Vous êtes sur la page 1sur 8

Journal of Thrombosis and Haemostasis, 5: 266273

ORIGINAL ARTICLE

The decrease of fibrinogen is an early predictor of the severity of


postpartum hemorrhage
B. CHARBIT,* L. MANDELBROT, E. SAMAIN, G. BARON, B. HADDAOUI, H. KEITA,
O. SIBONY,** D. MAHIEU-CAPUTO, M. F. HURTAUD-ROUX,** M. G. HUISSE,
M . H . D E N N I N G E R , and D . D E P R O S T F O R T H E P P H S T U D Y G R O U P
*AP-HP, Hopital Saint-Antoine, Clinical Investigation Center, Paris; AP-HP, Hopital Beaujon, Clichy; AP-HP, Hopital Louis Mourier, Colombes;
Hopital Jean Minjoz, Besancon; AP-HP, Hopital Bichat, Paris; **AP-HP, Hopital Robert Debre, Paris; INSERM U698, Paris; and AP-HP, CIB
PhenoGen, Paris, France

To cite this article: Charbit B, Mandelbrot L, Samain E, Baron G, Haddaoui B, Keita H, Sibony O, Mahieu-Caputo D, Hurtaud-Roux MF, Huisse MG,
Denninger MH, de Prost D, for the PPH Study Group. The decrease of fibrinogen is an early predictor of the severity of postpartum hemorrhage.
J Thromb Haemost 2007; 5: 26673.

Introduction
Summary. Background: Postpartum hemorrhage (PPH) is a
major source of maternal morbidity. Objectives: This studys
objective was to determine whether changes in hemostasis
markers during the course of PPH are predictive of its
severity. Patients and methods: We enrolled 128 women with
PPH requiring uterotonic prostaglandin E2 (sulprostone)
infusion. Two groups were dened (severe and non-severe
PPH) according to the outcome during the rst 24 hours.
According to our criteria, 50 of the 128 women had severe PPH.
Serial coagulation tests were performed at enrollment (H0), and
1, 2, 4 and 24 hours thereafter. Results: At H0, and through
H4, women with severe PPH had signicantly lower brinogen,
factor V, antithrombin activity, protein C antigen, prolonged
prothrombin time, and higher D-dimer and TAT complexes
than women with non-severe PPH. In multivariate analysis,
from H0 to H4, brinogen was the only marker associated with
the occurence of severe PPH. At H0, the risk for severe PPH was
2.63-fold higher for each 1 gL)1 decrease of brinogen. The
negative predictive value of a brinogen concentration
>4 gL)1 was 79% and the positive predictive value of a
concentration 2 gL)1 was 100%. Conclusion: These ndings
indicate that a simple brinogen measurement can anticipate
the risk of severe bleeding in PPH.
Keywords: brinogen, postpartum hemorrhage.
Correspondence: Dominique de Prost, Service dHematologie
Biologique, Immunologie et Transfusion, Hopital Louis Mourier,
178 rue des Renouillers, 92701 Colombes cedex, France.
Tel.: +33 14760 6113; fax: +33 14760 6278; e-mail: dominique.
de-prost@lmr.aphp.fr
This work is attributed to: AP-HP, Hopital Louis Mourier, Colombes,
France; University Paris 7 Denis Diderot, Paris, France.
Received 27 June 2006, accepted 23 October 2006

Postpartum hemorrhage (PPH) remains a major cause of


maternal morbidity and mortality related to childbirth [1,2]. In
most cases, PPH is due to bleeding from the placental site,
which is due to uterine atony [3]. Because the ow of blood is
high in the uterine arteries at the end of pregnancy, uterine
atony can rapidly result in severe hemorrhage. Protocols for
stepwise active management of PPH improve outcome [4,5]. As
soon as excessive bleeding is observed, the rst-line measures
are to manually explore the uterus and to inject oxytocin. In
case of persistent atony, more potent uterotonic prostaglandin
analogs are recommended. In severe cases of PPH, further
interventions are required, including hemodynamic resuscitation, blood products, uterine artery embolization and/or
surgical arterial ligation or hysterectomy.
As reported in other circumstances of major bleeding [6],
PPH may alter the subtle physiological changes in coagulation
brinolysis equilibrium that are observed in the peripartum
period and that may lead to profound and rapid changes in
hemostasis. At the end of normal pregnancy, the concentrations
of several clotting factors increase and several of the natural
anticoagulants, as well as brinolytic activity, decrease [7,8].
These physiological changes result in an apparent hypercoagulability state, described by some authors as a low-grade
compensated disseminated intravascular coagulation (DIC),
that may be important for minimizing blood loss at delivery
[7,9]. During the rst hours following delivery, a marked
increase in clotting system activity and an increase in brinolytic
activity have been reported [7,10]. Several factors may favor
postpartum DIC, including obstetrical complications such as
abruptio placentae and amniotic-uid embolism. DIC may also
be related to endothelial cell damage secondary to hemorrhagic
shock [11,12]. DIC is therefore a well-known phenomenon in
the course of PPH. However, the coagulation changes that may
occur in the early stage of PPH are poorly described. We
hypothesized that hemostatic alterations could contribute to the
 2006 International Society on Thrombosis and Haemostasis

