To describe transfusion practices and anemia in women with postpartum hemorrhage (PPH), according to the clinical context.
To describe transfusion practices and anemia in women with postpartum hemorrhage (PPH), according to the clinical context.
Population-based cohort study.
A total of 106 French maternity units (146 781 deliveries, December 2004 to November 2006).
All women with PPH (n = 9365).
Description of the rate of red blood cell (RBC) transfusion in PPH overall and compared with transfusion guidelines.
Transfusion practices and postpartum anemia by mode of delivery and cause of PPH in women given RBCs within 12 h after PPH.
A total of 701 women received RBCs (0.48 ± 0.04% of all women and 7.5 ± 0.5% of women with PPH). Half the women with clinical PPH and hemoglobin lower than 7.0 g/dL received no RBCs. In the group with clinical PPH and transfusion within 12 h (n = 426), operative vaginal delivery was associated with a larger maximal hemoglobin drop, more frequent administration of fresh-frozen plasma (FFP) and pro-hemostatic agents [odds ratio (OR) 3.54, 95% confidence interval (95% CI) 1.12–11.18], transfusion of larger volumes of RBCs and FFP, a higher rate of massive RBCs transfusion (OR 5.22, 95% CI 2.12–12.82), and more frequent use of conservative surgery (OR 3.2, 95% CI 1.34–7.76), compared with spontaneous vaginal delivery.
The RBC transfusion for PPH was not given in a large proportion of women with very low hemoglobin levels despite guidelines to the contrary. Operative vaginal delivery is characterized by higher blood loss and more transfusions than spontaneous vaginal delivery.
fresh frozen plasma
red blood cell
Blood transfusion for postpartum hemorrhage is given to one of every 200 women in France. Red blood cell transfusion for postpartum hemorrhage was not administered in a large proportion of women with low hemoglobin level, despite guidelines to the contrary. Operative vaginal delivery is characterized by higher blood loss and more transfusions than spontaneous vaginal delivery.
Postpartum hemorrhage (PPH) is a common complication of delivery, the incidence of which has increased recently in several countries [1-4]. PPH is associated with substantial maternal morbidity and mortality . In high-resource countries, PPH is the main obstetric reason for intensive care unit admission, and the rates of severe adverse outcomes after PPH, such as hysterectomy, have also increased [2, 5]. Progression from moderate to severe hemorrhage is probably dependent not only on individual characteristics of women and on the delivery situation, but also on factors related to health care . Transfusion is an important part of PPH management, especially ongoing hemorrhage. The goals of appropriate blood product transfusion are to maintain circulating blood volume and tissue oxygenation and to prevent or reverse coagulopathy. The lack of access to blood products can result in death of the mother . Better knowledge of transfusion practices in PPH may help to understand the impact on maternal outcomes.
Several studies have evaluated transfusion practices in women with PPH [3, 8-15]. However, these studies either used a single-centre retrospective design [8-11], and therefore had limited external validity, or relied on hospital discharge databases, which were not designed for clinical research [3, 15]. Transfusion practices were also described in recent nationwide prospective observational studies, but only as secondary outcomes, and in women requiring invasive second-line treatments for PPH [12-14]. Information on transfusion practices is needed to identify situations associated with specific blood product needs and to understand the reasons for these specific needs. This information is also necessary to optimize health-care resource allocation and to improve PPH management.
Our objective was to describe transfusion practices and anemia in women with PPH in general and in subgroups by obstetric interventions in a large population-based cohort.
The source population was the cohort of women included in a cluster-randomized trial performed in 106 French maternity units grouped into six regional perinatal networks  and accounting for 20% of all deliveries in France. Among the 106 maternity units, 11% were in university hospitals, 56% were in non-teaching public hospitals, and 33% were in non-teaching private hospitals; 56% were level 1, 36% were level 2, and 8% were level 3. The annual number of deliveries was less than 500 in 14% of the maternity units, between 500 and 1500 in 46%, and greater than 1500 in 40%. Data were collected over 1 year in each unit during December 2004 to November 2006. The aim of the trial was to evaluate a multifaceted educational intervention for reducing the rate of severe PPH. As no significant differences in severe PPH rates were found between the groups , all participants were pooled in a single cohort of women with PPH.
