TRANSFUSION CATEGORY
Hemovigilance network in France: organization and analysis of
immediate transfusion incident reports from 1994 to 1998
Georges Andreu, Pascal Morel, François Forestier, Joëlle Debeir, Danielle Rebibo, Gérard Janvier,
and Patrick Hervé
BACKGROUND: Hemovigilance networks have been
introduced in several countries to improve knowledge of
blood transfusion-related morbidity and mortality. The
general organization of the French network and its re-
sults from 1994 through March 1999 are presented
here.
STUDY DESIGN AND METHODS: The hemovigilance
network relies on blood transfusion centers and hospital
correspondents, who analyze unexpected and untoward
blood transfusion-related effects and transmit a Transfu-
sion Incident Report (TIR) to a national database
(Transfusion Incident Reports Electronic Data Manage-
ment [GIFIT]).
RESULTS: As of March 1, 1999, the GIFIT database
contained 24,234 TIRs related to incidents that occurred
from the start of the hemovigilance network until De-
cember 31, 1998. The network was not fully imple-
mented until 1996; but the reporting rate seems to have
since stabilized at approximately 7000 per year (2.5 re-
ports per 1000 blood components). The highest report-
ing rate is observed with platelet concentrates (4.02/
1000), followed by RBCs (1.71/1000) and FFP (0.34/
1000). Bacterial contamination quickly appeared as a
major cause of morbidity and mortality (185 cases and
18 fatalities). However, a general trend of reduction in
this type of incident was observed over time, which can
be attributed to adoption of several preventive mea-
sures. In contrast, major ABO mismatchings during
RBC transfusion remained at a constant rate throughout
this period and accounted for six fatalities. After the
implementation of universal WBC reduction, some inci-
dents known to be related to WBCs, such as nonhemo-
lytic febrile transfusion reactions (NHFTR) and HLA im-
munization, were dramatically reduced.
CONCLUSION: Hemovigilance is an important tool not
only to analyze blood transfusion incidents, but also to
measure the effects of new processes or corrective ac-
tions at a national level.
1356 TRANSFUSION Volume 42, October 2002
H
emovigilance networks have been either in-
troduced or reinforced in North America and
Europe in the past few years. The organiza-
tion of each network varies greatly from one
country to another.1-8 This heterogeneity is related to the
administrative organizations responsible for health sys-
tems in each country, as well as to the social and political
perception of transfusion-related risks.
France set up a hemovigilance network in 1994 that
covers every step of the transfusion chain from blood
collection through follow up with the recipients.1,2,7,9-11
The aim of the network is to collect and evaluate infor-
mation on unexpected or untoward effects related to
blood components in order to prevent their recurrence.
This organization differs from those where each step of
the transfusion chain is treated separately, as in the
United States and the United Kingdom.4-6 Moreover, re-
porting is mandatory, as opposed to the United Kingdom
Serious Hazards Of Transfusions (SHOT) network, which
works on a voluntary basis.4,5
This report analyzes information drawn from the ex-
perience of the first 4 years of the hemovigilance network,
ABBREVIATIONS: AFSSaPS = French Agency of Medical Safety
of Health Products; BTC(s) = blood transfusion center(s); EFS
= French Blood Establishment; GIFIT = Transfusion Incident
Reports Electronic Data Management; HaCs = hospitals and
clinics; NHFTR = nonhemolytic febrile transfusion reaction;
PC = platelet concentrate; SHOT = Serious Hazards of Transfu-
sions; TIR = Transfusion Incident Report.
From the French Blood Establishment Atlantique Region,
Tours; French Blood Establishment Bourgogne Franche-Comté
Region, Besançon; Anesthesiology and Intensive Care Depart-
ment II, Pessac; and French Blood Establishment, Paris,
France.
Address reprint requests to: Georges Andreu, MD, Etab-
lissement Français du Sang Center-Atlantique, 2 Bd Tonnellé,
37020 Tours Cedex, France; e-mail: georges.andreu@efs.sant.fr.
Received for publication July 7, 2001; revision received
March 27, 2002, and accepted May 20, 2002.
TRANSFUSION 2002;42:1356-1364.