Fibrinogen and severity of postpartum hemorrhage 267

severity of PPH, and that the early detection of these alterations


could predict cases with more severe bleeding. This study
compared changes in coagulation markers over a 24-h period
between two groups of women with PPH dened according to
the outcome of bleeding as severe or non-severe.
Materials and methods
This prospective multicenter study was conducted from
February 2002 to October 2004 in the maternity departments
of four Paris-area teaching hospitals of the Groupe HospitaloUniversitaire Nord (AP-HP, Paris, France) that provide both
primary and tertiary care. The study was approved by the
ethics committee of the French-Language Society of Intensive
Care Societe de Reanimation de Langue Francaise (Paris,
France). All patients gave oral informed consent to participate
in the study.
Inclusion criteria

Women were eligible if they had PPH, dened as uterine


bleeding, occurring in the rst 24 h after delivery, persisting
after manual exploration of the uterine cavity, and requiring
i.v. prostaglandin administration (sulprostone, Nalador,
Schering SA, Lys-Lannoy, France). Miscarriages (i.e. before
22 weeks of gestation) were not eligible.
PPH management

PPH was managed according to the same clinical protocol,


currently used in the four institutions, based on national
guidelines [4]. As soon as excessive bleeding was observed, the
rst-line measures were to manually explore the uterus and to
inject oxytocin. If hemorrhage persisted, sulprostone was
administered i.v. using an infusion pump. A rst 500 lg dose
was administered over 1 h, followed by a second 500 lg dose
administered over 35 h. Fluid therapy was managed to obtain
hemodynamic stability and normovolemia. Transfusion of
packed red blood cell (RBC) units was performed to maintain
the hemoglobin (Hb) level above 7 g dL)1, and fresh frozen
plasma (FFP) and platelet transfusions were decided by the
practitioners in charge of the patient. Hemostatic interventions,
including angiographic embolization, surgical arterial ligation
and hysterectomy, were performed in cases of persistent
bleeding. Angiographic embolizations are performed in
specialized radiology units in the Groupe Hospitalier Universitaire Nord of the APHP hospital system, and all participating
delivery sites are located within 10 min of these units by
medicalized transport. The use of antithrombin, activated
recombinant factor (F) VII and antibrinolytic agents was not
included in the therapeutic strategy.
Data collection

Enrollment (H0) was dened as the beginning of the i.v.


administration of sulprostone. All clinical events were
 2006 International Society on Thrombosis and Haemostasis

recorded prospectively and serial blood samples were collected at H0 and after 1, 2, 4 and 24 h. Importantly, no patient
received RBC units, FFP or platelet transfusion before the
H0 time-point.
The study protocol dened two groups of patients
according to the course of bleeding over the 24 h after H0.
The severe group included women meeting at least one of
the following criteria: peripartum decrease of Hb 4 g dL)1
(last Hb value before delivery considered as the reference);
transfusion of at least 4 RBC units; hemostatic intervention
(angiographic embolization, surgical arterial ligation or
hysterectomy); or death. The median (interquartile) time
period between the last predelivery Hb value and H0 was 0
(0;1) days. All women who did not meet any of these criteria,
over the rst 24 h postpartum, were assigned to the nonsevere group. Consequences of blood loss were assessed by
calculation of the organ dysfunction or infection (ODIN)
score [13].
Data on demographics, medical history, pregnancy and
delivery were collected, as was the last Hb level before labor or
Cesarean section.
Laboratory assays

For coagulation assays, blood was collected in vacutainer tubes


containing 0.129 mol L)1 sodium citrate, and plasma was
separated within 1 h. Analyses were performed in two steps:
rst, routine assays were performed immediately in the local
hospital laboratory and, second, the assay of additional
biomarkers was carried out centrally by one laboratory using
plasma samples stored at )80 C.
Routine analysis included blood cell count, Hb level,
prothrombin time (PT) expressed as International Normalized Ratio (INR) values, activated partial thromboplastin
time (APTT), brinogen, D-dimer, FII and FV, soluble
brin, and euglobulin clot lysis time for each sample. Serum
levels of creatinine, bilirubin and alkaline phosphatases were
obtained at H0 and H24 to assess renal or hepatic
dysfunction. Additional biomarkers included plasma levels
of antithrombin, protein C, thrombinantithrombin (TAT)
and plasminantiplasmin (PAP) complexes, and soluble
thrombomodulin.
Routine coagulation assays were performed on a STA-R
(Diagnostica Stago, Asnie`res, France) or BCS (Dade Behring,
Paris, France) analyzer. In three of the four laboratories, PT
was performed using the same analyzer (STA-R from Diagnostica Stago) and the same thromboplastin reagent (Neo CI
10; Diagnostica Stago) with a mean International Sensitivity
Index (ISI) of 1.80. The fourth laboratory used Thromborel
(Dade Behring) with an ISI of 1.07. Plasma levels of brinogen
were measured using STA Fibrinogen reagent (Diagnostica
Stago) or Multibren U (Dade Behring). D-dimer and protein
C antigen levels were measured by enzyme-linked immunoassays using the Vidas analyzer (BioMerieux, Marcy lEtoile,
France). Antithrombin activity was determined using a
chromogenic assay Stachrom ATIII (Diagnostica Stago). The

