uniform reporting of data following
Received: 16 December 2020 Revised: 1 June 2021 Accepted: 3 June 2021

DOI: 10.1111/vox.13174

ORIGINAL ARTICLE

The impact of COVID-19 outbreak on the Transfusion

Medicine Unit of a Northern Italy Hospital and Cancer Centre

Davide Schiroli1 | Lucia Merolle1 | Giuseppe Molinari1 |

Erminia Di Bartolomeo1 | Daniela Seligardi1 | Laura Canovi1 |
Thelma A. Pertinhez1,2 | Pamela Mancuso3 | Paolo Giorgi Rossi3 |
Roberto Baricchi1 | Chiara Marraccini1

1
Transfusion Medicine Unit, AUSL-IRCCS di
Reggio Emilia, Reggio Emilia, Italy Abstract
2
Department of Medicine and Surgery, Background and Objectives: The first wave of coronavirus disease-2019 (COVID-19)
Università di Parma, Parma, Italy
3
dramatically affected the Transfusion Medicine Unit of the Azienda Unità Sanitari
Epidemiology Unit, AUSL-IRCCS di Reggio
Emilia, Reggio Emilia, Italy Locale - Istituto di Ricovero e Cura a Carattere Scientifico (AUSL-IRCCS) di Reggio Emi-
lia, which faced a total rearrangement of the procedures for donors and patients. This
Correspondence
Lucia Merolle, Transfusion Medicine Unit,
study aims to assess the major implications of COVID-19 on our department, focusing
AUSL-IRCCS di Reggio Emilia, Reggio Emilia, on the blood transfusion chain and therapies, in order to support transfusion specialists
Italy.
Email: [email protected]
in seeking efficient ways to face similar future emergencies.
Materials and Methods: This retrospective study compares our Transfusion Medicine
Funding information
Unit data collected between February and May 2020 with the same period in 2017–
None
2019. Data on red blood cells and platelets donations, transfusions and clinical proce-
dures were collected as aggregates from our internal electronic database.
Results: During the lockdown, donor centres were re-organized to reduce the risk
of contagion and avoid unnecessary blood collection. Blood donations were re-scheduled
to meet the decrease in elective surgery; consequently, plateletapheresis was
implemented to supply the reduction of buffycoat-derived platelets. Transfusions
significantly decreased together with orthopaedic and vascular surgery, while they
were only marginally diminished for both cancer and onco-haematological patients.
Reduced procedures for inpatients and outpatients were matched by remote medi-
cine, addressing the need of a constant healthcare support for patients with chronic
diseases.
Conclusions: The described measures were adopted to avoid excessive blood collec-
tion and expiration, guarantee the safety of our ward (for both patients and staff)
and supply the necessary transfusion therapies. These measures may support the
development of appropriate risk management plans and safety procedures for other
hospitals and transfusion services that have to face similar events.

KEYWORDS
blood collection, donor recruitment, platelet transfusion, transfusion medicine (in general),
transfusion therapy

Vox Sanguinis. 2022;117:235–242. wileyonlinelibrary.com/journal/vox © 2021 International Society of Blood Transfusion. 235
236 SCHIROLI ET AL.

I N T R O D U CT I O N comparing data from February to May 2020 with the same period of
the previous triennium (2017–2019).
The first wave of coronavirus disease-2019 (COVID-19) significantly Since one-way analysis of variance (ANOVA) analysis did not reveal
impacted Northern Italy, where hospitals faced a rapid and profound significant difference (p > 0.05) in the mean transfusions among the
re-organization of the healthcare services supply [1, 2]. The Italian tested groups (2017–2019), total red blood cells (RBCs) transfusions per-
Government declared a lockdown period that, for the Reggio Emilia formed during the 2020 lockdown were compared with the mean trans-
province, started on 9 March 2020, with a gradual loosening from fusions performed during the same period of the previous triennium.
3 May 2020 [3]. During this period, all the wards of the AUSL-IRCCS PLTs transfusions’ comparison, instead, was performed on 2020
of Reggio Emilia rearranged their routine procedures to meet the versus 2019. In our OECI Clinical Cancer Centre, between 65% and
requirements imposed by the COVID-19 outbreak. 90% of PLTs usage is generally destined to haematological and
Reggio Emilia hospital is located in Emilia Romagna (Northern onco-haematological patients. The number of these patients increased
Italy) and experienced thousands of COVID-19 cases, with different since July 2017, when our hospital established the Cancer and
grades of severity during the lockdown period [4]. The impact of the Haematology Centre (CORE). This pavilion became fully functional by
pandemic on the healthcare system negatively affected a large num- the end of 2018 and now centralizes all the onco-haematological
ber of patients with different pathologies [5–9]. Emilia Romagna Local patients of the Reggio Emilia province.
Governance listed the healthcare services that could not be post- No ethical approval was necessary for collecting the aggregated,
poned during the lockdown, as well as those that could be delayed or anonymous data presented herein.
developed through remote medicine (i.e., telemedicine, electronic pre-
scriptions, phone consultations), in order to limit the access to hospi-
tals and avoid overcrowding [10]. Transfusion Medicine Unit procedures and
Despite not being involved in first-line treatment of COVID-19 organization
patients, even our Transfusion Medicine Unit faced many changes in
its routine clinical practice, due to the close connection with the other Our Transfusion Medicine Unit accounts for 24 collection points (donor
hospital departments [11]. To ensure high standards of blood donation centres) covering the whole Reggio Emilia province. The management is
and blood safety during the emergency, the Italian National Blood equally distributed between the hospital and the volunteer associations
Centre (Centro Nazionale Sangue, CNS), in compliance with the Minis- (Associazione Volontari Italiani del Sangue, AVIS). Blood processing and
try of Health and the COVID-19 Technical Scientific Committee (CTS), distribution are exclusively managed by the Transfusion Medicine Unit.
issued a series of measures that our ward promptly adopted [12, 13]. Our procedures involve therapies and consultations for patients
In the present study, we investigate the effects of the COVID-19 coming from other wards, for inpatients temporarily hospitalized
pandemic and lockdown-related restrictions on the activity of our within our ward and for outpatients. Medical offices that compose
Transfusion Medicine Unit, providing a description of the adopted our Transfusion Medicine Unit are dedicated to patient blood man-
measures and of their effectiveness on blood collection, blood transfu- agement, bloodlettings/phlebotomies, transfusions, apheresis and
sions, transfusion medicine and immune-haematology clinical practice. sideropaenic anaemia treatment.
Our aim is to point out, through an evidence-based analysis, the
practices that may help transfusion specialists worldwide to deal with
similar emergencies in the future. Data management and statistical analysis

Data concerning hospitalizations and COVID-19 patients were col-

MATERIALS AND METHODS lected as aggregates from our hospital discharge database. Data con-
cerning the Transfusion Medicine Unit procedures were collected as
Reggio Emilia is one of the nine provinces located in the Emilia-Romagna aggregates using our internal database interface (Eliot, Engineering
region, accounting for about 532,000 inhabitants. The Emilia-Romagna Ingegneria Informatica S.p.A., Italy) and downloaded as Excel spread-
region coordinates blood availability and distribution between the prov- sheets. Subsequent data analysis was performed on Excel (Microsoft
inces via the Regional Blood Centre (Centro Regionale Sangue [CRS]). Office 2010, Microsoft Corp., Redmond, WA), Python (3.8) and
Reggio Emilia province harbours five hospitals. Nevertheless, there is GraphPad Prism 7.02 (GraphPad Software, San Diego, CA). Data were
only a Transfusion Medicine Unit that centralizes the necessity for trans- expressed as number of patients or blood/platelets (PLTs) units and,
fusions, therapies and medical consultations. when possible, as a mean of three different years (2017, 2018 and
2019) with its maximum–minimum range. Statistical significance was
assessed using a t-test when comparing the same time lapse of two
Study design different years, while one-way ANOVA was used to determine statis-
tical significance of the same time lapse in the last 3 years (from 2017
The study was carried out at the Transfusion Medicine Unit of the to 2019) either including 2020 or not within the comparison. The sig-
Azienda USL-IRCCS di Reggio Emilia. This is a retrospective study nificance was set at *p ≤ 0.05, **p ≤ 0.001 and ***p ≤ 0.0001.
COVID-19 IMPACT IN AN ITALIAN TRANSFUSION MEDICINE UNIT 237

RESULTS Plateletapheresis was, thus, increased to allow a constant PLTs supply

to our blood bank (Figure 1(b)).
Management of the donor centres Whole blood donations dedicated to group 0 negative (0 )
(Supplementary Figure S1) and rare phenotypes reserved for specific
During the national lockdown period, Reggio Emilia donor centres were cohorts of patients (i.e., thalassaemics and myelodysplastics) were pre-
re-organized to: served. Rare group donors were scheduled by direct phone call, while
all the other appointments were cancelled by SMS.
1. Lower the risk of contagion. Since plateletapheresis is a complex and invasive procedure, PLTs
2. Avoid unnecessary blood collection. donors are generally fewer than blood donors. The collection of a suf-
ficient amount of plateletapheresis-derived units, therefore, represen-
To lower the risk of contagion, two small donor centres were ted a challenge for our recruiting personnel. To overcome this
temporarily closed and converted into COVID clinics. In the remaining problem, our Transfusion Medicine Unit established a bank of frozen
operating centres, some of the personnel were appointed to check PLTs, which can be stored for up to 2 years. Between August and
the temperature and health status of all donors immediately before October 2020, our personnel collected and stored over 50 frozen
their access. Social distancing measures were adopted to avoid over- PLTs units, according to Italian standards [14].
crowding, and, after each donation, rooms and seats were sanitized. Furthermore, the CRS coordinated blood availability and distribu-
Donations were suspended for students and first donors; when not tion between the provinces according to the epidemic diffusion. In
necessary, eligibility and control exams were halted. order to determine how this internal and local re-organization of
blood supply and transfusions affected the coordination with the
other transfusion centres of our region, we also analysed the influx/
Blood collection efflux of blood units. Results are reported in Figure S2 of the
Supplementary information.
Blood, PLTs and plasma donations are routinely planned with the Finally, since plasma can be stored frozen for several months, its
donors by SMS alert or phone call and scheduled in order to ade- collection was not interrupted. The donor centres extended the open-
quately manage the blood bank. During the 2020 lockdown, the ing time, while appointments for plasma donations were re-scheduled
schedule was deeply modified to meet the hospital requirements. to avoid overcrowding and to allow the necessary sanitization (data
According to the Italian Government instructions [3], our hospital not shown).
re-scheduled a large majority of surgical procedures. When possible,
non-urgent orthopaedics, cardiovascular and cancer surgeries were
postponed or moved to non-COVID hospitals. This had an impact on Blood transfusion—RBCs and PLTs
the total number of surgical interventions, especially during the most
critical lockdown phase (i.e., March and April 2020, data not shown). To understand how the pandemic affected the transfusion therapies,
As a consequence, the Transfusion Medicine Unit faced a decrease in we compared the percentages of hospitalized patients transfused
blood components’ requests, and blood donation schedules were all between February and May 2020 with those of 2017–2019 (Table 1).
re-considered in order to avoid excessive stored blood units. Taking into account the number of non-COVID patients, the 2020
Figure 1(a) shows that the number of red blood cells (RBCs) units RBCs transfusion rate increased from 18.5% to 22.5%, while only a
prepared from whole blood donations between February and May small number of COVID-19 patients underwent a transfusion
2020 decreased, particularly between March and April. Consequently, (161 over 1720, 9.4%). These data are in agreement with the hospital-
the available buffycoat-derived PLTs units also diminished. izations that, during the lockdown period, dropped from a 15,771

T A B L E 1 Patients transfused and total patients hospitalized by

the AUSL-IRCCS di Reggio Emilia between February and May 2020
(absolute number), compared to the same period of 2017–
2019 (mean)

Transfused Hospitalized % Transf./

February–May patients patients Hosp.
2017–2019 2920 15,771 18.5
(mean value)
2020 COVID 2373 10,520 22.5
F I G U R E 1 (a) Whole blood–derived RBCs, weekly donations
negative
between February and May 2020, and (b) platelets apheresis
donations collected in the same period. LD, lockdown; RBCs, red 2020 COVID 161 1720 9.4
blood cells; PLTs, platelets positive
238 SCHIROLI ET AL.

F I G U R E 3 (a) Patients transfused with PLTs (left panel) and PLTs

units transfused in 2020, compared to the same period of 2019 and
(b) buffycoat-derived (pool) and apheresis-derived (PLT apheresis)
PLTs units transfused between February and May 2020. *p ≤ 0.05,
**p ≤ 0.001 and ***p ≤ 0.0001. LD, lockdown; PLT, platelet

F I G U R E 2 RBCs transfusions in 2020 (black lines), compared

with the same period of the previous triennium. (a) Patients
transfused with RBCs (left panel) and RBCs units transfused (right
panel); (b) patients and units transfused for orthopaedics surgery;
(c) patients and units transfused for cardiovascular surgery and
(d) patients and units transfused for cancer other than haematological.
*p ≤ 0.05, **p ≤ 0.001 and ***p ≤ 0.0001. RBCs, red blood cells

mean in 2017–2019 to 10,520 in 2020 (that becomes 12,240 if we

include COVID-19 patients, Table SI).
F I G U R E 4 Patients and units transfused for onco-haematological
Following the suspension of elective and non-urgent surgical
diseases: (a) RBCs and (b) PLTs. *p ≤ 0.05, **p ≤ 0.001. LD, lockdown;
interventions, RBCs transfusions significantly decreased when PLTs, platelets; RBCs, red blood cells
compared to the previous triennium, both in terms of patients
and total units transfused (Figure 2(a)). Starting from this evi-
dence, we further dissected the RBCs transfusions according to patients and, to a lesser extent, for cardiovascular surgery
the applicant ward. Our data revealed that the most significant (Figure 2(b),(c), respectively). Interestingly, the decrease of RBCs
decrease of RBCs transfusions was observed for orthopaedic usage in the orthopaedic interventions’ accounts for about 35%
COVID-19 IMPACT IN AN ITALIAN TRANSFUSION MEDICINE UNIT 239

of the total reduction and started around 15 days before the 2019. Therefore, the effects of the lockdown on transfusion therapies for
national lockdown. Conversely, there was no significant decrease this sub-group of patients are described in comparison with 2019 only.
in RBCs transfusion of cancer patients during the lockdown In our hospital, onco-haematological patients transfused with
period and was only a slight decrease of the transfused units in RBCs account for about 20% of the total, and their caregiving was
the post-lockdown period (Figure 2(d)). preserved as much as possible during the first wave of the pandemic.
In parallel, we investigated PLTs transfusions, we observed a Nevertheless, during the lockdown, we also observed a slight
slight decrease in patients and units transfused during the lockdown decrease of RBCs transfusions for these patients (Figure 4(a)).
(Figure 3(a)), although only these last data are significant (p < 0.01). Although these data were not highly significant, it is supported by a
PLTs transfusions started decreasing 3 weeks after the beginning of similar observation on PLTs transfusions (Figure 4(b)).
lockdown and came back to the 2019 level within 1 month. The larg-
est amount of PLTs units transfused during the second month of lock-
down (i.e., April 2020) came from plateletapheresis, as a consequence Transfusion medicine and immune-haematology
of the reduction of buffycoat-derived pools following the decrease of clinical practice
whole blood donations (Figure 3(b)).
Following the official resolution of the Emilia-Romagna region [14],
our Transfusion Medicine Unit totally re-organized its services to:
Blood transfusion—RBCs and PLTs in onco-
haematological patients • Limit the access of patients to emergencies and unavoidable
therapies;
As mentioned earlier, the establishment of the CORE pavilion implied an • Guarantee transfusions and other therapies to patients coming
increase of hospitalized onco-haematological patients between 2017 and from other wards;

F I G U R E 5 Transfusion medicine and immune-haematology clinical practice procedures performed for: (a) inpatients coming from other
wards; (b) outpatients and (c) medical consultations for outpatients. Black line indicates 2020 procedures, while grey line indicates 2019
procedures. LD, lockdown; LDL, low density lipoprotein; PLS, plasma
240 SCHIROLI ET AL.

• Extend the opening times of medical offices to ensure an adequate blood transfusion as a consequence of the pathology (161 over 1720,
number of procedures and, at the same time, allow the necessary 9.4%) [16].
room sanitization after each procedure; However, the temporary suspension of elective surgeries, along-
• Encourage medical phone consultations to guarantee therapeutic side the need to keep unaltered the clinical routine in some other
continuity for non-urgent and chronic patients. departments (in particular, oncology and haematology), drastically
impacted on our Unit activity. If we consider the non-COVID patients
These measures affected all the inpatients’ and outpatients’ hospitalized between February and May 2020 (Table 1), the RBCs
rooms that compose our Transfusion Medicine Unit: patient blood transfusion rate rose from 18.5% to 22.5%. This evidence suggests
management; bloodlettings/phlebotomy; transfusion; apheresis and that, in a scenario where the hospitalization of non-serious patients
sideropaenic anaemia. was postponed or even diminished, it was chosen to prioritize the
Figure 5 resumes the main procedures, procedures and medical consul- emergencies and the most severe cases that often require blood
tations performed between February and May 2020, compared with the transfusions.
same period of 2019. The large majority of our procedures decreased dur- In terms of organization, the main changes for our Transfusion
ing the lockdown. A significant decrease was observed in particular for Medicine Unit involved blood donations, which were significantly
medical consultations and plasma -exchange procedures required for inpa- reduced and re-scheduled to meet the decrease need of blood com-
tients coming from other hospital wards, as a consequence of the ponents, as similarly described in other blood centres [17–19]. To
decreased hospitalizations. Despite Reggio Emilia province trying to pre- avoid a surplus of RBCs, with the risk of discarding a large amount of
serve the healthcare services provided to oncological patients, the number expired units, all blood donations were temporarily reduced, except
of hospitalized cancer patients decreased, especially during the first weeks for those of rare groups. Adopting this strategy, our Transfusion Med-
of lockdown. Moreover, no autologous or allogeneic stem cell collection for icine Unit avoided RBCs units expiration for almost the whole lock-
bone marrow transplantation was performed in March 2020 (Figure 5(a)). down. Moreover, the regional blood supply-chain system, set up by
Procedures on outpatients decreased as well, particularly iron CRS, buffered the distribution of RBCs units through the Emilia-
infusions on anaemic pre-operative patients (Figure 5(b)). Between Romagna provinces from the least affected to the most affected by
April and May, our physicians started implementing medical phone the pandemic (Figure S2 panels B and C). Reggio Emilia, for instance,
consultations for the outpatients (Figure 5(c)). The Patient Blood Man- received blood units from the CRS in April, after having faced the
agement medical office, which routinely performs medical consulta- peak of COVID-19 diffusion in the previous month.
tions to pre-operative patients, temporarily closed due to the lack of After the initial reduction of the routine procedures, the
elective surgeries imposed by the lockdown. healthcare supply for cancer and onco-haematological patients was
The treatment of chronically ill patients (thalassaemic, polycy- gradually restored following the national and regional indications
thaemic, familial hypercholesterolaemia patients among all) was pre- [10, 15, 20]. Accordingly, a decrease in blood transfusions for the
served as much as possible, albeit the patients’ appointments were onco-haematological patients was mainly concentrated in the most
spread over a longer range of time (from 8 AM–12 noon to 8 AM–6 PM). critical weeks (between the 20th of March and the 5th of April,
All patients were only allowed to stay for the exact time of the Figure 4), while PLTs transfusion (mainly administered to prevent or
hospital procedure. Visitors were not allowed, except when accompa- treat bleeding episodes of onco-haematological patients) remained
nying minors or fragile patients. When possible, on-site appointments constant for the whole lockdown duration.
were substituted by remote consultations. Although not shown in this To compensate for the decrease in buffycoat-derived PLTs units,
study, medical assistance by phone was implemented during the we increased the number of PLTs donations (Figure 2(c)). The latter
whole summer in order to be prepared for a second pandemic wave. are not easily collected in an adequate quantity to satisfy the request,
and we thus adopted a procedure to store frozen PLTs [21]. PLTs
cryopreservation is an alternative to the conventional storage temper-
DISCUSSION ature (25 C) that allows the product shelf-life to increase from 5 days
to up to 2 years.
Between March and May 2020, the Italian Government declared a Following the decrease of inpatients hospitalized by other wards,
national lockdown, accompanied by further regional indications on our physicians faced a significant decrease of medical consultations
the management of healthcare services that significantly changed the (Figure 5(a)) and plasma-exchange procedures.
routine and emergency procedures all over the country [3, 10, 15]. In summary, the COVID-19 pandemic deeply impacted the
The COVID-19 outbreak also severely affected the procedures of the healthcare systems at all levels, with important implications also for
AUSL-IRCCS di Reggio Emilia: the number of non-COVID patients blood donation and transfusion [12, 22]. The mechanisms and proce-
accessing the hospital between February and May 2020 was only dures adopted to face this serious emergency were necessary to meet
10,520, which is over 33% less than the same period of the previous the safety requirements for donors, patients and staff. Our experience
triennium (15,771). has, actually, many points in common with other Transfusion
In accordance with what was previously reported, in our hospital, Medicine Units in Italy and other parts of the world [11, 12, 23, 24].
only a few hospitalized COVID-19 patients might have received a Among the strategies developed to meet this health emergency, we
COVID-19 IMPACT IN AN ITALIAN TRANSFUSION MEDICINE UNIT 241

believe that, alongside the consolidated practices such as sanitizing ORCID

procedures, re-modulation of access and the use of personal protec- Lucia Merolle https://orcid.org/0000-0001-5712-4272
tive equipment [25], a provision of frozen blood components that can Chiara Marraccini https://orcid.org/0000-0002-8053-498X
be stored for several months could be helpful during emergency
periods and should be promptly planned. Moreover, remote appoint- RE FE RE NCE S
ments were revealed as effective, especially for chronically ill patients, 1. Prezioso C, Pietropaolo V. COVID-19: update of the Italian situation.
and it has been implemented after the lockdown. J Neurovirol. 2020;26:834–7.
2. Conti S, Ferrara P, Mazzaglia G, et al. Magnitude and time-course of
Although our data are in agreement with recent literature about
excess mortality during COVID-19 outbreak: population-based
transfusion in COVID-19 patients, [16] we must point out that they empirical evidence from highly impacted provinces in northern Italy.
come from our internal database, which is not directly connected to clini- ERJ Open Res. 2020;6:00458.
cal records and other hospital databases. Therefore, it is plausible that 3. Gazzetta Ufficiale della Repubblica Italiana Anno 161 Numero 62 –
Edizione Straordinaria. Decreto Legge 9 Marzo 2020, n. 14: Disposizioni
some data are underestimated, especially those relating to cancer
urgenti per il potenziamento del Servizio sanitario nazionale in relazione
patients (other than onco-haematological). The latter can be hospitalized all’emergenza COVID-19. Decreto del Presidente del Consiglio dei
by several wards in our hospital, depending on tumour localization, and Ministri 9 Marzo 2020: 23 febbraio 2020, n. 6, recante misure urgenti in
this implies higher tracing complexity; furthermore, when a physician materia di contenimento e gestione dell’emergenza epidemiologica da
COVID-19, applicabili sull’intero territorio nazionale [in Italian].
prescribes a blood component, the diagnosis can sometimes not be well
4. Giorgi Rossi P, Marino M, Formisano D, Venturelli F, Vicentini M, Grilli R,
specified, thus generating missing data in our internal database. Never-
et al. Characteristics and outcomes of a cohort of COVID-19 patients in
theless, our electronic system blocks blood component requirements in the province of Reggio Emilia. Italy PLoS One. 2020;15:e0238281.
case of missing diagnosis, and we are confident that, despite a slight 5. De Filippo O, D’Ascenzo F, Angelini F, et al. Reduced rate of hospital
underestimation, the data presented herein are close to reality. admissions for ACS during Covid-19 outbreak in northern Italy. N
Engl J Med. 2020;383:88–9.
Finally, between February and May 2020, a ‘COVID-19 diagnosis’
6. Lazzerini M, Barbi E, Apicella A, Marchetti F, Cardinale F, Trobia G.
was not present in our database. Therefore, we cannot discriminate Delayed access or provision of care in Italy resulting from fear of
between RBCs and PLTs units transfused for COVID-19 or for medi- COVID-19. Lancet Child Adolesc Health. 2020;4:e10–1.
cal/surgical reasons. Nevertheless, we traced the requests 7. Benazzo F, Rossi SMP, Maniscalco P, Moretti B, Vaienti E, Ruggieri P,
et al. The orthopaedic and traumatology scenario during Covid-19 out-
for transfusions coming from the wards that hosted COVID-19
break in Italy: chronicles of a silent war. Int Orthop. 2020;44:1453–9.
patients, and we did not find any significant variation (data not shown). 8. Lambertini M, Toss A, Passaro A, Criscitiello C, Cremolini C,
In conclusion, during the first wave of the pandemic, our Transfu- Cardone C, et al. Cancer care during the spread of coronavirus dis-
sion Medicine Unit both avoided unnecessary blood collection and ease 2019 (COVID-19) in Italy: young oncologists’ perspective.
ESMO Open. 2020;5:e000759.
excessive stored blood units, preserving donor and staff safety. This
9. Rosenbaum L. Facing Covid-19 in Italy – ethics, logistics, and thera-
result was reached by coordinating with the internal (hospital wards peutics on the epidemic’s front line. N Engl J Med. 2020;382:1873–5.
and donor centres) and external (i.e, CRS) stakeholders. However, 10. Direzione Generale Cura della Persona. Salute e Welfare Regione
these measures come from an emergency phase, and, despite their Emilia Romagna. PG/2020/176519 del 28 febbraio 2020;
PG/2020/179766 del 29 febbraio 2020; PG/2020/191369 del
effectiveness, they should be implemented with other strategies
4 marzo 2020; PG/2020/0210546 del 10 marzo 2020 [in Italian].
(such as prioritizing blood use for hospitalized patients during blood 11. Stanworth SJ, New HV, Apelseth TO, et al. Effects of the COVID-19
shortage events) and integrated in appropriate risk management pandemic on supply and use of blood for transfusion. Lancet
plans and safety procedures. At the time of writing, the spread of Haematol. 2020;7:e756–64.
12. Franchini M, Farrugia A, Velati C, Zanetti A, Romanò L, Grazzini G,
COVID-19 is still beyond control in many countries worldwide; the
et al. The impact of the SARS-CoV-2 outbreak on the safety and
knowledge acquired during the emergency should not be dispersed, availability of blood transfusions in Italy. Vox Sang. 2020;115:603–5.
but exploited, in order to define a more efficient healthcare 13. Baron D, Franchini M, Goobie S, et al. Patient blood management
management. during the COVID-19 pandemic: a narrative review. Anaesthesia.
2020;75:1105–13.
14. Grazzini G, Berti P, Boccagni P & Standard in Medicina Trasfusionale.
ACKNOWLEDGEMEN TS
3 ed. Milano, Italy: Società Italiana di Medicina Trasfusionale
The authors wish to thank all the staff, donors and blood associations e Immunoematologia; 2017 [in Italian].
volunteers who demonstrated their commitment to patients care 15. Regione Emilia Romagna Atti Amministrativi Giunta Regionale.
during this serious health emergency. Delibera N. 404 del 2/04/2020, seduta n.16 [in Italian].
16. Barriteau CM, Bochey P, Lindholm PF, et al. Blood transfusion utilization in
Conceptualization: D.S., L.M., C.M.; Data curation: D.S., G.M.,
hospitalized COVID-19 patients. Transfusion (Paris). 2020;60:1919–23.
D.S., L.M., Formal analysis: D.S., P.M., P.G.R.; Writinig-original draft: 17. Leung JN, Lee CK. Impact of the COVID-19 – a regional blood cen-
D.S., L.M., C.M.; Review and editing: L.M., C.M.; E.D.M.; Visualization: ter’s perspective. ISBT Sci Ser. 2020;15:362–4.
L.C., T.A.P.; Resources: R.B.; Supervision: R.B., C.M. All authors have 18. Wang Y, Han W, Pan L, Wang C, Liu Y, Hu W, et al. Impact of COVID-19
on blood centres in Zhejiang Province China. Vox Sang. 2020;115:502–6.
read and agreed to the published version of the manuscript.
19. Pandey HC, Coshic P, Chippy CS, et al. Blood supply management in
times of SARS-CoV-2 pandemic – challenges, strategies adopted,
CONF LICT OF IN TE RE ST and the lessons learned from the experience of a hospital-based
The authors declare no conflict of interest. blood Centre. Vox Sang. 2020;116:497–503.
242 SCHIROLI ET AL.