under the control of the French Blood Agency. After a
brief description of the hemovigilance network organiza-
tion, we present an overview of the Transfusion Incident
Reports (TIRs) according to blood component type, as
well as a more detailed analysis related to immediate in-
cidents, with a special focus on major ABO mismatching,
bacterial contamination, and the impact of universal
WBC reduction implementation.
HEMOVIGILANCE NETWORK
ORGANIZATION
The hemovigilance network was created by law.10,11 It
includes hemovigilance correspondents in hospitals and
clinics (HaCs) as well as in blood transfusion centers
(BTCs) and other members belonging to regional and na-
tional health authorities. It was set up in 1994, with the
national database of TIRs under the responsibility of the
French Blood Agency. All the data analyzed in the follow-
ing sections was gathered by this initial network organi-
zation, which changed in March 1999 when the French
Agency of Medical Safety of Health Products replaced the
French Blood Agency. However, because the basic steps
Fig. 1. Members of the hemovigilance network in France. 1 = Paper document transmitted by facsimile and/or mail; 2 = elec-
tronic document transmitted by e-mail.
HEMOVIGILANCE NETWORK IN FRANCE
of transfusion incident reporting and analysis were simi-
lar in the two periods, we describe the current organiza-
tion in this section (Fig. 1).
Hemovigilance correspondents
In all instances, hemovigilance correspondents are phy-
sicians or pharmacists. In HaCs, approximately 2000 cor-
respondents (i.e., nearly 95% of overall HaCs with trans-
fusion activity) have been appointed. Hemovigilance
correspondents are mostly anesthesiologists devoting a
small part of their time to this activity. However, some
large university hospitals have enrolled full-time physi-
cians in this function.
In BTCs, there is a local hemovigilance correspon-
dent in each blood component distribution site. As in
HaCs, correspondents usually work for the network part
time and usually are responsible for blood components
released to HaCs and/or for the immunohematology
laboratory. In addition, since January 2000, each of the 18
French BTCs has a hemovigilance correspondent who co-
ordinates the action of the local correspondents and is
directly connected with the regional and national health
authorities.
Volume 42, October 2002 TRANSFUSION 1357
ANDREU ET AL.
Regional hemovigilance coordinators
Regional hemovigilance coordinators are physicians be-
longing to the regional health authority (Direction Régio-
nale des Affaires Sanitaires et Sociales). They coordinate
the action of the local HaC and BTC correspondents.
The French Agency of Medical Safety of
Health Products
The French Agency of Medical Safety of Health Products
(Agence Française de Sécurité Sanitaire des Produits de
Santé [AFSSaPS]) was created on March 4, 1999, in accor-
dance with a 1998 law that completely reorganized the
blood transfusion supply in France.12 Its main objectives
in regard to the hemovigilance network are as follows:13
1. Provide the general orientation for hemovigilance.
2. Lead and coordinate the actions between the differ-
ent parties, HaCs, and BTCs.
3. Take appropriate measures to ensure transfusion
safety.
4. Provide the Ministry of Health with the epidemio-
logic information necessary to make decisions about
transfusion recommendations.
5. Carry out or organize (under the control of the Min-
istry of Health) epidemiologic investigations.
In addition, AFSSaPS is responsible for setting up regu-
lation and inspecting BTCs. This institution replaced the
French Blood Agency, which was the regulatory body in
charge of the hemovigilance network between 1994 and
March 4, 1999.
French Blood Establishment
The French Blood Establishment (Etablissement Français
du Sang [EFS]) is a national public institution responsible
for providing all French HaCs with the blood components
that they need. Its organization includes a headquarters
and 18 regional BTCs, with a total of 209 blood compo-
nent distribution sites.
In relation to the hemovigilance network, the main
objectives of EFS are the following:
1. Participate in the hemovigilance network via the BTC
Hemovigilance correspondents (as described ear-
lier).
2. Manage the computerized national database of un-
toward and unexpected transfusion reactions
(Transfusion Incident Reports Electronic Data Man-
agement [GIFIT]) in collaboration with AFSSaPS.
3. Communicate all the relevant epidemiologic data
from donors to the Institute of Sanitary Surveillance.
REPORTING OF
TRANSFUSION-RELATED INCIDENTS
The founding decree of the hemovigilance network states
that “Any person, physician, pharmacist, dental surgeon,
1358 TRANSFUSION Volume 42, October 2002
midwife or nurse who observes an effect unexpected or
undesirable due or likely due to Labile Blood Compo-
nents must report it without delay.”10 The reporting of
transfusion incidents is compulsory, whatever their se-
verity. This does not apply to near misses, which are not
reported to the hemovigilance network.
TIR document
HaC and BTC correspondents fill out and sign a stan-
dardized anonymous TIR that includes items to identify
the type of blood component involved, to classify the
incident according to its possible cause, and to provide
information about a patient’s follow up.
The TIR includes an indication of the severity of the
incident according to four categories as defined by WHO:
Grade 1 indicates an absence of immediate or long-term
vital threat, Grade 2 indicates long-term morbidity, Grade
3 indicates immediate vital threat, and Grade 4 indicates
death of the patient.
The TIR also evaluates the contribution of transfu-
sion to the incident, according to five scales: 0 excludes
any causal relationship between blood transfusion and
the incident, 1 is doubtful, 2 is possible, 3 is likely, and 4
is unquestionable.
Mode of transmission of TIRs
TIRs are transmitted to the regional hemovigilance net-
work coordinator, and there is national data gathering at
the AFSSaPS and the EFS headquarters. Both institutions
have to look out for incidents that could involve multiple
recipients or reveal a potential serial risk, leading to
prompt action at the national level. TIRs that belong to
this category are usually transmitted to AFFSaPS and EFS
via facsimile transmission or immediately after the ob-
servation of the incident.
As a rule, all TIRs are transmitted within 48 hours to
the regional and national levels in paper form (usually a
facsimile transmission). Then, TIRs are computerized by
the HaC and BTC correspondents for inclusion in a na-
tional database (GIFIT). Frequently, at the time of trans-
mission, a given TIR is not fully finalized and some major
investigations are not yet documented. Therefore, TIRs in
the database can be modified or completed by the HaC
and BTC correspondents after investigations are com-
plete.
ANALYSIS OF THE GIFIT DATABASE
The data presented below are extracted from the national
database of computerized management of the transfu-
sion incident reports (GIFIT) on March 1, 1999. They
cover the period during which the French Blood Agency
 HEMOVIGILANCE NETWORK IN FRANCE