In the trial, PPH was assessed clinically by the medical staff or defined as a greater than 2.0 g/dL decline in the hemoglobin level. The clinical definition of PPH was blood loss greater than 500 mL or excessive blood loss prompting manual removal of the placenta or examination of the uterine cavity (or both). The antenatal hemoglobin level was determined in routine antenatal care during the last few weeks before delivery. Postpartum hemoglobin was the lowest hemoglobin found within 3 days of delivery (nadir of hemoglobin), whether the woman had been transfused before or not. It was neither measured routinely, nor as part of the study protocol. Instead, the decision was left to the clinicians. Birth attendants in each unit identified all women with PPH and reported them to the research team. A research assistant reviewed the delivery-suite logbook of each unit monthly. For each woman with a note of PPH, uterine cavity examination, or manual removal of the placenta, the obstetric ward file was reviewed to verify the diagnosis of PPH. During the study period, among 146 781 deliveries, 9365 (6.4%) were complicated by PPH, including 6660 (71.1%) diagnosed clinically and 2705 (28.9%) diagnosed only on a hemoglobin decline (Figure 1).
Women with PPH defined by a hemoglobin decline, but with no clinical diagnosis of PPH, did not receive specific acute care for PPH and were excluded. Our analysis was therefore restricted to women with clinical PPH requiring red blood cell (RBC) transfusion within 12 h after the diagnosis (n = 426). Indeed, this situation indicates significant acute PPH, where transfusion is an essential part of the management and can be lifesaving.
We looked at compliance with 2002 national French transfusion guidelines  and 2004 French guidelines on PPH management . These guidelines indicate that RBCs should be transfused when the hemoglobin level is <7.0 g/dL, especially in case of acute anemia. The hemoglobin concentration should be interpreted according to blood loss and vital signs and should be kept between 7.0 and 10.0 g/dL as long as the hemorrhage continues. RBC transfusion is usually unnecessary when the hemoglobin level is >10.0 g/dL . French transfusion guidelines are available on the websites of the French Society of Anesthesiology and the National College of Obstetricians and Gynecologists.
The following baseline characteristics were recorded: maternal age in years, body mass index as weight (kg)/[height (m)]², primiparity, previous PPH, previous cesarean section and multiple pregnancy, all handled as binary variables; gestational age; and prenatal hemoglobin level (g/dL). We also recorded the characteristics of the delivery: epidural or spinal analgesia/anesthesia (binary variable); mode of delivery in four categories (spontaneous vaginal delivery, operative vaginal delivery, cesarean section before labor, cesarean section during labor); birthweight (g), and postpartum hemoglobin measurement (binary variable). PPH was documented using the time from delivery to PPH as a continuous variable. The cause of PPH was recorded using five categories: coagulation disorders, trauma, abnormal placenta insertion, uterine atony or retained tissues, and unidentified cause. In PPH due to multiple causes, only one cause was recorded, in the order reported above.
Transfusion was studied during the whole postpartum hospital stay. We evaluated the nature of the blood product [RBC, fresh frozen plasma (FFP), platelets] or blood-derived product (fibrinogen concentrates), use of the RBC + FFP + Platelet + Fibrinogen combination, and massive transfusion (10 or more RBC units), all studied as binary variables. The transfused volume of each blood product (in units) and the fibrinogen dose (g) administered were studied as continuous variables. In women who received both RBC and FFP, the FFP/RBC ratio was calculated and analyzed both as a continuous variable and as a categorical variable (FFP/RBC ratio ≥0.5 and FFP/RBC ratio <0.5). The administration of pro-hemostatic agents (recombinant activated factor VII, other synthetic coagulation factors, tranexamic acid, anti-thrombin III, aprotinin) was handled as a binary variable. Time from PPH diagnosis to initiation of RBC transfusion was analyzed as a continuous variable. Acute adverse events of transfusion were looked for specifically in the medical chart, where they were spontaneously reported.