20. Gazzetta Ufficiale Serie Generale n. 45 del 23/02/2020. Decreto del 25. Granero V, Manzini PM, Bordiga AM. A multimodal practical
Presidente del Consiglio dei Ministri. Disposizioni attuative approach in a transfusion medicine Centre to face the COVID-19
del decreto-legge 23 febbraio 2020, n. 6, recante misure urgenti in outbreak. Blood Transfus. 2020;18:235–6.
materia di contenimento e gestione dell’emergenza epidemiologica
da COVID-19 [in Italian].
21. Valeri CR, Feingold H, Marchionni LD. A simple method for freezing SUPPORTING INFORMATION
human platelets using 6% dimethylsulfoxide and storage at-80 C. Additional supporting information may be found online in the
Blood. 1974;43:131–6.
Supporting Information section at the end of this article.
22. Yazer MH, Jackson B, Pagano M, Rahimi-Levene N, Peer V, Bueno JL,
et al. Vox Sanguinis international forum on transfusion services’
response to COVID-19: summary. Vox Sang. 2020;115:536–42.
23. Murphy C, Jackson B, Fontaine M. Tools for rapid analysis of blood How to cite this article: Schiroli D, Merolle L, Molinari G, et al.
usage and inventory during the COVID-19 pandemic. Transfusion. The impact of COVID-19 outbreak on the Transfusion
2020;60:2199–202. Medicine Unit of a Northern Italy Hospital and Cancer Centre.
24. Colpo A, Astolfi L, Tison T, de Silvestro G, Marson P. Impact of
Vox Sang. 2022;117:235–242.
COVID-19 pandemic in the activity of a therapeutic apheresis unit in
Italy. Transfus Apher Sci. 2020;59:102925.
Received: 16 February 2021 Revised: 22 June 2021 Accepted: 26 June 2021

DOI: 10.1111/vox.13181

ORIGINAL ARTICLE

Immediate intravenous iron administration improves

anaemia recovery following total knee arthroplasty:
A propensity-matched analysis

Hee-Sun Park1 | Seong-Il Bin2 | Ha-Jung Kim1 | Jinsun Kim1 | Hyungtae Kim1 |
Youngjin Ro1 | Won Uk Koh1

1
Department of Anesthesiology and Pain
Medicine, Asan Medical Center, University of Abstract
Ulsan College of Medicine, Seoul, South Korea
Background and Objectives: Patients who undergo total knee arthroplasty (TKA)
2
Department of Orthopaedic Surgery, Asan
Medical Center, University of Ulsan College of
have a risk of postoperative anaemia. This observational study evaluated whether
Medicine, Seoul, South Korea single-dose intravenous ferric carboxymaltose (FCM) administered immediately after
TKA facilitates the correction of anaemia.
Correspondence
Won Uk Koh, Department of Anesthesiology Materials and Methods: We retrospectively analysed 722 patients who underwent
and Pain Medicine, Asan Medical Center,
primary TKA. The FCM group receiving 1000 mg intravenous FCM within one post-
University of Ulsan College of Medicine, 88
Olympic Ro 43-gil, Songpa-gu, Seoul 05505, operative hour was compared with the non-FCM group that did not receive the
South Korea.
medication. A propensity score matching with multiple logistic regression analysis
Email: [email protected]
was used to minimize intergroup differences in the baseline characteristics and
Funding information postoperative blood loss. The rate and severity of postoperative anaemia were
None
compared between the groups, along with haemoglobin (Hb) value, transfusion rate
and complications.
Results: After propensity score matching, 231 patients were included in each group.
In the FCM group, the rate of anaemia at postoperative day (POD) 7 (p = 0.021)
and postoperative week (POW) 5 (p < 0.001) and the transfusion rate were signifi-
cantly lower (p = 0.008). The rate of moderate to severe anaemia at POW-5 was
also significantly lower in the FCM group (p < 0.001). In patients without preopera-
tive anaemia (n = 322), the transfusion rate and rate and severity of anaemia at
POD-7 and POW-5 were significantly lower in the FCM group than in the
non-FCM group.
Conclusion: Postoperative intravenous FCM administration facilitated recovery of
surgery-related anaemia by improving Hb and may reduce the need for transfusion in
TKA patients. Preoperative non-anaemic patients could also benefit from accelerated
recovery by intravenous iron treatment.

KEYWORDS
anaemia, ferric carboxymaltose, intravenous iron, iron deficiency, patient blood management,
total knee arthroplasty

Vox Sanguinis. 2022;117:243–250. wileyonlinelibrary.com/journal/vox © 2021 International Society of Blood Transfusion. 243
244 PARK ET AL.

I N T R O D U CT I O N postoperative blood loss exceeding 1000 ml and (6) patients with

popliteal artery injury during surgery. This study compared patients
Postoperative anaemia is common after total knee arthroplasty (TKA), who received IV ferric carboxymaltose (FCM; Ferinject®, Vifor Int.,
due to considerable postoperative blood loss of >500 ml [1, 2]. Anae- St. Gallen, Switzerland) (the FCM group) with a non-FCM group, who
mia is associated with increased cardiovascular risk as well as did not receive this treatment.
increased morbidity and mortality after non-cardiac surgery and
among elderly patients [3–5]. Allogeneic blood transfusion is among
the methods used for rapidly correcting anaemia, but it can be associ- Data collection
ated with adverse outcomes, such as surgical site infection (2.88%)
[6], transfusion-related acute lung injury (0.08%–15%), and circulatory We reviewed the electronic medical records and collected data on
overload (1%–11%) [7]. age, sex, height, weight, body mass index, hypertension, diabetes,
To correct preoperative anaemia, iron preparations with or with- underlying disease, and American Society of Anaesthesiologists (ASA)
out erythropoiesis-stimulating agents (ESAs) are recommended for physical status. Preoperative medication included the use of anticoag-
patients who have iron deficiency with preoperative haemoglobin ulants or non-vitamin K–antagonist oral anticoagulants.
(Hb) less than 13.0 g/dl [8]. This process should be worked up and ini- Intraoperative variables included the use of tranexamic acid, type
tiated 4–6 weeks before surgery. Multiple publications have reported of anaesthesia (general or regional anaesthesia), duration of operation
on the impact of preoperative iron treatment in the context of elec- and tourniquet use. Postoperative data included the rate and volume
tive surgery [9–12]. of perioperative allogenic red blood cell transfusion, oral iron adminis-
Many patients who have undergone major surgery are at risk of tration (200 mg iron acetyl transferrin or 256 mg dried ferrous sul-
being discharged with moderate to severe anaemia. The presence phate), postoperative blood loss (the amount of postoperative drains
of moderate or severe anaemia in elderly patients who have multiple during 48 h), postoperative clinical complications, readmission rate
comorbidities is known to negatively affect clinical and functional out- until 3 months after discharge, 1-year mortality and overall mortality.
comes [13]. As part of the Patient Blood Management (PBM) [8] for The length of hospital stay was calculated from the day of surgery
managing individualized care for each patient, iron replacement can until hospital discharge. We collected serious adverse events associ-
be used to improve Hb levels and may reduce the need for allogeneic ated with FCM infusion [21].
blood transfusion. However, few studies have focused on the efficacy Postoperative Hb and anaemia were evaluated at three desig-
of postoperative intravenous (IV) iron therapy to improve postopera- nated time-points: PODs 1 and 7 and POW-5. For some patients
tive anaemia in patients who undergo elective surgery [14–19]. (if they did not have Hb results from POD-7), Hb results from PODs
We retrospectively evaluated whether immediate postoperative 4 to 6 were considered in POD-7s results. The POW-5 Hb results
single high-dose IV iron treatment could facilitate recovery of postop- included Hb results from between PODs 29 and 45.
erative anaemia, mitigate the severity of postoperative anaemia and
decrease transfusion rate among patients who have undergone
unilateral TKA. Perioperative patient blood management and iron
preparation administration strategy

METHODS Intraoperatively, a tourniquet was applied to all patients, and it was

removed at the end of surgery after the application of the surgical
This retrospective observational study was approved by the institu- wound dressing. Since the end of 2016, a bolus dose of tranexamic
tional review board of the study centre (2019-0223). Data of patients acid (TXA) (10 mg/kg) has been administered intraoperatively to
who underwent elective primary unilateral TKA between January patients at our institution. However, TXA was not used in patients
2016 and December 2018 at a single tertiary teaching hospital were allergic to TXA and in case of other contraindications, such as a past
collected. The requirement for informed consent was waived. medical history of cerebral ischaemia or infarction, ischaemic heart
disease, deep vein thrombosis and other thromboembolic conditions.
Patients in the FCM group received 1000 mg IV FCM for 30–
Study population 60 min on the operation day 1 h after surgery. The patients were
closely monitored and evaluated during and after FCM injection for
Among 1195 identified patients, the following were excluded: any discomfort or possible side effects. If the severity of reaction was
(1) patients who underwent bilateral staggered TKA (the two severe/life-threatening, it was recorded as a serious adverse event
arthroplasty are performed on different days, within 7 days after the [21]. We calculated the estimated mean iron requirement according
first procedure) [20], (2) patients with haematological and renal disor- to the adopted Ganzoni formula, with a target Hb of 13.0 g/dl [22].
ders (serum creatinine level > 1.5 mg/dl), (3) patients who received Surgeons and the attending anaesthesiologists were encouraged
iron preparations or transfusion preoperatively, (4) patients with no to follow the restrictive transfusion triggers of Hb <8.0 g/dl through-
Hb results from postoperative week (POW) 5, (5) patients with 48 h out the entire perioperative period. If patients presented with
POSTOPERATIVE HIGH-DOSE IV IRON INFUSION 245

symptoms of significant hemodynamic instability despite adequate Statistical analysis

fluid administration and the use of vasopressor was essential, alloge-
neic transfusion of packed red blood cells was permitted even for Hb Continuous variables are reported as mean with standard deviation or
≥8.0 g/dl. Every patient received postoperative oral iron during the median with interquartile range (IQR), as appropriate. Student’s t-test
hospital stay unless they could not tolerate it. or Mann–Whitney test was used to compare continuous variables.
Categorical variables are expressed as frequencies and percentages
and were analysed using the chi-square test or Fisher’s exact test, as
Outcome variables appropriate.
To reduce the effect of potentially confounding factors, we used
The primary outcomes included the difference in Hb and rate of propensity score (PS) matched analysis to modify intergroup differ-
anaemia at each time-point. The secondary outcome of the study ences between patients with and without FCM administration. PS
was the difference in transfusion rate between two groups. Anae- was calculated through multiple logistic regression, including the base-
mia was defined as an Hb concentration < 12.0 g/dl in women line characteristics shown in Table 1 and anaesthetic method as
and < 13.0 g/dl in men. Anaemia was further categorized into mild covariates. To minimize the effect of surgery-associated blood loss,
(women: Hb 11.0–11.9 g/dl; men: Hb 11.0–12.9 g/dl), moderate postoperative blood loss—whether above 500 ml or not—was also
(Hb 8.0–10.9 g/dl for women and men) and severe (Hb < 8.0 g/dl included as a covariate. There were no missing values in the baseline
for women and men) according to World Health Organization characteristics. Model discrimination was assessed with a c-statistic of
(WHO) guidelines [23]. We also performed the analysis of postop- 0.732, and model calibration was assessed with Hosmer–Lemeshow
erative Hb, perioperative transfusion and rate of postoperative statistics (chi-square = 4.017, degrees of freedom = 8, p = 0.856).
anaemia in preoperatively non-anaemic patients between both We subsequently used the derived PS to create PS-matched pairs at a
study groups. ratio of 1:1 using greedy matching with a calliper width of 0.1 without

TABLE 1 Baseline characteristics of the entire cohort

Total set After PS matching

Non-FCM group FCM group Non-FCM group FCM group

N = 386 N = 336 p-Value Stdiff N = 231 N = 231 p-Value Stdiff

Age (y) 70.1  6.11 70.0  6.40 0.786 0.020 70.2  5.96 70.1  6.68 0.848 0.018
Female 333 (86.3) 309 (92.0) 0.021 0.184 213 (92.2) 211 (91.3) 0.866 0.032
Height (cm) 153.8  7.00 153.4  6.37 0.514 153.1  6.72 153.6  6.46 0.323
Weight (kg) 66.5  14.66 64.5  14.52 0.073 64.7  13.69 65.4  16.28 0.654
Body mass index (kg/m2) 28.1  5.62 27.3  5.64 0.090 0.127 27.6  5.19 27.6  6.41 0.939 0.007
Diabetes 106 (27.5) 78 (23.2) 0.222 0.099 55 (23.8) 54 (23.4) 1.000 0.010
Hypertension 260 (67.4) 211 (62.8) 0.228 0.096 145 (62.8) 152 (65.8) 0.560 0.063
Underlying disease 85 (22.0) 94 (28.0) 0.078 0.138 61 (26.4) 48 (20.8) 0.189 0.133
Ischaemic heart disease 35 (9.1) 35 (10.4) 0.628 24 (10.4) 18 (7.8) 0.418
Cerebrovascular disease 33 (8.55) 32 (9.5) 0.744 26 (11.3) 18 (7.8) 0.267
Pulmonary disease 11 (2.85) 21 (6.2) 0.042 8 (3.5) 9 (3.9) 1.000
Nephrotic disease 16 (4.15) 19 (5.7) 0.442 11 (4.8) 18 (7.8) 0.418
ASA 0.056 0.154 1.000 <0.001
1 or 2 344 (89.1) 314 (93.5) 216 (93.5) 216 (93.5)
3 42 (10.9) 22 (6.5) 15 (6.5) 15 (6.5)
Anticoagulants/NOACs 126 (32.6) 100 (29.8) 0.452 0.062 72 (31.2) 66 (28.6) 0.611 0.057
Preoperative Hb (g/dl) 12.7  1.27 12.6  1.16 0.212 0.093 12.5  1.21 12.6  1.15 0.419 0.075
a
Postoperative blood loss <0.001 0.325 0.626 0.054
≤500 ml 200 (51.8) 227 (67.6) 155 (66.2) 149 (63.7)
>500 ml 186 (48.2) 109 (32.4) 78 (33.8) 84 (36.4)

Note: Data are expressed as number of patients (%), mean  standard deviation or median [interquartile range].
Abbreviations: ASA, American Society of Anesthesiologists physical status; FCM, ferric carboxymaltose; NOACs, non-vitamin K-antagonist oral
anticoagulants; PS, propensity score; Stdiff, standardized difference.
a
48 h postoperative drainage amount.
246 PARK ET AL.

FIGURE 1 Flowchart of the study

TABLE 2 Intraoperative and postoperative variables and outcomes after propensity score matching

Total set After PS matching

Non-FCM group FCM group Non-FCM group FCM group

N = 386 N = 336 p-Value N = 231 N = 231 p-Value Stdiff

Intraoperative variables
Tranexamic acid use 235 (60.9) 231 (68.8) 0.033 152 (65.8) 167 (72.3) 0.159 0.141
General anaesthesia 304 (78.8) 154 (45.8) <0.001 152 (65.8) 145 (62.8) 0.56 0.063
Operation time (min) 106.2  18.9 124.8  14.1 <0.001 105.8  19.3 124.8  13.4 <0.001 1.137
Tourniquet time (min) 105.1  20.2 118.5  18.4 <0.001 105.0  19.4 118.7  19.1 <0.001 0.713
Postoperative variables
Transfusion 46 (11.9) 20 (6.0) 0.008 31 (13.4) 12 (5.2) 0.008 0.286
Red blood cell (units) 0.26  0.77 0.09  0.46 0.001 0.28  0.75 0.07  0.32 <0.001 0.360
Postoperative blood lossa (ml) 496.3  285.7 402.1  245.3 <0.001 423.4  279.7 418.5  244.2 0.842 0.019
Postoperative oral ironb 365 (94.6) 322 (95.9) 0.534 214 (92.6) 225 (97.4) 0.032 0.220
Length of stay (days)c 8.4  2.9 7.9  2.2 0.003 8.2  2.9 7.6  1.4 0.004 0.220
Readmission rates 4 (1.0) 4 (1.2) 1.000 0 2 (0.9) 0.479 0.132
1-Year mortality 0 0 1.000 0 0 1.000 <0.001
Overall mortality 2 (0.5) 2 (0.6) 1.000 2 (0.8) 1 (0.4) 1.000 0.054
Complication 83 (21.5) 67 (19.9) 0.671 51 (22.1) 45 (19.5) 0.566 0.064
Nephrotic 13 (3.4) 18 (5.4) 0.276 8 (3.5) 13 (5.6) 0.834
Cerebrovascular 20 (5.2) 19 (5.7) 0.946 10 (4.3) 13 (5.6) 0.834
Pulmonary 14 (3.6) 7 (2.1) 0.297 10(4.3) 6 (2.6) 0.786
Vascular 12 (3.1) 0 0.010 8 (3.5) 0 0.013
Psychiatric 12 (3.1) 15 (4.5) 0.471 8 (3.5) 11 (4.7) 0.323
Neurologic 16 (4.2) 9 (2.7) 0.364 11(4.8) 5 (2.2) 0.574
Infection 8 (2.1) 9 (2.7) 0.796 5 (2.2) 7 (3.0) 0.770
Urologic 12 (3.1) 0 0.010 8 (3.5) 0 0.013

Note: Data are expressed as number of patients (%), mean  standard deviation or median [interquartile range].
Abbreviations: FCM, ferric carboxymaltose; PS, propensity score; Stdiff, standardized difference.
a
48 h postoperative drainage amount.
b
Oral iron administration from postoperative day 1.
c
Length of hospital stay after surgery.
POSTOPERATIVE HIGH-DOSE IV IRON INFUSION 247

T A B L E 4 Comparison of variables among preoperative non-

anaemic patients after propensity score matching

Non-FCM
group FCM group

N = 161 N = 161 p-Value

Preoperative Hb (g/dl) 13.1  0.83 13.2  0.79 0.438
Postoperative Hb (g/dl)
POD-1 10.8  1.12 11.2  1.07 0.001
POD-7 9.7  1.16 10.5  1.31 <0.001
POW-5 12.1  0.90 12.9  1.01 <0.001
Transfusion 11 (6.8) 2 (1.2) 0.024
Red blood cells (units) 0.12  0.47 0.01  0.11 0.010
F I G U R E 2 Haemoglobin alteration after unilateral total knee
Postoperative anaemia
arthroplasty. *p < 0.05
POD-1 137 (85.1) 127 (78.9) 0.192
POD-7 155 (96.3) 143 (88.8) 0.020
POW-5 74 (46.0) 30 (18.6) <0.001
T A B L E 3 The rate of postoperative anaemia and its severity after Ratio moderate to severe anaemia
propensity scored matching POD-1 112 (69.6) 96 (59.6) 0.080
Non-FCM group FCM group POD-7 132 (82.0) 98 (60.9) <0.001

N = 231 N = 231 p-Value POW-5 19 (11.8) 4 (2.5) 0.002

Overall anaemia Length of stay daysa 8.1  2.4 7.7  1.3 0.107

POD-1 207 (89.6) 197 (85.3) 0.206 Postoperative blood 442.5  279.9 437.6  255.1 0.870
lossb (ml)
POD-7 225 (97.4) 213 (92.2) 0.021
POW-5 140 (60.6) 75 (32.5) <0.001 Note: Data are expressed as number of patients (%), mean  standard
deviation or median [interquartile range].
Moderate to severe anaemia
Abbreviations: FCM, ferric carboxymaltose; POD, postoperative day;
POD-1 180 (77.9) 161 (69.7) 0.057 POW, postoperative week.
a
POD-7 197 (85.3) 162 (70.1) <0.001 Length of hospital stay after surgery.
b
48 h postoperative drainage amount.
POW-5 54 (23.4) 19 (8.2) <0.001

Note: Data are expressed as number of patients (%). Anaemia grade

according to WHO guidelines: mild 11.0–11.9 g/dl for women and 11.0–
12.9 g/dl for men, moderate 8.0–10.9 g/dl and severe <8.0 g/dl. Intraoperative and postoperative outcome variables are pres-
Abbreviations: FCM, ferric carboxymaltose; POD, postoperative day; ented in Table 2. The non-FCM group had a significantly higher rate
POW, postoperative week. of transfusion in both the crude (11.9% and 6.0%, p = 0.008) and PS-
matched set (13.4% and 5.2%, p = 0.008). The duration of hospital
stay was significantly shorter in the FCM group than in the non-FCM
replacement. After all PS matches were completed, we assessed the group (p = 0.004). Readmission within 3 months after discharge and
balance of baseline covariates between PS-matched pairs using paired mortality did not differ significantly.
t-tests or McNemar’s tests for continuous and categorical variables, as The changes in Hb at each time-point are shown in Figure 2. At
appropriate. p-Values < 0.05 were considered statistically significant. each time-point in the postoperative period, the Hb level was signifi-
Statistical analyses were performed using R software version 3.5.0 cantly higher in the FCM group (10.4  1.3 vs. 10.7  1.25,
(R Foundation for Statistical Computing, Vienna, Austria). 9.4  1.17 vs. 10.1  1.36 and 11.7  1.05 vs. 12.6  1.12,
p = 0.008, <0.001 and <0.001), respectively for POD-1, POD-7 and
POW-5 (Figure 2).
RESULTS The overall rate and the severity of postoperative anaemia are
summarized in Table 3. The rate of anaemia was significantly lower in
The data of 722 patients who underwent primary unilateral TKA the FCM group at POD-7 (97.4% vs. 92.2%, p = 0.021) and POW-5
were analysed in this study (Figure 1). The baseline characteristics (60.6% vs. 32.5%, p < 0.001). In addition, the FCM group showed sig-
included in the PS matching are shown in Table 1. The calculated nificantly lower moderate to severe anaemia at POD-7 (85.3%
mean estimated iron requirement was 847.7  219.8 mg and vs. 70.1%, p < 0.001) and POW-5 (23.4% vs. 8.2%, p < 0.001). Among
899.2  192.5 mg at POD-1 in the FCM and non-FCM groups preoperative non-anaemic patients in the matched set (Table 4), the
(p = 0.63), respectively. rate of anaemia and its severity were significantly lower in the FCM
248 PARK ET AL.

group. No serious adverse events associated with IV FCM administra- transfusion rate [28]. This combination method has recently been rec-
tion were noted in the electronic medical records. ommended at a PBM conference for major orthopaedic surgeries with
preoperative Hb less than 13 g/dl [8].
The mortality and readmission rate in this study were not differ-
DISCUSSION ent, but we did not present other clinical outcomes such as anaemia-
associated morbidity/mortality and quality of life (QoL)/functional
This study evaluated the effect of immediate postoperative 1000 mg outcomes. The effect of postoperative iron therapy on
IV iron supplementation on the recovery of anaemia after unilateral QoL/functioning has been assessed in several previous studies; how-
TKA. Iron supplementation was associated with a lower rate of anae- ever, the result was inconclusive for improving these patient-centred
mia due to increase in Hb. The rate of moderate to severe anaemia at outcomes [14, 16, 19, 33]. For anaemia correction, postoperative IV
POD-7 and POW-5 was significantly lower in the FCM group. Fur- iron administration increased the mean Hb level only by 0.34 g/dl
thermore, postoperative IV iron therapy was associated with a (95% CI: 0.118–0.562) [29], which is a small increase and is not likely
reduced requirement for perioperative transfusion. to have much clinical meaning. Moreover, the number and quality of
Even patients without preoperative anaemia will frequently studies investigating the effects of postoperative anaemia treatment
develop anaemia following major surgery due to blood loss. In addi- on patient-centred outcomes are inadequate, as they were relatively
tion, non-anaemic patients may have undetected iron deficiency [2]. small-sized studies and the timing of evaluation was different. Further
Surgery-associated inflammation also hinders normal erythropoiesis in studies are necessary to evaluate recovery from anaemia and its
patients. Inflammation increases the synthesis of hepcidin, which related outcomes following postoperative IV iron treatment.
inhibits the mobilization of iron from macrophages and enterocytes The mean estimated iron requirement was 847.7  219.8 mg and
into the systemic circulation [24]. Therefore, despite an adequate iron 899.2  192.5 for the FCM and non-FCM groups at POD-1, respec-
store, it cannot be used effectively for erythropoiesis; this is known as tively. Thus, if low-dose iron was administered (e.g., 200 mg of iron
iron sequestration, which delays anaemia recovery even in patients sucrose), patients would require repeated doses to replenish body
without pre-existing anaemia or iron deficiency. iron stores. Study findings that did not support postoperative iron
Iron replacement following surgery has been studied for the man- therapy for correcting anaemia after surgery [17, 34] may be attrib-
agement of postoperative anaemia and iron deficiency. Postoperative uted to relative low doses (100–600 mg) of iron. Patients who
oral iron administration has been reported to be ineffective for received high dose (≒1000 mg) IV iron have yielded favourable results
treating functional iron deficiency or iron sequestration [25]. In con- of improved postoperative anaemia [19, 27]. This suggests that a suf-
trast, parenteral iron administration bypasses hepcidine-mediated ficient dose of iron is required to facilitate recovery from anaemia
pathways. Thus, IV iron therapy is recommended to increase Hb levels after major surgery. Therefore, a single high dose (up to 1000 mg) of
in patients with functional iron deficiency [26]. Published studies [14– IV FCM is appropriate in the postoperative setting.
16, 18, 27, 28] and a systematic review and meta-analysis [29] This study had several limitations. First, this was a retrospective
reported that postoperative IV iron treatment can improve Hb levels, observational study, and patients were given IV iron without pre-
particularly at 2–4 weeks after surgery. Our results are consistent assessing iron metabolism test. This test reflects the cause of anaemia,
with those of previous studies. For this reason, the international con- which can be an important covariate with regards to the outcome of
sensus statement [26] recommends the early use of postoperative IV postoperative IV iron treatment [26]. The PBM recommendations
iron for patients with iron deficiency anaemia and functional iron include evaluating iron metabolism preceding surgery. If there are no
deficiency. preoperative tests, testing is recommended within 24 h after surgery
The impact of iron replacement alone, particularly administered when ferritin has not yet been elevated in response to inflammation
within 7 days before or after surgery, on reducing the requirement for [26]. Second, ESAs can be useful in iron sequestration and functional
transfusion is unclear [16, 18, 19, 27, 29]. Although IV iron therapy iron deficiency, particularly in the context of surgery-associated anae-
can rapidly replenish body iron stores, it still requires a sufficient dura- mia. Although IV iron with short-acting ESAs has been recommended
tion of treatment before surgery [30]. In addition, protocols and for major orthopaedic surgery only [8], ESA use in these patients is
guidelines often call for restrictive transfusion thresholds based on not yet licenced in some country. Furthermore, the possible side
evidence regarding the risks associated with transfusion [31]. These effects and associated costs should be considered. Lastly, we used an
factors may have contributed to the lack of difference in transfusion anaemia criteria <12.0 g/dl for women, a lower cut-off than <13.0 g/
rates. According to the systematic review and meta-analysis in 2018, dl for men. Some studies have questioned this definition and shown
postoperative IV iron alone did not significantly decrease transfusion that mildly low and even low-normal Hb have been adversely associ-
requirement (relative risk, 0.75; 95% confidence interval [CI]: 0.53– ated with mobility function in elderly women [35]. We agree on the
1.07) [29]. In our previous randomized controlled trial (RCT), IV iron necessity for re-evaluating this criteria; however so far, the WHO
therapy alone showed no difference in transfusion rate among guideline for anaemia has been used in most studies, even studies
patients who underwent TKA and total hip arthroplasty (THA) [32]. with older participants. Despite these limitations, our data demon-
However, in other prospective RCT, IV iron combined with ESAs on strated some benefit of single high-dose iron replacement given
the day before cardiac surgery was effective in decreasing the immediately postoperatively, without serious adverse reactions
POSTOPERATIVE HIGH-DOSE IV IRON INFUSION 249