was the regulatory body in charge of the hemovigilance
network.
Quantitative evolution of TIRs
As of March 1, 1999, the GIFIT database contained 24,234
TIRs related to incidents that occurred from the begin-
ning of the hemovigilance network until December 31,
1998. During the first 2 years, the network was not fully
implemented; 436 TIRs were registered in 1994 and 2127
in 1995. The reporting rate seems more stable since 1996
(approx., 7000/year), which corresponds to a ratio of 2.5
reports per 1000 blood components, as shown in Table 1.
Although the reporting rate has been stable at the
national level since 1997, there is a wide heterogeneity of
reporting according to the different regions, from 1.1 to
4.64 per 1000 blood components, indicating clearly that
not all the incidents are reported.
Allogeneic versus autologous blood components
The percentage of incidents with autologous blood com-
ponents is 1.7 percent. This is to compare with the fact
that autologous blood components represent 8 percent of
all blood components delivered, as shown in Table 1.
Although the overall rate of incidents is lower with au-
tologous components, some categories such as ABO mis-
matching are as frequent as for allogeneic blood compo-
nents (see below, ABO mismatching).
Severity of and contribution of transfusion to TIRs
Severity of incident. In the entire GIFIT database,
74.5 percent of the incidents were Grade 1, 21.6 percent
were Grade 2, 3.2 percent were Grade 3, and 0.7 percent
were Grade 4 (death).
Contribution of transfusion. Among the 24,230
evaluable TIRs, 6.3 percent were not attributed to trans-
fusion, whereas 13.9 percent were doubtful to have re-
sulted from transfusion, 33.7 percent were possible, 29.1
percent were likely, and 17 percent were unquestionably
due to transfusion.
Incidents possibly, likely, and unquestionably due
to transfusion according to the type of blood
component: analysis of 1997 and 1998
Tables 2 and 3 summarize the overall data of TIRs ac-
cording to the type of component transfused. FFP has the
lowest reporting rate (0.34/1000), whereas for RBCs the
reporting rate is 1.71 per 1000 and for platelet concen-
trates (PCs) it is 4.02 per 1000.
For almost half of the transfusion incidents observed
within 8 days of transfusion, a cause has not been iden-
tified. In the vast majority of these cases, symptoms con-
sisted of fever and/or chills with no identified immuno-
logic incompatibility or bacterial contamination. The role
of cytokines, whether active or passive, remains to be
properly investigated. Anaphylactoid symptoms ac-
counted for 31.6 percent of the TIRs, with most of the
observed symptoms consisting of rash and/or urticaria.
Immunologic conflict (RBCs or platelet antigens) ac-
counted for 11.8 percent. ABO mismatchings are ana-
lyzed in the following section. Incidents where bacterial
contamination is suspected are also analyzed separately.
Blood volume overload, almost exclusively observed
with RBCs as expected, represents 4 percent of the overall
TIRs and 6.5 percent of the RBC-related TIRs. The fre-
quency of this untoward effect highlights the fact that
blood transfusion is a complex medical treatment. Crite-
ria must be set for surveillance of the volume therapy
every time transfusion is decided on, by noninvasive or
invasive hemodynamic measurements as required.
Moreover, it is mandatory that physicians who prescribe
RBC transfusion know the physicochemical, pharmaco-
kinetic, and pharmacodynamic properties of the plasma
substitutes that are often associated with blood compo-
nents. Some of these substitutes, such as HES solutions of
high in vivo molecular weight and high concentration,
are likely to interfere with blood component transfusion,
as a result of their considerable intravascular persistence,
and to lead to blood volume overload.14,15 TRALI has not
been identified in the GIFIT database as an entity. It is
highly probable that at least some minor forms of TRALI

TABLE 1. Number of blood components used in France between 1994 and 1998 and the progressive introduction of
TIRs through the Hemovigilance network
1994 1995 1996 1997 1998 Total (%)
Allogeneic blood components 3,214,943 2,973,388 2,853,813 2,722,205 2,706,383 14,470,732
RBCs 2,375,357 2,193,000 2,139,000 2,055,270 2,053,807 10,816,434 (68.62)
PCs from whole blood 473,723 422,000 345,000 289,509 262,702 1,792,934 (11.37)
Single donor PCs 120,144 129,000 134,000 135,425 147,423 665,992 (4.23)
FFP (quaranteened) 127,774 131,853 123,766 119,454 123,608 626,455 (3.97)
FFP (SD) 117,945 97,535 112,047 122,547 118,843 568,917 (3.61)
Autologous blood components 244,876 264,688 271,193 261,728 249,733 1,292,218
RBCs 132,446 140,565 141,222 135,031 126,970 676,234 (4.29)
FFP 112,430 124,123 129,971 126,697 122,763 615,984 (3.91)
Total blood components 3,459,819 3,238,076 3,125,006 2,983,933 2,956,116 15,762,950
TIRs 436 2127 7414 7464 6793 24,234
TIR/1000 blood components 0.13 0.66 2.37 2.50 2.30 1.54