Concerning blood loss severity, the nadir of hemoglobin and the greatest hemoglobin (g/dL) drop versus baseline were studied as continuous variables. Secondary disseminated intravascular coagulation, defined as a coagulation disorder not present before the diagnosis of PPH (platelet count less than 50 × 103 mm−3, or prothrombin time less than 50%, or combination of platelet count between 50 × 103 and 100 × 103 mm−3 and/or prothrombin time between 50% and 65% and/or fibrinogen level less than 1 g/L), was studied as a binary variable. Components of the second-line management of PPH were handled as binary variables; these components included arterial embolization, conservative surgery (vascular ligation and/or uterine suture), hysterectomy, and intensive care unit admission.
The rate of blood transfusion was calculated for all deliveries and for all PPH cases during the study period, overall and according to the mode of diagnosis of PPH and to the mode of delivery (vaginal or cesarean delivery). We calculated the rates of hemoglobin measurements and of RBC transfusion among women with hemoglobin levels <7.0 g/dL and among those with hemoglobin levels <6.0 g/dL.
The characteristics of the women, pregnancies, and deliveries were compared between the clinical PPH with early transfusion group and the other women with clinical PPH (not transfused or transfused more than 12 h after PPH diagnosis). In the clinical PPH with early transfusion group, we compared transfusion practices according to mode of delivery and according to cause of PPH, separately in the women with vaginal delivery and cesarean section.
Normality plots were constructed to assess normality of distribution of continuous data. Categorical variables were compared using the chi-squared test or Fisher's exact test as appropriate. For continuous variables, parametric tests (unpaired t-test or analysis of variance followed by Bartlett's test) or non-parametric tests (Mann–Whitney test or Kruskal–Wallis test) were used as appropriate. Analyses were performed using STATA v10.1 software (Stata Corporation, College Station, TX, USA).
The Sud Est III Institutional Review Board and the French Data Protection Authority approved this study (QH 04 2005). The ethics committee waived the requirement for informed consent.
The RBC transfusion rate for PPH was 0.48 ± 0.04% overall, 0.34 ± 0.03% after vaginal delivery (n = 117 606), and 1.03 ± 0.11% after cesarean delivery (n = 29 175). PPH with transfusion of at least four RBC units occurred in 0.17% of deliveries. The 701 recipients of RBC transfusions accounted for 7.5 ± 0.5% of women with PPH (n = 9365), 5.5% of women with PPH after vaginal birth and 14.4% of women with PPH after cesarean delivery. The RBC transfusion rate was 9.7 ± 0.7% (n = 647) among women with clinically diagnosed PPH (n = 6660) (Figure 1).
Transfusions of RBC were given to less than half of the women with clinical PPH and hemoglobin levels <7.0 g/dL and to three-quarters of women with clinical PPH and hemoglobin levels <6.0 g/dL (Table 1). Five transfusion-related adverse events were recorded. Only one was severe, with pulmonary edema requiring intensive care unit admission.
|Clinical PPH (n = 6660)||Clinical PPH and Hb <7.0 g/dL (n = 858)a||Clinical PPH and Hb <6.0 g/dL (n = 289)a|
|Postpartum Hb measurements, n (%)||5776 (86.7)||858 (100)||289 (100)|
|RBC transfusion, n (%)||647 (9.7)||423 (49.3)||219 (75.8)|
|Nadir of hemoglobin in case of RBC transfusion (g/dL), mean (SD)||6.6 (±1.4)||5.8 (±0.8)||5.2 (±0.7)|
|Number of RBC units transfused, median (IQR)||3 (2–5)||3 (2–5)||3 (2–6)|
Of the 647 women with clinical PPH who required RBC transfusions, 426 (65.8%) received RBCs within 12 hours of the diagnosis (clinical PPH with early transfusion group) and 157 (24.3%) received RBCs later. In the remaining 64 (9.9%) women, the time from PPH diagnosis to RBC transfusion was unknown. Women with early transfusion were significantly older and had significantly lower prenatal hemoglobin, higher prevalence of previous PPH, previous cesarean section, multiple pregnancy, cesarean delivery, as well as lower gestational age at delivery and lower birthweight than the other women with clinical PPH (Table 2).