268 B. Charbit et al

presence of soluble brin monomer complexes was semiquantitatively evaluated using the FS test (Diagnostica Stago).
Plasma levels of TAT (Dade Behring), PAP (Montes et al. [14])
and soluble thrombomodulin (Asserachrom Thrombomodulin, Diagnostica Stago) were measured by enzyme-linked
immunoassay.
Statistical analysis

Study sample size was calculated in order to obtain 2948


patients in the severe group, so as to test three to ve covariates
in the multivariate analysis. We hypothesized that 35% of cases
would be in the severe group, and thus we required the
enrollment of 110 cases of PPH.
Quantitative variables are presented as mean 1 SD or
the median (25th percentile; 75th percentile) for variables
not normally distributed. To compare the distribution of
baseline characteristics between the two groups, v2 and
MannWhitney U-tests were performed. A linear mixedeffect model taking into account the analysis of repeated
measurements was used to compare the kinetics of each
hemostatic biomarker during the rst 24 h between the two
groups. P-values of the group effect and the interaction
between groups and time were computed. The predictive
value of each biological parameter in relation to the severity
of hemorrhage was evaluated by using the area under the
receiver operating characteristic curves (AUC). Sensitivity,
specicity, positive and negative predictive values at various
cut-offs were also calculated. At each time point, a logistic
multiple regression with a stepwise procedure (a signicance
level of 0.05 was required to allow a variable into the model,
and a signicance level of 0.05 was required for a variable to
stay in the model) was performed to identify independent
predictors of the severity of hemorrhage. All statistical
analyses were performed using SAS version 9.1 (SAS Institute,
Cary, NC, USA).

Fig. 1. Study prole.

one uterine rupture; all were in the severe group. In the two
groups, we analyzed the incidence of factors previously
described as being associated with PPH [15]: body mass index
> 30; age > 38 years; parity > 4; Cesarean section; premature delivery; birthweight > 4,000g; twin pregnancy; polyhydramnios; previous PPH; labor duration > 12 h; third stage of
labor > 30 min; abnormal placental insertion; and pre-existent
coagulation disturbances. Forty per cent, 31%, 16% and 13%
of all patients had no, one, two, and three or more PPH risk
factors, respectively. The incidence of these risk factors did not
differ between the two groups (P 0.8).
Clinical history of hemorrhage

During the study period, there were 13 402 deliveries in the


participating centers and 129 patients were enrolled in the
study. One case was excluded from the analysis because blood
samples were lacking after H0. This hemorrhage did not
require transfusion or hemostatic intervention. Among the 128
cases of PPH, 50 (39%) presented at least one criterion of
severity over the rst 24 h postpartum, forming the severe
group (Fig. 1).

Among the 50 women in the severe group, 16 (32%) were


transfused with a median of 4.0 (2.2; 5.7) RBC units (range 1
9). Thirteen (26%) received 4.0 (1.9; 6.1) units (range 28) of
FFP. One patient received a platelet transfusion, brinogen
concentrate and aprotinin, 3 h after enrollment. Two women
required uterine arterial ligation and required angiographic
uterine artery embolization. No hysterectomy was performed.
The ODIN score was equal to 0 in 74% of patients in the severe
group. In 11 cases, the score was equal to 1 because of transient
clinical symptoms of hypovolemia; none required continuous
infusion of catecholamines. We observed an increase in serum
creatinine level in two women, one of whom developed
transient renal insufciency related to severe HELLP
syndrome.
In the non-severe group, only one patient was transfused
with 2 RBC units and 2 units of FFP. One woman had an
ODIN score equal to 1 (transient hypotension). The time from
delivery to hospital discharge was longer in the more severe
group 6.3 4.2 vs. 4.7 2.0 days (P < 0.01). All women
had a favorable nal outcome.

Clinical characteristics

Laboratory parameters at enrollment (H0)

Demographic and obstetric characteristics of the 128 patients


are presented in Table 1. Prepregnancy body weight and body
mass index were signicantly lower in the severe group. A
specic etiology of hemorrhage was found in four patients: one
severe hemolysis, elevated liver enzymes, low platelets
(HELLP) syndrome, two cases of abruptio placentae, and

Laboratory values are shown in Table 2. Several coagulation


parameters, including PT expressed in seconds (data not
shown), percentage activity and INR, brinogen, FII and FV,
D-dimer, antithrombin, protein C, TAT and soluble brin
monomer complexes differed signicantly between the two
groups, suggesting the early onset of acquired coagulopathy in

Results

 2006 International Society on Thrombosis and Haemostasis

Fibrinogen and severity of postpartum hemorrhage 269


Table 1 Clinical characteristics of the patients in relation to outcome

Maternal age (years)