Volume 42, October 2002 TRANSFUSION 1359

ANDREU ET AL.

and 18 in 1998. Therefore, the inci-

TABLE 2. Number and classification of TIR according to the responsible dence of major ABO mismatching may
blood components in 1997 and 1998
be estimated for the entire period
Attributed to transfusion
reporting rate >1 RBCs (%) PCs (%) FFP (%) Total (%) data. Considering that during the years
Immediate incidents 4706 (62.8) 3196 (95.2) 216 (86.4) 8118 (73.1) 1994 through 1998, 10,816,434 alloge-
Delayed incidents 492 (6.6) 18 (0.5) 26 (10.4) 536 (4.8) neic RBC units were transfused in
Occurrence of RBC antibody 2241 (29.9) 140 (4.2) 7 (2.8) 2388 (21.5) France, the incidence of major ABO
Other 55 (0.7) 3 (0.1) 1 (0.4) 59 (0.5)
Total 7494 3357 250 11,101 mismatching is estimated at 1 in
TIR/1000 blood components 1.71 4.02 0.34 1.87 138,672 RBC units (95% CI, 1/174,458 to
1/117,569).17
Because 676,234 autologous RBC
units were used during the same pe-
TABLE 3. Number and classification of immediate incidents according to riod, the incidence of ABO-mismatched
the responsible blood components in 1997 and 1998
RBC delivery in patients who have been
Classification of
immediate incidents RBCs (%) PCs (%) FFP (%) Total (%) in an autologous predeposit program
Unknown 2632 (55.9) 1079 (33.8) 65 (30.1) 3776 (46.5) may be estimated at 1 in 96,604 autolo-
Anaphylactoid 788 (16.7) 1651 (51.7) 127 (58.8) 2566 (31.6) gous RBC units (95% CI, 1/225,411 to
Immunologic incompatibility 667 (14.2) 285 (8.9) 3 (1.4) 955 (11.8) 1/45,082). It is important to note that
Bacterial infection 73 (1.6) 24 (0.8) 1 (0.5) 98 (1.2)
Blood volume overload 305 (6.5) 11 (0.3) 5 (2.3) 321 (4.0) there is no significant difference in ma-
Inefficient transfusion 94 (2.0) 114 (3.6) 0 (0.0) 208 (2.6) jor ABO mismatching incidence be-
Other 147 (3.1) 32 (1.0) 15 (6.9) 194 (2.4) tween patients scheduled for allogeneic
Total 4706 3196 216 8118
TIR/1000 blood components 1.08 3.83 0.29 1.37 or autologous transfusion (chi-square
test, p < 0.35), despite the fact that the
latter blood components have a special
did occur but were not differentiated from vascular over- label system that avoids ABO group indication and clearly
load. This may be related to poor awareness of TRALI in indicates the recipient’s identity.
the French medical community. More recently, observa- Finally, the incidence of ABO mismatch-related
16
tion of a TRALI case led to more scrutiny on this com- death may be estimated at 1/1,802,739 allogeneic RBC
plication in the hemovigilance network. units transfused (95% CI, 1/5,000,000 to 1/772,602).
Estimation of the incidence of RBC delivery errors.
According to the ABO blood group distribution in the
French population, a random delivery of RBCs without
ABO mismatching after RBC transfusion: considering this blood group would lead to a safe trans-
analysis of 1994 through 1998 fusion, either ABO-identical or -compatible, in 65 percent
Since the beginning of the hemovigilance network, there of the transfusions. Therefore, considering that ABO mis-
have been 135 reports of ABO mismatchings (i.e., 0.55% matching is observed at random, there are approximately
of all TIRs). Among them, 50 were minor mismatchings two undetected errors in RBC delivery for each observed
related to the presence of ABO antibodies in RBCs, PCs, ABO mismatching, and we can speculate that the actual
or FFP and are not analyzed in this text. The remaining 85 incidence of errors in RBC delivery is approximately 1 in
reports were major ABO mismatchings after RBC trans- 50,000. This incidence is in the magnitude of that ob-
fusion, including seven cases of patients scheduled to served in other studies done on more limited geographic
receive autologous RBCs who received allogeneic mis- areas.18
matched units. Origin of the initial error. Among the 85 cases, 84
Overall, these 85 major ABO mismatchings were re- were fully evaluable for this criterion. The origin of the
sponsible for six fatalities, all being patients scheduled to initial error came from the BTCs in 13 cases and from the
receive allogeneic blood components. Twenty-four cases HaCs in 70 cases. In one case, a blood-typing error by a
were considered to be of Grade 3 severity (vital threat): private laboratory acting for a clinic was involved.
among them, the main clinical symptoms were shock Initial error at the BTC. Two errors of patient ABO
(n = 13), hemoglobinuria (n = 6), anuria (n = 2), pulmo- grouping were identified. One was an error in the tran-
nary edema (n = 2), and disseminated intravascular co- scription of results, and the other was a misinterpretation
agulation (n = 1). of laboratory data. The remaining 11 cases were related to
Incidence of major ABO mismatching. Although the the delivery of RBCs to a patient other than one originally
hemovigilance network was not fully implemented in scheduled, mainly due to partial homonymy. It is impor-
1994 and 1995, ABO mismatchings were actively reported tant to note that in these 11 cases, neither a serologic nor
from 1994: 16 in 1994, 11 in 1995, 18 in 1996, 22 in 1997, electronic crossmatch was done and that delivery of RBCs