associated with iron administration. Furthermore, this study used a 10. Moon T, Smith A, Pak T, Park BH, Beutler SS, Brown T, et al. Preop-
relatively homogeneous study sample consisting of only TKA patients erative anemia treatment with intravenous iron therapy in patients
undergoing abdominal surgery: a systematic review. Adv Ther. 2021;
and minimized the potential effects of confounding factors through
38:1447–1469.
PS matching. 11. Ng O, Keeler BD, Mishra A, Simpson JA, Neal K, Al-Hassi HO, et al.
In conclusion, the administration of high-dose IV iron immediately Iron therapy for preoperative anaemia. Cochrane Database Syst Rev.
after primary unilateral TKA in patients was associated with improved 2019;12:CD011588.
12. Richards T, Baikady RR, Clevenger B, Butcher A, Abeysiri S, Chau M,
Hb recovery and reduced the severity of postoperative anaemia.
et al. Preoperative intravenous iron to treat anaemia before major
Transfusion requirement was also reduced in the iron therapy group. abdominal surgery (PREVENTT): a randomised, double-blind, con-
This finding was also observed in patients without preoperative anae- trolled trial. Lancet. 2020;396:1353–1361.
mia, suggesting the possible expansion of the clinical application of IV 13. Woodman R, Ferrucci L, Guralnik J. Anemia in older adults. Curr Opin
Hematol. 2005;12:123–128.
iron therapy to this group of patients.
14. Bisbe E, Molto  L, Arroyo R, Muniesa JM, Tejero M. Randomized trial
comparing ferric carboxymaltose vs oral ferrous glycine sulphate for
ACKNOWLEDGEMEN TS postoperative anaemia after total knee arthroplasty. Br J Anaesth.
We thank the workforce and staff at our medical centre for their 2014;113:402–409.
15. Muñoz M, Go  mez-Ramírez S, Martín-Montañez E, Naveira E, Seara J,
assistance during this study as well as the patients who participated in
Pavía J. Cost of post-operative intravenous iron therapy in total
the study.
lower limb arthroplasty: a retrospective, matched cohort study.
H.-S.P. analysed data and wrote the first draft and final draft of Blood Transfus. 2014;12:40–49.
the manuscript. S.-I.B. designed study and wrote the manuscript. H.-J. 16. Khalafallah AA, Yan C, Al-Badri R, Kirkby BE, Ingram E, Gray Z,
K. and J.K. collected data and analysed of data. H.K. and Y.R. revised et al. Intravenous ferric carboxymaltose versus standard care in
the management of postoperative anaemia: a prospective, open-
the manuscript. W.U.K. supervised the conduct of the study. All
label, randomised controlled trial. Lancet Haematol. 2016;3:
authors read and revised the manuscript. e415–425.
17. Karkouti K, McCluskey SA, Ghannam M, Salpeter MJ, Quirt I,
CONF LICT OF IN TE RE ST Yau TM. Intravenous iron and recombinant erythropoietin for the
treatment of postoperative anemia. Can J Anaesth. 2006;53:11–19.
The authors declare that they have no conflicts of interest associated
18. Kim SK, Seo WY, Kim HJ, Yoo JJ. Postoperative intravenous ferric
with this study. carboxymaltose reduces transfusion amounts after orthopedic hip
surgery. Clin Orthop Surg. 2018;10:20–25.
ORCID 19. Kim YW, Bae JM, Park YK, Yang HK, Yu W, Yook JH, et al. Effect of
intravenous ferric carboxymaltose on hemoglobin response among
Won Uk Koh https://orcid.org/0000-0003-4881-1884
patients with acute isovolemic anemia following gastrectomy: the
FAIRY randomized clinical trial. JAMA. 2017;317:2097–2104.
RE FE R ENC E S 20. Koh WU, Kim H-J, Park H-S, Jang MJ, Ro YJ, Song JG. Staggered
1. Larocque BJ, Gilbert K, Brien WF. Prospective validation of a point rather than staged or simultaneous surgical strategy may reduce the
score system for predicting blood transfusion following hip or knee risk of acute kidney injury in patients undergoing bilateral TKA. JBJS.
replacement. Transfusion. 1998;38:932–937. 2018;100:1597–1604.
2. Muñoz M, Laso-Morales M, Go  mez-Ramírez S, Cadellas M, Núñez- 21. Rampton D, Folkersen J, Fishbane S, Hedenus M, Howaldt S,
Matas MJ, García-Erce JA. Pre-operative haemoglobin levels and iron Locatelli F, et al. Hypersensitivity reactions to intravenous iron: guid-
status in a large multicentre cohort of patients undergoing major ance for risk minimization and management. Haematologica. 2014;
elective surgery. Anaesthesia. 2017;72:826–834. 99:1671–1676.
3. Spahn DR. Anemia and patient blood management in hip and knee 22. Ganzoni AM. Intravenous iron-dextran: therapeutic and experimental
surgery. Anesthesiology. 2010;113:482–495. possibilities. Schweiz Med Wochenschr. 1970;100:301–303.
4. Beattie WS, Karkouti K, Wijeysundera DN, Tait G. Risk associated 23. World Health Organization. Haemoglobin concentrations for the
with preoperative anemia in noncardiac surgery: a single-center diagnosis of anaemia and assessment of severity. Vitamin and Min-
cohort study. J Am Soc Anesthesiologists. 2009;110:574–581. eral Nutrition Information System. Geneva: World Health Organiza-
5. Culleton BF, Manns BJ, Zhang J, Tonelli M, Klarenbach S, tion; 2011.
Hemmelgarn BR. Impact of anemia on hospitalization and mortality 24. Biesma D, Avd W, Beguin Y, Kraaijenhagen RJ, Marx JJ. Post-
in older adults. Blood. 2006;107:3841–3846. operative erythropoiesis is limited by the inflammatory effect of sur-
6. Kim JL, Park JH, Han SB, Cho IY, Jang KM. Allogeneic blood transfu- gery on iron metabolism. Eur J Clin Invest. 1995;25:383–389.
sion is a significant risk factor for surgical-site infection following 25. Vaglio S, Prisco D, Biancofiore G, Rafanelli D, Antonioli P, Lisanti M,
total hip and knee arthroplasty: a meta-analysis. J Arthroplasty. et al. Recommendations for the implementation of a patient blood
2017;32:320–325. management programme. Application to elective major orthopaedic
7. Ponnusamy KE, Kim TJ, Khanuja HS. Perioperative blood transfu- surgery in adults. Blood Transfus. 2016;14:23–65.
sions in orthopaedic surgery. JBJS. 2014;96:1836–1844. 26. Muñoz M, Acheson AG, Bisbe E, Butcher A, Go  mez-Ramírez S,
8. Mueller MM, Van Remoortel H, Meybohm P, Aranko K, Aubron C, Khalafallah AA, et al. An international consensus statement on the
Burger R, et al. Patient blood management: recommendations from the management of postoperative anaemia after major surgical proce-
2018 Frankfurt consensus conference. JAMA. 2019;321:983–997. dures. Anaesthesia. 2018;73:1418–1431.
9. Petis SM, Lanting BA, Vasarhelyi EM, Naudie DD, Ralley FE, 27. Johansson P, Rasmussen A, Thomsen L. Intravenous iron iso-
Howard JL. Is there a role for preoperative iron supplementation in maltoside 1000 (Monofer®) reduces postoperative anaemia in preop-
patients preparing for a total hip or total knee arthroplasty? eratively non-anaemic patients undergoing elective or subacute
J Arthroplasty. 2017;32:2688–2693. coronary artery bypass graft, valve replacement or a combination
250 PARK ET AL.

thereof: a randomized double-blind placebo-controlled clinical trial quality of life during the first two weeks postoperatively. Transfu-
(the PROTECT trial). Vox Sang. 2015;109:257–266. sion. 2011;51:71–81.
28. Spahn DR, Schoenrath F, Spahn GH, Seifert B, Stein P, 34. Garrido-Martín P, Nassar-Mansur MI, de la Llana-Ducro s R, Virgos-
Theusinger OM, et al. Effect of ultra-short-term treatment of 
Aller TM, Fortunez PM, Avalos-Pinto R, et al. The effect of intrave-
patients with iron deficiency or anaemia undergoing cardiac surgery: nous and oral iron administration on perioperative anaemia and
a prospective randomised trial. Lancet. 2019;393:2201–2212. transfusion requirements in patients undergoing elective cardiac sur-
29. Perelman I, Winter R, Sikora L, Martel G, Saidenberg E, gery: a randomized clinical trial. Interact Cardiovasc Thorac Surg.
Fergusson D. The efficacy of postoperative iron therapy in 2012;15:1013–1018.
improving clinical and patient-centered outcomes following sur- 35. Chaves PH, Ashar B, Guralnik JM, Fried LP. Looking at the relationship
gery: a systematic review and meta-analysis. Transfus Med Rev. between hemoglobin concentration and prevalent mobility difficulty
2018;32:89–101. in older women. Should the criteria currently used to define anemia in
30. Besarab A, Hörl WH, Silverberg D. Iron metabolism, iron deficiency, older people be reevaluated? J Am Geriatr Soc. 2002;50:1257–1264.
thrombocytosis, and the cardiorenal anemia syndrome. Oncologist.
2009;14:22–33.
31. Padhi S, Kemmis-Betty S, Rajesh S, Hill J, Murphy MF. Blood transfu-
sion: summary of NICE guidance. BMJ. 2015;351:h5832. How to cite this article: Park H-S, Bin S-I, Kim H-J, Kim J,
32. Park H-S, Kim T-Y, Kim H-J, Ro YJ, Jang HY, Koh WU. The effect of Kim H, Ro Y, et al. Immediate intravenous iron administration
intraoperative ferric carboxymaltose in joint arthroplasty patients: a
improves anaemia recovery following total knee arthroplasty:
randomized trial. J Clin Forensic Med. 2019;8:1674.
33. So-Osman C, Nelissen R, Brand R, Brand A, Stiggelbout AM. Postop-
A propensity-matched analysis. Vox Sang. 2022;117:243–50.
erative anemia after joint replacement surgery is not related to
Received: 14 April 2021 Revised: 25 June 2021 Accepted: 26 June 2021

DOI: 10.1111/vox.13180

ORIGINAL ARTICLE

Impacts of COVID-19 and elective surgery cancellations

on platelet supply and utilization in the Canadian Province
of British Columbia

Kevin E. Shopsowitz1 | Christina Lim2 | Andrew W. Shih1,3,4 | Nick Fishbane2 |

Brian R. Berry5 | Mark Bigham6 | Tanya Petraszko6,7 | Jacqueline Trudeau8 |
Maureen Wyatt9 | Matthew T. S. Yan1,6,10 | Douglas Morrison1,2,11

1
Department of Pathology and Laboratory
Medicine, University of British Columbia, Abstract
Vancouver, British Columbia, Canada
Background and Objectives: The coronavirus disease 2019 (COVID-19) pandemic
2
BC Provincial Blood Coordinating Office,
Vancouver, British Columbia, Canada
raised concerns about the vulnerability of platelet supply and the uncertain impact of
3
Department of Pathology and Laboratory the resumption of elective surgery on utilization. We report the impact of COVID-19
Medicine, Vancouver Coastal Health on platelet supply and utilization across a large, integrated healthcare system in the
Authority, Vancouver, British Columbia,
Canada Canadian province of British Columbia (BC).
4
Centre for Blood Research, University of Materials and Methods: Historical platelet use in BC by indication was compiled for
British Columbia, Vancouver, British Columbia,
fiscal year 2010/2011–2019/2020. Platelet collections, initial daily inventory and
Canada
5
Department of Pathology, Vancouver Island disposition data were assessed pre-COVID-19 (1 April 2018–15 March 2020) and
Health Authority, Victoria, British Columbia, for two COVID-19 time periods in BC: a shutdown phase with elective surgeries hal-
Canada
6
ted (16 March–17 May, 2020) and a renewal phase when elective surgeries resumed
Canadian Blood Services, Vancouver, British
Columbia, Canada (18 May–27 September 2020); comparisons were made provincially and for individ-
7
Department of Medicine, University of British ual health authorities.
Columbia, Vancouver, British Columbia, Canada
8
Results: Historically, elective surgeries accounted for 10% of platelets transfused in
Department of Anesthesia, Pharmacology, and
Therapeutics, University of British Columbia, BC. Initial daily supplier inventory increased from baseline during both COVID-19
Vancouver, British Columbia, Canada periods (93/90 units vs. 75 units pre-COVID-19). During the shutdown phase, plate-
9
Department of Pathology and Laboratory
let utilization decreased 10.4% (41 units/week; p < 0.0001), and remained signifi-
Medicine, Interior Health Authority, Kelowna,
British Columbia, Canada cantly decreased during the ensuing renewal period. Decreased platelet utilization
10
Department of Pathology and Laboratory was attributed to fewer transfusions during the shutdown phase followed by a
Medicine, Fraser Health Authority, New
Westminster, British Columbia, Canada decreased discard/expiry rate during the renewal phase compared to pre-COVID-19
11
Department of Pathology and Laboratory (15.2% vs. 18.9% pre-COVID-19; p < 0.0001). Differences in COVID-19 platelet utili-
Medicine, Children’s and Women’s Health
zation patterns were noted between health authorities.
Centre of BC, Vancouver, British Columbia,
Canada Conclusion: Decreased platelet utilization was observed in BC compared to pre-
COVID-19, likely due to a transient reduction in elective surgery as well as practice
Correspondence
Douglas Morrison, BC Children’s Hospital, and policy changes triggered by pandemic concerns.
4500 Oak St., Vancouver, BC V6H 3N1,
Canada. KEYWORDS
Email: [email protected] platelet transfusion, transfusion medicine, transfusion strategy

Funding information
None

Vox Sanguinis. 2022;117:251–258. wileyonlinelibrary.com/journal/vox © 2021 International Society of Blood Transfusion. 251
252 SHOPSOWITZ ET AL.

I N T R O D U CT I O N collections. Buffy coat platelet units, each manufactured from four

whole blood units, comprise 70% of platelet production in the prov-
Blood products rely on a complex infrastructure of donors, distribu- ince. Pre-hospital bacterial testing by CBS allows for an extended
tion networks and hospital transfusion services to meet clinical 7-day platelet shelf life. CBS’s BC inventory fluctuates during the day
demand. Platelets, with a relatively short shelf-life of 5–7 days, are as hospital orders are filled and donor collections from previous days
particularly vulnerable to perturbations to the supply–demand equilib- become available for release to hospitals. At the start of each day, ini-
rium. Concerns over potential platelet shortages are not new, for tial daily CBS inventory is communicated to hospitals.
example, related to ageing donor and recipient populations [1]; how-
ever, the COVID-19 pandemic had the potential to exacerbate these
vulnerabilities in ways not previously experienced by healthcare sys- Data sources
tems. Potential problems included the disruption of donor networks
and blood collection, production and logistics networks (reduced sup- The BC Provincial Blood Coordinating Office Central Transfusion Reg-
ply) and/or by increasing platelet use through large numbers of criti- istry (CTR) is a comprehensive database of all transfusions that have
cally ill COVID-19 patients (increased demand). On the other hand, occurred in the province of BC since 1999. We performed a retro-
decreases in non-COVID-19-related healthcare visits due to both spective analysis of BC platelet utilization for pre- and COVID-19 time
behavioural changes and policies like halting elective surgeries [2, 3], periods. Historical data on platelet use, by indication, were analysed
or policies to extend platelet shelf life, for example, through cold stor- for fiscal years 2010/2011–2019/2020 (FY: 1 April–31 March). Plate-
age [4], could counter-balance these stressors by reducing demand. let transfusions were linked to surgical procedures via the provincial
Understanding the multifaceted influence of COVID-19 on platelet Surgical Patient Registry and considered related to the surgical event
supply and demand is crucial for ongoing resource management and when a patient received platelets from 1 day prior to surgery to
future planning. Data are emerging on the effects of the pandemic on 30 days after the surgery. Surgeries were further differentiated
platelet utilization [5]. Studies from diverse geographical regions have between elective and urgent/emergency procedures. All other platelet
suggested that although COVID-19 infection is associated with transfusions were considered to be non-surgical. Surveillance data on
thrombocytopaenia [6], significant bleeding is uncommon, and hospital- provincial COVID-19 cases and hospitalizations were obtained from
ized patients have relatively low platelet transfusion requirements, with the publicly available COVID-19 Tracker Canada [12]. Total hospital
some centres reporting decreased overall platelet utilization during the admission, intensive care unit (ICU) admissions and surgical volumes
first wave of the pandemic [7–10]. In one study from a tertiary care cen- in BC from March to June 2020 with year-over-year comparisons to
tre in India, a diminished inventory caused by decreased donations was 2019 were available from the Canadian Institute for Health Informa-
offset by a lower demand for platelets during the early phase of the pan- tion [13]. Total population and age data for BC and its HAs were
demic [11]. Additional data from large health systems over longer time obtained from publicly available government data [14]. Median ages
periods are useful to further understand the manifold impacts of COVID- were estimated from population data capped at age 90 years
19 on platelet supply and utilization. (i.e., individuals aged ≥90 were binned together as 90+).
The Canadian province of British Columbia (BC) maintains a cen- Data related to blood donations and the number of platelet units
tralized transfusion database for six health authorities (HAs) that ser- supplied to hospitals within BC were obtained from CBS for the
vice a population of roughly 5 million residents. Here, we report on period from 1 December 2019 to 27 November 2020. Initial daily
BC’s baseline platelet use patterns and experience with platelet supply supplier platelet availability within BC (CBS BC inventory at midnight)
and utilization during the initial 6 months of the COVID-19 pandemic. was obtained for the period from 1 April 2018 to 27 September 2020
This time period encompasses an initial response phase when all elec- (last available data point at the time of writing). Platelet disposition is
tive surgeries were suspended in the province and, subsequently, reported as transfused units and expired/discarded units. The total
when elective surgeries were resumed. To our knowledge, this study number of units is the sum of transfused and expired/discarded units.
represents the largest analysis to-date of system-wide platelet use Daily platelet disposition data were obtained from CTR for 1 April
during the COVID-19 pandemic. 2018 to 27 September 2020. Three time periods are considered
throughout the study: pre-COVID-19 baseline (until 15 March 2020);
COVID-19 shutdown period, when elective surgeries in BC were hal-
MATERIALS AND METHODS ted (16 March 2020 to 17 May 2020), and COVID-19 renewal phase,
when elective surgeries resumed in BC (18 May 2020 to
Platelet supply in BC 27 September 2020).

BC receives blood components from Canadian Blood Services (CBS),

which operates the national blood system outside the province of Statistical analysis
Quebec. While the majority of platelets issued to hospitals in BC are
collected in the province, some are imported from other provinces. The mean of total weekly platelet use in BC was calculated for the
Platelets are produced from both whole blood and apheresis three time periods noted above (pre-COVID-19, shutdown, renewal),
COVID-19’S IMPACT ON PLATELET USE IN BC 253

and statistical testing was performed by analysis of variance with a cardiac surgery – were historically the greatest platelet users by a
post-hoc Tukey test for pairwise comparisons. Linear regression was large margin, each accounting for approximately 33% of surgical plate-
used to assess trend over time for historical annual platelet usage. Ini- let usage. We found there was a significant increase in platelet trans-
tial daily platelet inventory and platelet usage are reported as 7-day fusions over time from FY 2010/2011 to 2019/2020 (+248 units/
rolling average to remove the periodicity of daily counts related to dif- year; p = 0.001), but we did not detect a significant trend when cor-
ferent transfusion practices on weekend versus weekdays. Statistical recting for population growth (p = 0.7) nor in the proportion of plate-
values are all reported as mean with 95% confidence intervals (CIs) lets transfused for surgical events (p = 0.4) over this time period.
unless otherwise indicated. Weekly mean with 95% CI was also calcu-
lated for each of the three largest HAs within the province and for the
sub-categories of weekly transfused and expired platelet units (formal Effect of COVID-19 on provincial platelet supply
statistical testing not carried out for these sub-categories). Province- and demand
wide and HA-specific discard rates were calculated as expired/
discarded units divided by total units for the three time periods. For At the outset of the COVID-19 pandemic, several precautionary mea-
the total provincial data, these rates were statistically compared by sures were put into place in BC, including the suspension of elective
chi-squared test for both overall trend and pairwise comparisons with surgeries between 16 March 2020 and 17 May 2020, which led to a
Bonferroni correction. p-values <0.05 were considered to be statisti- large decrease in planned surgeries in the province from March to
cally significant for all comparisons. June 2020 compared to the previous year (Figure S2A). In March
2020, day surgeries decreased by 35% in BC compared to the previ-
ous year (although BC-specific data were not available for the months
RESULTS leading up to the shutdown, national surgical volumes had been stable
through the end of February 2020) [13]. Surgical volume reached a
Historical platelet use in BC nadir during the middle of the shutdown phase in April, when day sur-
geries, planned inpatient surgeries and total surgeries had decreased
Over the 10-year period beginning with FY 2010/2011, an average of by 81%, 59% and 70%, respectively, compared to April 2019. Of note,
15,642 platelet units per year were transfused in BC, with in addition to the expected decrease in planned surgeries, April also
17,120 units transfused over the last complete fiscal year with avail- saw a 27% decrease in unplanned surgeries. During this initial shut-
able data (2019/2020; Figure 1a). Over the past 10 years, non- down period, BC experienced its first wave of COVID-19, reaching a
surgical indications have accounted for 67.0% of platelet transfusions peak of 149 hospitalized COVID-19 patients during the first week of
(yearly range = 63.4%–72.5%; Figure 1b). The remaining third of April (including a peak of 72 patients in critical care; Figure S3).
platelet units were associated with surgical procedures, of which Concurrently, overall inpatient occupancy and ICU occupancy were
23.2% of the provincial total were transfused for emergency proce- down by 33% and 20%, year over year, respectively (Figure S2B).
dures (range = 19.4%–27.2%) and 9.8% for elective procedures Elective surgeries were resumed on 18 May 2020, and continued
(range = 7.4%–12.6%). Eighty percent of platelets transfused during unabated throughout the second wave of COVID-19 in BC (renewal
the peri-operative period occur within 5 days of surgery (Figure S1). phase), with surgeries beginning to approach baseline levels by June
Two surgical specialties – general surgery (which includes trauma) and (Figure S2A). In late August, COVID-19 hospitalizations again rose

F I G U R E 1 Platelets transfused by indication in BC from fiscal year 2010/2011–2019/2020. (a) Annual transfused platelet units in BC by
indication. (b) Cumulative percentage of transfused platelets by indication over the 10-year time period. BC, British Columbia
254 SHOPSOWITZ ET AL.

from a minimum of <10 to 69 by the end of September with 19 in crit- shown in Figure 2. There was a decrease in CBS national whole blood
ical care (Figure S3). donations from late March to early June, only partly compensated for
The trends in whole blood donations (a proxy for buffy coat– by an increase in apheresis platelet donations (Figure 2a). Towards the
derived platelets) and apheresis platelet donations for baseline and end of May 2020, there was a steep increase in whole blood dona-
COVID-19 time periods, both nationally and specifically for BC, are tions (similar to pre-COVID-19 levels) that was sustained until the end

F I G U R E 2 Weekly platelet donations for (a) all of Canada (excluding Quebec) and (b) British Columbia, before and after the onset of the
coronavirus disease 2019 (COVID-19) pandemic. Whole blood donations are used as a proxy for buffy coat-derived platelets

F I G U R E 3 Initial daily CBS platelet inventory (solid) compared to daily usage (dashed). Both curves represent 7-day rolling averages. The
average initial daily supplier inventory for different time periods are 75.1 units (pre-COVID-19), 92.8 units (shutdown) and 90.1 units (renewal). The
average daily platelet usages (transfused + expired/discarded) for the same time periods are 56.4 units (pre-COVID-19), 50.5 units (shutdown) and
51.1 units (renewal), respectively. Note that for calculation of averages, a longer pre-COVID-19 period was used than what is shown on the graph
(April 2018–March 2020) CBS, Canadian Blood Services; COVID, coronavirus disease 2019
COVID-19’S IMPACT ON PLATELET USE IN BC 255

of 2020, whereas apheresis donations decreased towards baseline. In shutdown and renewal periods, respectively (Figure 3). Comparing
BC specifically, donations remained comparatively stable during the weekly platelet utilization to remove periodicity of variable weekday
initial shutdown phase and showed similar trends of increasing whole and weekend practices, this decrease in platelet usage was found to
blood and decreasing apheresis donations from June onwards be statistically significant (Table 1 and Figure 4a,e). Weekly platelet
(Figure 2b). Overall, the number of platelets issued to BC by CBS usage decreased from a pre-COVID-19 mean of 394.8 units/week
decreased during both the shutdown and renewal phases compared (95% CI, 391.2–398.5) to 353.7 units/week (95% CI, 337.2–370.1)
to the preceding months (Figure S4). during the shutdown period (p < 0.0001) and did not significantly
Initial CBS daily inventory and daily usage for the year leading up change from this level during the renewal period (357.7 units/week
to COVID-19, the initial shutdown period and the renewal period are [95% CI, 346.5–368.9; p = 0.9]).
shown in Figure 3. Initial daily CBS platelet inventory for BC was con-
siderably more variable than daily platelet use, with large oscillations
between 25 and 150 units at baseline. The initial daily CBS inven- Effect of COVID-19 on platelet disposition provincially
tory increased from a pre-COVID-19 baseline average of 75.1 units to and in separate health regions
92.8 units for the shutdown phase and remained elevated at 90.1 units
during the renewal phase. Overall, the platelet supply in BC was more To further understand the effects of the COVID-19 pandemic on
secure from the onset of COVID-19 with initial daily CBS inventory platelet disposition in BC, we broke down overall platelet use into
minima remaining higher than pre-COVID-19, both in absolute terms units transfused and units expired/discarded for the pre-COVID-19,
and relative to daily usage. Daily platelet usage decreased from a shutdown and renewal time periods (Table 1 and Figure 4). Figure 4a
baseline average of 56.4 units to 50.5 and 51.1 units during the shows that the decrease in platelet use throughout the pandemic was

TABLE 1 Pairwise statistical comparisons between time periods for platelet use in British Columbia (BC)

Comparison Mean of total weekly units used Expired/discarded rate (chi-squared)

Pre-COVID-19 versus shutdown 395 versus 354 (p < 0.0001) 18.9% versus 19.8% (p = 0.6)
Pre-COVID-19 versus renewal 395 versus 358 (p < 0.0001) 18.9% versus 15.2% (p < 0.0001)
Shutdown versus renewal 354 versus 358 (p = 0.9) 19.8% versus 15.2% (p < 0.0001)

Abbreviation: COVID, coronavirus disease 2019.