|Population characteristics||PPH with early RBC transfusion (n = 426)a||Other clinical PPH (n = 170)b||p-value|
|Women and pregnancies|
|Maternal age (years), n (%)|
|<25||65 (15.3)||970 (15.7)||0.009|
|25–35||257 (60.3)||4062 (65.8)|
|>35||104 (24.4)||1137 (18.4)|
|Body mass index (kg/m2), n (%)|
|≤18||22 (5.2)||284 (4.6)||0.29|
|19–25||255 (59.9)||3837 (62.2)|
|26–30||55 (12.9)||848 (13.8)|
|>30||25 (5.9)||415 (6.7)|
|Primiparous, n (%)||173 (40.6)||3130 (50.7)||<0.001|
|Prior PPH, n (%)||29 (6.8)||287 (4.7)||0.04|
|Prior cesarean delivery, n (%)||70 (16.4)||554 (9.0)||<0.001|
|Multiple pregnancy, n (%)||33 (7.8)||216 (3.5)||<0.001|
|Prenatal Hb level (g/dL), mean (SD)||11.5 (±1.4)||12.0 (±1.2)||<0.001|
|Labor and delivery|
|Gestational age < 37 weeks, n (%)||59 (13.8)||394(6.4)||0.0001|
|Mode of delivery, n (%)|
|Vaginal delivery||231 (54.2)||5345 (86.6)||<0.001|
|Spontaneous vaginal delivery||170 (73.6)||4147 (77.6)|
|Operative vaginal delivery||61 (26.4)||1198 (22.4)|
|Cesarean delivery||195 (23.1)||824 (13.4)|
|Cesarean delivery before labor||109 (55.9)||439 (53.3)|
|Cesarean delivery during labor||86 (44.1)||385 (46.7)|
|Epidural or spinal analgesia/anesthesia, n (%)||336 (78.9)||4984 (80.8)||0.51|
|Birthweight (g), mean (SD)||3236 (±704)||3377 (±568)||<0.001|
|Time from delivery to PPH diagnosis, median (IQR)||12 min (2–45)||15 min (9–30)||0.08|
In women with clinical PPH and early transfusion (n = 426), more than half received a combination of blood products (Table 3). In women with PPH after operative vaginal delivery, the median volumes of RBC and FFP units transfused were larger compared with the women with spontaneous vaginal delivery (p = 0.001 for RBC and p = 0.004 for FFP). Overall, 11% of the women with PPH after a vaginal delivery received 10 or more RBC units, and receiving 10 or more RBC units was significantly more common after operative than after spontaneous vaginal delivery (p < 0.001, odds ratio 5.22, 95% confidence interval 2.12–12.82). The use of pro-hemostatic agents was significantly more common in the operative vaginal delivery group (p = 0.04 between spontaneous and operative vaginal delivery, odds ratio 3.54, 95% confidence interval 1.12–11.18), with five of the nine women who received recombinant activated factor VII included in this group.