1360 TRANSFUSION Volume 42, October 2002

 HEMOVIGILANCE NETWORK IN FRANCE

was done according to blood group result documents
treated manually. Therefore, delivery of RBCs in patients
without RBC alloantibodies using manual procedures
without electronic crossmatch account for 13 percent of
observed major ABO mismatchings, which could have
been prevented by an adequate electronic crossmatch
procedure.
The main reason for initial error in the HaCs was
misidentification of a patient either during blood sam-
pling for ABO grouping, during the ordering of blood
components, or at the time of transfusion.
Bacterial contamination of blood components
From the beginning of the hemovigilance network (April
1994) until December 31, 1998, there were 730 reported
incidents with a suspicion of posttransfusion bacterial
contamination. After careful analysis of each TIR, 545
(74%) of these 730 records were not confirmed to be re-
lated to bacterial contamination: in 219 cases, the bacte-
rial origin could be formally excluded, and in 326 cases,
the exclusion of bacterial contamination was based on
the existence of invalid investigations, notably bacterial
analysis.
Among the 185 confirmed cases, 18 (9.7%) patients
died. These deaths due to bacterial contamination rep-
resent 22 percent of the 82 overall deaths related to trans-
fusion reported in this period.
Blood components involved. As shown in Table 4,
PCs are responsible for 37.3 percent of the incidents of
bacterial contamination; 24.9 percent are attributed to
single-donor PCs and 12.4 percent to pooled PCs. RBCs
are implicated in 62.7 percent of the cases. There is not a
single confirmed bacterial incident related to plasma.
Overall incidence of bacterial contamination is 12.6
per 1 million allogeneic blood components and is much
more important for PCs (12.8 per 1 million from whole
blood and 69.1 per 1 million single-donor PCs) than for
RBCs (10.7 per 1 million). Note that these calculated in-
cidences may be underestimated because the reported
level of suspected bacterial contamination was lower in
the years 1994 and 1995. However, during this period, the
number of reported deaths attributed to bacterial con-
tamination was high: five in 1994 and five in 1995, in
comparison to eight between 1996 and 1998.
The observed death rate during the years 1994
through 1998 is 0.7 per 1 million RBCs, 1.1 per 1 million
PCs prepared from whole blood, and 12 per 1 million
single-donor PCs.
Characterization of bacteria. With RBCs, cocci
(staphylococcus and streptococcus) represent more than
half of the cases (58%), Gram-negative bacilli represent
32 percent of cases, and “other” bacteria represent 10
percent. With PCs, Gram-negative bacilli represent 36
percent of the cases, cocci represent 42 percent, and
“other” bacteria represent 22 percent.
Severity. Over the 5 years, 63.8 percent (118/185) of
the cases were considered minor and 26.5 percent (49/
185) were considered a vital threat. A total of 18 cases
(9.7%) had a fatal outcome.
During the first 2 years after the implementation of
the hemovigilance network in France (i.e., between April
1994 and March 1996), 10 fatalities related to bacterial
contamination of blood components were reported, lead-
ing to the awareness of bacterial contamination as a ma-
jor blood transfusion complication in the French trans-
fusion community.
Action plan to reduce bacterial contamination of
blood components. A national multidisciplinary work-
group (working party) was created in 1995. It included
experts in transfusion medicine, bacteriology, hygiene,
epidemiology, and pathology. These experts analyzed all
the reports of transfusion incidents due or suspected to
be due to bacterial contamination.
This working party introduced several active mea-
sures. Guidelines for investigating a suspected bacterial
contamination were produced in October 1995, followed
by guidelines for cutaneous asepsy procedures before do-
nation in September 1996.19,20

TABLE 4. TIRs related to bacterial contamination during the period 1994 through 1998 reported to the
hemovigilance network: incidence according to the type of blood components transfused
Number of blood TIR confirmed as related Deaths related to
components transfused to bacterial contamination bacterial contamination
Blood component 1994-1998 Number (Number/106) Number (Number/106)
RBCs 10,816,434 113 10.4 8 .7
PCs from whole blood 1,792,934 23 12.8 2 1.1
Single-donor PCs 665,992 46 69.1 8 12.0
FFP (quaranteened) 626,455 0 NA 0 NA
FFP (SD) 568,917 0 NA 0 NA
Total allogeneic blood components 14,470,732 182 12.6 18 1.2
RBC (autologous) 676,234 3 4.4 0 NA
FFP (autologous) 615,984 0 NA 0 NA
Total autologous blood components 1,292,218 3 2.3 0 NA
Total blood components 15,762,950 185 11.7 18 1.1