Maternal weight (kg)
Gestational age (weeks)
Parity
Complications of pregnancy (%)
None
Hypertension/pre-eclampsia
Diabetes
Other
Chorioamnionitis (%)
Twin deliveries (%)
Mode of delivery (%)
Spontaneous vaginal
Instrumental vaginal
Emergent cesarean
Elective cesarean
Length of labor (hours)
Third stage of labor (minutes)
Management of third stage (vaginal deliveries)
Natural (%)
Assisted (%)
Manual (%)
Time between delivery and infusion of sulprostone (minutes)
Predelivery hemoglobin value (g dL)1)
Decrease in hemoglobin level between predelivery and H0 (g dL)1)

Severe group n 50

Non-severe group n 78

P-value

28 [26; 33]
56 [50; 66]
40 [39; 41]
2 [1; 2]

40
4
4
3
4.0
4

62
18
18
2
6 [3; 9]
11.9 11.9

12.5
65.5
25.0
87 [40; 150]
12.1 [11.0; 13.0]
2.5 [)0.4; 5.3]

30 [26; 34]
62 [56; 72]
40 [38; 41]
2 [1; 3]

61
7
5
7
2.6
11

67
17
9
8
6 [4; 9]
11.1 9.1

32.3
46.2
21.5
60 [35; 125]
11.3 [10.7; 12.4]
1.9 [)1.1; 4.3]

0.34
0.01
0.29
0.20
0.9

0.65
0.06
0.27

0.94
0.24
0.07

0.08
0.04
0.46

Results are given as median [25th percentile; 75th percentile].

Table 2 Laboratory values at enrollment (H0) in the severe and non-severe groups

Platelets (109 L)1)


PT (%)
INR
APTT ratio
Fibrinogen (g L)1)
Factor II (%)
Factor V (%)
D-dimer (lg mL)1)
Antithrombin activity (%)
Protein C antigen (%)
Soluble brin monomer complexes (% positive)
TAT complexes (lg L)1)
Euglobulin lysis time (% < 180min)
PAP complexes (ng mL)1)
Soluble thrombomodulin (ng mL)1)
Blood creatinine (lmol L)1)
Alkaline phosphatase (IU L)1)

Severe group n 50

Non-severe group n 78

P-value

AUC

173
81
1.16
1.0
3.3
83
72
9
72
69
55
39
26
304
9
63
142

181
88
1.10
1.0
4.4
93
90
6
79
75
36
20
19
199
9
69
145

0.40
0.02
0.02
0.05
< 0.0001
0.005
0.004
0.007
0.005
0.038
0.04
0.014
0.36
0.22
0.83
0.13
0.57

0.544
0.625
0.625
0.601
0.753
0.653
0.655
0.648
0.656
0.615

0.639

0.430

0.468
0.462

[141;209]
[70;90]
[1.08;1.30]
[1.0;1.2]
[2.5;4.2]
[72;94]
[60;94]
[5;22]
[62;79]
[57;83]
[40;69]
[19;57]
[15;41]
[162;803]
[9;20]
[48;74]
[112;166]

[139;231]
[80;96]
[1.04;1.18]
[0.9;1.1]
[3.7;5.1]
[83;102]
[75;109]
[3;9]
[71;84]
[66;85]
[24;48]
[13;38]
[11;31]
[94;645]
[9;20]
[60;79]
[113;193]

Results are given as median [25th percentile; 75th percentile] or percentage [95% condence interval]. AUC is the area under ROC curve used to
test the predictive power of each laboratory parameter for the severity of bleeding.

the severe group. The distribution of individual brinogen


levels between the two groups is shown in Fig. 2. In contrast,
the mean platelet count was within the normal range in both
groups (intergroup difference, P 0.05). Euglobulin lysis time
and PAP complex levels, both reecting brinolysis, and
soluble thrombomodulin, reecting endothelial injury, were
not signicantly different between the two groups.
 2006 International Society on Thrombosis and Haemostasis

Predictive values of coagulation parameters at H0

Univariate analysis The predictive power of each laboratory


parameter in relation to the severity of bleeding was evaluated
using AUCs. As shown in Table 2, the highest value was found
for brinogen (AUC 0.75; P < 0.0001). ROC curve
analysis for brinogen (Fig. 3) showed that the cutoff

270 B. Charbit et al
Table 3 Multivariate logistic regression of laboratory parameters associated with severe PPH
Time

Laboratory
parameter

H0*
H1*
H2
H4

brinogen
brinogen
brinogen
brinogen

(g
(g
(g
(g

L)1)
L)1)
L)1)
L)1)

Number of
patients

OR [95% CI]

P-value

113
114
114
115

2.63
2.70
3.70
5.00

<
<
<
<

[1.664.16]
[1.754.16]
[2.176.25]
[2.639.09]

0.0001
0.0001
0.0001
0.0001

For brinogen, odds ratios (OR) were calculated for each 1 g L)1
decrease in its plasma concentration. *data introduced in the model:
INR, APTT, brinogen, factor II, factor V, D-dimer, antithrombin,
protein C and TAT. data introduced in the model: INR, APTT,
brinogen, factor II, factor V, D-dimer, antithrombin and protein C.

data introduced in the model: platelets, INR, APTT, brinogen,


factor II, factor V, D-dimer and protein C.