|Total (n = 426)||Spontaneous vaginal delivery (n = 170; 40.0%)||Operative vaginal delivery (n = 61; 14.3%)||Cesarean delivery before labor (n = 109; 25.5%)||Cesarean delivery during labor (n = 86; 20.1%)||p-valueb|
|RBC only||168 (39.4)||65 (38.2)||17 (27.9)||46 (42.2)||40 (46.5)||0.13|
|FFP||248 (58.1)||102 (60.0)||44 (72.1)||59 (54.1)||43 (50.0)||0.04|
|Fibrinogen||83 (19.5)||31 (18.2)||12 (19.7)||23 (21.1)||17 (19.8)||0.95|
|Platelets||52 (12.2)||18 (10.6)||13 (21.3)||15 (13.8)||6 (7.0)||0.06|
|RBC + FFP + Platelets + Fibrinogen||32 (7.5)||14 (8.2)||8 (13.1)||7 (6.4)||3 (3.5)||0.17|
|Median transfused quantity (IQR)|
|RBC (units)||3 (2–6)||3 (2–5)||4 (3–9)||3 (2–6)||4 (2–5)||0.01|
|FFP (units)||4 (2–6)||3 (2–4)||4 (2–6)||4 (3–6)||3 (2–4)||0.004|
|Fibrinogen (g)||3 (3–4.5)||3 (1.5–4.5)||3 (3–7.5)||4 (3–5.5)||3 (2–4.5)||0.37|
|Platelets (units)||1 (1,2)||1 (1,2)||1 (1,2)||1 (1,2)||2 (1,2)||0.39|
|≥10 RBC units||46 (10.8)||10 (5.9)||15 (24.6)||16 (14.7)||5 (5.8)||<0.001|
|Median (IQR)||0.8 (0.5–1)||0.7 (0.6–1)||0.8 (0.5–1)||0.8 (0.6–1)||0.6 (0.5–1)||0.38|
|FFP/RBC ≥0.5||209 (84.3)||85 (83.3)||39 (88.6)||52 (88.1)||33 (76.7)||0.36|
|Median time from PPH diagnosis to RBC administration, hours (IQR)||2 h 18 min (1 h 18 min to 3 h 54 min)||2 h 30 min (1 h 24 min to 4 h 18 min)||2 h 12 min (1 h 18 min to 3 h 48 min)||2 h 00 min (48 min to 3 h 36 min)||2 h 12 min (1 h 06 min to 3 h 48 min)||0.12|
|Use of pro-hemostatic agents||17 (4.0)||6 (3.5)||7 (11.5)||2 (1.8)||2 (2.3)||NA|
The maximal hemoglobin drop was significantly greater after operative than after spontaneous vaginal delivery (p = 0.003), as was the rate of conservative surgical procedures (p = 0.006, odds ratio 3.22, 95% confidence interval 1.34–7.76) (Table 4).
|Total (n = 426)||Spontaneous vaginal delivery (n = 170; 40.0%)||Operative vaginal delivery (n = 61; 14.3%)||Caesarean delivery before labor (n = 109; 25.5%)||Caesarean delivery during labor (n = 86; 20.1%)||p-valueb|
|Mean maximal Hb drop (g/dL) (SD)||4.7 (1.9)||4.6 (1.8)||5.6 (2.0)||4.1 (1.9)||4.8 (1.8)||<0.001|
|Secondary DIC||110 (25.8)||42 (24.7)||19 (31.2)||22 (20.2)||27 (31.4)||0.23|
|Embolization||106 (24.9)||49 (28.8)||19 (31.2)||22 (20.2)||16 (18.6)||0.12|
|Conservative surgery||58 (13.6)||12 (7.1)||12 (19.7)||23 (21.1)||11 (12.8)||0.004|
|Hysterectomy||64 (15.0)||23 (13.5)||13 (21.3)||23 (21.1)||5 (15.0)||0.01|
|ICU admission||180 (42.3)||64 (37.7)||33 (54.1)||45 (41.3)||38 (44.2)||0.16|
Uterine atony or retained tissues were the most common causes of PPH overall (47.4%) and in case of vaginal or cesarean delivery. The distribution of PPH causes significantly differed between vaginal and cesarean deliveries, abnormal placentation and coagulation disorders being more frequent in cesarean delivery. Among vaginal delivery, the distribution of PPH causes differed significantly between spontaneous and operative vaginal deliveries (p = 0.002). In operative vaginal delivery, trauma was the leading cause of PPH requiring transfusion within the first 12 h in 42.6%.
In PPH after vaginal delivery, coagulation disorders and abnormal placentation were associated with the highest rates of combined blood products transfusion and of massive transfusion, the largest blood product volumes, the greatest maximal hemoglobin drops, and the highest disseminated intravascular coagulation rate (see Supporting Information, Table S1). In PPH after cesarean delivery, these two causes were associated with larger RBC volumes, larger FFP volumes, and higher rates of massive transfusion (see Supporting Information, Table S2).