Volume 42, October 2002 TRANSFUSION 1361

ANDREU ET AL.
A research program was initiated at the end of 1995,
covering: 1) evaluation of bacteria detection by auto-
mated culture system in PCs before distribution; 21
2) phagocytosis and bactericidal assessment of whole
blood; 3) evaluation of diverting the first 30 mL of blood
collection out of the blood bag as a measure to reduce the
incidence of initial contamination;22 and 4) development
of a case-control study to identify risk factors for this
transfusion complication (BACTHEM study).23
Apart from the research program, an intensive sen-
sitization program on bacterial contamination was con-
ducted inside the hemovigilance network with blood col-
lection staff, with a reassessment of bacteria risk factors
for donors and blood collection.
For the next 3 years (April 1996 to March 1999), the
number of deaths reported through the hemovigilance
network were three, three, and one, respectively. The
small number of reports makes any interpretation of
these results difficult. However, it is tempting to hypoth-
esize that the sensitization to this complication, along
with many new procedures introduced by most blood
centers between the end of 1995 until 1999, including
cutaneous asepsy procedures (1995), revised donor ques-
tioning (1996), introduction of collection bags specifically
designed for diverting the first 30 mL of blood collected
(start of use in 1999 and fully generalized at the end of
2000),22 and implementation of universal WBC reduction
at the blood center (1998),24 may have played a role in
reducing the incidence of fatal reaction due to bacterial
contamination in France. Long-term reporting through
the hemovigilance network will provide useful informa-
tion to confirm this trend.
Major findings about bacterial contamination from
the Hemovigilance network
Four major findings were drawn from incidents related to
bacterial contamination of blood components:
1. The findings of the French hemovigilance network
confirm the incidence of this type of reaction.
2. Minor reactions are frequent, which emphasizes the
necessity of making the prescriber aware that even
fever or chills can be the only symptoms of the pres-
ence of bacteria in a blood component, as suggested
by Blajchman.25
3. Gram-positive cocci bacilli are very often involved in
RBC contamination, which is consistent with the fre-
quency of whole-blood contamination by skin flora
as described by Bruneau et al.22
4. Investigations carried out either to demonstrate
transfusion involvement in the septic episode or to
document the origin of blood component contami-
nation must be improved.
This last finding is in agreement with those of SHOT,
which reports inconclusive findings in 32 percent of 22
1362 TRANSFUSION Volume 42, October 2002
cases.26 Over the period between 1995 and 2000, SHOT
report 15 cases of bacterial contamination incidents with
five fatalities. PCs are generally involved (12/15). Thus,
there are almost no minor reactions recorded by SHOT.
The difference of reporting method (mandatory vs. vol-
untary) could certainly explain the difference in terms of
incidence and severity.
Impact of universal WBC reduction on TIRs
Universal WBC reduction was implemented in France on
April 1, 1998.24 Many advantages of WBC reduction had
been identified in the previous 15 years and supported
the argument for this implementation. However, most of
these advantages were documented through in vitro or
clinical studies that involved only a few components and
patients. Therefore, a general overview on the actual ef-
fect of universal WBC reduction at the national level was
lacking.
Analysis of RBC-related TIRs. To analyze the impact
of WBC reduction, it was necessary to evaluate the level
of WBC-reduced blood components used before the
implementation of universal WBC reduction. With more
than 90 percent of PCs being WBC reduced in France
since 1996, these blood components were not taken into
account when investigating the impact of WBC reduction
through the hemovigilance network.
Conversely, it was estimated from the national data-
base that WBC-reduced RBCs were approximately 50 per-
cent of all RBC units in 1997, 40 percent in 1996, and 20
percent in 1995. Therefore, we decided to compare TIRs
related to RBC transfusions during an 18-month period
(July 1, 1996-December 31, 1997) before implementation
of universal WBC reduction with an equivalent period
starting April 1, 1998. We were aware that this analysis of
the GIFIT database probably underestimates the effect of
WBC reduction due to the fact that 40 to 50 percent of
RBCs were already WBC reduced during the first 18-
month period analyzed, as opposed to 100 percent during
the second period.
Method of analysis. During the two periods, we con-
sidered RBC-related TIRs with severity Grades 1, 3, and 4
as being highly attributable to transfusion (likely and un-
questionable) for two categories of incidents that could
be expected. Category 1 comprised incidents related to
WBCs, such as nonhemolytic febrile transfusion reactions
(NHFTR) and HLA allo-immunization, and Category 2
comprised incidents not related to WBCs, such as ana-
phylactoid reactions, hemolysis, major ABO mismatch-
ing, or incompatibility due to anti-D.
Results. The number of transfused RBCs during the
two periods have been estimated to be 3,124,649 before
universal WBC reduction and 3,053,807 after universal
WBC reduction (i.e., a reduction of 2%).
The total number of TIRs with severity Grades 1, 3,
and 4 was 5559 before universal WBC reduction and 5306
 HEMOVIGILANCE NETWORK IN FRANCE