F I G U R E 4 Platelet usage in BC and its three largest health authorities before and after COVID-19. (a–d) Total platelet use and transfused
platelets for the pre-COVID-19, shutdown and renewal periods (weekly mean  95% CI). (e–h) Expired/discarded rate for the same three time
periods. For statistical comparisons: *** indicates p < 0.0001, ns indicates p ≥ 0.05. BC, British Columbia; CI, confidence interval; COVID,
coronavirus disease 2019
256 SHOPSOWITZ ET AL.

due to a combination of fewer platelet transfusions and fewer expired renewal phase, platelet transfusions increased in HA-1 from the shut-
units; however, the timing of these influences has been different. The down nadir but still remained below baseline levels by 8%. The rate of
decrease in transfused units reached a nadir during the shutdown expired/discarded units in HA-1 decreased from a baseline of
phase when elective surgeries were halted (11% decrease from base- 10.4% to 2.1% during the renewal phase, compared to virtually no
line), whereas a decrease in expired units was observed only during change from baseline during the initial shutdown (Figure 4f ).
the renewal phase (15.2% discard/expiry rate vs. 18.9% at baseline; HA-2 comprises 37% of the provincial population and contains
p < 0.0001). a level-1 trauma centre and a cardiovascular surgery service. HA-2
Key features of the three HAs in BC that utilize the most platelets accounted for 20% of provincial platelet use at baseline. In contrast to
(86% of total) are summarized in Table 2. These HAs vary by popula- HA-1, HA-2 saw an increase in total platelet use (18% increase from
tion, geographical breadth and clinical services. Over the past baseline) and transfused platelet units (22% increase from baseline)
10 years, the three HAs had a similar proportion of platelet transfu- during the initial shutdown but returned to baseline levels during the
sions related to elective surgeries (range 8.3%–10.1%), with more var- renewal phase, while expired/discarded units remained largely
iability for platelet use by emergency surgeries and non-surgical unchanged throughout both periods in HA-2 (Figure 4c,g).
indications (see Figure S5). HA-3 comprises 15% of the provincial population and includes a
Although HA-1 encompasses only 20% of the population, it regional hospital with a cardiovascular surgery service and other geo-
includes two cardiovascular surgery centres and contains the prov- graphically dispersed hospital services with significant resupply logistics
ince’s major quaternary referral centre, which provides level-1 trauma challenges. HA-3 was responsible for 13% of provincial baseline platelet
care and a bone marrow transplant (BMT) program. At baseline, HA-1 use. Similar to HA-1, HA-3 experienced a decrease in total platelet use
accounted for 53% of platelet use in the province. During the shut- starting with the shutdown (12% decrease from baseline) that was
down phase, HA-1 experienced a mean decrease of 52.1 platelet units sustained during the renewal (Figure 4d). However, in HA-3, this was
used per week, which was greater than the province-wide decrease of driven by decreased platelet transfusions during both periods, whereas
41.1 units per week (Figure 4b). This was primarily driven by a mean the expired/discarded platelet rate in HA-3 increased relative to baseline
decrease in transfused platelets in HA-1 of 45.8 units per week during during both the shutdown and renewal phases (Figure 4d,h).
the shutdown phase (24% decrease from baseline). During the

DI SCU SSION
T A B L E 2 Description of the three health authorities in British
Columbia (BC) with the highest platelet demand The COVID-19 pandemic has placed an incredible strain on healthcare
Health authority systems around the world, including sustaining a stable inventory of
(baseline platelets and other blood products. Initial studies have noted reduced
platelet use) Populationa Major services
platelet demands for critically ill COVID-19 patients compared to
HA-1 (53%) • Predominantly urban • Level-1 trauma other ICU patients [8], and several tertiary care centres in New York
• Pop. 1,193,977 centre
[9], Washington [10] and India [11] reported decreased platelet trans-
• Median age: • Quaternary
41 years referral centre fusions during the first 1–2 months of the pandemic, possibly due to a
• Two combination of low platelet requirements for COVID-19 patients and
cardiovascular policies that decreased healthcare utilization (e.g., halting elective sur-
surgery centres
geries). In our healthcare jurisdictions, we found similar findings that
• Bone marrow
transplantation platelet transfusions decreased concurrently with decreased surgeries
(BMT) service and hospital admissions. We also found that platelet use remained
• Solid organ decreased even when elective surgeries resumed, largely driven by
transplant service
decreased expired/discarded units. Our comprehensive study of
HA-2 (20%) • Mixed urban/rural • Level-1 trauma system-wide platelet use during the COVID-19 pandemic also demon-
• Pop. 1,906,933 centre
strates differential effects on platelet use by health region, potentially
• Median age: • Cardiovascular
39 years surgery related to their different demographics and characteristics.
• No BMT service The overall balance of blood product supply and demand depends
HA-3 (13%) • Dispersed with • Cardiovascular on many interacting factors: donors, collection services, distribution
mixture of rural and surgery networks, hospitals and patients requiring transfusions. All of these
small urban centres • No level-1
are potentially impacted directly and indirectly by COVID-19 and
• Significant resupply trauma or BMT
logistics challenges service related policies. The Canadian province of BC, with a large integrated
• Pop. 827,314 healthcare system and transfusion database, provides a unique oppor-
• Median age: tunity to examine these various factors. We found that platelet use
47 years
significantly decreased compared to baseline during the initial shut-
a
Population data from 2019[14]. down phase of BC’s response and remained decreased during the
COVID-19’S IMPACT ON PLATELET USE IN BC 257

subsequent renewal phase when elective surgeries were resumed. On compared to baseline during the renewal phase, possibly related to
the other hand, platelet supply declined during the first 2 months of persistent decreases in healthcare utilization, the main driver of
the pandemic but subsequently rebounded to levels at or slightly decreased platelet utilization after the shutdown ended was a signifi-
above the pre-COVID-19 baseline. The initial decrease in supply was cant decrease in the platelet expiration/discard rate. This was mainly
driven by fewer whole blood collections and reduced capacity to pro- driven by one HA (HA-1) that also implemented specific policies
cess blood, related to cancellations of mobile clinics along with including a shared platelet inventory among different hospitals and
decreased capacity at donation and manufacturing sites due to public prospective isoagglutinin titres of all group O platelets to facilitate
health orders. Apheresis donations were less impacted by these fac- their transfusion into non-group O patients [15]. The degree to which
tors because they largely occur at fixed sites and could be increased the significant decrease in platelet expiration/discard rate observed is
during the shutdown phase to partially compensate for the decrease explained by these policy changes warrants further study.
in buffy coat platelets. Possible explanatory factors for the subse- Strengths of this study include the availability of historical platelet
quent increase in whole blood collections during the renewal phase transfusion data as well as the availability of platelet disposition data
include greater donor awareness of the need for blood and the safety across a large population/geographical region throughout distinct
of donor collection sites, operational changes by CBS such as phases of the COVID-19 pandemic. An important limitation of this
increased clinic hours/staffing and comfort with efficient donor turn- study is that the indications for platelet transfusions (i.e., surgical
around time in the new, physically distanced environment. Impor- vs. non-surgical) were only available for the pre-COVID-19 period. In
tantly, by this time point, efficient COVID-19 safety procedures were addition, data related to healthcare utilization during the pandemic
in place at manufacturing sites, allowing production capacity to be were only available until the end of June 2020. Given these limitations
safely increased. As a result, at no point during the first 6 months of and the inherent confounding factors in a retrospective observational
COVID-19 emerging in BC, did the province observe a platelet short- study, we cannot definitively ascribe changes to platelet demand to
age; if anything, there was a greater average platelet reserve in the specific causes or factors. Another limitation is that our findings may
province compared to the year prior to the pandemic. not be generalizable to other regions that had different strains on
The decreased platelet utilization in BC during COVID-19 was their healthcare system due to COVID-19 or with different patient
driven by different factors during the distinct periods of the province’s demographics or resources/organizational structures with regards to
response to the pandemic. During the initial shutdown phase, when blood product distribution.
elective surgeries were halted and overall healthcare utilization was In conclusion, to our knowledge, this is the largest analysis to-
dramatically decreased from baseline, reduced platelet utilization date of platelet supply and utilization during the COVID-19 pan-
was mostly due to fewer platelet transfusions. Based on provincial demic. We found that the platelet supply in BC remained stable dur-
historical data, halting elective surgeries was expected to decrease ing the first 6 months of COVID-19, even with an initial decrease in
platelet transfusions by 10% – close to the observed provincial blood donations. Platelet utilization was significantly decreased in
decrease of 11%. However, the considerable differences seen BC during the first 6 months of the pandemic compared to baseline:
between HAs, which have similar baseline proportions of platelet utili- this appears to have been due to province-wide policies that led to
zation from elective surgeries, suggest that additional factors also decreased healthcare utilization and platelet transfusions during the
played a role. In BC, HA-1 saw a considerably larger decrease in plate- initial phase of the pandemic, followed by local practice changes that
let transfusions during the shutdown compared to the provincial aver- may have contributed to fewer wasted platelet units. The relatively
age. This difference might be explained by decreased utilization of modest contribution of elective surgery cancellations to decreased
non-surgical services with high platelet requirements, only offered in platelet utilization is noteworthy while other clinical and laboratory
HA-1, such as a BMT/dedicated haematology–oncology service. On practice changes also likely contributed to significant reductions in
the other hand, HA-2 saw an increase in platelet transfusions during platelet utilization in the HA with the highest baseline utilization. In
the shutdown phase. Given the retrospective nature of our study and particular, laboratory practice changes implemented in response to
our inability to account for a multitude of complex confounders the threat of shortages appear to have contributed to a sustained
including clinician practice patterns and other contingency plans, reduction in utilization, extending for at least a 4-month period fol-
which may have affected platelet usage, we are unable to determine lowing the resumption of elective surgery and BMT activity. These
the exact reasons why HAs were differentially affected during the findings may assist others in determining suitable practices for secur-
pandemic. ing platelet supply throughout the ongoing pandemic and in other
In addition to shutting down surgeries, various provincial efforts challenging scenarios.
were made across different HAs to preserve blood products. Led pri-
marily by HA-1 physicians, these initiatives included clinical advisories AC KNOW LEDG EME NT S
and education to encourage appropriate blood utilization, screening K.E.S. wrote the original manuscript and contributed to data collection
platelet orders for appropriateness and improving inventory manage- and analysis; C.L. and N.F. contributed to data collection, statistical
ment to reduce wastage [15]. Of note, platelet utilization remained analysis and preparation of the manuscript; A.W.S. and D.M. initiated
significantly less in BC during the renewal phase, when elective sur- this work and contributed to the preparation of the manuscript;
geries resumed. While platelet transfusions remained slightly less B.R.B., M.B., T.P., J.T., M.W. and M.T.S.Y. were also involved in the
258 SHOPSOWITZ ET AL.

conception of this work and provided comments and revisions that transfusion activity in a general hospital during the COVID-19
were incorporated in the final manuscript. pandemic. Vox Sang. 2020;116:1–7.
9. DeSimone RA, Costa VA, Kane K, Sepulveda JL, Ellsworth GB,
Gulick RM, et al. Blood component utilization in COVID-19 patients
CONF LICT OF IN TE RE ST in New York City: transfusions do not follow the curve. Transfusion.
No conflicts of interest were identified in the preparation of this 2021;61:692–8.
manuscript. 10. Pagano MB, Cataife G, Fertrin KY, Gernsheimer T, Hess JR, Staley E,
et al. Blood use and transfusion needs at a large health care system
in Washington state during the SARS-CoV-2 pandemic. Transfusion.
ORCID 2020;60:2859–66.
Kevin E. Shopsowitz https://orcid.org/0000-0003-3988-0837 11. Pandey HC, Coshic P, Chippy CS, Arcot PJ, Kumar K. Blood supply
management in times of SARS-CoV-2 pandemic – challenges, strate-
gies adopted, and the lessons learned from the experience of a
RE FE R ENC E S
hospital-based blood centre. Vox Sang. 2020;116:497–503.
1. Stubbs JR, Homer MJ, Silverman T, Cap AP. The current state of the 12. Little N. COVID-19 Tracker Canada. 2020. https://covid19tracker.
platelet supply in the US and proposed options to decrease the risk ca. Accessed 11 Mar 2021.
of critical shortages. Transfusion. 2021;61:303–12. 13. COVID-19’s Effect on Hospital Care Services. Canadian Institute for
2. Ojha S, Gupta AM, Nagaraju P, Poojary M, Sumathi SH. Challenges in Health Information. https://www.cihi.ca/en/covid-19-resources/
platelet inventory management at a tertiary care oncology center impact-of-covid-19-on-canadas-health-care-systems/covid-19s-
during the novel coronavirus disease (COVID-19) pandemic lock- effect-on-hospital. Accessed 2 Mar 2021.
down in India. Transfus Apher Sci. 2020;59:102868. 14. British Columbia-Population Estimates. https://bcstats.shinyapps.io/
3. Gniadek TJ, Mallek J, Wright G, Saporito C, AbiMansour N, popApp/. Accessed 1 Mar 2021.
Tangazi W, et al. Expansion of hospital-based blood collections in the 15. McGinnis E, Guo RJ, Marcon KM, Berry B, Coupland R,
face of COVID-19 associated national blood shortage. Transfusion. Meneghetti V, et al. Adaptations of transfusion systems to the
2020;60:1470–5. COVID-19 pandemic in British Columbia, Canada: early experiences
4. Warner MA, Kurian EB, Hammel SA, van Buskirk CM, Kor DJ, of a large tertiary care center and survey of provincial activities.
Stubbs JR. Transition from room-temperature to cold-stored plate- Transfusion. 2021;61:1102–11.
lets for the preservation of blood inventories during the COVID-19
pandemic. Transfusion. 2021;61:72–7.
5. Doyle AJ, Danaee A, Furtado CI, Miller S, Maggs T, Robinson SE, SUPPORTING INFORMATION
et al. Blood component use in critical care in patients with COVID- Additional supporting information may be found online in the
19 infection: a single-centre experience. Br J Haematol. 2020;191:
Supporting Information section at the end of this article.
382–5.
6. Zong X, Gu Y, Yu H, Li Z, Wang Y. Thrombocytopenia is associ-
ated with COVID-19 severity and outcome: an updated meta-
analysis of 5637 patients with multiple outcomes. Lab Med. 2021; How to cite this article: Shopsowitz KE, Lim C, Shih AW,
52:10–5. Fishbane N, Berry BR, Bigham M, et al. Impacts of COVID-19
7. Wiersinga WJ, Rhodes A, Cheng AC, Peacock SJ, Prescott HC. Path- and elective surgery cancellations on platelet supply and
ophysiology, transmission, diagnosis, and treatment of coronavirus
utilization in the Canadian Province of British Columbia. Vox
disease 2019 (COVID-19): a review. JAMA. 2020;324:782–93.
8. Marín-Mori K, González-Gasco n y Marín I, Foncillas-García MA,  Sang. 2022;117:251–8.
Muñoz-Novas C, Infante M, Churruca-Sarasqueta J, et al. Blood
Received: 11 January 2021 Revised: 1 June 2021 Accepted: 8 July 2021

DOI: 10.1111/vox.13187

ORIGINAL ARTICLE

Improvement of transfusion practice and reduction in red

blood cell utilization in Belgian hospitals: Results of a national
survey and benchmarking

Jana Vanden Broeck1,2 | Katrien Beeckman3,4,5 | Evelyne Van Gastel6 |

Luc De Keersmaecker7 | Timothy Devos8 | Christiane Gérard9 | Lucien Noens10 |
11 12 2
Dominique Putzeys | Karin Van Poucke | Margareta Haelterman |
Véronique Deneys13 | Rik Schots1
1
Department of Hematology, Universitair Ziekenhuis Brussel, Vrije Universiteit Brussel (VUB), Brussels, Belgium
2
Federal Public Service Health, Food Chain Safety and Environment, Brussels, Belgium
3
Nursing and Midwifery Research Unit, University Hospital Brussels, Brussels, Belgium
4
Nursing and Midwifery Research Unit, Faculty of Medicine and Pharmacy, Universitair Ziekenhuis Brussel, Vrije Universiteit Brussel (VUB), Brussels, Belgium
5
Centre for Research and Innovation in Care, Midwifery Research Education and Policymaking (MIDREP), Verpleeg- en vroedkunde, Universiteit Antwerpen,
Antwerp, Belgium
6
Federal Agency for Medicines and Health Products, Brussels, Belgium
7
Clinical Laboratory, Jan Yperman Hospital, Ypres, Belgium
8
Department of Hematology, Universitair Ziekenhuis Gasthuisberg, Katholieke Universiteit Leuven, Leuven, Belgium
9
Blood Bank, Centre Hospitalier Universitaire Liège, Liège, Belgium
10
Blood Bank, Universitair Ziekenhuis Gent, Ghent, Belgium
11
Department of Nursing, Centre Hospitalier Régional de la Citadelle Liège, Liège, Belgium
12
Clinical Laboratory, Algemeen Ziekenhuis Nikolaas, Sint-Niklaas, Belgium
13
Blood Bank, Département des Laboratoires Cliniques, Cliniques universitaires Saint-Luc, Université Catholique de Louvain Brussels, Brussels, Belgium

Correspondence
Jana Vanden Broeck, Secretariaat Abstract
Hematologie, UZ Brussel, Laarbeeklaan
Background and Objectives: Belgian health authorities launched a national platform
101, 1090 Brussels, Belgium.
Email: [email protected] in 2011 to improve the quality of transfusion practices and blood use in Belgian hos-
pitals. No data were available about the quality of hospital transfusion practice at the
national level.
Materials and Methods: Three consecutive national surveys (2012, 2014 and 2016)
were performed in all 111 Belgian hospitals to assess the degree of implementation
of standards in four process domains related to red blood cell (RBC) transfusion: gen-
eral quality aspects, ordering of RBC, electronic traceability and reporting of adverse
events. The surveys were part of a methodology based on informing, feedback and
benchmarking. Responses to questions were analysed semi-quantitatively, and hospi-
tals could score 10 points on each of the domains.
Results: The proportion of hospitals scoring below 5 per domain decreased from
16%, 70%, 14% and 11% (2012) to 2%, 17%, 1% and 1% (2016), respectively. Simi-
larly, scores above 7.5 increased from 25%, 1%, 23% and 36% (2012) to 64%, 30%,

Vox Sanguinis. 2022;117:259–267. wileyonlinelibrary.com/journal/vox © 2021 International Society of Blood Transfusion. 259
260 VANDEN BROECK ET AL.

68% and 81% (2016), respectively. In 2016, overall quality of transfusion practices,
including the four pre-specified domains, improved continuously with an average
total score (max = 40) increasing from 24.2 to 30.5 (p = 0.0005). In addition, there
was a decrease in the number of distributed and transfused RBC per 1000 popula-
tion between 2011 and 2019 from 47.0 to 36.5 and 43.5 to 36.1, respectively.
Conclusion: These data show that the applied methodology was a powerful tool to
improve quality of transfusion practices and to optimize utilization of RBC at the
national level.

KEYWORDS
benchmark, national survey, red blood cell utilization, transfusion practice

I N T R O D U CT I O N The primary aims of this study were to measure and improve the
quality of the transfusion practices in Belgian hospitals in a period of
Transfusion of red blood cells (RBCs) is a frequent medical interven- 6 years with a 2-year interval. The secondary aim was to examine
tion in many fields of medicine. In a typical European context, more whether the measured quality had an impact on national transfusion
than 50% of RBC are administered to patients older than 70 years rates of RBC.
and around 25%–30% to patients older than 80 years. Major indica-
tions for RBC transfusion are haemato-oncology and cancer
followed by traumatology-orthopaedics, gastrointestinal–liver dis- M A T E R I A L S A N D M ET H O D S
eases and cardiovascular disorders [1–3]. According to a national
survey performed in England and North Wales in 2014, 67.4% of General aspects
RBC transfusions were for medical indications including 20% for
haematological and 10% for non-haematological cancers [3]. Taken BeQuinT includes a central steering committee and several working
into account the ageing of the Western population and a higher inci- parties with the participation of experts and stakeholders in transfu-
dence of cancer in the elderly, it might be expected that there will be sion. The methodology used was based on repetitive national sur-
a growing gap between supply and demand of RBC. On the other veys in combination with feedback and information sessions and
hand, the overall improvement in the quality of transfusion practice, benchmarking as depicted in Figure 1. The surveys were sent to the
the implementation of more restrictive transfusion strategies and chairs of the transfusion committees of all 111 Belgian acute hospi-
the gradual introduction of patient blood management (PBM) tals, including university hospitals. The chairs of the transfusion com-
programmes have led to a decrease in RBC transfusions over the mittees received a link and an individual code to complete the
past decade [4]. However, there is substantial heterogeneity, ranging survey online in a period of 3 months with a deadline at the end of
between 27 and 55, in the number of RBC transfused per 1000 September 2012, 2014 and 2016, respectively. The Red Cross of
population in developed countries suggesting that the degree of Flanders (Dutch speaking part) and the Red Cross of Belgium
implementation of these measures at the national level is highly (French speaking part) were trusted third parties and used the coding
variable [5]. In several countries, national surveys have been con- system to ensure an anonymous data analysis. Participation in the
ducted to assess the quality of transfusion practice processes [6–9]. survey was mandatory and promoted by a Royal Decree, a govern-
These surveys used variable quality indicators and showed incom- ment decision, which twice a year allocates a budget to hospitals to
plete participation rate among hospitals, whereas the impact on establish and maintain a quality system relating to transfusion. The
nation-wide RBC utilization rate was not studied. Some nation- or results of the benchmarking were sent to all chairs of the transfusion
community-wide studies could show that the implementation of committees, individually, in a closed envelope by the trusted third
transfusion guidelines [10] or PBM measures [11, 12] had a positive parties. The overall results of each survey were disseminated
impact on RBC utilization. In Belgium, the RBC utilization rate in through a report published on the website of BeQuinT and pres-
2011 was 47.0 per 1000 population, and there were no data on the ented during a yearly information session organized by BeQuinT.
quality of hospital transfusion practice at the national level. BeQuinT The information sessions were combined with a scientific sympo-
stands for Belgian Quality in Transfusion and is an initiative of the sium and interactive workshops. Based on the overall results, a num-
Federal Public Service (Ministry) of Public Health, Food Chain Safety ber of general recommendations related to the most frequently
and Environment since 2011. Its mission is to support the transfu- observed deficiencies were included in the report, whereas the
sion and haemovigilance policy and to optimize the utilization of benchmarking allowed hospitals to prioritize on areas of deficiencies
blood components. at the local level.
TRANSFUSION PRACTICE AND BLOOD UTILIZATION 261

FIGURE 1 Summary of methodology based on national surveys

F I G U R E 2 Benchmarking report: example of a report as sent to each hospital individually. Results are shown for hospital X of which scores
are represented by the vertical red bar. Other vertical bars represent the scores of all hospitals participating to the survey. The median scores for
each process domain are indicated by the horizontal bars

Survey content general’ relates to the organizational structure of the transfusion

practice including the number of meetings of the transfusion com-
Legal requirements, guidelines, expertise and consensus were used mittee, the appointment of a transfusion practitioner and/or a multi-
by the steering committee of BeQuinT to identify structural and pro- disciplinary transfusion team. In addition, training of physicians and
cess measures associated with best practices (‘standards’) in transfu- nurses in transfusion practice, the availability of transfusion proce-
sion. Several issues of transfusion practices were grouped into four dures and the availability of data on the number of transfused and
process domains, and for each domain, standards were defined invoiced blood components were included into this domain. Stan-
(the complete survey is available in the Supplementary Appendix in dards in the domain ‘Ordering’ related to the availability of proce-
the online version of the manuscript). The domain ‘Transfusion dures regarding blood sampling for pre-transfusion testing, the use
262 VANDEN BROECK ET AL.

TABLE 1 Reported frequencies (%) for (a) ‘transfusion general’ and ‘ordering’ and (b) ‘traceability’ and ‘reporting’

(a) ‘Transfusion general’ and ‘ordering’

2012 2014 2016

(n = 111) (n = 111) (n = 107)
Transfusion general
Meeting transfusion committee >2 times/year 89.2 95.5 98.1
a
Multidisciplinary transfusion team appointed 67.6 82.2
Transfusion practitioner appointed 82.8 80.2 92.5
Digital transfusion handbook with procedures 86.5 96.4 96.2
available
Educational programme physicians (min 1/year) 13.5 26.1 32.7
Transfusion history available in patient record 33.3 50.5 57.0
Ordering
Electronic order implemented 6.3 7.2 15.0
Electronic order in development 12.6 28.8 39.3
Ordering always by Medical Doctor 29.7 29.7 43.0
Blood samples identified with barcode 34.2 52.2 59.8
Procedure pre-transfusion testing available 87.4 94.6 97.2
Procedure urgent transfusion available 56.8 77.5 79.4
(hospital wide)
Procedure urgent transfusion available (some units) 14.4 8.1 11.2
a a
Indication for transfusion systematically mentioned 42.1
in order (hospital wide)
a
Systematic analysis prescriptive behaviour 11.7 13.1
(hospital wide)
a
Systematic analysis prescriptive behaviour 27.9 32.7
(some units)

(b) ‘Traceability’ and ‘reporting’

2012 2014 2016

(n = 111) (n = 111) (n = 107)
Traceability
Emergency stock of RBC in separate refrigerator 66.1b 61.2b 60.8b
with T monitoring, data storage and alarm in
hospital care unit(s)
24 h/24 h intervention in case of alarm for stored 53.6b 71.4b 86.3
RBC in remote refrigerator in hospital care unit(s)
Person responsible for follow-up emergency stock in 69.6b 83.6b 82.4b
hospital care unit(s)
Identification with wristband upon admission 92.8 98.2 99.1
Partial electronic blood tracking system 53.1 35.1 41.1
Complete electronic blood tracking system 9.9 18.9 25.2
Scanning identification wristband before transfusion 15.3 22.5 32.7
Responsible medical doctor for transfusion 58.6 69.4 78.5
identifiable in patient record in >90% of cases
Administrator blood component identifiable in file in 55.9 67.6 86.0
>90% of cases
Procedure for transport and storage of blood 57.6 85.6 95.3
components
Reporting
Procedure administration blood components 91.0 95.5 98.2
available
(Continues)
TRANSFUSION PRACTICE AND BLOOD UTILIZATION 263

TABLE 1 (Continued)

(b) ‘Traceability’ and ‘reporting’

2012 2014 2016

(n = 111) (n = 111) (n = 107)
Procedure patient follow-up 24 h after transfusion 27.9 36.9 47.7
available
Procedure patient long-term follow-up available 14.4 19.8 23.3
Registration vital signs in patient record in >90% of 55.0 70.3 84.1
cases
Registration serious transfusion reactions in patient 71.2 81.1 90.7
record in >90% of cases
Reporting of serious transfusion reactions to staff 81.1 94.6 94.4
member responsible for notification to the
national haemovigilance office
Reporting of serious transfusion incidents to staff 81.1 93.7 92.5
member responsible for notification to the
national haemovigilance office
Serious transfusion reactions and incidents discussed 91.9 99.1 99.1
by transfusion committees at least once a year
Methodology for analysis of serious adverse events 37.8 70.3 85.0

Abbreviations: FAMHP, Federal Agency for Medicines and Health Products; RBC, red blood cell.
a
Not questioned in the first survey.
b
Only applicable in 56 (in 2012), 49 (in 2014) and 51 (in 2016) hospitals, respectively.

T A B L E 2 Reported frequencies (%) of hospitals performing internal audits to check the implementation of some important transfusion-
related procedures at least once a yeara

2012 2014 2016

(n = 111) (n = 111) (n = 107)
Audit of procedure on blood sampling for pre-transfusion 20.2 (18/89) 45.5 (46/101) 64.1 (66/103)
testing (e.g., patient identification and labelling
of blood samples)
Audit of procedure on pre-transfusion testing 32.0 (31/97) 34.3 (36/105) 56.0 (56/100)
(type of analyses)
Audit of procedure on patient identification upon 46.6 (48/103) 67.0 (73/109) 90.6 (96/106)
admission
Audit of procedure on transport and storage blood 29.3 (22/75) 44.2 (42/95) 53.9 (55/102)
components from delivery by the blood establishment
to the patient
Audit of procedure on patient identification before 18.9 (21/111) 50.5 (56/111) 74.8 (80/107)
starting the transfusion
a
Not all hospitals responded to the questions, and, therefore, absolute numbers are mentioned below the %.

of an electronic method to prescribe blood components, the evalua- blood bank and care units and the number of wasted RBC are
tion of the prescription behaviour and adherence to transfusion assessed. Finally, the availability of a procedure about the adminis-
guidelines and the use of O-negative RBC. In addition, the legal tration of blood components and the patient monitoring post
requirements regarding the ordering of RBC are questioned such as transfusion, the registration of vital signs during transfusion and
mentioning of the indication for transfusion on the order and adher- transfusion reactions in the patient record and the notification and
ence to the exclusive restriction of the act of ordering to physicians. analysis of transfusion reactions and incidents are grouped in the
In the domain ‘Traceability’ patient identification, the implementa- domain ‘Reporting’. On several occasions throughout the survey,
tion of pre-transfusion identity checks, the availability of an elec- the implementation of internal auditing procedures is assessed. The
tronic blood tracking system, the storage of RBC in the hospital survey was repeated twice with an interval of 2 years in 2012, 2014
264 VANDEN BROECK ET AL.

F I G U R E 3 Percentage of hospitals scoring low (<5), average (5–7.5) or high (>7.5) for the four transfusion process domains with comparison
of scores between the three surveys

and 2016. Consecutive survey contents were identical although differences between the first (2012) and last survey (2016) were
some terms were clarified in the questions and answer categories in analysed with the paired samples t-test using SPSS Statistics.
the second and third survey by means of an explanatory glossary.

Utilization of red blood cells

Scoring and data analysis
The data on the total national RBC distribution by the transfusion
For each hospital, scores were attributed to several answer categories establishments as well as the total number of RBC transfused in
and converted to a score out of 10 for each of the four domains and a Belgian hospitals are provided in an official yearly report by the
total score on 40 for the entire survey, being the sum of the four scores Federal Agency for Medicines and Health Products. Data on Belgian
for each domain. Scoring could be 2, 1 or 0 depending on whether or population are obtained at the Belgian be.STAT official website. The
not a standard was implemented or, for example, a written procedure distribution and transfusion rates per 1000 capita are calculated by
was available hospital-wide (2 points), in some departments (1 point) or the ratio of total number of transfused RBC during a particular year
not (0 points), respectively. These scores provided information on the divided by the total population (1/1000) as reported in January of
degree of implementation of standards per domain and enabled to ana- the next year.
lyse trends over time and to perform benchmarking of the hospitals on
a national scale. The results of the anonymous benchmarking of the
scores for the four domains for all of the hospitals for each survey were RE SU LT S
sent individually to all the chairs of the transfusion committees. An
example of such reporting per hospital is illustrated in Figure 2. For the Characteristics of the hospitals
scores on each domain, the frequencies of hospitals scoring low (below
5), intermediate (5–7.5) or high (above 7.5) were calculated. As for the All the Belgian hospitals represented by 111 transfusion committees
total score, average, median and percentiles were calculated, and participated in the first two surveys. Due to three fusions of hospitals,
TRANSFUSION PRACTICE AND BLOOD UTILIZATION 265

there were 107 responders to the third survey. In 2016, 25.2% of the identification of blood samples with barcode (34.2% ⇨ 59.8%), the
hospitals had less than 250 acute beds, 44.1% between 250 and availability of a hospital-wide procedure for urgent transfusion
500 beds and 33.6% more than 500 beds; seven hospitals were aca- (56.8% ⇨ 79.4%); the availability of a complete electronic tracking
demic hospitals. In terms of activities with higher RBC transfusion system (9.9% ⇨ 25.2%), procedures for transport and storage of
needs, 23.4% had a clinical haematology department with transplanta- blood components (28.8% ⇨ 54.2%); the implementation of a sys-
tion, 9.3% performed solid organ transplantation, 29.0% had cardiac tematic methodology to analyse serious adverse events in the trans-
surgery and 20.6% had neonatal intensive care. fusion chain (37.8% ⇨ 85%). Also, an increase in the number of
hospitals performing internal audits related to several aspects of the
transfusion chain was observed (Table 2). The frequency of scores
Survey data analysis per process domain was calculated for each survey (Figure 3). In
2016, except for the domain ‘Ordering’, less than 3% of hospitals
For each of the surveys, the degree of implementation of the most had scores below 5 and 64%, 30%, 68% and 81% instead of 25%,
important standards per domain is shown in Table 1. Major and con- 1%, 23% and 36% had high (>7.5) scores on each of the domains,
tinuous progress could be noticed between 2012 and 2016, for respectively. The average scores per domain were significantly
instance an increase in: the existence of multidisciplinary transfusion better for ‘Transfusion general’, ‘Ordering’, ‘Traceability’ and
teams (only measured since 2014) (67.6% ⇨ 82.2%), a readily avail- ‘Reporting’ (Table 3) and the average total scores were significantly
able documented patient transfusion history (33.3% ⇨ 57.0%); the better, increasing from 24.2 in 2012 to 30.5 in 2016 (p = 0.0005)
development of an electronic ordering system (12.6% ⇨ 39.3%), (Table 4).