This study shows that one woman in every 200 who delivered, received an RBC transfusion for PPH in France. Omission of RBC transfusion for PPH, in contradiction of French transfusion guidelines, was observed in a large proportion of women with low hemoglobin level. In women with early RBC transfusion, operative vaginal delivery was associated with a higher rate of FFP transfusion, larger volumes of blood products, higher rate of massive transfusion, and greater severity of hemorrhage than spontaneous vaginal delivery.
The RBC transfusion rates we report are consistent with findings from several previous studies [2, 15, 19, 20]. In a population-based retrospective study from the USA, the overall RBC transfusion rate in the obstetric population was 0.48% (1994–2004) . Another retrospective population-based cohort study from Canada reported a national rate of PPH with RBC transfusion of 0.39%, without any significant change over time during the study period (1991–2004) . In a prospective cohort study performed in several hospitals from Uruguay and Argentina, the rate of transfusion for PPH after vaginal birth was 0.35% . A recent population-based Danish study from Holm et al.  reported an RBC transfusion rate of 1.92% of all deliveries. This higher rate may to some extent be a result of the inclusion of all cases of RBC transfusion given within 7 days of delivery—whether related to PPH or to other causes of postpartum anemia, or a different blood transfusion strategy.
Our data have documented omission of RBC transfusion in a significant proportion of women with PPH whose hemoglobin levels were lower than the recommended trigger. These results suggest that under-transfusion may exist in this context. Moreover, our results suggest that women with PPH in our study may have received smaller volumes of RBCs compared with those in the LEMMoN study, a recent nationwide study of severe acute maternal morbidity performed in the Netherlands . In this study, PPH with transfusion of at least four RBC units occurred for 0.6% of deliveries, three times more frequently than in our study. Failure to recognize severe hemorrhage may result in less frequent RBC transfusion. Omission of RBC transfusion for PPH with low hemoglobin level and small RBC volumes was also found in a previous study of maternal death secondary to PPH in France, where they certainly contributed to the fatal result . In contrast, a trend towards over-transfusion of women has been reported in the USA , the UK  and the Netherlands . That the RBC transfusion rate in our study was comparable to that in other countries despite omission of transfusion for PPH with low hemoglobin level may also be a result of more severe PPH in our population. Variation in management of labor and delivery, and delayed initiation of PPH treatment, as well as place of delivery, have been shown to influence the risk of severe blood loss in women with PPH . The low transfusion threshold in France may be ascribable to greater concern among physicians about the risk of maternal alloimmune reactions compared with the risk of acute anemia in women who usually have no history of cardiovascular disease. In addition, the human immunodeficiency virus epidemic probably made a major contribution to the persistent reluctance to use blood transfusion in France . The impacts of a low hemoglobin trigger for RBC transfusion on women's health has to be evaluated, to determine if transfusion guidelines should be followed or if the physicians' attitudes are justified.
In our study, almost 20% of the women with PPH and early RBC transfusion received fibrinogen concentrates. This proportion appears to be high considering the absence of scientific evidence for an efficacy of this treatment. The use of fibrinogen concentrates could have been influenced by results from experimental laboratory and animal studies that strongly suggested a potent hemostatic effect of fibrinogen substitution [24, 25], and from few observational studies. Virally inactivated fibrinogen concentrate offers rapid restoration of fibrinogen levels, with a small volume infusion and minimal preparation time . It is effective in treating patients with congenital hypofibrinogenemia , but there are very few reports of its use in obstetric hemorrhage. The 2004 French recommendations stated that the use of fibrinogen concentrates in PPH was controversial . Before recommending the early use of fibrinogen concentrates in PPH, prospective studies designed to assess its efficacy and tolerance, such as the FIB-TRIAL, are required .