TABLE 5. Occurrence of TIR before and after implementation of universal WBC reduction in France*
Before universal After universal
WBC reduction WBC reduction
Number % Number % Variation after/before (%) p value†
RBCs transfused 3,124,649 3,053,807
Total TIRs 7437 NA 7849 NA 6
Severity Grades 1, 3, and 4 5559 75 5306 68 −5
Imputability 1, 2, 3, and 4 5159 69 4819 61 −7
Imputability 2, 3, and 4 3988 54 3502 45 −12
Imputability 3 and 4 1872 25 1408 18 −25
Severity Grades 1, 3, and 4,
and Imputability 3 and 4 1872 100 1408 NA 100 <0.001
Anti-HLA 190 10.1 94 6.7 −51 <0.001
HLA antibodies and NHFTR 112 6.0 33 2.3 −71 <0.001
Isolated NHFTR 615 32.9 363 25.8 −41 <0.001
Bacterial contaminations 71 3.8 24 1.7 −66 <0.001
Anaphylactoid reactions 266 14.2 224 15.9 −16 0.10
Hemolytic reactions
(excluding ABO and D) 26 1.4 23 1.6 −12 0.73
D incompatibilities 12 0.6 11 0.8 −8 0.88
ABO incompatibilities 30 1.6 33 2.3 10 0.64
* Analysis of RBC transfusion 18 months before and after implementation of universal WBC reduction shows as expected no significant im-
pact on anaphylactoid reactions, ABO and D incompatibilities and hemolytic reactions due to other RBC antibodies. Conversely, there is a
dramatic reduction of TIRs related to HLA immunization, NHFTR and bacterial contamination.
† Comparison of rates (i.e., 1408/3,053,807 vs. 1872/3,124,649) using the chi-square test.

after universal WBC reduction (chi-square test, p = 0.2).
Among these TIRs, the number with an attributible-to-
transfusion value greater than 2 were 1872 before univer-
sal WBC reduction and 1408 after universal WBC reduc-
tion (p < 0.0001).
The analysis of immediate incidents that were not
supposed to be influenced by WBC reduction showed no
significant variation in the two periods, as indicated in
Table 5. Statistical comparison has been made between
the observed rates (i.e., number of events/number of
RBCs transfused during the period) using the chi-square
test.
In contrast, NHFTR was reduced by 41 percent (p <
0.001) and HLA immunization by 51 percent (p < 0.001).
Moreover, immediate incidents where NHFTR could be
related to a documented HLA immunization were re-
duced by 71 percent (p < 0.001), as indicated in Table 5.
The dramatic reduction, by 66 percent, of bacterial
contamination may be related only in part to WBC re-
duction. As indicated in the preceding section, several
preventive measures were taken simultaneously. How-
ever, most of the preventive measures had already been
adopted in July 1996. Therefore, the role of WBC reduc-
tion alone is probably important, at least in reducing the
preparation process risk for RBCs and pooled PCs.27,28
CONCLUSION
Eighteen months after its implementation, the Hemo-
vigilance network was fully functional, and the number of
reported transfusion incidents remained stable since 1996.
Results of the hemovigilance network increased
awareness of known transfusion risks. ABO mismatching
and bacterial contamination were quickly identified as
the most important factors in immediate post-
transfusion serious morbidity and mortality, requiring
the implementation of specific measures to prevent their
occurrence.
We described in detail all the corrective measures
taken to prevent bacterial contamination. Although they
do not eliminate the risk, we think that they lead to its
reduction. However, despite a large increase in informa-
tion among hospital staff and the implementation in 1997
of a specific educational kit for nurses in charge of the
final check before transfusion, the rate of major ABO mis-
matchings remained stable during the 5 year period
(1994-1998).
Grade 1 transfusion incidents (i.e., immediate inci-
dents that are not life threatening) constitute the majority
of reports and are mostly fever and chills or anaphylac-
toid symptoms without serious clinical incidence on re-
cipients. We think it is useful to collect them as our prac-
tices are changing, and it at least may modify their
incidence: for instance, the impact of universal WBC re-
duction is mostly visible on NHFTR reactions, and it will
be interesting to observe the evolution of anaphylactoid
reactions when platelet storage solutions are introduced.
The reporting of transfusion incidents is compulsory.
However, the danger for the hemovigilance network
would be for it to be perceived of as repressive. Care is
taken to make sure that it remains descriptive and con-
structive, the ultimate goal being to make transfusion in-
cident reporting a natural component of medical prac-