TABLE 3 Average scores (max 10) (SD) per domain (first and third TABLE 4 Total scores (max 40) (first, second and third survey)
survey)
2012 2014 2016
2012 2016 (n = 111) (n = 111) (n = 107)a
Process domain (n = 111) (n = 107)a p-Value
Average (SD) 24.2 (4.1) 28.1 (4.4) 30.5 (3.7)b
Transfusion in general 6.3 (2.1) 8.1 (1.5) 0.001
Median 24.4 28.4 30.6
Ordering 4.4 (1.3) 6.7 (1.5) 0.397
Range 14.5–36.5 13.8–36.5 14.1–38.1
Traceability 6.4 (1.4) 8.2 (1.1) 0.003
P25 20.8 25.5 28.5
Reporting 7.0 (1.6) 8.4 (1.1) 0.001
P75 27.2 31.4 32.8
Abbreviation: SD, standard deviation.
a
Abbreviation: SD, standard deviation.
Four hospitals less due to fusion. a
Four hospitals less because of fusion.
b
p = 0.0005.

FIGURE 4 Utilization of red blood cells (RBC) in Belgian hospitals (expressed as units per 1000 population) and indicator for wastage of RBC
266 VANDEN BROECK ET AL.

Utilization of red blood cells reporting. Hence, the data were not verified by external audits, and this
is mainly due to the practical problems to perform such audits in more
Figure 4 shows that from 2011 (the year before the launch of than 100 hospital sites. In addition, the impact of the surveys on the
BeQuinT) to 2017 (the year after the last survey was performed), the number of transfusion errors or near misses in Belgian hospitals was not
number of distributed and transfused RBC per 1000 population examined, although this can be considered as an important measure of
decreased from 47.0 to 36.6 and from 43.5 to 36.1, respectively. This transfusion quality. Data of Federal Agency for Medicines and Health
decrease was sustained but did not further improve in 2018 and Products showed that in the period 2012–2016, 19 hospitals did not
2019. In addition, the difference between distributed and transfused report any adverse event to the national haemovigilance system, and in
RBC decreased from 7.4% in 2011 to 1.4% in 2017 and further down 2016, 45 hospitals did not report any adverse event, introducing a bias
to 1.2% in 2019. for the reliability of this indicator. More action is needed to encourage
hospitals to report serious errors and near misses.
An albeit indirect but objective evidence of improved transfusion
DISCUSSION practices in Belgium is the RBC distribution and transfusion rate,
which decreased significantly from 2011 to 2017 (Figure 4). Most of
BeQuinT conducted three consecutive national surveys to study the the decrease occurred from 2013 to 2017, concurrently with the
quality of transfusion practices in Belgian hospitals by measuring BeQuinT surveys. We showed that a substantial part of decreased
the degree of implementation of numerous standards related to the RBC utilization is related to a marked reduction in RBC wastage since
major processes involved in RBC transfusion: the functioning of the the difference between distributed and transfused RBC decreased
transfusion committees, the presence or not of dedicated personnel from 7.4% (38,097 units) in 2011 to 1.2% (4858 units) in 2019. In
involved in transfusion, the processes of ordering, transport, storage addition to less wastage, we hypothesized that our survey-based
and tracing of blood components and the management of transfusion- methodology and associated increased awareness also led to a more
related adverse events and reactions. The surveys were part of a rational transfusion policy and associated reduction in RBC transfu-
methodology based on providing information through yearly informa- sion. The introduction of PBM has reduced RBC utilization in some
tion sessions, individualized feedback to the hospitals and anonymized areas of clinical practice over the past decade in several countries.
benchmarking (Figure 1). This methodology was very successful to This was particularly the case in cardiovascular and orthopaedic sur-
increase awareness and involvement of transfusion committees and gery. However and based on data (2016) provided by the Data and
hospital managements leading to numerous structural measures at the Strategic Analysis, Directorate-General for Health Care and Federal
local level to improve transfusion policy. The result of this approach Public Service of Public Health, RBC transfusion in surgical areas only
was that the quality of transfusion practice in Belgian hospitals accounted for 30% of the total. Even a 50% reduction of RBC transfu-
increased significantly and continuously with an average score increas- sion in surgical areas would have accounted for much less than the
ing from 24.2 (2014) to 30.5 (2016). Also, the maximum scores reduction of the total utilization rate as was observed between 2011
increased overtime, whereas the minimum scores remained the same, and 2017. We, therefore, concluded that the BeQuinT survey-based
indicating that some hospitals at the bottom of the range had difficul- methodology has significantly contributed to less wastage of RBC and
ties in improving their transfusion processes. A major strength was the a more rational transfusion policy introduced in a broad field of medi-
100% participation rate of the Belgian hospitals, so that the results of cal and surgical disciplines. Data from 2017 to 2019 showed no fur-
the surveys generate a complete set of data on national transfusion ther decrease but stabilization, suggesting that the impact of
practices over a 6-years’ period. The endorsement by national health improved transfusion practices on RBC utilization is durable but level-
authorities, the possibility to complete the survey online and the finan- ling off and that other measures are needed for further optimization.
cial support to the individual hospitals were factors contributing to the This should be possible since in many European countries lower than
success. In 2016, the majority of hospitals had high scores for the 35 per 1000 population utilization rates were noted [5, 14]. Further
implementation of general transfusion measures, traceability improvement will require a more direct impact on clinical-oriented
and reporting of adverse events and reactions. Although markedly as opposed to process-oriented transfusion practices. PBM is an
improved, only 30% of the hospitals had high scores on the process of evolving multidisciplinary concept shifting from a ‘product-centred’
ordering of RBC (Figure 3). BeQuinT is currently facilitating the imple- towards a ‘patient-centred’ evidence-based approach to minimize
mentation of a computerized physician order entry (CPOE) system in unnecessary exposure of patients to blood components. Recently,
Belgian hospitals. A CPOE should include immediate access to key PBM recommendations have been published including diagnosing
medical information related to the indication for transfusion, incorpo- and medical treatment of pre-operative anaemia, minimizing peri-
rate a clinical decision support system and facilitate data collection to operative blood loss and using restrictive triggers to transfuse in
document compliance with evidence-based transfusion guidelines. most clinical situations [15, 16]. A survey in some European univer-
Clinical decision support systems have been shown to reduce transfu- sity hospitals showed that the degree of implementation of PBM
sions and lead to substantial cost savings in clinical practice [13]. was highly variable [17]. The implementation of PBM in clinical
A major limitation of our methodology is that the surveys were not practice has a significant effect on RBC utilization. The combination
based on existing electronic sources but through web-based self- of measuring, auditing and providing feedback resulted in a marked
TRANSFUSION PRACTICE AND BLOOD UTILIZATION 267

decrease RBC utilization in intraoperative RBC usage in a cardiac 5. Carson JL, Triulzi DJ, Ness PM. Indications for and adverse effects of
surgery setting [18]. Although some PBM elements were red-cell transfusion. N Engl J Med. 2017;377:1261–72.
6. Verlicchi F, Facco G, Macrì M, Antoncecchi S, Bonomo P. Blood
questioned, the BeQuinT surveys were not designed to assess the
transfusion practice: a nationwide survey in Italy. Blood Transfus.
degree of PBM implementation in the hospitals. Based on the suc- 2011;9:430–5.
cess of the methodology used, BeQuinT has launched a national 7. Zijlker-Jansen PY, Janssen MP, van Tilborgh-de Jong AJ,
PBM survey in all Belgian hospitals in 2020 with the aim of further Schipperus MR, Wiersum-Osselton JC. Quality indicators for the
hospital transfusion chain: a national survey conducted in 100 dutch
improving transfusion practices and optimizing RBC utilization.
hospitals. Vox Sang. 2015;109:287–95.
We concluded that this nation-wide survey–based 8. Taylor CJ, Murphy MF, Lowe D, Pearson M. Changes in practice and
benchmarking initiative combining information, data collection and organisation surrounding blood transfusion in NHS trusts in England
individualized feedback has contributed to a measurable and durable 1995-2005. Qual Saf Health Care. 2008;17:239–43.
9. Romon I, Lozano M. Quality indicators for transfusion medicine in
improvement of the quality of transfusion practices in Belgian hospi-
Spain: a survey among hospital transfusion services. Blood Transfus.
tals. We also demonstrated that improving transfusion quality is 2017;15:207–14.
associated with a marked and structural decrease in RBC 10. Qiang JK, Thompson T, Callum J, Pinkerton P, Lin Y. Variations in
utilization rate. RBC and frozen plasma utilization rates across 62 Ontario commu-
nity hospitals. Transfusion. 2019;59:545–54.
11. Whitaker B, Rajbhandary S, Kleinman S, Harris A, Kamani N. Trends
ACKNOWLEDGEMEN TS
in United States blood collection and transfusion: results from the
The project received general support of the Directorate-General for 2013 AABB blood collection, utilization, and patient blood manage-
Health Care, Federal Public Service Health, Food Chain Safety and ment survey. Transfusion. 2016;56:2173–83.
Environment, Belgium. 12. Sherliker L, Pendry K, Hockley B. Patient blood management: how is
implementation going?: a report comparing the 2015 survey with the
All authors have actively contributed to the conception and design
2013 survey of PBM in England. Transfus Med. 2018;28:92–7.
of the survey. K.B. realized the statistical analysis. J.V.B., K.B., R.S., 13. Swart N, Morris S, Murphy MF. Economic value of clinical decision
C.G. and V.D. were involved in the interpretation of the data. J.V.B., support allied to direct data feedback to clinicians: blood usage in
K.B. and R.S. drafted the manuscript. All authors did a critical haematology. Vox Sang. 2020;115:293–302.
14. Laurén E, Vikatmaa L, Kerkelä E, Kivipuro T, Krusius T, Syrjälä M,
revision of the manuscript and gave final approval of the version to be
et al. Red blood cell transfusion in southern Finland from 2011 to
published. 2016: a quality audit. Transfus Med. 2019;29:41–7.
15. Mueller MM, Van Remoortel H, Meybohm P, Kivipuro T, Krusius T,
CONF LICT OF IN TE RE ST Syrjälä M, et al. Patient blood management: recommendations from the
2018 Frankfurt consensus conference. JAMA. 2019;321:983–97.
The authors declare that they have no conflicts of interest.
16. Sadana D, Pratzer A, Scher LJ, Saag HS, Adler N, Volpicelli FM, et al. Pro-
moting high-value practice by reducing unnecessary transfusions with a
FUND ING INFORMATION patient blood management program. JAMA Intern Med. 2018;178:
None. 116–22.
17. Bruun MT, Pendry K, Georgsen J, Manzini P, Lorenzi M, Wikman A, et al.
Patient blood management in Europe: surveys on top indications for red
ORCID blood cell use and patient blood management organization and activities
Jana Vanden Broeck https://orcid.org/0000-0002-8514-9765 in seven European university hospitals. Vox Sang. 2016;111:391–8.
18. Hensley NB, Cho BC, Visagie M, Lester LC, Abernathy JH III,
Frank SM. How do I audit intraoperative blood component utilization
RE FE R ENC E S
in cardiac surgery? Transfusion. 2019;59:3058–64.
1. Bosch MA, Contreras E, Madoz P, Ortiz P, Pereira A, Pujol MM, et al.
The epidemiology of blood component transfusion in Catalonia,
northeastern Spain. Transfusion. 2011;51:105–16. SUPPORTING INFORMATION
2. Fillet AM, Desmarets M, Assari S, Quaranta JF, François A, Pugin A,
Additional supporting information may be found online in the
et al. Blood products use in France: a nationwide cross-sectional sur-
vey. Transfusion. 2016;56:3033–41.
Supporting Information section at the end of this article.
3. Tinegate H, Pendry K, Murphy M, Babra P, Grant-Casey J,
Hopkinson C, et al. Where do all the red blood cells (RBCs) go?
Results of a survey of RBC use in England and North Wales in 2014. How to cite this article: Vanden Broeck J, Beeckman K, Van
Transfusion. 2016;56:139–45. Gastel E, De Keersmaecker L, Devos T, Gérard C, et al.
4. Leahy MF, Hofmann A, Towler S, Trentino KM, Burrows SA, Improvement of transfusion practice and reduction in red
Swain SG, et al. Improved outcomes and reduced costs associated
blood cell utilization in Belgian hospitals: Results of a national
with a health-system-wide patient blood management program: a
retrospective observational study in four major adult tertiary-care survey and benchmarking. Vox Sang. 2022;117:259–67.
hospitals. Transfusion. 2017;57:1347–58.
Received: 20 December 2020 Revised: 5 May 2021 Accepted: 16 May 2021

DOI: 10.1111/vox.13161

ORIGINAL ARTICLE

A comparative study on perinatal outcomes of red blood

cell–alloimmunized pregnancies with anti-RhD in
combination and anti-RhD alone in China

Muxue Yu | Tonghui Tang | Rujiang Zheng | Miaoqiong Situ | Jiali Feng

Department of Pediatrics, The First Affiliated

Hospital, Sun Yat-sen University, Guangzhou, Abstract
China
Background and Objectives: The advent of intrauterine transfusion (IUT) has
Correspondence improved the survival of severe foetal anaemia. The aim of this study was to compare
Muxue Yu, Department of Pediatrics, The First the perinatal outcomes of red blood cell (RBC)–alloimmunized pregnancies with anti-
Affiliated Hospital, Sun Yat-sen University,
58 Zhongshan Road 2, Guangzhou 510080, RhD in combination and anti-RhD alone in China.
China. Materials and Methods: A retrospective study was conducted involving RBC–
Email: [email protected]
alloimmunized pregnancies with anti-RhD in combination and anti-RhD alone admit-
ted to The First Affiliated Hospital, Sun Yat-sen University, between January 2007
and December 2019. Obstetric data and neonatal outcomes were compared.
Results: A total of 165 alloimmunized pregnancies were identified, with 32 pregnan-
cies in the anti-RhD-in-combination group (25 pregnancies with anti-RhD + anti-
RhC and 7 pregnancies with anti-RhD + anti-RhE) and 133 pregnancies in the
anti-RhD-alone group. The anti-RhD-in-combination group had significantly higher
frequency of IUTs than the anti-RhD-alone group (59.4% [19/32] vs. 30.1%
[40/133]; p < 0.01). The postnatal frequency of top-up transfusions was significantly
higher in the anti-RhD in combination group than the anti-RhD-alone group (90.6%
[29/32] vs. 70.7% [94/133]; p = 0.02). There was no significant difference in the
frequency of exchange transfusions (ETs) between the two groups (15.6% [5/32]
vs. 17.3% [23/133]; p = 0.82).
Conclusions: Compared to alloimmunized pregnancies with anti-RhD alone, pregnan-
cies with anti-RhD in combination with anti-RhC or anti-RhE have an increased
requirement for antenatal IUTs and postnatal top-up transfusions but do not have an
increased need for ETs.

KEYWORDS
alloimmunization, anti-RhD alone, anti-RhD in combination, pregnancy, red blood cell antibodies

I N T R O D U CT I O N Anti-RhD is the most common antibody implicated in foetal or new-

born haemolysis [1, 2]. Antibodies to other Rh antigens, such as C, c, E
Rh red blood cell (RBC)-alloimmunized pregnancies are associated and e, also contribute to the haemolytic effect [3, 4].
with foetal or newborn haemolysis due to incompatibility of the Rh Studies [5–7] from the United States and Europe have shown that
RBC antigen. The Rh antigens are highly immunogenic. RhD-negative pregnancies with anti-RhD combined with other RBC antibody speci-
individuals will produce anti-RhD if they encounter the D antigen. ficities are likely to develop more aggressive immune responses than

268 © 2021 International Society of Blood Transfusion. wileyonlinelibrary.com/journal/vox Vox Sanguinis. 2022;117:268–274.
Anti-RhD IN COMBINATION VERSUS anti-RhD ALONE 269

pregnancies with anti-RhD alone. The ethnicities of the pregnant The derivation of the included pregnancies and the reasons for
women were mainly Native Americans, Caucasians and blacks. The exclusion are shown in Figure 1. Of the 181 potentially eligible preg-
frequency of Rh antigens and phenotypes varies across ethnicities nancies with Rh RBC alloimmunization, five pregnancies with non-
throughout the world [1]. No published studies have compared Rh anti-RhD (two pregnancies with anti-RhE alone, one pregnancy with
RBC alloimmunization perinatal outcomes caused by anti-RhD in com- anti-RhC + anti-RhE, one pregnancy with anti-RhE + anti-RhC
bination with anti-RhD alone among Asians. + anti-Jkb and one twin gestation with anti-RhC + anti-RhE) were
In mainland China, anti-RhD immunoglobulin is not available excluded. There were 35 pregnancies with anti-RhD in combination.
because of import restrictions. The prevalence of RhD-negative Two pregnancies with anti-RhD + non-anti-Rh (one pregnancy with
women varies widely around the globe, and it is only approximately anti-RhD + anti-M and one pregnancy with anti-RhD + anti-A) were
0.5% in China. Nevertheless, maternal RhD alloimmunization occurs excluded. One pregnancy with anti-RhD in combination with anti-RhC
frequently, given the high number of births and the very high likeli- was excluded because an emergency caesarean section was per-
hood that the father is RhD-positive. In addition, the implementation formed for foetal distress 8 days after an IUT. We also excluded three
of the ‘more than one child’ policy is expected to increase the occur- pregnancies with twins, one pregnancy with an incompetent cervix,
rence of Rh disease because the response to Rh incompatibility one pregnancy requiring emergency delivery (due to foetal distress
increases in severity with subsequent pregnancies [8]. The advent of 2 days after an IUT), one pregnancy with rupture of membranes
intrauterine transfusion (IUT) has improved the perinatal outcomes (occurred 4 days after an IUT that resulted in a preterm delivery) and
and survival of severe foetal anaemia [9]. IUT is carried out to treat two pregnancies with severe neonatal complications (congenital
foetal anaemia due to maternal alloimmunization in our hospital cytomegalovirus infection and intestinal perforation complicated with
[10–12]; as a result, the healthcare team at our hospital peritonitis, respectively) among 141 pregnancies with anti-RhD alone.
manages most patients with RBC alloimmunization in pregnancy In summary, twin gestations, pregnancies with other RBC antibodies
throughout China. (except for anti-RhC and anti-RhE) in combination with anti-RhD and
The aim of this study was to compare perinatal outcomes of pregnancies requiring emergency delivery for foetal distress during
RBC-alloimmunized pregnancies that received routine antenatal and antenatal management were excluded, as well as pregnancies with
neonatal management caused by anti-RhD in combination with anti- severe neonatal complications. Altogether, 32 pregnancies with anti-
RhD alone in China. RhD in combination (anti-RhD + anti-RhC [n = 25] or anti-RhD
+ anti-RhE [n = 7]) and 133 pregnancies with anti-RhD alone were
eligible to enrol our study.
MATERIALS AND METHODS

Study population Obstetric data collection

This retrospective single-centre study was conducted at The First Obstetric data were recorded on the following: maternal age, number
Affiliated Hospital, Sun Yat-sen University, Guangzhou, China. Follow- of pregnancies, maternal indirect anti-globulin tests (IATs), maximum
ing approval of the Institutional Review Board, the medical records antibody titre, major obstetric complications, past obstetric history,
were searched to identify pregnant patients diagnosed with Rh RBC gestational age at the first IUT, foetal haemoglobin (Hb) concentration
alloimmunization and their infants who received routine antenatal and and haematocrit (Hct) levels at the first IUT, number of IUTs, foetal
neonatal management in The First Affiliated Hospital, Sun Yat-sen direct agglutination test (DAT) and elution analysis and foetal RBC
University, between January 2007 and December 2019. antigen type.

FIGURE 1 Diagram of inclusion of pregnancies for anti-RhD-in-combination versus anti-RhD-alone study

270 YU ET AL.

The alloimmunized pregnant women had routine antibody screen- After delivery, the newborns were transferred to the Neonatal
ing at the first antenatal visit. The Rh phenotypes of the pregnant Unit and received neonatal intensive care. Phototherapy was initiated
women were available. They were RhD-negative and were shown not and intravenous immunoglobulin (IVIG) was administered to the new-
to be D variants by IATs with a monoclonal human IgM + IgG blend. borns at the time of admission to the Neonatal Unit. Top-up RBC
The Rh phenotypes of the husbands were also available. Repeat anti- transfusions were performed when Hb levels were < 8.0 g/dl or at
RhD, anti-RhC and anti-RhE titres were performed each month until higher levels if clinical symptoms of anaemia (lethargy, feeding prob-
24–28 weeks’ gestation and every 2 weeks thereafter. For pregnan- lems or failure to thrive) were present. Data on the number of top-up
cies that had reached 16–20 weeks’ gestation, the middle cerebral transfusions during the 3 months of life were collected. The indication
artery peak systolic velocity (MCA-PSV), as measured by ultrasound for ET was based on the consensus from the Society of Pediatrics,
Doppler interrogation, was used to detect foetal anaemia. Measure- Chinese Medical Association [13, 14].
ments of the MCA-PSV were repeated every 1–2 weeks and were
performed more frequently in pregnancies with higher multiples of
the mean (MoM) levels. Cordocentesis was performed to determine Statistical analysis
the foetal Hct when the MCA-PSV value was ≥1.5 MoM. An IUT was
indicated if the foetal Hct was ≤30%. With the exception of two preg- Data are presented as mean  standard deviation (SD), median (inter-
nancies, gravidas presenting at >34 weeks’ gestation were typically quartile range [IQR]) or numbers and percentage as appropriate. For
offered delivery. The two gravidas (one in the anti-RhD-in- comparisons, the t-test was used if the values were normally distrib-
combination group and one in the anti-RhD-alone group) who had uted, otherwise the Mann–Whitney U test was applied. The chi-
undergone seven or six IUTs, respectively, were delivered at square test or Fisher’s exact test were used for distribution-based
<33 weeks’ gestation because the risk of continued monitoring and comparison. A p-value <0.05 was considered statistically significant.
IUT was balanced. All statistical analyses were performed using SPSS (version 20; IBM
Corp., Armonk, NY).

Neonatal data collection

RE SU LT S
Neonatal data were recorded on the following: gender, gestational age at
birth, Apgar score at 1 and 5 min, birth weight, Hb concentration, Hct, bil- Altogether 165 alloimmunized pregnant women and their infants
irubin levels at birth, blood DAT and elution analysis, RBC antigen status, were included in the study. Anti-RhD with anti-RhC or anti-RhD with
maximum bilirubin level during hospitalization, duration of phototherapy, anti-RhE antibodies were identified in pregnant women in the anti-
number of exchange transfusions (ETs) required and number of top-up RhD-in-combination group, and only anti-RhD antibodies were identi-
transfusions received during the first 3 months of life. fied in pregnant women in the anti-RhD-alone group.

TABLE 1 Maternal characteristics of the anti-RhD-in-combination and anti-RhD-alone groups

Anti-RhD in combination (n = 32) Anti-RhD alone (n = 133) p-value

Maternal age (years)a 31.1  4.3 31.1  4.0 0.92
b
Median number of pregnancies 4 (3.8–5) 3 (2–4) <0.01
Maximum anti-RhD titre – n (%)
4-1024 26 (81.3) 121 (91.0) 0.20
≥2048 6 (18.7) 12 (9.0)
Gestational diabetes – n (%) 4 (12.5) 18 (13.5) 1.00
Pregnancy-induced hypertension – n (%) 1 (3.1) 0 (0) 0.19
Placental abruption – n (%) 0 (0) 2 (1.5) 1.00
Intrauterine growth restriction – n (%) 1 (3.1) 1 (0.8) 0.35
Caesarean section – n (%) 24 (75.0) 99 (74.4) 0.95
History of IUT – n (%) 1 (3.1) 3 (2.3) 0.58
History of IUFD – n (%) 14 (43.8) 17 (12.8) <0.01
History of neonatal death – n (%) 12 (37.5) 22 (16.5) <0.01

Abbreviations: IUFD, intrauterine foetal death; IUT, intrauterine transfusion.

a
Value given as mean  SD.
b
Value given as median (IQR).
Anti-RhD IN COMBINATION VERSUS anti-RhD ALONE 271

TABLE 2 IUT characteristics of the anti-RhD-in-combination and anti-RhD-alone groups

Anti-RhD in combination (n = 32) Anti-RhD alone (n = 133) p-value

IUT required – n (%) 19 (59.4) 40 (30.1) <0.01
a
Median number of IUTs 1.5 (0–5) 0 (0–1) <0.01
Median number of IUTs in subgroup requiring IUTa,b 4 (2–6) 3 (1–5) 0.05
Gestational age at first IUT (weeks) b,c
26.1  4.8 29.2  4.0 0.01
Foetal Hb level at first IUT (g/dl)b,c 6.6  2.5 7.4  2.1 0.22
Foetal Hct level at first IUT (%)b,c
20.9  7.2 22.5  6.0 0.36

Abbreviations: Hb, haemoglobin; Hct, haematocrit; IUT, intrauterine transfusion.

a
Value given as median (IQR).
b
Assessed in pregnancies requiring IUT.
c
Value given as mean  SD.

TABLE 3 Neonatal outcomes and management of the anti-RhD-in-combination and anti-RhD-alone groups

Anti-RhD in combination (n = 32) Anti-RhD alone (n = 133) p-value

Male – n (%) 17 (53.1) 69 (51.9) 0.90
Gestational age at birth (weeks)a 36.0  1.7 36.6  1.4 0.04
Birth weight (g) a
2672  480 2804  361 0.08
Apgar score < 7 at 1 min – n (%) 2 (6.3) 6 (4.5) 1.00
Apgar score < 7 at 5 min – n (%) 1 (3.1) 0 (0) 0.19
RDS – n (%) 4 (12.5) 6 (4.5) 0.20
Hypoglycaemia – n (%) 1 (3.1) 2 (1.5) 0.48
Hb level at birth (g/dl)a 11.8  2.4 13.1  2.5 <0.01
Hct level at birth (%)a 34.7  6.9 38.9  6.8 <0.01
Bilirubin level at birth (μmol/L)a 99.0  32.8 84.2  33.8 0.03
Maximum bilirubin level (μmol/L) a
203.3  62.2 207.1  67.8 0.78
Duration of phototherapy (days)a 5 (4–6) 5 (4–7) 0.88
Top-up transfusion required – n (%) 29 (90.6) 94 (70.7) 0.02
Median number of top-up transfusionsb 2 (1–3) 1 (0–2) 0.02
ET required – n (%) 5 (15.6) 23 (17.3) 0.82
Median number of ETsb 1 (0–0) 0 (0–0) 0.87
Median number of ETs in subgroup requiring ETb,c 1 (1–2) 1 (1–2) 0.81

Abbreviations: ET, exchange transfusion; Hb, haemoglobin; Hct, haematocrit; RDS, respiratory distress syndrome.
a
Value given as mean  SD.
b
Value given as median (IQR).
c
Assessed in newborns requiring ET.