Fig. 2. Individual brinogen plasma concentrations at H0 in women with


severe (d) or non-severe (s) postpartum hemorrhage. Mean SD values are reported for both groups.

one underwent angiographic embolization. Finally, of the ve


patients requiring massive transfusion (more than 6 RBC
units), four had an H0 brinogen level < 2g L)1.
Changes in coagulation parameters over the first 24 h

The changes in the laboratory parameters in the two groups


from H0 to H24 are shown in Figs 4 and 5. The changes over
time of INR, APTT, brinogen, FII and FV, D-dimer, protein
C, antithrombin and TAT complexes differed signicantly
between the groups. Moreover, for platelet count, INR, APTT,
brinogen, FV and antithrombin, a signicant interaction over
time was found, indicating a different time course between the
two groups. However, because the H24 point in both groups
differed substantially from the H0H4 period, this analysis was
redone excluding H24. The interaction over time then remained
signicant only for platelets (P 0.003), D-dimers
(P < 0.001) and TAT complexes (P < 0.001).

Fig. 3. ROC curve of brinogen plasma concentration at H0 for the


diagnosis of severe postpartum hemorrhage.

point of 4.0 g L)1 had the highest discriminatory power, with a


sensitivity of 74% and a specicity of 65%. By contrast,
PT and APTT were of little help to predict the severity of
bleeding.
Multivariate analysis In the logistic regression model,
brinogen concentration at H0 was the only parameter
independently associated with progress toward severe
bleeding. For each 1 g L)1 decrease in brinogen, the
calculated odds ratio and 95% condence interval were equal
to 2.63 (1.664.16; P < 0.0001; Table 3). Importantly, at H0,
while brinogen > 4g L)1 had a reassuring negative predictive
value of 79% (6889%), brinogen 2 g L)1 had a positive
predictive value of 100% (71100%). Among the 11 women
with brinogen below 2g L)1 at H0, eight (73%) were
transfused, uterine artery ligation was performed in two, and

Fig. 4. Serial coagulation parameters from H0 to H24 in the severe and


non-severe postpartum hemorrhage (PPH) groups. Data are presented as
median and interquartile range. The respective P-values refer to the difference between the two groups evaluated by repeated measurement
analysis. Open circles, non-severe PPH; full circles, severe PPH.
 2006 International Society on Thrombosis and Haemostasis

Fibrinogen and severity of postpartum hemorrhage 271

Fig. 5. Serial specialized coagulation and brinolysis biomarkers from H0


to H24 in the severe and non-severe postpartum hemorrhage (PPH)
groups. Data are presented as median and interquartile range. The
respective P-values refer to the dierence between the two groups evaluated by repeated measurement. Open circles, non-severe PPH; full circles,
severe PPH.

Predictive values of laboratory parameters from H1 to H4

Univariate analysis Among all hemostatic biomarkers, from


H1 to H4, brinogen displayed the highest AUC at all time
points studied (data not shown). The AUC of brinogen
increased from 0.75 at H0 to 0.78 at H1 (P 0.21 vs. H0), 0.82
at H2 (P 0.003 vs. H0) and 0.84 at H4 (P < 0.0001 vs. H0).
This increase in AUC of brinogen reected an increased
sensitivity (89% at H2 and H4), with a decreased specicity.
Multivariate analysis The multivariate logistic regression
model showed that, at each time point from H1 to H4,
brinogen was the only parameter independently and constantly
associated with the occurrence of severe bleeding (Table 3).
The statistical analysis was redone using expressions of PT
other than INR, i.e. time expressed in seconds and as activity
percentage. The same level of statistical signicance was
obtained with all three-expression modalities.
Discussion
This is the rst study to compare coagulation changes during
the course of PPH, according to the outcome. At H0 time
point, several coagulation parameters (INR, brinogen, FII
and FV, D-dimer, antithrombin, protein C, soluble brin
monomer and TAT complexes) differed signicantly between
patients who developed severe hemorrhage and those who did
not. These differences were already evident at the time of
diagnosis of hemorrhage and persisted at least for 4 h. Our
 2006 International Society on Thrombosis and Haemostasis