Volume 42, October 2002 TRANSFUSION 1363

ANDREU ET AL.
tice, as a procedure of quality assurance among many
others.
Finally, we feel that this network acts at the local
hospital and transfusion blood center level as a major
tool for sensitization of the medical and nurse commu-
nity about blood transfusion, and at the national level as
a tool to evaluate the actions developed to reduce blood
transfusion complications.
ACKNOWLEDGMENT
The authors thank Catherine Frette for her active participation
in the analysis of GIFIT database.
REFERENCES
1. Debeir J, Noël L, Aullen J, et al. The french hæmovigi-
lance system. Vox Sang 1999;77:77-81.
2. Noël L, Debeir J, Cosson A. The french hæmovigilance
system. Vox Sang 1998;74(Suppl 2):441-5.
3. McClelland B, Love E, Scott S, Williamson LM. Hæmo-
vigilance. Concept, Europe UK Initiatives Vox Sang
1998;74(Suppl 2):431-9.
4. Williamson LM. Systems contributing to the assurance of
transfusion safety in the United Kingdom. Vox Sang 1999;
77:82-7.
5. Menitove JE. Hemovigilance in the United States of
America. Vox Sang 1998;74(Suppl 2):447-55.
6. Busch M, Chamberland M, Epstein J, Kleinman S, Khab-
baz R, Nemo G. Oversight and monitoring of blood safety
in the United States. Vox Sang 1999;77:67-76.
7. Salmi LR, Hervé P. L’organisation de l’hémovigilance en
France. Transfus Clin Biol 1994;3:252-6.
8. Morel P, Hervé P. Surveillance of blood transfusion safety.
Transf Med Rev 1998;12:109-27.
9. Quaranta JF, Canivet N. A quoi sert l’hémovigilance?
Transfus Clin Biol 1998;5:415-21.
10. Décret no. 94-68 du 24 janvier. Relatif aux règles
d’hémovigilance pris pour application de l’article L.666-12
du code de la santé publique et modifiant ce code (deuxi-
ème partie: décrets en Conseil d’Etat), 1994.
11. Circulaire DGS/DH no. 40 du 7 juillet. Relative au décret
no. 94-68 du 24 janvier 1994 sur l’hémovigilance pris
pour application de l’article L.666-12 du code de la santé
publique et modifiant ce code (deuxième partie: décrets
en Conseil d’Etat), 1994.
12. Loi no. 98-535 du 1er juillet. Relative au renforcement de
la veille sanitaire et du contrôle de la sécurité sanitaire
des produits destinés à l’homme. J Officiel la République
Française, 2 juillet 1998;15:10056.
13. Décret no. 99-142 du 4 mars. Relatif à l’Agence Française
de Sécurité Sanitaire des produits de santé et modifiant le
code de la santé publique (deuxième partie: décrets en
Conseil d’Etat), 1999.
1364 TRANSFUSION Volume 42, October 2002
14. Forestier F, Janvier G. Actualités sur les solutés de rem-
plissage en anesthésie. Conférences D’actualisation 2000,
42ème Congrès National D’anesthésie et de Réanimation.
Paris: Elsevier, 2000:151-63.
15. Hardy JF, Belisle S, Janvier G, Samama M. Reduction in
requirements for allogeneic blood products: nonpharma-
cological methods. Ann Thorac Surg 1996;62:1935-43.
16. Odent-Malaure H, Gallon P, Dumontier P, Foulhoux AC,
Lamy B, Le Petit JC. Transfusion-related acute lung injury
(TRALI) during plasmapheresis and transfusion. Paris:
ISBT, 15-8 July 2001.
17. Newcombe RG. Two sided confidence intervals for the
single proportion. Comparison of seven methods. Stat
Med 1998;17:857-72.
18. Linden JV, Wagner K, Voytovich AE, Sheehan J. Transfu-
sion errors in New York State. An analysis of 10 years’ ex-
perience. Transfusion 2000;40:1207-13.
19. Circulaire DGS/DH/AFS no. 85 du 10 OCT. Relative à la
conduite à tenir en cas d’accident bactérien lié à la trans-
fusion sanguine. BO Ministère des Affaires Sociales 1995;
46:93-107.20.
20. Recommandation AFS/DMS du 20 SEPT. Relative à la re-
commandation sur l’hygiène et l’antisepsie cutanée du
préleveur et de la peau au site de ponction dans le cadre
du prélèvement en transfusion sanguine, 1996.
21. Morel P, Masse M, Talon D, Leprat R. Systematic micro-
biological control of platelet concentrates (PC) using an
automated culture system: analysis of 5015 results.
Transfusion 1997;37(Suppl.):56S.
22. Bruneau C, Perez P, Chassaigne M, et al. Efficacy of new
collection procedure for preventing bacterial contamination
of whole-blood donations. Transfusion 2001;41:74-81.
23. Perez P, Salmi LR, Folléa G, et al. Determinants of trans-
fusion-associated bacterial contamination: results of the
French BACTHEM Case-Control Study. Transfusion 2001;
41:862-72.
24. Circulaire DGS/DH/AFS no. 98/118 du 20 février 1998.
Relative à la mise en place de la déleucocytation systéma-
tique des PSL (concentrés de globules rouges et concen-
trés de plaquettes de base), et aux mesures nécessaire à
la montée en charge de ce dispositif. BO Ministère des
Affaires Sociales (1998).
25. Blajchman MA. Bacterial contamination and proliferation
during the storage of cellular blood products. Vox Sang
1998;74(Suppl 2):155-9.
26. Serious Hazards of Transfusions. Annual Report 1999-
2000. Brentwood (UK): National Blood Service; 2001.
27. Högman CF, Gong J, Eriksson L, Hambraeus A, Johansson
CS. White cells protect donor blood against bacterial con-
tamination. Transfusion 1991;31:620-6.
28. Wenz B, Ciavarella D, Freundlich L. Effect of prestorage
white cell reduction on bacterial growth in platelet con-
centrates. Transfusion 1993;33:520-3.