Maternal and foetal data pregnancies in the anti-RhD-alone group were the first pregnancy;
the two gravidas had histories of RBC transfusions. The other
The maternal characteristics of the anti-RhD-in-combination and anti- 131 pregnancies in the anti-RhD-alone group were subsequent preg-
RhD-alone groups are depicted in Table 1. The anti-RhD-in- nancies, and 9 of the 131 gravidas had histories of RBC transfusions.
combination and anti-RhD-alone groups were comparable with No significant difference was demonstrated in the history of IUT fre-
respect to maternal age, distribution of maximum anti-RhD titre, fre- quency between the two groups. The frequencies of intrauterine foe-
quencies of major obstetric complications and caesarean section rate. tal death (IUFD) and neonatal death histories were significantly higher
The median number of pregnancies was higher in the anti-RhD-in- in the anti-RhD-in-combination group than the anti-RhD-alone group
combination group than the anti-RhD-alone group (4 [IQR 3.8–5, (43.8% [14/32] vs. 12.8% [17/133]; p < 0.01 and 37.5% [12/32]
range 2–9] vs. 3 [IQR 2–4, range 1–7]; p < 0.01). All pregnancies in vs. 16.5% [22/133]; p < 0.01).
the anti-RhD-in-combination group were subsequent pregnancies, The IUT characteristics are presented in Table 2. The anti-RhD-
and 2 of the 32 gravidas had histories of RBC transfusions. Two in-combination group had significantly higher frequency of IUTs
272 YU ET AL.

during pregnancy than the anti-RhD-alone group (59.4% [19/32] There were alloimmunized pregnancies with anti-RhD plus anti-
vs. 30.1% [40/133]; p < 0.01). The anti-RhD-in-combination group RhC or anti-RhE, while there were no pregnancies with anti-RhD plus
also had significantly higher median number of IUTs during pregnancy two other Rh antibody specificities in our study. Other authors have
than the anti-RhD-alone group (1.5 [IQR 0–5, range 0–8] vs. 0 [IQR reported alloimmunized pregnancies by anti-RhD plus anti-RhC plus
0–1, range 0–7]; p < 0.01). In the subgroup analysis of pregnancies anti-RhE [6, 7, 15, 16]. This is due to the low occurrence of the RhD-
requiring IUT, the median number of IUTs tended to be higher in the negative ccee phenotype in Asians. Anti-RhD plus anti-RhC plus anti-
anti-RhD-in-combination group than the anti-RhD-alone group RhE could be produced through pregnancy with the RhD-negative
(4 [IQR 2–6, range 1–8] vs. 3 [IQR 1–5, range 1–7]; p = 0.05). The ccee phenotype. The frequency of the RhD-negative ccee phenotype
gestational age at the first IUT was significantly lower in the anti- in Caucasians and blacks is 15.1% and 6.8%, respectively. Neverthe-
RhD-in-combination group compared to the anti-RhD-alone group less, the frequency of the RhD-negative ccee phenotype in Asians is
(26.1 weeks vs. 29.2 weeks; p = 0.01). No significant differences were only 0.1% [17].
demonstrated in foetal mean Hb concentration and Hct at the first In our study, alloimmunized pregnancies with anti-RhD in com-
IUT between the anti-RhD-in-combination and anti-RhD-alone group. bination with one Rh antibody specificity were more likely to have
foetal and neonatal anaemia compared with pregnancies with anti-
RhD alone. As a result, there was an increased requirement for IUTs
Neonatal outcomes and management and postnatal top-up transfusions. These findings could account for
anti-RhD in combination with anti-RhC or anti-RhE enhancing the
Table 3 shows the neonatal outcomes and management of the anti- immune response. IUT appears to increase and prolong the risk of
RhD-in-combination and anti-RhD-alone groups. The neonatal gender anaemia, probably by reducing erythropoiesis [2, 18]. These find-
and frequencies of Apgar score < 7 at 1 min, Apgar score < 7 at 5 min, ings are in accordance with the results of previous studies [5–7]. In
respiratory distress syndrome (RDS) and hypoglycaemia were not sig- the study conducted by Markham et al. [5], alloimmunized pregnan-
nificantly different between the two groups. A lower gestational age cies with anti-RhD alone versus alloimmunized pregnancies with
at birth existed in the anti-RhD-in-combination group compared to anti-RhD plus one or more additional RBC antibodies were com-
the anti-RhD-alone group (36.6 weeks vs. 36.0 weeks; p = 0.04). The pared. In the study of Nordvall et al. [6], alloimmunized pregnancies
birth weight was not significantly different between the anti-RhD-in- with anti-RhD alone versus alloimmunized pregnancies with anti-
combination group and the anti-RhD-alone group, but a trend was RhD plus one other RBC antibody were compared. In Spong et al.’s
noted (2672 g vs. 2804 g; p = 0.08). study [7], alloimmunized pregnancies with anti-RhD alone versus
The Hb concentration and Hct at birth were significantly lower in alloimmunized pregnancies with two to four RBC antibodies were
the anti-RhD-in-combination group than the anti-RhD-alone group compared.
(11.8 g/dl vs. 13.1 g/dl; p < 0.01 and 34.7% vs. 38.9%; p < 0.01). The Phung et al. [15] and Walsh et al. [16] focussed on those pregnan-
bilirubin levels at birth were significantly higher in the anti-RhD-in- cies requiring IUT. Phung et al. [15] showed that in pregnancies
combination group than the anti-RhD-alone group (99.0 μmol/L requiring IUT, the daily decrease in Hb between the first and second
vs. 84.2 μmol/L; p = 0.03). IUT did not detect a significant difference between alloimmunized
The anti-RhD-in-combination group had significantly higher fre- pregnancies with anti-RhD alone and anti-RhD plus one other RBC
quency of top-up transfusions than the anti-RhD-alone group (90.6% antibody while the daily decrease in Hb were significantly lower in
[29/32] vs. 70.7% [94/133]; p = 0.02). The median number of top-up alloimmunized pregnancies with anti-RhD alone compared with anti-
transfusions per infant was also higher in the anti-RhD-in-combination RhD plus two RBC antibodies. Walsh et al. [16] reported that there
group compared to the anti-RhD-alone group (2 [IQR 1–3, range 0–5] were no differences in the number of IUTs per affected alloimmunized
vs. 1 [IQR 0–2, range 0–8]; p = 0.02). There were no significant differ- pregnancy when comparing anti-RhD alone with anti-RhD plus one
ences in the mean maximum bilirubin level, median duration of photo- other RBC antibody.
therapy, frequency of ETs and median number of ETs between the There was a lack of data on postnatal ETs in the studies of
two groups. In the subgroup analysis of newborns requiring ET, no sig- Markham et al. [5] and Sponges et al. [7]. The Nordvall et al.’s study
nificant difference in the median number of ETs between the two [6] showed that alloimmunized pregnancies with anti-RhD in combi-
groups was found. nation required more postnatal ETs compared with anti-RhD alone.
Indeed, we detected no difference in the postnatal maximum bilirubin
level, duration of phototherapy, frequency of ETs and median number
DISCUSSION of ETs in both groups. In our department, phototherapy was initiated,
and IVIG was administered when the newborn was admitted to the
This study showed that compared to alloimmunized pregnancies with Neonatal Unit because Rh alloimmunization is a risk factor for
anti-RhD alone, alloimmunized pregnancies with anti-RhD-in-combi- unfavourable neonatal outcomes. The American Academy of Pediat-
nation with anti-RhC or anti-RhE increased the need for antenatal rics suggests that intensive phototherapy and IVIG administration
IUTs and postnatal top-up transfusions. The need for ETs was similar may reduce the need for ETs in infants with Rh haemolytic disease
in both groups. [19]. Studies of infants with Rh incompatibility supported a reduction
Anti-RhD IN COMBINATION VERSUS anti-RhD ALONE 273

in the use of ET with IVIG treatment [20]. Intensive phototherapy for infants with anaemia in both groups were performed in our study;
neonatal hyperbilirubinaemia has resulted in a worldwide decrease in however, the initial infant’s age at the time of EPO administration,
the need for ETs during the last two to three decades [21–23]. duration and dosages of EPO varied, and a number of infants were
In addition, replacement of foetal RBCs by antigen-negative donor treated by EPO after discharge. We did not collect completed
RBCs during the IUT could result in less active postnatal neonatal information regarding the use of EPO in both groups. Finally, our
alloimmune haemolysis [18]. Therefore, although alloimmunized preg- study was retrospective and was limited to obstetric and neonatal
nancies with anti-RhD in combination compared to alloimmunized outcomes. There was also a lack of data about long-term growth
pregnancies with anti-RhD alone had more severe foetal haemolysis development of the infants.
and an increased need for antenatal IUTs, an increased need for post- In conclusion, alloimmunized pregnancies with anti-RhD in combi-
natal ETs was not required. Currently, most foetal therapy centres opt nation with anti-RhC or anti-RhE need more perinatal treatment than
for postnatal top-up transfusions rather than postnatal ETs [9, 24]. alloimmunized pregnancies with anti-RhD alone, including IUTs and
Theoretically, more bilirubin is produced after birth because of more top-up transfusions. More severe haemolysis in pregnancies with anti-
severe haemolysis in alloimmunized pregnancies with anti-RhD in RhD in combination with anti-RhC or anti-RhE resulting in foetal
combination [25]. However, we did not demonstrate a difference anaemia leads to an increased need for IUTs, and pregnancies with
in the postnatal maximum bilirubin level and duration of phototherapy anti-RhD in combination with anti-RhC or anti-RhE require more post-
between the two groups. This finding could be explained that the natal transfusions for neonatal anaemia. Pregnancies with anti-RhD in
severe haemolysis in pregnancies underwent antenatal IUTs and neo- combination with anti-RhC or anti-RhE, however, do not have an
nates in both groups received aggressive phototherapy. Previous increased need for ETs.
studies did not show the neonatal bilirubin level and phototherapy
data [5–7]. AC KNOW LEDG EME NT S
Although alloimmunized pregnancies with anti-RhD in combina- We would like to thank Dr. Yanli Ji from Guangzhou Blood Center for
tion increased the need for antenatal IUTs and postnatal top-up trans- her advice on datainterpretation.
fusions, there were no differences in type of delivery and major M.Y. contributed to study design, data collection, statistical analy-
obstetric and neonatal complications in both groups. RDS is associ- sis, manuscript writing/revising and final approval. T.T. contributed to
ated with pre-term delivery [26]. Gestational age at birth was lower in study design, data collection, statistical analysis and manuscript writ-
the anti-RhD-in-combination group than the anti-RhD alone group ing. R.Z., M.S. and J.F. contributed to data collection and manuscript
in our study; however, the mean gestational age at birth in both writing.
groups was >36 weeks. RDS is more common in neonates born at a
gestational age < 35 weeks when compared to a gestational CONFLIC T OF INT ER E ST
age > 35 weeks [27]. Therefore, the risk of RDS decreased in the anti- The authors declare no conflict of interests.
RhD-in-combination and anti-RhD-alone groups. A fetus with immune
haemolysis destined to become hydropic after 32–34 weeks’ gesta- FUNDING INF ORMATI ON
tion might survive after early delivery and prompt ET, before the None.
development of hydrops [28]. Currently, the therapeutic options for
pregnancies with RBC alloimmunization consist of controlled early ORCID
delivery and IUT. Pregnancies with a foetus at significant risk for foe- Muxue Yu https://orcid.org/0000-0002-1122-3918
tal anaemia are delivered at 37–38 weeks’ gestation unless indications
develop prior to this time [29]. The last foetal transfusion is given at RE FE RE NCE S
approximately 35 weeks with delivery at 37–38 weeks’ gestation in 1. Dean L. Blood groups and red cell antigens [internet]. Bethesda, MD:
preference to early delivery and ET because the foetus tolerates larger National Center for Biotechnology Information (US); 2005.
transfusion volumes than the neonate due to the large capacity of the 2. Rath ME, Smits-Wintjens VE, Walther FJ, Lopriore E. Hematolog-
ical morbidity and management in neonates with hemolytic dis-
placenta [30].
ease due to red cell alloimmunization. Early Hum Dev. 2011;87:
There are several limitations to this study. First, these data 583–8.
were limited to alloimmunized pregnancies with anti-RhD alone 3. Moise KJ. Fetal anemia due to non-Rhesus-D red-cell
and anti-RhD plus one other anti-Rh antibody specificity. The num- alloimmunization. Semin Fetal Neonatal Med. 2008;13:207–14.
4. Moise KJ, Argoti PS. Management and prevention of red cell
ber of pregnancies with anti-RhD plus non-anti-Rh or with multiple
alloimmunization in pregnancy: a systematic review. Obstet Gynecol.
RBC antibodies was less, thus these pregnancies were not included
2012;120:1132–9.
in our study. We collected data after the introduction of MCA-PSV 5. Markham KB, Rossi KQ, Nagaraja HN, O’Shaughnessy RW. Hemo-
in our hospital, which could cause a limited number of pregnancies lytic disease of the fetus and newborn due to multiple maternal anti-
with multiple RBC antibodies. Additionally, we did not compare the bodies. Am J Obstet Gynecol. 2015;213:61–8.
6. Nordvall M, Dziegiel M, Hegaard HK, Bidstrup M, Jonsbo F,
use of erythropoietin (EPO) in both groups. Small case series
Christensen B, et al. Red blood cell antibodies in pregnancy and their
reported the use of EPO to prevent late anaemia in neonates with clinical consequences: synergistic effects of multiple specificities.
Rh haemolytic disease [2, 31]. EPO administration to some of the Transfusion. 2009;49:2070–5.
274 YU ET AL.

7. Spong CY, Porter AE, Queenan JT. Management of isoimmunization 20. Zwiers C, Scheffer-Rath ME, Lopriore E, de Haas M, Liley HG. Immu-
in the presence of multiple maternal antibodies. Am J Obstet noglobulin for alloimmune hemolytic disease in neonates. Cochrane
Gynecol. 2001;185:481–4. Database Syst Rev. 2018;3:D3313.
8. Visser G, Di Renzo GC, Spitalnik SL. The continuing burden of Rh dis- 21. Steiner LA, Bizzarro MJ, Ehrenkranz RA, Gallagher PG. A decline in the
ease 50 years after the introduction of anti-Rh(D) immunoglobin pro- frequency of neonatal exchange transfusions and its effect on
phylaxis: call to action. Am J Obstet Gynecol. 2019;221:221–7. exchange-related morbidity and mortality. Pediatrics. 2007;120:27–32.
9. Zwiers C, van Kamp I, Oepkes D, Lopriore E. Intrauterine transfusion 22. Murki S, Kumar P. Blood exchange transfusion for infants with severe
and non-invasive treatment options for hemolytic disease of the neonatal hyperbilirubinemia. Semin Perinatol. 2011;35:175–84.
fetus and newborn - review on current management and outcome. 23. Siddappa AM, Prekker FL, Slusher TM. Improving effectiveness of
Expert Rev Hematol. 2017;10:337–44. phototherapy in an academic center: a quality improvement project.
10. Li L, Huang L, Luo G, Luo Y, Fang Q. Prenatal treatment of severe fetal Glob Pediatr Health. 2020;7:2333794X20969275X.
hemolytic disease due to anti-M alloimmunization by serial intrauterine 24. Lindenburg IT, van Kamp IL, Oepkes D. Intrauterine blood transfu-
transfusions. Taiwan J Obstet Gynecol. 2017;56:379–81. sion: current indications and associated risks. Fetal Diagn Ther.
11. Wu J, Huang LH, Luo YM, Fang Q. Discordant intrauterine transfu- 2014;36:263–71.
sion in dichorionic twin pregnancy with Rh isoimmunization. Trans- 25. Volpe JJ, Inder TE, Darras BT, de Vries LS, du Plessis AJ, Neil J, et al.
fus Apher Sci. 2017;56:883–5. Volpe’s neurology of the newborn. 6th ed. Philadelphia, PA: Elsevier;
12. Li S, Mo C, Huang L, Shi X, Luo G, Ji Y, et al. Hemolytic disease of 2018.
the fetus and newborn due to alloanti-M: three Chinese case reports 26. Sweet DG, Carnielli V, Greisen G, Hallman M, Ozek E, te Pas A, et al.
and a review of the literature. Transfusion. 2019;59:385–95. European consensus guidelines on the Management of Respiratory
13. Subspecialty Group of Neonatology, The Society of Pediatrics, Distress Syndrome - 2019 update. Neonatology. 2019;115:432–50.
Chinese Medical Association. The experts consensus on the manage- 27. Imkesorn R, Hluangdansakul W, Prommas S, Chotedelok A,
ment of neonatal hyperbilirubinemia. Chin J Pediatr. 2014;52:745–8. Luangwilai T. Comparison respiratory distress syndrome (RDS) rate
14. Subspecialty Group of Neonatology, The Society of Pediatrics, Chi- in late pre-term infants between gestational age 34 weeks and
nese Medical Association. The management of neonatal of 35-36 weeks. Thai J Obstetr Gynaecol. 2015;23:8–14.
neonatal jaundice. Chin J Pediatr. 2001;39:185–7. 28. Bowman J. The management of hemolytic disease in the fetus and
15. Phung TV, Houfflin-Debarge V, Ramdane N, Ghesquière L, newborn. Semin Perinatol. 1997;21:39–44.
Delsalle A, Coulon C, et al. Maternal red blood cell alloimmunization 29. Mari G, Norton ME, Stone J, Berghella V, Sciscione AC, Tate D, et al.
requiring intrauterine transfusion: a comparative study on manage- Society for Maternal-Fetal Medicine (SMFM) clinical guideline #8:
ment and outcome depending on the type of antibody. Transfusion. the fetus at risk for anemia–diagnosis and management. Am J Obstet
2018;58:1199–205. Gynecol. 2015;212:697–710.
16. Walsh CA, Russell N, McAuliffe FM, Higgins S, Mahony R, Carroll S, 30. Urbaniak SJ, Greiss MA. RhD haemolytic disease of the fetus and the
et al. Relationship between maternal antibody type and antenatal newborn. Blood Rev. 2000;14:44–61.
course following intrauterine transfusion for red cell alloimmunisation. 31. Alaqeel AA. Hyporegenerative anemia and other complications of
Eur J Obstet Gynecol Reprod Biol. 2013;171:235–9. rhesus hemolytic disease: to treat or not to treat is the question. Pan
17. Reid ME, Lomas-Francis C, Olsson ML. RH - RH blood group system. Afr Med J. 2019;32:120.
In: Reid ME, Lomas-Francis C, Olsson ML, editors. The blood group
antigen FactsBook. 3rd ed. Boston, MA: Academic Press; 2012.
p. 147–262.
18. De Boer IP, Zeestraten EC, Lopriore E, van Kamp IL, Kanhai HH,
How to cite this article: Yu M, Tang T, Zheng R, Situ M,
Walther FJ. Pediatric outcome in rhesus hemolytic disease treated
with and without intrauterine transfusion. Am J Obstet Gynecol. Feng J. A comparative study on perinatal outcomes of red
2008;198:51–4. blood cell–alloimmunized pregnancies with anti-RhD in
19. American Academy of Pediatrics Subcommittee on Hyper- combination and anti-RhD alone in China. Vox Sang. 2022;
bilirubinemia. Management of hyperbilirubinemia in the newborn
117:268–274.
infant 35 or more weeks of gestation. Pediatrics. 2004;114:
297–316.
Received: 31 March 2021 Revised: 26 May 2021 Accepted: 2 June 2021

DOI: 10.1111/vox.13167

ORIGINAL ARTICLE

A first WHO reference reagent for the detection of

anti-human platelet antigen-15b

Giles Sharp1 | Anthony Poles2 | Lucy Studholme1

1
Biotherapeutics Division, National Institute
for Biological Standards and Control (NIBSC), Abstract
Potters Bar, UK
Background and Objectives: Alloantibodies to human platelet antigen-15b (anti-
2
Histocompatibility and Immunogenetics,
NHSBT, Filton, UK
HPA-15b) have been detected in mothers with foetal–neonatal alloimmune thrombo-
cytopenia and in multiply transfused patients. Assays used to detect this antibody,
Correspondence
which aids in disease diagnosis, can be unreliable and vary in sensitivity. The objec-
Lucy Studholme, Biotherapeutics Division,
National Institute for Biological Standards and tive was to generate a stable, lyophilized anti-HPA-15b preparation and evaluate its
Control (NIBSC), Potters Bar, Herts, EN6 3QG,
suitability as a World Health Organization (WHO) reference reagent for use in the
UK.
Email: [email protected] quality control of platelet alloantibody detection assays. Results from an international
collaborative study to evaluate the preparation were used to assign a minimum
Funding information
None. potency at which laboratories can be expected to detect the antibody.
Materials and Methods: Recalcified plasma containing anti-HPA-15b was aliquotted,
lyophilized and coded 18/220. Twenty-five laboratories in 16 countries tested dou-
bling dilutions of the reconstituted material in glycoprotein-specific assays such as
the monoclonal antibody–specific immobilization of platelet antigen assay and
reported the last positive (or endpoint) dilution.
Results: Twenty-four laboratories (96%) detected antibodies with HPA-15b specific-
ity in preparation 18/220. Reported endpoint dilutions were normally distributed
with a modal dilution of 1 in 16 and ranged from 1 in 2 to 1 in 128. Only two labora-
tories (8%) failed to detect anti-HPA-15b at 1 in 8 dilution.
Conclusions: When diluted 1 in 8, most laboratories detected anti-HPA-15b in prep-
aration 18/220 using HPA-15bb platelets but not with HPA-15aa platelets. The par-
ticipants agreed this to be an appropriate dilution for assignment as the minimum
potency. In October 2020, the WHO Expert Committee on Biological Standardiza-
tion approved 18/220 as an International Reference Reagent.

KEYWORDS
CD109, HPA-15b, MAIPA, platelet alloantibody detection, reference reagent

I N T R O D U CT I O N antigens; HPA -1, -2, -3, -4, -5, -15) [1, 2]. The molecular basis of the
41 antigens has been resolved, and in all but one (HPA-14b),
To date, a total of 41 platelet alloantigens have been defined serologi- the difference between self and non-self is defined by a single
cally, of which 12 are grouped in six bi-allelic systems (human platelet nucleotide polymorphism in the gene encoding the relevant membrane

This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium,
provided the original work is properly cited.
© 2021 The Authors. Vox Sanguinis published by John Wiley & Sons Ltd on behalf of International Society of Blood Transfusion.

Vox Sanguinis. 2022;117:275–281. wileyonlinelibrary.com/journal/vox 275

276 SHARP ET AL.

glycoprotein [2]. Alloantibodies against HPAs are involved in foetal– laboratories testing the established reference reagent at the
neonatal alloimmune thrombocytopenia (FNAIT), a disease in which assigned minimum potency should expect a positive result if the
foetal/neonatal platelets are destroyed by IgG alloantibodies from the sensitivity of their method is acceptable.
incompatible, HPA-sensitized mother. HPA antibodies are also involved
in platelet transfusion refractoriness (PTR), the repeated failure to
achieve the desired level of blood platelets following transfusion and in M A T E R I A L S A N D M ET H O D S
post-transfusion purpura (PTP), the delayed reaction to a transfusion
caused by recipient antibodies to platelet antigens. Identification of the Candidate material
HPA antibody specificity is essential to the diagnosis and treatment of
the patient. In severe cases of thrombocytopenia requiring treatment The material was provided to NIBSC by the National Blood Service,
with a platelet transfusion, it is important that the transfused platelets Oxford, United Kingdom. All donations of recalcified plasma came
are negative for the target alloantibody specificity [3–6]. from just one consenting donor, and each was screened using the
A variety of techniques are in use across the world to detect HPA MAIPA to identify the anti-HPA-15b titre. Those with the highest
antibodies. The monoclonal antibody–specific immobilization of plate- titres were pooled to make a bulk (coded 18/220) prior to filling
let antigen (MAIPA) assay is considered the gold standard method for 0.5 ml/ampoule into 1896 glass ampoules (size 2.5 ml). Filled mate-
the determination of platelet (IgG) alloantibody specificity. The rial was freeze-dried under conditions established in house for the
method involves using specific HPA-typed platelets to capture alloan- routine lyophilization of WHO serum standards/reference reagents;
tibodies in human serum/plasma. Mouse monoclonal antibodies spe- a summary of the product information is shown in Table 1. The indi-
cific for glycoproteins on which the platelet antigens of interest are vidual donations from which the candidate reference material was
located are used to capture the platelet antigen—alloantibody com- prepared and the pooled bulk were found to be negative for HBsAg,
plexes after solubilization of the platelets. This design ensures the anti-HIV1 + 2 and anti-hepatitis C virus (HCV). HCV PCR testing of
accurate identification of alloantibody specificity and avoids false- the pooled bulk was also negative, and no microbial contaminants
positives resulting from, for example, HLA antibodies. The MAIPA were detected.
does not distinguish between IgG alloantibody subclasses in patient A trial-sized plasma pool containing the same proportions of
samples. To date, there are three World Health Organization (WHO) plasma donations as for the definitive bulk (18/220) and selected sin-
reference reagents established (anti-HPA-5b, anti-HPA-3a and anti- gle donations used to prepare 18/220 were evaluated by two inde-
HPA-1a) [7–9], which are used for assay quality control. These pendent clinical laboratories; both laboratories were able to detect
reagents are used to validate the minimum sensitivity of tests for the anti-HPA-15b antibodies but could not detect antibodies with any
respective HPA antibodies. other HPA specificity. Both laboratories also confirmed the presence
In addition to HPA-1a, -3a and -5b, HPA-15 is of clinical rele- of anti-HLA class 1 antibodies, and the following specificities were
vance. Studies have shown this antigen to be as immunogenic as detected: A2, A68, A69, C5, C8 and C15. Furthermore, constituent
HPA-5, and alloantibodies against HPA-15b can be detected in
patients receiving multiple transfusions and mothers with FNAIT TABLE 1 Product summary
[10–12]. The platelet-based methods used for the detection of anti-
Code number 18/220
HPA-15 antibodies can be unreliable and vary in their sensitivity
Presentation Heat sealed, 2.5-ml
because CD109, the glycoprotein on which HPA-15b is located, is
glass ampoules
expressed in low numbers by platelets and is labile [13, 14]. An anti-
Number available 1830
HPA-15b minimum potency reference reagent would allow clinical
Date filled 1 March 2019
laboratories to validate their methods. The need for this important
Mean fill mass (n = 96) 0.52 g
reference material was emphasized in the proficiency testing
Fill mass CV (n = 96) 0.65%
scheme organized by National Institute for Biological Standards and
Control (NIBSC) in 2017 showing that anti-HPA-15b detection in Residual moisture by coulometric Karl 0.20%
Fischer titration (n = 6)
the samples distributed was generally poor. Furthermore, it was con-
Residual moisture CV (n = 6) 24.8%
cluded in a report by the Platelet Immunobiology Working Party of
Mean dry weight (n = 6) 0.04 g
the International Society of Blood Transfusion (ISBT) in 2018 [15]
that a more standardized approach to the CD109 MAIPA is required. Dry weight CV (n = 6) 0.44%

The purpose of this study was to produce and evaluate a candidate Mean oxygen in head space by lighthouse 0.20%
FMS670 (n = 6)
WHO anti-HPA-15b reference reagent (minimum potency) for use
as a quality control reagent in glycoprotein-specific assays used to Oxygen in head space CV (n = 6) 44.69%

detect/identify alloantibody specificity. The authors describe the Storage conditions 20 C

evaluation of an anti-HPA-15b lyophilized preparation coded Address of processing facility and NIBSC, Potters Bar, UK
custodian
18/220 in an international collaborative study, the results of which
are used to assign a minimum potency to this preparation. Clinical Abbreviation: CV, Coefficient of Variation.
ANTI-HPA-15B REFERENCE REAGENT 277

TABLE 2 List of participating laboratories quality assessment scheme organized by the National External Quality
Institute City, Country Assessment Site UK (NEQAS). In total, 27 clinical laboratories located
across the globe accepted the invitation to participate (see Table 2).
Australian Red Cross Blood Service, Melbourne, Australia
Victoria
Australian Red Cross Blood Service, Brisbane, Australia
Queensland Study design
Hospital Sirio Libanês S~
ao Paulo, Brazil
Canadian Blood Services Winnipeg, Canada Four ampoules of the definitive freeze-dried material (coded 18/220)

Institute of Blood Transfusion, Zhejiang Hangzhou, China were sent to each laboratory. Participants were asked to reconstitute the
Blood Centre material immediately before testing and to titrate at doubling dilutions in
Institute of Blood Transfusion, Guangzhou, China one or more assay method(s) routinely used in clinical practice. Partici-
Guangzhou Blood Centre pants were required to use a CD109 glycoprotein–specific method such
French Blood Establishment (EFS), HFNO Lille, France as the MAIPA because the material contains anti-HLA antibodies, which
French Blood Establishment (EFS), Rennes, France could otherwise cause false-positive results. Clinical laboratories would
Brittany normally be required to use such a method since patient samples can also
French Blood Establishment (EFS), Aura Lyon, France contain HLA antibodies. Two ampoules were to be tested, each on a dif-
Centre Hospitalier Universitaire de Nantes, France ferent day, with HPA-15bb and HPA-15aa platelets or similar on both
Nantes days. Ideally, different platelet donors with the same HPA-15 type were
Institut National de la Transfusion Paris, France to be used over 2 days. Laboratories were asked to report their interpre-
Sanguine tation of the results for each test by recording ‘positive’ or ‘negative’ for
Red Cross Blood Transfusion Services, Dessau, Germany each dilution tested. The endpoint dilution (antibody titre) for 18/220 in
NSTOB
each test was assigned as the largest (maximum) dilution, which the labo-
Zentrum für Transfusionsmedizin und Berlin, Germany
ratory reported to be ‘positive.’ Laboratories reporting results as ‘weak
Zelltherapie
positive’ or results that were defined as borderline in ‘grey zones’ of
Institute of Clinical Immunology and Giessen, Germany
defined optical density ranges were also deemed negative.
Transfusion Medicine
National Blood Centre Kuala Lumpur, Malaysia
Sanquin Diagnostic Services Amsterdam, Netherlands
Stability studies
University Hospital of North Norway Tromsø, Norway
Institute of Haematology and Transfusion Warsaw, Poland
To predict loss in stability over time, accelerated thermal degradation
Medicine
studies were performed at NIBSC using ampoules of lyophilized
Banc de Sang i Teixits Barcelona, Spain
18/220 stored at 70, 20, +4, +20, +37 and +45 C for 13 months.
Karolinska University Hospital Stockholm, Sweden
Reconstituted material from two ampoules at each temperature was
University Hospital Geneva Geneva, Switzerland tested in duplicate at a range of twofold serial dilutions in the MAIPA
The Thai Red Cross Society Bangkok, Thailand assay with HPA-15bb platelets. Absorbance readings were used to
National Health Service Blood and Filton, UK calculate the relative potencies of the accelerated thermal degradation
Transplant
samples by parallel-line analysis using the 70 sample as the refer-
Welsh Blood Service Pontyclun, UK ence. Relative potency calculations were performed using the
National Institute for Biological Standards Potters Bar, UK European Directorate for the Quality of Medicines software Com-
and Control
biStats, version 6.0, using a sigmoid curve model and logit transforma-
Versiti Milwaukee, USA tion of responses. The Arrhenius equation, relating degradation rate
Bloodworks Seattle, USA to absolute temperature assuming first-order decay [16], was used to
predict the degradation rates for each storage temperature.

donations of the definitive pool have been evaluated in previous pro-

ficiency testing schemes organized by NIBSC; no other antibody spec- RE SU LT S
ificities were reported other than anti-HPA-15b and anti-HLA.
A total of 27 clinical laboratories accepted the invitation to partici-
pate in the study, however, only 25 laboratories returned results.
Participants Each laboratory was assigned a code number, which does not
reflect the order of listing shown in Table 2. The methods used by
The invitation was distributed to participants of the ISBT Platelet Immu- participants are summarized in Table 3. All 25 laboratories per-
nology Workshop and to participants of the HPA antibody detection formed glycoprotein-specific assays, all of which were a version of
278 SHARP ET AL.