main nding was that throughout the rst 4 h the brinogen


level was consistently lower in cases of severe hemorrhage, and
was the only parameter independently and signicantly
associated with the worsening of bleeding.
The classical denition of PPH is a blood loss of more
than 500 mL within 24 h after vaginal delivery or 1000 mL
after Cesarean section. These criteria are now challenged
because of the inaccuracy in postpartum blood loss measurement [16]. We chose to enroll women with PPH requiring
the use of uterotonic prostaglandin infusion, because this is a
well-dened and reproducible step in the current guidelines
for PPH management, as applied in the four participating
centers [4,17]. In France, the uterotonic prostaglandin
licensed for this indication is the synthetic prostaglandin E2
sulprostone. Sulprostone infusion is recommended in cases of
bleeding from the placental site, after failure of manual
exploration of the uterus and administration of oxytocin.
This criterion excluded mild cases of PPH responding to rstline therapeutic measures. The number of patients enrolled
over the study period is compatible with the incidence of
PPH of 6.7 per 1000 reported by Waterstone et al. [15] in
nearly 50000 deliveries in south England, when using criteria
similar to ours.
Furthermore, we enrolled patients with PPH related mostly
to uterine atony. Uterine atony could be attributed to a specic
obstetric complication in only 3% of the patients, and in nearly
one-half of the cases no PPH risk factor was identied,
conrming that every delivery carries a risk of PPH [18].
Because our main objective was to compare coagulation
parameters according to the outcome of hemorrhage, the
criteria used to dene the severity of PPH are important to
consider. In the absence of consensus, we adapted the criteria
for severe PPH used by Waterstone et al. [15]. Despite the fact
that according to this denition 39% of our patients had severe
PPH, some having a critical clinical course, none of them
developed severe organ failure or died. This fortunate outcome
may be related in part to the early and intensive management of
PPH, according to a well-dened protocol, in use in our
institutions. It has been shown that delayed treatment of PPH
is one of the strongest predictors of poor nal outcome,
including maternal death [19].
The coagulation abnormalities we observed are compatible
with the early occurrence of DIC [20,21]. The association of
DIC with PPH requiring massive transfusions has been
previously described [12]. However, the kinetics of such
changes have not yet been reported during the initial stages
of PPH. Following normal delivery, a rapid activation of
coagulation occurs, as reected by a sharp increase in TAT
complexes [22] and D-dimer levels [23]. These changes are
transient, peaking 23 h after delivery. At that time, platelets,
brinogen and antithrombin have been reported to remain
unchanged [10] as compared with predelivery levels. We
observed that, at enrollment, the patients who went on to
develop severe PPH had signicantly decreased brinogen, FV,
antithrombin and protein C levels, together with increased PT
and levels of TAT complexes and D-dimers, as compared with

272 B. Charbit et al

those of the non-severe group. At H0 time point, the platelet


count was similar in the two groups; however, interestingly,
while the platelet count remained quite stable over time in the
non-severe group, it tended to decrease in the severe group, as
shown by the signicance of the interaction between groups
and time (P 0.003). These differences are in favor of an
excessive activation of coagulation in the more severe group, as
further indicated by the presence of soluble brin monomer
complexes. In contrast, the PAP complex concentration, which
sensitively reects the activation of the brinolytic system, did
not differ between the two groups. This nding is of interest as
it suggests that systemic brinolysis, which is sometimes
advocated, is not the primary mechanism leading to PPH in
patients with uterine atony. Our most striking nding was the
value of brinogen levels in predicting the severity of PPH.
While a number of hemostatic markers differed signicantly
between the two groups, only brinogen was independently
associated, at all time points from enrollment on, with the
course of the hemorrhage. Fibrinogen concentration increases
physiologically during pregnancy. Mean brinogen concentration upon admission to the labor ward for delivery has been
reported to be 4.8 g L)1 (SD 1.0 g L)1; range 2.19.0) [24].
Interestingly, in this study, brinogen concentrations were
signicantly lower in women who went on to develop PPH in
comparison with women who did not. Our results further show
that, once PPH is diagnosed, the decrease in brinogen level
can be used as an indicator of its severity in the hours to come.
Decrease in brinogen concentration may result from several
mechanisms. (i) Fibrinogen abnormalities can be either quantitative or qualitative. In the present study, brinogen was
initially measured using a functional test that cannot discriminate between these two mechanisms. However, further assays
using an immunological method gave similar results (data not
shown), thus excluding this hypothesis. (ii) Decreases in
concentrations of all blood components during the early phase
of hemorrhage are in part related to hemodilution, because of
redistribution of water from the extracellular space and to uid
infusion. In our study, at H0, the decrease in brinogen
concentration (expressed as the percentage of its value in the
severe vs. the non-severe group) was two times higher than the
drop in Hb level, suggesting that hemodilution cannot totally
account for the difference observed between the groups. (iii)
Our preferred hypothesis, which is sustained by the biological
context of marked activation of coagulation, is that in the case
of bleeding because of uterine atony, intravascular brin
deposits increase and are associated with an increased brinogen consumption. Activation of coagulation appears to be
mainly responsible for this brinogen decrease because, as
previously discussed, brinolytic markers were not different
between the two groups, suggesting that brinolysis is not
preponderant. The marked increase in D-dimer levels also
sustain this mechanism because, as reported by Dempe et al.,
D-dimers display a high degree of correlation with soluble
brin, which is regarded as an indicator of acute brin
formation [25]. In our study, this correlation was conrmed
(r 0.78, P < 0.0001) in a subgroup of 98 patients in whom