TABLE 3 Laboratory method summary the MAIPA using a monoclonal antibody specific for CD109. Some
Number of laboratories provided supplementary datasets from additional
Method Laboratory code entries platelet donors or from a second method. A full summary of the
Rapid MAIPA a a,b c d
3, 4 , 4a , 5, 7, 7a , 9, 11, 17, 17a , 13 results from each laboratory and details of their assay protocols are
18, 21, 23 shown in Appendix S1.
2-Day MAIPA 1, 10b, 14, 16 4
In-house MAIPA 2, 6, 12, 13, 15, 19, 20, 22 8
Other 8b,e, 18ae 2 Reported endpoint dilutions against HPA-15bb
antigen
Note: Labs providing supplementary datasets from a second method have
an ‘a’ suffix.
Abbreviation: MAIPA, monoclonal antibody–specific immobilization of The majority of laboratories tested two ampoules of the material on
platelet antigen. two separate occasions, using different donors of HPA-15bb plate-
a
Frozen platelets. lets, as described in the study protocol. Laboratory 20 reported
b
CD109 antibody other than mAb TEA 2/16.
c results from four different HPA-15bb donors using three of the
K562 Recombinant cells used.
d
Lyophilized platelets. ampoules provided and laboratory 1 reported results using three dif-
e
Modified MAIPA method. ferent HPA-15bb donors from three ampoules provided. Out of

F I G U R E 1 Maximum (endpoint) dilutions of anti-HPA-15b reference reagent (18/220) giving a positive result, as reported by international
collaborative study participants for HPA-15bb platelets/cells in CD109-specific assays for the detection of anti-HPA antibodies. Laboratory codes
are shown in boxes; box colours refer to assay repetitions with different platelet donors except for 7 where the same donor for each repetition
was used and 7a where recombinant cells were used
ANTI-HPA-15B REFERENCE REAGENT 279

F I G U R E 2 Maximum (endpoint) dilutions of anti-HPA-15b reference reagent (18/220) giving a positive result, as reported by international
collaborative study participants for HPA-15aa platelets/cells in CD109-specific assays for the detection of anti-HPA antibodies. Laboratory codes
are shown in boxes; box colours refer to assay repetitions with different platelet donors except for 4, 4a and 7 where the same donor for each
repetition was used and for 7a where recombinant cells were used

25 laboratories, 24 were indeed able to detect antibodies with HPA- significant in this study. Several laboratories (13 methods in total)
15b specificity in preparation 18/220 with all HPA-15bb platelet reported different endpoint dilutions for 18/220 in each indepen-
donors/cells used and in all methods performed. The remaining one dent assay using the same method but where different HPA-15bb
laboratory reported a positive result for undiluted material using platelet donors were used. For example, three of the five laborato-
both HPA-15aa and HPA-15bb platelets but failed to titrate the ries that reported poor endpoint dilutions (i.e., 1 in 2 or 1 in 4), only
material, and so endpoint dilutions could not be recorded. Another did so for one of two donors, with endpoint dilutions for a second
laboratory which did detect antibodies with HPA-15b specificity donor of either 1 in 8 or 1 in 16. This difference could be attributed
failed to titrate the material enough; consequently, an endpoint dilu- to the levels of CD109 expression on the platelets [13, 14]. The
tion could not be recorded for this laboratory either. Figure 1 shows remaining two of the five laboratories reported 1 in 2 or 1 in 4 for
the endpoint dilutions reported by 23 laboratories for each method both donors used.
and for each donor of HPA-15bb platelets/cells. The reported end- The majority of laboratories used CD109-specific, mouse mono-
point dilutions were normally distributed with a modal dilution of clonal antibody TEA 2/16 from BD Pharmingen. There was no clear
1 in 16 and a range from 1 in 2 to 1 in 128 as shown in Figure 1. The trend in reported endpoint dilutions for 18/220 relating to the differ-
intra-laboratory variation in reported endpoint dilutions was ent CD109-specific monoclonal antibodies used as shown in Figure 1.
280 SHARP ET AL.

TABLE 4 Accelerated degradation studies dilution of 1 in 32 (result not shown in Figure 2); this appears to
Storage Relative 95% lower be anomalous in comparison with the results from all other labora-
temperature potency ( 70 C confidence 95% upper tories. The anomaly could have arisen from the incorrect typing of
( C) reference, n = 4) limit confidence limit the alleged HPA-15aa platelets or may be attributed to the fact
20 1.00 0.92 1.09 that these platelets were 9 days old. Indeed, the same laboratory
+4 1.28 1.18 1.40 reported ‘neat’ as the endpoint for a second donor of HPA-15aa
+20 1.13 1.04 1.23 platelets which were only 6 days old. While some laboratories
+37 0.98 0.90 1.07 have reported a positive result with HPA-15aa platelets at high

+45 0.72 0.66 0.79 concentrations of 18/220, there still remains a clear distinction
between reported endpoint dilutions with HPA-15bb and HPA-
Note: Ampoules stored for 13 months at each temperature.
15aa types.

As expected, most laboratories (22/23) used fresh HPA-15bb plate-

lets. Laboratory 4 used only frozen platelets characterized as having Stability
high CD109 expression levels prior to freezing and achieved an end-
point dilution in a MAIPA of 1 in 8 for 18/220 using the anti-CD109 Estimates of the potency of 18/220 stored at elevated tempera-
clone TEA 2/16 and an endpoint dilution of 1 in 16 using the anti- tures for a period of 13 months relative to 18/220 stored at
CD109 clone HU17. Laboratory 17 compared lyophilized and fresh 70 C for 13 months are summarized in Table 4. Tests for non-
platelets in the MAIPA, with comparable endpoint dilutions being parallelism and non-linearity in the parallel-line analysis to esti-
reported for both types (1 in 8 and 1 in 16 for two donors of mate relative potencies were not statistically significant (p = 0.14
lyophilized platelets and 1 in 16 and 1 in 32 for two donors of fresh and 0.59, respectively). There was insufficient degradation at the
platelets). Laboratory 7a used recombinant K562 cells with a reported elevated temperatures, even after 13 months of storage, to fit the
endpoint dilution of 1 in 16. Arrhenius model, with only a clear relative potency loss at +45 C.
Participants were asked to provide technical information This indicates that 18/220 will be stable for long-term storage at
on their protocols; this is summarized in Appendix S1. Indeed, 20 C and sufficiently stable to allow for shipment of ampoules at
there was considerable variation among the MAIPA methods ambient temperature.
performed, the most notable differences (further to those
described earlier) being the number of platelets and sample
volume used. Platelet numbers ranged from 0.4  10 6 to DI SCU SSION
300  10 6 per well and sample volumes ranged from 20 to 120 μl
per well. The aim of the study was to prepare and evaluate a stable, reference
reagent for anti-HPA-15b detection that clinical laboratories can use to
assess and validate the sensitivity of their routine assays. The collabora-
Reported endpoint dilutions against HPA-15aa antigen tive study has shown that candidate preparation 18/220 contains anti-
HPA-15b antibody that could be detected at a dilution of 1 in 8 by
To confirm the specificity of the HPA-15 antibodies within the 21 of 23 laboratories with at least one donor. Setting the minimum
candidate material, laboratories were also asked to test 18/220 potency at 1 in 8 dilution removes ambiguity resulting from positivity
with HPA-15aa platelets from two different donors where possi- with HPA-15aa platelets or alike at lower dilutions. The results of the
ble at doubling dilutions and to report the endpoint dilution; these study in general show good consistency across most laboratories but
results are shown in Figure 2. There were only two laboratories also indicate that some laboratories should consider further improving
that reported results from just one HPA-15aa donor. In summary, the sensitivity of their assay method. Since the material contains an
as expected, nearly all laboratories reported that for all HPA-15aa anti-HLA component (as may any patient clinical sample), it should only
platelet donors (or alike) used with each method performed; they be used in techniques that are glycoprotein-specific (i.e., for CD109)
were unable to detect HPA-15a antibodies or reported a positive such as the ‘gold standard’ MAIPA which all laboratories used in this
result only for the undiluted (neat) material. However, three labs study or where it can be ensured that the anti-HLA antibodies will not
did report endpoint dilutions of 1 in 2, 1 in 4 or 1 in 8 with one or cause a false-positive reaction (i.e., through chloroquine treatment of
more donors. This could be explained by the presence of anti-HLA platelets to remove HLA-class 1 epitopes).
antibodies at high serum concentrations and incomplete solubili- A report of the 19th ISBT Platelet Immunology Workshop gave a
zation of the platelet membrane during the MAIPA procedure. comprehensive summary of the variations in the MAIPA procedure
Indeed, of both laboratories which reported endpoint dilutions of used by workshop participants and concluded that the MAIPA is far
1 in 4 or 1 in 8, each did so for different HPA-15aa platelet donors from harmonized which may contribute to variations in results [15].
suggesting that it is not a rare donor-specific phenomenon but While our study also showed a lack of inter-laboratory harmonization of
rather a procedural effect. One laboratory reported an endpoint the CD109 MAIPA procedure, we were also able to demonstrate
ANTI-HPA-15B REFERENCE REAGENT 281

notable intra-assay variability in reported endpoint dilutions that are 3. Kroll H, Kiefel V, Santoso S. Clinical aspects and typing of
likely due to the differential levels in CD109 expression by platelets platelet alloantigens. Vox Sang. 1998;74:345–54.
4. Davoren A, Curtis BR, Aster RH, McFarland JG. Human platelet
from different donors, as previously reported [14]. Therefore, the har-
antigen-specific alloantibodies implicated in 1162 cases of neo-
monization of test methods using optimized assay conditions may well natal alloimmune thrombocytopenia. Transfusion. 2004;44:
improve the sensitivity of testing across clinical laboratories to some 1220–5.
degree but cannot overcome test variability as a result of donor–donor 5. Aster R. Post-transfusion purpura. Baillieres Clin Immunol Allergy.
1987;1:453–61.
differences. The use of the anti-HPA-15b reference reagent will at least
6. Metcalfe P, Allen D, Chapman J, Ouwehand WH. Interlaboratory
provide some guarantee as to the sensitivity of the platelets used. variation in the detection of clinically significant alloantibodies
All participants of the international collaborative study were invited against human platelet alloantigens. Br J Haematol. 1997;97:
to comment on the final study report, and their approval for use of 204–7.
7. Metcalfe P, Ouwehand WH, Sands D, Barrowcliffe TW. Collabora-
18/220 as a reference reagent for human IgG antibodies against HPA-
tive studies to establish the first WHO reference reagent for detec-
15b with a minimum potency of 1 in 8 was sought. No objections were tion of human antibody against HPA-5b. Vox Sang. 2003;84:
received from the participants, and, in addition, this proposal was 237–40.
endorsed by the ISBT Working Party on Platelet Immunology. In October 8. Berry J, Allen D, Porcelijn L, de Haas M, Kekomaki R, Kaplan C, et al.
Collaborative studies to establish the first WHO international stan-
2020, all data were reviewed by the WHO Expert Committee on Biologi-
dard for detection of human antibody against HPA-3a. Vox Sang.
cal Standardization, and 18/220 was approved for use as a first WHO
2007;93:309–15.
anti-HPA-15b reference reagent. This material should be used at a dilu- 9. Metcalfe P, Allen D, Kekomaki R, Kaplan C, de Haas M,
tion of 1 in 8 for assay validation (i.e., the minimum potency which should Ouwehand WH. An international reference reagent (minimum sensi-
test positive) and may be used to qualify ‘in-house’ controls. It is hoped tivity) for the detection of anti-human platelet antigen 1a. Vox Sang.
2009;96:146–52.
that wide use of this reference reagent will improve the sensitivity of test
10. Berry JE, Murphy CM, Smith GA, Ranasinghe E, Finberg R, Walton J,
methods giving more confidence to the results generated. et al. Detection of Gov system antibodies by MAIPA reveals an
immunogenicity similar to HPA-5 alloantigens. Br J Haematol. 2000;
ACKNOWLEDGEMEN TS 110:735–42.
11. Kroll H, Ertel K, Al-Tawil SS. Relevance of the HPA-15 (Gov) poly-
We thank the Platelet & Leucocyte Department, Sanguin Diagnostic Ser-
morphism on CD109 in alloimmune thrombocytopenic syndromes.
vices, the Netherlands for their laboratory support and advice throughout Transfusion. 2005;45:366–73.
the study. We also thank the ISBT Platelet Immunobiology Working Party 12. Bordin JO, Kelton JG, Warner MN, Smith JW, Denomme GA,
and UK NEQAS Histocompatibility & Immunogenetics, Wales, for distrib- Warkentin TE, et al. Maternal immunization to Gov system alloanti-
gens on human platelets. Transfusion. 1997;37:823–8.
uting the collaborative study invitations. Finally, we thank the Standards
13. Schuh AC, Watkins A, Nguyen Q, Harmer NJ, Lin M, Prosper JYA,
Processing Division, NIBSC, for lyophilizing the bulk and sample dispatch et al. A tyrosine703serine polymorphism of CD109 defines the Gov
and the participants of the collaborative study for providing the data. platelet alloantigens. Blood. 2002;99:1692–8.
G.S. and L.S. were responsible for the generation of the reference 14. Maslanka K, Michur H, Guz K, Wro  bel A, Uhrynowska M, Misiak A,
et al. The relevance of HPA-15 antigen expression for anti-HPA-15
material, internal evaluation of the material, running the collaborative
antibody detection. Int J Lab Hematol. 2012;34:65–9.
study and report and manuscript preparation; A.P. contributed signifi- 15. Lewin A, Khan S, Beaudin L, Meilleur L, Clarke G, Richard L. The 19th
cantly to pre-collaborative study experimental work. International Platelet Immunology Workshop of ISBT. 2018. http://
www.isbtweb.org/fileadmin/user_upload/Workshop_report.pdf
16. Kirkwood TBL. Predicting the stability of biological standards and
CONF LICT OF IN TE RE ST
products. Biometrics. 1977;33:736–42.
The authors declare no conflicts of interest.

ORCID SUPPORTING INFORMATION

Lucy Studholme https://orcid.org/0000-0002-2935-2048 Additional supporting information may be found online in the
Supporting Information section at the end of this article.

RE FE R ENC E S
1. Metcalfe P, Watkins NA, Ouwehand WH, Kaplan C, Newman P, How to cite this article: Sharp G, Poles A, Studholme L. A first
Kekomaki R, et al. Nomenclature of human platelet antigens (HPA).
WHO reference reagent for the detection of anti-human
Vox Sang. 2003;85:240–5.
2. Human Platelet Antibody (HPA) Database. https://www.versiti. platelet antigen-15b. Vox Sang. 2022;117:275–281.
org/hpa
Received: 22 March 2021 Revised: 3 May 2021 Accepted: 2 June 2021

DOI: 10.1111/vox.13168

ORIGINAL ARTICLE

Estimation of Lewis-negative alleles by high-resolution melting

analysis of three tag SNPs of FUT3

Mikiko Soejima | Yoshiro Koda

Department of Forensic Medicine, Kurume

University School of Medicine, Kurume, Japan Abstract
Background and Objectives: The expression of type 1 chain Lewis blood group anti-
Correspondence
Yoshiro Koda, Department of Forensic gens is regulated by secretor-type α(1,2)fucosyltransferase, encoded by FUT2, and
Medicine, Kurume University School of Lewis α(1,3/1,4)fucosyltransferase, encoded by FUT3. Accumulating evidence has
Medicine, Kurume 830-0011, Japan.
Email: [email protected] linked Lewis phenotypes or genotypes to various clinical conditions. Thus, in addition
to FUT2, large-scale FUT3 genotyping is important. Because FUT3 has two
Funding information
paralogous genes (FUT5 and FUT6) with high DNA sequence similarity, we should
None.
select the polymerase chain reaction (PCR) primers carefully for FUT3 genotyping.
Previously, we suggested that 13G>A (rs28362458), 59T>G (rs28362459) and
202T>C (rs812936) could be selected as tag single nucleotide polymorphisms (SNPs)
for detection of Lewis-negative alleles (le).
Materials and Methods: In this study, three high-resolution melting (HRM) analyses
for genotyping these SNPs were developed and applied for 140 Japanese, eight
Ghanaians and four Sinhalese subjects.
Results: Each of three genotypes of 13G>A (G/G, G/A, A/A), 59T>G (T/T, T/G, G/G)
and 202T>C (T/T, T/C, C/C) was discriminated clearly. Although we need to be care-
ful in interpretation of results due to SNPs other than the 59T>G in the amplicon,
the results of 59T>G genotyping were in full agreement with the results by a previ-
ous PCR-restriction fragment length polymorphism analysis in 140 Japanese. In addi-
tion, three heterozygotes of 202C substitution were identified, and no one having a
13A substitution was found in 140 Japanese.
Conclusion: The present HRM analyses are useful and reliable methods for large-
scale estimation of le alleles.

KEYWORDS
FUT3, high-resolution melting, Lewis-negative allele, rs28362458, rs28362459, rs812936

I N T R O D U CT I O N FUT3 [1]. Functional FUT2 and FUT3 alleles (Se and Le) are dominant
over non-functional alleles (se and le). Accordingly, individuals who
Type 1 chain Lewis blood group antigens are composed of Lewis a lack the Lewis enzyme (le/le) have Le (a-b-) red cells irrespective of
(Lea) and Lewis b (Leb) antigens and belong to the ABH blood group– secretor status, while among individuals who have the active
related antigens. The expression of these antigens is regulated by Lewis enzyme (Le/Le or Le/le), red cells of secretors (Se/Se or Se/se)
secretor-type α(1,2)fucosyltransferase (Se enzyme), encoded by FUT2, are Le(a-b+), those of non-secretors (se/se) are Le(a+b-) and those of
and Lewis α(1,3/1,4)fucosyltransferase (Lewis enzyme), encoded by weak secretors are Le(a+b+) [2, 3]. However, since the determination

282 © 2021 International Society of Blood Transfusion. wileyonlinelibrary.com/journal/vox Vox Sanguinis. 2022;117:282–287.
HRM GENOTYPING FOR LE ALLELES ESTIMATION 283

of secretor status by traditional serological Lewis phenotyping is three HRM analyses for detection, each of 13G>A, 59T>G and
difficult [3], a reliable method of genotyping of FUT2 and FUT3 is 202T>C, were developed.
important for estimation of the Lewis phenotype as an alternative to
phenotyping. Previous studies suggested that the expression of Lewis
blood group antigens is associated with Helicobacter pylori infection, M A T E R I A L S A N D M ET H O D S
ischemic heart disease, ulcerative colitis, ankylosing spondylitis and
obesity [4–8]. DNA samples
According to previous reports, a 1000-genome browser (https://
www.ncbi.nlm.nih.gov/variation/tools/1000genomes/) and Erythro- The study protocol was approved by the Ethical Committee of
gene v0.8 (27 November 2017) (http://www.erythrogene.com/) Kurume University (Bioethics approval No. 342).
[9, 10], four single nucleotide polymorphisms (SNPs), 202T>C We used genomic DNAs from 140 Japanese whose genotypes of
(rs812936), 484G>A (Trp68Arg, rs28362463), 508G>A (Gly170Ser, three SNPs (59T>G, 508G>A and 1067T>A of FUT3) had been already
rs3745635) and 1067T>A (Ile356Lys, rs3894326) are SNPs that fre- determined by PCR-restriction fragment length polymorphism
quently inactivate Lewis enzyme. They often accompany other SNPs (PCR-RFLP) [17]. We also used eight Ghanaians (in Ghana) and fours
respectively, and many le alleles have been reported [11–14]. 484G>A Sinhalese (in Sri Lanka) whose FUT3 haplotypes had been already
is in almost absolute linkage disequilibrium (a complete dependence determined by DNA sequencing of PCR amplicons [14, 18]. The
of allele distribution, e.g., there are G–G and A–A combinations but 202T>C polymorphism of selected Japanese samples was determined
not G–A and A–G combinations at the positions of 13 and 484) with by Sanger sequencing methods as described previously [14].
13G>A (Gly5Ser, rs28362458) and in complete linkage disequilibrium
(there are three combinations of two SNPs, e.g., there are G–G, G–A
and A–A combinations but not A–G combination at the positions of PCR and HRM conditions
13 or 484 and 667) with 667G>A (Gly223Arg, rs144569478), while
508G>A and 1067T>A are in almost complete linkage disequilibrium Primers for amplification of only FUT3, but not paralogous genes
with 59T>G (Leu20Arg, rs28362459), and the 59G allele without FUT5 and FUT6, were designed by Primer3 software (https://bioinfo.
508A or 1067A is quite rare (Table 1). In addition, the 508A or 1067A ut.ee/primer3-0.4.0/, [19]). Primer sequences for detection of 13G>A,
allele without 59G is also rare. Accordingly, we suggested that three 59T>G and 202T>C and amplicon length are indicated in Table 2.
tag SNPs, 13G>A, 59T>G and 202T>C, are useful for estimation of le Real-time PCR and HRM analyses were performed in a single run on a
alleles in many populations in a previous study [14]. On the other LightCycler 480 instrument II and gene scanning software (Roche
hand, FUT3 has two paralogous genes, FUT5 and FUT6, with high Diagnostics, Tokyo, Japan) as described previously with a slight
sequence similarity [15]. Thus, we need to design polymerase chain modification [20]. The reaction was performed in a mixture containing
reaction (PCR) primers to amplify FUT3 specifically. 2–20 ng of genomic DNA, 5 μl of Premix Ex Taq (Probe qPCR)
High-resolution melting (HRM) analysis is a simple and relatively (Takara, Tokyo, Japan), 125 nM of each primer and 0.1 μl of
high-throughput method for the detection of variations such as SNPs, LightCycler 480 ResoLight Dye (Roche Diagnostics, Tokyo, Japan)
small insertion/deletion in PCR amplicons. This method does not with PCR-grade water adjusted to a final volume of 10 μl. The temper-
require post-PCR processing and is based on detecting subtle differ- ature conditions of all three analyses are identical and as follows:
ences in the melting curve and melting temperature of short PCR initial denaturation and enzyme activation at 95
C for 30 s, followed
amplicons, using saturated fluorescent dyes [16]. In the present study, by 45 cycles of denaturation at 95
C for 5 s and annealing/extension

TABLE 1 Major non-functional alleles of FUT3 and their tag single nucleotide polymorphism (SNP) and distributions

Frequency (%)

Allele Tag SNP Africa America E. Asia Europe S. Asia

le202,314 202 7.7 12.0 3.1 16.2 11.2
47,202,314
le 202 1.1 1.4 0 1.3 1.1
le13,484,667 13 8.7 0.9 0 0 0
13,484,667,807
le 13 3.8 0.4 0 0 0
le59,1067 59 3.9 4.2 13.5 6.2 16.0
le59,508 59 18.9 19.9 13.6 1.5 5.4
le 59,508,858 59 3.7 0.1 0.2 0 0.1
59,61,508
le 59 3.1 1.3 0 0 0

Note: Allele frequencies were obtained from Erythrogene. Allele frequency of 1% or more in global population was indicated.
Abbreviations: E. Asia, East Asia; S. Asia, South Asia.
284 SOEJIMA AND KODA

TABLE 2 Primer for detection of single nucleotide polymorphisms (SNPs)

Differences Differences Amplicon

Primer sequences Position of FUT3 with FUT5 with FUT6 length
Detection of 13G>A
Forward primer: 50 -CCTCTCTCCTCTCTTCCCAGA-30 32 to 12 bp 1 2 63 bp
0 0
Reverse primer: 5 -ATTGTGGCTTGGCTGCAC-3 14 to 31 bp 3 4
Detection of 59T>G
Forward primer: 50 -GTGCAGCCAAGCCACAAT-30 14 to 31 bp 3 4 68 bp
Reverse primer: 50 -AGCCACCAGCAGCTGAAATA 30 62 to 81 bp 1 2
Detection of 202T>C
Forward primer: 50 -CCACCCTCCTGATCCTGCTA-30 182 to 201 bp 2 3 50 bp
0 0
Reverse primer: 5 -AGCCACAGGGATGTGGAA-3 214 to 231 bp 5 7

Note: ‘Differences with FUT5 (or FUT6)’ indicate the number of base differences in each of the primer with corresponding FUT5 and FUT6 sequences.

T A B L E 3 Relationships of nucleotide positions, used DNA

samples and belonging groups by high-resolution melting (HRM)
analysis for 59T>G

47 59 61 Group
Ghanaian 1 G/G T/T C/C I
Ghanaian 2 G/G T/G C/C II
Ghanaian 3 G/G G/G C/C III
Ghanaian 4 G/G T/G C/T IV
Ghanaian 5 G/G G/G C/T II
Ghanaian 6 G/G G/G T/T I
Sinhalese 1 C/G T/T C/C V
DNA is not available C/G T/G C/C
DNA is not available C/C T/T C/C
DNA is not available C/G T/G C/T

at 60
C for 20 s. Before HRM analysis, the samples were heated to

95
C for 1 min and rapidly cooled to 60
C for 1 min, and fluorescence
data for HRM analysis were collected over the range from 72 to 92
C,
increasing at 0.02
C/s with 25 acquisitions/s. The raw melting curve
data were normalized by manual adjustment of linear regions pre- and
post-melt signals of all samples. Temperature shift correction for all
analyses to clearly discriminate wild-type and mutated allele homozy-
gotes was not performed.

RESULTS F I G U R E 1 Normalized melting curves (a) and difference plot

curves (b) for genotyping of 59T>G obtained from six selected
Ghanaians and a Sinhalese (Ghanaian 1: red, Ghanaian 2: blue,
59T>G genotyping
Ghanaian 3: green, Ghanaian 4: pink, Ghanaian 5: brown and
Ghanaian 6: yellow, Sinhalese 1: grey, see Table 2). High-
According to the 1000 Genome Project and Erythrogene databases resolution melting (HRM) analysis separated seven individuals into
[10], two other SNPs, 47G>C (Cys16Ser, rs145362171) and 61C>T five groups. Individuals having 47G/G, 59T/T, 61C/C (Ghanaian 1)
(synonymous SNP, rs28362460), in addition to 59T>G, were present and 47G/G, 59G/G, 61T/T (Ghanaian 6) and individuals having
47G/G, 59T/G, 61C/C (Ghanaian 2) and 47G/G, 59G/G, 61C/T
within the short amplicon sequence (68 bp, Table 2) with a frequency
(Ghanaian 5) were not separated
of more than 1%. The SNP 47G>C seems to be present in many
HRM GENOTYPING FOR LE ALLELES ESTIMATION 285

F I G U R E 2 Normalized melting curves (a) and difference plot

curves (b) for genotyping of 59T>G obtained from 60 to
140 randomly selected Japanese. The individuals having genotypes of
T/T (red), T/G (blue) and G/G (green) at 59T>G were completely
separated

populations but not East Asians, while 61C>T seems to be specific to

Africans and Americans (probably African origin). The 47C allele is in
complete linkage disequilibrium with the 59T allele, while the 61C
allele is in complete linkage disequilibrium with the 59G allele. There-
fore, 10 genotypes by combinations of three sites (haplotypes) are
F I G U R E 3 Normalized melting curves (a) and difference plot
expected. However, we did not have DNA samples with 3 of 10 haplo- curves (b) for genotyping of 202T>C obtained from three selected
types (see Table 3); HRM analysis was able to be performed only in Sinhalese. Normalized melting curves (c) for genotyping of 202T>C
seven samples. As shown in Figure 1(a, b), HRM analysis separated obtained from 60 to 140 randomly selected Japanese. The individuals
seven individuals into five groups. Group I included Ghanaian 1 and having genotypes of T/T(blue), T/C (red) and C/C (green) at 202T>C
were completely separated
Ghanaian 6, group II included Ghanaian 2 and Ghanaian 5, group III
included Ghanaian 3, group IV included Ghanaian 4 and group V
included Sinhalese 1.
We then analysed 140 Japanese whose genotypes of 59T>G, 61C>T such as East Asians. We confirmed the repeatability of the
508G>A and 1067T>A of FUT3 were previously determined by PCR- method by two independent HRM analyses.
RFLP [17]. Figure 2(a, b) show results of normalized melting curves
and difference plots of 60 of 140 Japanese subjects. Three genotypes
of 59T>G could be separated clearly (T/T = 55, T/G = 63, G/G = 22). 202T>C genotyping
The present results were completely in agreement with previous PCR-
RFLP results [17]. Thus, the HRM analysis for 59T>G seems to be We next performed the HRM analysis for 202T>C using a 50-bp
valid and reliable, especially in a population without 47G>C and amplicon (Table 2) on three Sinhalese having genotypes of T/T
286 SOEJIMA AND KODA

amplicon. As shown in Figure 4(a, b), we clearly distinguished three

genotypes from each other by two independent assays. We then
analysed 140 Japanese and did not find any individuals containing the
13G>A polymorphism, as expected from the databases (data not
shown).

DI SCU SSION

Accumulating evidence suggested that FUT3 polymorphisms (Lewis

blood group status) seemed to be associated with various disease
[4–8]. Large-scale replication studies are desirable to confirm these
associations, and thus, a reliable method of genotyping FUT3 is
required. The findings of this study achieved with the application of
the HRM method to validate FUT3 SNPs are in agreement with those
of other previous studies such as PCR-RFLP, PCR using sequence-
specific primers (PCR-SSP), an allele-specific oligonucleotide hybridi-
zation method, hydrolysis probe-based real-time PCR (TaqMan) assay,
nucleotide sequencing of PCR products and a multiplex SNaPshot
assay [6, 14, 21–23]. Compared with other methods, the HRM
analysis does not require post-PCR processing as a TaqMan assay
does. In addition, this method is simple, cost-effective, accurate, high-
throughput and faster than the conventional PCR.
As mentioned earlier, PCR primers for the detection of SNPs of
FUT3 should be selected to avoid amplification of two paralogous
genes, FUT5 and FUT6. Because all three HRM analyses could distin-
F I G U R E 4 Normalized melting curves (a) and difference plot
curves (b) for genotyping of 13G>A obtained from three selected guish three genotypes of FUT3, primers using the present three HRM
Ghanaians. The individuals having genotypes of G/G (blue), G/A (red) analyses seemed to amplify FUT3 sequence specifically.
and A/A (green) at 13G>A were completely separated The presence of SNPs other than the target SNP in the amplified
region is likely to affect the melting curve profiles of amplicons
obtained by HRM analysis. Thus, we investigated whether the three
amplicons contain SNPs with a high frequency or not by searching the
(wild type), T/C and C/C at 202T>C. No SNP with a frequency of 1000 Genome Browser and found two SNPs (47G>C and 61C>T)
more than 1% seems to be present within this amplicon according to with the frequency of more than 1% in the amplicon only for 59T>G
the database. As shown in Figure 3(a, b), we clearly distinguished detection. The SNP 47G>C was found in many populations (1.2%–
the three genotypes from each other. 1.5% and 1.02% in global populations) except East Asians, and 61C>T
Although we have never examined this polymorphism in 140 Japa- was found in African and Americans (4.24% in Africans, 1.30% in
nese subjects, the 1000 Genome Project and Erythrogene databases Americans and 1.30% in global populations). 47C was in complete
indicated that this SNP is present in East Asian populations including linkage disequilibrium with 202C and in almost complete linkage dis-
Japanese with a frequency about 1%–6% [9, 10]. Therefore, we then equilibrium with 314T, while a synonymous SNP, 61T, was in com-
analysed 140 Japanese and found three heterozygotes of 202T>C plete linkage disequilibrium with both 59G and 508A. However, there
(Figure 3(c)) by two independent HRM analyses. In addition, we dem- is no East Asian having these two SNPs in the 1000 Genomes Project
onstrated this SNP in these three individuals by Sanger sequencing of database.
the PCR products of coding region of FUT3. Accordingly, the fre- Although these SNPs, particularly 61C>T, have an impact on the
quency of 202C is estimated to be 1.07% in our Japanese samples. melting curve profiles in the HRM analysis for detection of 59T>G,
the influence of these SNPs seems to be limited even in large-scale
association studies in an East Asian population considering their allele
13G>A genotyping frequencies. In addition, a reliable estimate of le could be made by
combining HRM analyses for 202T>C and 59T>G in European sub-
Finally, we performed the HRM analysis for 13G>A using a 63-bp jects because 47C was in complete linkage disequilibrium with 202C.
amplicon (Table 2) on three Ghanaians having genotypes of G/G (wild Thus, for large-scale association studies of FUT3, examination of
type), G/A and A/A at 13G>A. As well as the 202T>C amplicon, no 59T>G and 202T>C might be sufficient for East Asian and European
SNP with a frequency more than 1% seems to be present within this populations. On the other hand, in addition to 59T>G and 202T>C,
HRM GENOTYPING FOR LE ALLELES ESTIMATION 287

we should consider 13G>A, which contains the second highest le 9. Sudmant PH, Rausch T, Gardner EJ, Handsaker RE, Abyzov A,
allele in African populations, with a frequency of 18.4% (and in Huddleston J, et al. An integrated map of structural variation in
2,504 human genomes. Nature. 2015;526:75–81.
American populations with a frequency of 1.7%).
10. Möller M, Jöud M, Storry JR, Olsson ML. Erythrogene: a data-
In conclusion, the present HRM analyses for 59T>G and 202T>C base for in-depth analysis of the extensive variation in 36 blood
are valid and feasible for association studies of FUT3, particularly in group systems in the 1000 genomes project. Blood Adv. 2016;
East Asian populations. Addition of HRM analysis for 13G>A to those 1:240–9.
11. Koda Y, Kimura H, Mekada E. Analysis of Lewis fucosyltransferase
for 59T>G and 202T>C is desirable for a more correct estimation of
genes from the human gastric mucosa of Lewis-positive and
the le alleles in African and American populations, although the pres- -negative individuals. Blood. 1993;82:2915–9.
ence of the 61T>C polymorphism slightly complicates the results of 12. Pang H, Liu Y, Koda Y, Soejima M, Jia J, Schlaphoff T, et al. Five novel
HRM analysis for 59T>G genotyping. missense mutations of the Lewis gene (FUT3) in African (Xhosa) and
Caucasian populations in South Africa. Hum Genet. 1998;102:675–80.
13. Pang H, Koda Y, Soejima M, Kimura H. Significance of each of three
ACKNOWLEDGEMEN T missense mutations, G484A, G667A, and G808A, present in an inac-
We thank Katherine Ono for editing the English in this manuscript. tive allele of the human Lewis gene (FUT3) for alpha(1,3/1,4)
M.S. investigated and wrote the original draft; Y.K. supervised the fucosyltransferase inactivation. Glycoconj J. 1998;15:961–7.
14. Soejima M, Munkhtulga L, Iwamoto S, Koda Y. Genetic variation of
methodology, validated the results and reviewed and edited
FUT3 in Ghanaians, Caucasians, and Mongolians. Transfusion. 2009;
the manuscript.
49:959–66.
15. Oulmouden A, Wierinckx A, Petit JM, Costache M, Palcic MM,
CONF LICT OF IN TE RE ST Mollicone R, et al. Molecular cloning and expression of a bovine
The authors declare they have no conflict of interest. alpha(1,3)-fucosyltransferase gene homologous to a putative ances-
tor gene of the human FUT3-FUT5-FUT6 cluster. J Biol Chem. 1997;
272:8764–73.
ORCID 16. Wittwer CT, Reed GH, Gundry CN, Vandersteen JG, Pryor RJ. High-
Mikiko Soejima https://orcid.org/0000-0001-9721-5881 resolution genotyping by amplicon melting analysis using LCGreen.
Yoshiro Koda https://orcid.org/0000-0003-1368-0116 Clin Chem. 2003;49:853–60.
17. Liu Y, Koda Y, Soejima M, Uchida N, Kimura H. PCR analysis of
Lewis-negative gene mutations and the distribution of Lewis alleles
RE FE R ENC E S in a Japanese population. J Forensic Sci. 1996;41:1018–21.
1. Daniels G. ABO, H, and Lewis systems. In: Daniels G, editor. Human 18. Soejima M, Kimura H, Koda Y. Two novel FUT3 alleles responsible for
blood groups. West Sussex: Wiley-Blackwell; 2013. p. 11–95. Lewis-null phenotypes in Sri Lanka. Transfusion. 2004;44:1534–5.
2. Yu LC, Yang YH, Broadberry RE, Chen YH, Chan YS, Lin M. Correlation of a 19. Untergasser A, Cutcutache I, Koressaar T, Ye J, Faircloth BC,
missense mutation in the human secretor alpha 1,2-fucosyltransferase gene Remm M, et al. Primer3-new capabilities and interfaces. Nucleic
with the Lewis(a+b+) phenotype: a potential molecular basis for the weak Acids Res. 2012;40:e115.
secretor allele (sew). Biochem J. 1995;312:329–32. 20. Soejima M, Koda Y. Estimation of secretor status of ABO antigens by
3. Koda Y, Soejima M, Liu Y, Kimura H. Molecular basis for secretor high-resolution melting analysis of rs601338 (428G > a). Clin Chim
type alpha(1,2)-fucosyltransferase gene deficiency in a Japanese Acta. 2021;517:86–91.
population: a fusion gene generated by unequal crossover responsi- 21. Grahn A, Elmgren A, Åberg L, Svensson L, Jansson PA, Lönnroth P,
ble for the enzyme deficiency. Am J Hum Genet. 1996;59:343–50. et al. Determination of Lewis FUT3 gene mutations by PCR using
4. Borén T, Falk P, Roth KA, Larson G, Normark S. Attachment of sequence-specific primers enables efficient genotyping of clinical
helicobacter pylori to human gastric epithelium mediated by blood samples. Hum Mutat. 2001;18:358–9.
group antigens. Science. 1993;262:1892–5. 22. Cakir B, Pankow JS, Salomaa V, Couper D, Morris T, Brantley K, et al.
5. Cakir B, Heiss G, Pankow JS, Salomaa V, Sharrett AR, Couper D, Distribution of Lewis (FUT3)genotype and allele: frequencies in a
et al. Association of the Lewis genotype with cardiovascular risk fac- biethnic United States population. Ann Hematol. 2002;81:558–65.
tors and subclinical carotid atherosclerosis: the atherosclerosis risk in 23. Djoussé L, Karamohamed S, Herbert AG, D’Agostino RB, Cupples LA,
communities (ARIC) study. J Intern Med. 2004;255:40–51. Ellison RC. Fucosyltransferase 3 polymorphism and ath-
6. Hu D, Zhang D, Zheng S, Guo M, Lin X, Jiang Y. Association of Ulcer- erothrombotic disease in the Framingham offspring study. Am Heart
ative Colitis with FUT2 and FUT3 polymorphisms in patients from J. 2007;153:636–9.
Southeast China. PLoS One. 2016;11:e0146557.
7. Jiang G, Han R, Chen M, Liu R, Wu M, Zhang X, et al. Associations
between fucosyltransferase 3 gene polymorphisms and ankylosing
spondylitis: a case-control study of an east Chinese population. PLoS
How to cite this article: Soejima M, Koda Y. Estimation of
One. 2020;15:e0237219.
8. Enevold C, Nielsen CH, Molbo D, Lund R, Bendtzen K, Fiehn NE,
Lewis-negative alleles by high-resolution melting analysis of
et al. Lewis and AB0 blood group-phenotypes in periodontitis, car- three tag SNPs of FUT3. Vox Sang. 2022;117:282–287.
diovascular disease, obesity and stroke. Sci Rep. 2019;9:6283.
Received: 28 April 2021 Revised: 25 May 2021 Accepted: 25 May 2021

DOI: 10.1111/vox.13165

LETTER TO THE EDITOR

Using blood services platforms to facilitate COVID-19

vaccination programs

Implementing an effective coronavirus disease-19 (COVID-19) Since March 2020, Magen David Adom (MDA) is Israel’s National
vaccination program presented several challenges, including Blood Bank and Blood Services organization in addition to being the
approaches to convince people to get vaccinated, setup of convenient national Red Cross organization and Emergency Medical Services orga-
vaccination locations, building and operating the logistics involved in nization. MDA participated in the national effort to mitigate the COVID-
the cold-chain needed standards and the optimal use of thawed vac- 19 pandemic in the 9.33 million population, through more than 4.5 mil-
cines to prevent wastage. lion swabs collection for polymerase chain reaction testing and by pro-
Most vaccination plans prioritized first responders and other viding over 1 million vaccines, carefully matching expected number of
medical teams, followed by the at-risk elderly people. Currently, a sig- individuals and vaccines needed, to prevent wastage of precious
nificant number of younger working individuals, initially not eligible resources. This includes successful vaccination events in workplaces,
for vaccination, show hesitancy to get the vaccine, expressing doubts cooperating with the Manufacturers Association and Unions. A small
regarding its safety and efficacy, fear of serious side effects and reluc- survey showed that employees desired to be vaccinated at work (scale
tance to invest time and efforts, even when the vaccine is offered ‘for 6.2/7). MDA Blood Services’ experience in operating mobile blood
free.’ While the trade-off of their health is important, the public has drives was utilized to conduct an efficacious program nationwide.
already adjusted to live with the pandemic. An ‘out-of-the-box’
approach can be useful to overcome such issues. CONFLIC T OF INT ER E ST
Blood services worldwide use existing platforms for blood There are no conflicts of interests relevant to this letter for all
drives operation in easily accessible locations, including working authors.
places, educational institutions and social or religious gathering
places (churches, synagogues or mosques). The blood services’ per- FUNDING INF ORMATI ON
sonnel involved in blood drives organization is familiar with the None.
leading figures in the communities, who are known to positively
influence the public to donate blood and can be easily recruited to Eli Jaffe PhD1
support the vaccination programs. Chief executive officers hosting Itamar Abramovich MA1
a vaccination event ‘at the expense of the employer’ where Roman Sonkin MD1
employees can get vaccinated at work may be viewed as a tribute Eilat Shinar MD2
to the employees, both by them, by the workers’ union organiza-
1
tions and by the general population, as participating in an important Community Division, Magen David Adom, Or-Yehuda, Israel
2
life-saving project. Blood Services, Magen David Adom, Or-Yehuda, Israel
Additional numerous factors that affect the success of a blood
drive, such as accessibility, location and time, approval of family mem- Correspondence
bers, friends, managers and colleagues, may be used for the vaccina- Roman Sonkin, Community Division, Magen David Adom, Ha-Plada 5,
tion program. Or-Yehuda, 6021805 Israel.
Organizing ‘vaccination drives’ in these convenient, familiar loca- Email: [email protected]
tions may provide a possible solution. Moreover, blood drives sites
can be easily adapted as suitable vaccination locations, since the setup ORCID
for both operations is similar, including proper sites for assessment of Eli Jaffe https://orcid.org/0000-0002-8380-0219
individuals’ eligibility before vaccination, the actual procedure and the Roman Sonkin https://orcid.org/0000-0001-8662-3273
post-vaccination rest period needed. Eilat Shinar https://orcid.org/0000-0002-5658-248X

288 © 2021 International Society of Blood Transfusion. wileyonlinelibrary.com/journal/vox Vox Sanguinis. 2022;117:288.
Received: 27 April 2021 Revised: 18 May 2021 Accepted: 24 May 2021

DOI: 10.1111/vox.13166

LETTER TO THE EDITOR

The association of B blood group with SARS-CoV-2-induced

death in Babol, north of Iran

Relationship between blood types and incidence and fatality of prevalence of RhD were found between the different groups
coronavirus disease 2019 (COVID-19), caused by SARS-CoV-2, analysed.
has not been consistent in the studies performed around the world In conclusion, these results propose that blood type B may possi-
[1–5]. Here, we investigate the association between ABO and RhD bly be connected to SARS-CoV-2-induced fatality in homogeneous
blood types with incidence and fatality of COVID-19 in the ethni- population of Babol, north of Iran.
cally homogenous population of Babol, north of Iran. The ABO and
RhD distributions of 955 RT-PCR-confirmed COVID-19 patients, CONFLIC T OF INT ER E ST
hospitalized between 20 March and 21 September of the year The authors declare no conflict of interests.
2020 at the affiliated hospitals of Babol university of medical sci-
ences, and 639 local COVID-19 convalescent plasma (CCP) donors
were compared with 618 randomly selected pre-COVID-19 era FUNDING INF ORMATI ON
hospitalized patients and 19,668 pre-COVID-19 era local healthy This work was supported by the Babol University of Medical Sciences,
blood donors. Blood groups’ frequencies were compared using Babol, Iran (Grant No: 9909604).
IBM SPSS Statistics, Version 23.0 (Armonk, NY) via chi-square and
logistic regression tests. In our study region, blood type B individ- Saeed Hassani1,2
uals are more frequent than group A (Table 1), in contrast to the Narges Kalantari3
most other parts of the world [2–5]. Blood type B was associated Seyedeh Farzaneh Jalali1
with highest odds of testing positive for COVID-19 compared to Mahmood Shams1
the local healthy blood donors (OR = 1.142, CI = 0.99–1.32, Zahra Ahmadnia4
p = 0.06). Further, B type was significantly higher in the deceased Meysam Aghajani Daroonkola1
COVID-19 patients compared to the both non-COVID-19 patients Masomeh Bayani5
and local healthy blood donors (p = 0.048 and p = 0.028, respec- Housein Ghorbani4
tively). In agreement, B-type individuals were significantly lower in
1
the CCP donors compared to the local healthy blood donors Department of Medical Laboratory Sciences, Faculty of Paramedical
(p = 0.03). This reduction of blood type B in the CCP donors could Sciences, Babol University of Medical Sciences, Babol, Iran
2
also be attributed to more men than women in the deceased group Department of Medical Laboratory Sciences, School of Allied Medical
and male sex of all CCP donors. No significant changes in the Sciences, Arak University of Medical Sciences, Arak, Iran

TABLE 1 ABO and RhD distributions among COVID-19 patients, COVID-19 convalescent plasma (CCP) donors and controls in the north of
Iran, Babol

Number Male Female O (%) A (%) B (%) AB (%) RhD+ (%)

COVID-19 patients 955 471 (49.3) 484 (50.7) 371 (38.8) 227 (23.8) 295 (30.9)a 62 (6.5) 877 (91.8)
Deceased COVID-19 124 70 (56.5) 54 (43.5) 45 (36.3) 25 (20.2) 46 (37.1)b 8 (6.5) 109 (87.9)
patients
CCP donors 639 639 (100) 0 (0) 254 (39.7) 176 (27.5) 154 (24.1)c 55 (8.6) 594 (93.0)
Non-COVID-19 patients 618 247 (40) 371 (60) 242 (39.2) 164 (26.5) 174 (28.2) 38 (6.1) 565 (91.4)
Healthy blood donors 19,668 18,681 (95) 987 (5) 7773 (39.5) 5028 (25.6) 5531 (28.1) 1336 (6.8) 17,845 (90.7)
a
COVID-19 patients versus healthy blood donors: OR = 1.142, CI = 0.99–1.32, p = 0.06.
b
Deceased COVID-19 patients versus non-COVID-19 patients and healthy blood donors: OR = 1.51, CI = 1.004–2.255, p = 0.048 and OR = 1.507,
CI = 1.046–2.173, p = 0.028, respectively.
c
CCP donors versus healthy blood donors: OR = 1.23, CI = 1.025–1.481, p = 0.03.

Vox Sanguinis. 2022;117:289–290. wileyonlinelibrary.com/journal/vox © 2021 International Society of Blood Transfusion. 289
290 LETTER TO THE EDITOR

3
Cellular and Molecular Biology Research Center, Health Research RE FE RE NCE S
Institute, Babol University of Medical Sciences, Babol, Iran 1. Goel R, Bloch EM, Pirenne F, Al-Riyami AZ, Crowe E, Dau L, et al.
4
Pathology Department, Babol University of Medical Sciences, Babol, Iran ABO blood group and COVID-19: a review on behalf of the ISBT
5 COVID-19 working group. Vox Sang. 2021. https://doi.org/10.
Infectious Diseases and Tropical Medicine Research Center, Health
1111/vox.13076.
Research Institute, Babol University of Medical Sciences, Babol, Iran 2. Kotila TR, Alonge TO, Fowotade A, Famuyiwa OI, Akinbile AS. Asso-
ciation of the ABO blood group with SARS-CoV-2 infection in a com-
Correspondence munity with low infection rate. Vox Sang. 2021. https://doi.org/10.
1111/vox.13077.
Saeed Hassani, Department of Laboratory Sciences, Faculty of
3. Latz CA, DeCarlo C, Boitano L, Png CYM, Patell R, Conrad MF, et al.
Paramedical Sciences, Babol University of Medical Sciences, Ganj- Blood type and outcomes in patients with COVID-19. Ann Hematol.
Afroz Ave. Babol, Mazandaran, Iran. 2020;99:2113–8.
Email: [email protected]; [email protected] 4. Dzik S, Eliason K, Morris EB, Kaufman RM, North CM. COVID-19
and ABO blood groups. Transfusion. 2020;60:1883–4.
5. Zietz M, Zucker J, Tatonetti NP. Associations between blood type
ORCID
and COVID-19 infection, intubation, and death. Nat Commun. 2020;
Saeed Hassani https://orcid.org/0000-0001-5444-9987 11:5761.
Received: 30 March 2021 Revised: 24 May 2021 Accepted: 2 June 2021

DOI: 10.1111/vox.13169

LETTER TO THE EDITOR

Risk of future haemolytic disease of the fetus and newborn

following the transfusion of Rh(D)-positive blood products
to Rh(D)-negative children

Recently, the risks of an Rh(D)-negative female of child-bearing

potential (FCP) developing haemolytic disease of the newborn
(HDFN) following receipt of Rh(D)-positive red blood cells (RBCs)
or low-titre group O whole blood during her trauma resuscitation
was modelled using her age at the time of transfusion and several
other important societal factors that impact the development of
HDFN [1]. In that study, the FCP age range was 18–49 years. Since
its publication, questions have arisen about the future HDFN
potential following the transfusion of Rh(D)-positive units to
injured Rh(D)-negative children. The previously published model
was adapted for, and applied to, patients between 0 and 17 years
[1]. For this new model, the Rh(D) alloimmunization risk had to be
derived for a paediatric population as it has not been published
and might be difficult to obtain as it is uncommon to provide
Rh(D)-positive RBCs to Rh(D)-negative/type unknown children in F I G U R E 1 Anti-Rh(D)-mediated HDFN risk estimate based on
trauma. Age-specific paediatric alloimmunization rates to all RBC the patient’s age and age-specific RhD alloimmunization rates. The
HDFN risk was modelled using Rh(D)-negative alloimmunization rates
antigens were obtained from a study of paediatric patients in
of 7.8% (low estimate, lower solid line) and 42.7% (high estimate,
Japan; these age-specific rates might be different in populations
upper solid line), with an age correction factor for patients between
with different antigen frequencies [2]. These age-specific rates 0 and 17 years of age as described in the text. The hatched area
were divided by the overall alloimmunization rate to all RBC anti- between these lines indicates the range of risk for HDFN as defined
gens in adults who received a median of four RBC units (2.2%) [3], by these two Rh(D)-alloimmunization rates. Locally estimated
then separately multiplied by 7.8% (low estimate) or 42.7% (high scatterplot smoothing (LOESS) was used for HDFN rates between
ages 0 and 17 years
estimate) as these rates reflect the ends of the spectrum of
Rh(D) alloimmunization that have been reported in trauma
populations where the patients’ ages reflected the child-bearing
age range [4, 5]. These mathematical manipulations attempted to females from 0 to 49 years of age who received an Rh(D)-positive-
correct the paediatric-specific overall RBC alloimmunization risk unit during trauma resuscitation and estimated her risk of develop-
for the adult Rh(D)-alloimmunization risk, although this created the ing HDFN of any severity. The rate of future HDFN increased as
potential for slightly underestimating the paediatric Rh(D)- the child’s age increased, largely due to the increasing rates of
alloimmunization rate. The age-specific Rh(D)-alloimmunization alloimmunization as the child’s age increased, reaching a plateau
rates are shown in the inset table in the Figure 1. For the current between approximately 18–20 years (Figure 1); at this age, the
model, 7.8% and 42.7% were also used for the adult (18–49 years) alloimmunization rate is at its maximum, and the length of time for
Rh(D)-alloimmunization rates instead of 21.2% from the previous future pregnancies is approximately 30 years.
model [1]. The 28-day mortality rate used for injured patients, These data indicate that if Rh(D)-negative children, especially very
0–17 years, was 36% as recently described [6]. Although it is young children, are resuscitated with Rh(D)-positive units, the overall
known that injured adult females have lower mortality than males, risk of developing HDFN is relatively small compared to the mortality
it is not currently known if this finding also applies to injured girls. reduction of receiving pre- and early in-hospital transfusions. How-
All other assumptions were unchanged from the previous model. ever, Rh(D)-positive transfusions should only be administered when
The model, implemented in Microsoft R Open version 3.5.3 the Rh(D)-negative patient’s survival would be compromised by
(Microsoft, Redmond, WA), simulated 1 million Rh(D)-negative waiting for Rh(D)-negative units.

Vox Sanguinis. 2022;117:291–292. wileyonlinelibrary.com/journal/vox © 2021 International Society of Blood Transfusion. 291
292 LETTER TO THE EDITOR

CONF LICT OF IN TE RE ST RE FE RE NCE S

The authors have disclosed no conflicts of interest. 1. Seheult JN, Stram M, Pearce T, Bub CB, Emery SP, Kutner J, et al.
The risk to future pregnancies of transfusing Rh(D)-negative females
of childbearing potential with Rh(D)-positive red blood cells during
Mark H. Yazer1,2
trauma resuscitation is dependent on their age at transfusion. Vox
Philip C. Spinella3 Sang. 2021. doi: 10.1111/vox.13065.
Jansen N. Seheult1,2 2. Tamai Y, Ohto H, Takahashi H, Kitazawa J, Pediatric RBC
Alloimmunization Consortium. Transfusion-related alloimmunization
1 to red blood cell antigens in Japanese pediatric recipients. Transfus
Department of Pathology, University of Pittsburgh School of Medicine,
Med Rev. 2021;35:29–36.
Pittsburgh, Pennsylvania, USA 3. Evers D, Middelburg RA, de Haas M, Zalpuri S, de Vooght KMK, van
2
Vitalant, Pittsburgh, Pennsylvania, USA de Kerkhof D, et al. Red-blood-cell alloimmunisation in relation to
3 antigens’ exposure and their immunogenicity: a cohort study. Lancet
Department of Pediatrics, Division of Critical Care Medicine,
Haematol. 2016;3:e284-92.
Washington University in St Louis, St Louis, Missouri, USA
4. Yazer MH, Triulzi DJ, Sperry JL, Corcos A, Seheult J. Rate of RhD-
alloimmunization after the transfusion of RhD-positive red blood cell
Correspondence containing products among injured patients of childbearing age: sin-
Mark H. Yazer, Department of Pathology, University of Pittsburgh gle center experience and narrative literature review. Hematology.
School of Medicine, 3636 Blvd of the Allies, Pittsburgh, 2021;26:321–7.
5. Raval J, Madden K, Neal MD, Moore, S. Anti-D alloimmunization in Rh
PA 15213, USA.
(D) negative adults with severe traumatic injury. Transfusion. in press.
Email: [email protected] 6. Leonard JC, Josephson CD, Luther JF, Wisniewski SR, Allen C,
Chiusolo F, et al. Life-threatening bleeding in children: a prospective
ORCID observational study. Crit Care Med. 2021. doi: 10.1097/CCM.
0000000000005075.
Philip C. Spinella https://orcid.org/0000-0003-1721-0541
DOI: 10.1111/vox.13197

CORRIGENDUM

In the article by Bell et al. 2020, there were typographical errors in Table 1. The overall data in the ‘Weight at time of donation’ (kg) row, under
the ‘Yes’ and ‘No Skin Marker’ columns, should be (84.8  19.4) and (85.2  19.0) instead of (48.6  66.9) and (46.5  65.9).

The authors apologize for this error.

Reference:
Bell B, O’Donovan J, Wright ST, Gemelli CN, Knight E, Hirani R. Evaluation of a sterile surgical skin marker to indicate the optimal vein for
venepuncture in the blood donation setting. Vox Sang. 2020;115:377-387.

Vox Sanguinis. 2022;117:293. wileyonlinelibrary.com/journal/vox © 2021 International Society of Blood Transfusion 293
DOI: 10.1111/vox.13138

DIARY OF EVENTS

See also http://www.isbtweb.org/congresses/

10.2.2022 The European Hematology Association (EHA) and the European Society for Blood and Marrow Transplantation (EBMT) -
4th edition of the jointly organized European CAR T-cell Meeting.
15-16.3.2022 The IPFA/EBA Symposium on Plasma Collection and Supply will take place fully physical in Amsterdam,
the Netherlands on March 15 - 16, 2022.
23.3.2022 Eye Drops from Human Origin - First EDHO Workshop on Current Standards and Future Developments organized
by the ISBT Working Party Cellular Therapies.

294 © 2022 International Society of Blood Transfusion. wileyonlinelibrary.com/journal/vox Vox Sanguinis. 2022;117:294.