soluble brin monomers were measured using a quantitative


assay (Diagnostica Stago, data not shown). Lastly, we cannot
exclude the role of some differences in patient management, as
there was a trend towards a longer delay between delivery and
the start of sulprostone infusion in the severe vs. the non-severe
group, although these differences did not reach statistical
signicance (P 0.08).
Fibrinogen level had an overriding impact when predicting
the severity of bleeding in our PPH patients. However, whether
such low brinogen values by themselves play a role in the
evolution of PPH to severity, or only reect the severity of
hemorrhage, is not clear. Further research should be performed
to investigate the relative contributions to the bleeding risk of
coagulation disturbances occurring before, during and after
delivery. Finally, our study shows that changes over time in the
various coagulation parameters failed to offer additional
insight into the mechanisms of PPH or additional prognostic
value.
Coagulation disturbances peaked between the rst and
second hour after the beginning of sulprostone infusion. The
predictive value of low brinogen for poor outcome was
consistently maintained during the rst 4 h. Our ndings have
implications for clinical practice. The use of antibrinolytic
agents is a matter of discussion; in fact, only one small,
randomized prospective study conducted on patients with
abruptio placentae complicated by intrauterine death of the
fetus showed an improvement in the consumption coagulopathy after treatment with trasylol, an antibrinolytic agent [26].
Our results do not sustain the use of these drugs in severe PPH.
By contrast, the brinogen level at the time of diagnosis of PPH
can be used to guide the management of PPH. When it is below
2 g L)1, clinicians should be aware that there is a high risk of
severe bleeding. This is particularly useful because the brinogen concentration is determined by a simple assay that is
available in most institutions on an emergency basis.
Addendum
The complete list of all participants in the PPH Study Group is
as follows. N. Ajzenberg, E. Angle`s-Cano, G. Baron, A.
Bonnet, L. Boudaoud, B. Charbit, M. H. Denninger, B. Deval,
A. M. Duprat, M. J. Duviquet, Y. El Houari, E. Foucher, F.
Genty, O. Goncalvez, B. Haddaoui, M. G. Huisse, M. F.
Hurtaud-Roux, H. Keita, M. Lamotte, Y. Laurian, D.
Mahieu-Caputo, L. Mandelbrot, E. Peynaud-Debayle, D. de
Prost, C. Roy, E. Samain, F. Schlemer, F. Shaller, O. Sibony
and P. Verpillat.
Author contributions
D. de Prost and E. Samain were responsible for study initiation
and design, B. Charbit collected and analyzed the data, E.
Samain, D. de Prost, L. Mandelbrot and B. Charbit analyzed
the data and wrote the manuscript. G. Baron performed the
statistical analysis, B. Haddaoui, M. F. Hurtaud-Roux, M. G.
Huisse, E. Peynaud-Debayle, E. Angle`s-Cano and M. H.
 2006 International Society on Thrombosis and Haemostasis

Fibrinogen and severity of postpartum hemorrhage 273

Denninger participated in biological data collection and


interpretation. H. Keita, O. Sibony, D. Mahieu-Caputo, E.
Foucher and F. Genty were involved in clinical data collection
and interpretation. All authors approved the nal version of
the manuscript.
Acknowledgements
We thank Y. Laurian (AP-HP, Hopital Jean Verdier, Bondy,
France) for his support and contribution to the study, P.
Verpillat (AP-HP, Hopital Louis Mourier, Colombes, France)
for contribution to the design of the study, and E. PeynaudDebayle (AP-HP, Hopital Louis Mourier) and E. Angle`s-Cano
(INSERM U698, Paris, France), for their invaluable contribution to the assay of PAP complexes and other biological
assays. We gratefully acknowledge M. Lamotte and A. M.
Duprat from the Laboratory of Hematology, Hopital Louis
Mourier for their technical support and for taking care of the
biological samples. We thank the physicians, nurses, midwives
and laboratory technicians of all four hospitals for their great
help in this investigation, notably E. Foucher and F. Schlemer
of the Hopital Louis Mourier, L. Boudaoud, Y. El Houari and
B. Deval of the Hopital Beaujon, N. Ajzenberg, O. Goncalvez
and M. J. Duviquet of the Hopital Bichat, and F. Genty, A.
Bonnet and F. Shaller of the Hopital Robert Debre. We thank
C. Roy for her help with the statistical analysis.
Disclosure of Conflict of Interests
This work was supported by a grant from Novo Nordisk
Pharmaceutical S.A.S. (La Defense, Paris, France).
References
1 Hazra S, Chilaka VN, Rajendran S, Konje JC. Massive postpartum
haemorrhage as a cause of maternal morbidity in a large tertiary
hospital. J Obstet Gynaecol 2004; 24: 51920.
2 Zhang WH, Alexander S, Bouvier-Colle MH, Macfarlane A. Incidence of severe preeclampsia, postpartum haemorrhage and sepsis as a
surrogate marker for severe maternal morbidity in a European population-based study: the MOMS-B survey. BJOG 2005; 112: 8996.
3 Alamia V, Meyer BA. Peripartum hemorrhage. Obstet Gynecol Clin
North Am 1999; 26: 38598.
4 Langer B, Boudier E, Haberstich R, Dreyfus M. Obstetrical management in the event of persistent or worsening postpartum hemorrhage despite initial measures [French]. J Gynecol Obstet Biol Reprod
2004; 33: 4S7379.
5 Rizvi F, Mackey R, Barrett T, McKenna P, Geary M. Successful
reduction of massive postpartum haemorrhage by use of guidelines
and sta education. BJOG 2004; 111: 4958.
6 Lynn M, Jeroukhimov I, Klein Y, Martinowitz U. Updates in the
management of severe coagulopathy in trauma patients. Intensive Care
Med 2002; 28: S2417.
7 Bremme KA. Haemostatic changes in pregnancy. Best Pract Res Clin
Haematol 2003; 16: 15368.

 2006 International Society on Thrombosis and Haemostasis

8 Brenner B. Haemostatic changes in pregnancy. Thromb Res 2004; 114:


40914.
9 Cerneca F, Ricci G, Simeone R, Malisano M, Alberico S, Guaschino
S. Coagulation and brinolysis changes in normal pregnancy. Increased levels of procoagulants and reduced levels of inhibitors during
pregnancy induce a hypercoagulable state, combined with a reactive
brinolysis. Eur J Obstet Gynecol Reprod Biol 1997; 73: 316.
10 Gerbasi FR, Bottoms S, Farag A, Mammen EF. Changes in hemostasis activity during delivery and the immediate postpartum period.
Am J Obstet Gynecol 1990; 162: 115863.
11 Levi M, Ten Cate H. Disseminated intravascular coagulation. N Engl J
Med 1999; 341: 58692.
12 Kobayashi T, Terao T, Maki M, Ikenoue T. Diagnosis and management of acute obstetrical DIC. Semin Thromb Hemost 2001; 27: 1617.
13 Fagon JY, Chastre J, Novara A, Medioni P, Gibert C. Characterization of intensive care unit patients using a model based on the presence
or absence of organ dysfunctions and/or infection: the ODIN model.
Intensive Care Med 1993; 19: 13744.
14 Montes R, Paramo JA, Angles-Cano E, Rocha E. Development and
clinical application of a new ELISA assay to determine plasmin-alpha2-antiplasmin complexes in plasma. Br J Haematol 1996; 92: 979
85.
15 Waterstone M, Bewley S, Wolfe C. Incidence and predictors of severe
obstetric morbidity: casecontrol study. BMJ 2001; 322: 108993.
16 Duthie SJ, Ven D, Yung GL, Guang DZ, Chan SY, Ma HK. Discrepancy between laboratory determination and visual estimation of
blood loss during normal delivery. Eur J Obstet Gynecol Reprod Biol
1991; 38: 11924.
17 Reyal F, Dearges J, Luton D, Blot P, Oury JF, Sibony O. Severe
post-partum hemorrhage: descriptive study at the Robert-Debre
Hospital maternity ward [French]. J Gynecol Obstet Biol Reprod 2002;
31: 35864.
18 Combs CA, Murphy EL, Laros Jr RK. Factors associated with
postpartum hemorrhage with vaginal birth. Obstet Gynecol 1991; 77:
6976.
19 Bouvier-Colle MH, Ould El Joud D, Varnoux N, Gonet F, Alexander S, Bayoumeu F, Beaumont E, Fernandez H, Lansac J, Levy G,
Palot M. Evaluation of the quality of care for severe obstetrical haemorrhage in three French regions. BJOG 2001; 108: 898903.
20 Levi M, ten Cate H, van der Poll T, van Deventer SJ. Pathogenesis of
disseminated intravascular coagulation in sepsis. JAMA 1993; 270:
9759.
21 Macphail S, Talks K. Massive post-partum haemorrhage and management of disseminated intravascular coagulation. Curr Obstetr
Gynaecol 2004; 14: 12331.
22 Andersson T, Lorentzen B, Hogdahl H, Clausen T, Mowinckel MC,
Abildgaard U. Thrombin-inhibitor complexes in the blood during and
after delivery. Thromb Res 1996; 82: 10917.
23 Epiney M, Boehlen F, Boulvain M, Reber G, Antonelli E, Morales M,
Irion O, De Moerloose P. D-dimer levels during delivery and the
postpartum. J Thromb Haemost 2005; 3: 26871.
24 Simon L, Santi TM, Sacquin P, Hamza J. Pre-anaesthetic assessment
of coagulation abnormalities in obstetric patients: usefulness, timing
and clinical implications. Br J Anaesth 1997; 78: 67883.
25 Dempe CE, Zips S, Ergul H, Heene DL. The brin assay comparison
trial (FACT): correlation of soluble brin assays with D-dimer.
Thromb Haemost 2001; 86: 12049.
26 Sher G. Trasylol in the management of abruptio placentae with consumption coagulopathy and uterine inertia. J Reprod Med 1980; 25:
1138.