Previous studies have shown that operative vaginal delivery is a risk factor for PPH [19, 29, 30]. We also found that operative vaginal delivery is a risk factor for transfusion among women with PPH. Hemorrhage after operative vaginal delivery required larger blood product volumes compared with spontaneous vaginal delivery, as previously reported by James et al., without any clear explanation . Decreased accuracy of visual blood loss assessment after operative vaginal delivery has been reported . Hence, delayed PPH management due to challenges in blood loss assessment, together with the high rate of secondary coagulopathy after operative vaginal delivery, as observed in our study, may contribute to increase blood loss severity in operative vaginal deliveries.
As previously reported, coagulation disorders and abnormal placenta insertion were associated with higher blood loss and more transfusion use than other causes of PPH [12, 14, 32]. These findings suggest that, in the event of operative vaginal delivery, coagulopathy, or abnormal placenta insertion, very close monitoring of postpartum blood loss and of its consequences is particularly needed. Point-of-care tests for hemoglobin level and coagulation may be useful tools in these contexts .
Due to the paucity of data, transfusion guidelines for women with PPH are often derived from data and recommendations for trauma patients. We found several similarities in transfusion practices between trauma and obstetric patients: as with trauma patients, the number of RBC units transfused per woman varied widely; the FFP/RBC ratio in trauma patients was similar to that in our study , suggesting that transfusion practices for acute hemorrhage with coagulopathy may be comparable in trauma and PPH . Nevertheless, most of the studies of transfusion practices in trauma patients found considerably higher transfusion rates (45–55%) [34, 36] and larger transfused volumes of RBC  than in our women with PPH. Morbidity and mortality rates due to hemorrhagic shock are higher in trauma patients than in women with PPH, with severe acute hemorrhage requiring aggressive transfusion therapy being more common among trauma patients than among women with PPH. Therefore, transfusion guidelines for trauma patients may be relevant only to women with excessive hemorrhage.
This study has several strengths. We used a large population-based cohort of women whose characteristics were comparable to those of the overall population of women delivering in France . Moreover, contemporary French guidelines on PPH management  and on transfusion in patients with acute hemorrhage  are similar to those from other high-resource countries [39, 40], making international comparisons of transfusion practices possible. In the present study, the prospective identification of PPH cases and review of delivery-suite logbooks and computerized patient charts probably ensured a high ascertainment rate. Finally, we collected detailed data on transfusion practices in PPH. Three previous prospective population-based studies described transfusion practices but were confined to women with severe PPH requiring invasive second-line treatments such as uterine compression suture, pelvic vessel ligature, interventional radiological techniques [12, 13], and/or hysterectomy [13, 14], which limited the total number of transfused women included. Moreover, in these studies, data on transfusion practices were limited, with no information on the FFP/RBC ratio, time from PPH diagnosis to transfusion, or use of fibrinogen, although these items may have a significant impact on maternal outcome [23, 41].
This study has some limitations. The hemoglobin trigger for RBC transfusion was not directly available in the collected data. However, it could be indirectly assessed through the postpartum nadir of hemoglobin, especially among women who were not transfused. Indeed, the absence of RBC transfusion in a great proportion of women with hemoglobin nadir <7 g/dL—the level recommended as a trigger for RBC transfusion—suggests that the actual hemoglobin trigger for RBC transfusion is frequently lower than recommended. Importantly, this trigger should be interpreted in conjunction with the clinical context. Hence, transfusion requirements differ between women with stable anemia and those with acute hemorrhage. In the event of active hemorrhage, clinical symptoms of acute anemia should be given more weight than the hemoglobin trigger. Consequently, using only the hemoglobin to select women for transfusion may be overly restrictive. Nevertheless, most of the studies evaluating the appropriateness of RBC transfusion both in women with stable anemia and in those with acute hemorrhage relied on the hemoglobin value [42, 43]. Designing the original trial might have changed transfusion practices in PPH. However, as no significant differences were found between the two trial arms of the original trial regarding the rates of severe PPH and of blood transfusion, transfusion practices were probably not influenced by the original trial.
Omission of RBC transfusion for PPH in contrast to transfusion guidelines was found in a large proportion of women with a low hemoglobin level. The poor compliance may be explained by poor integration of transfusion guidelines into everyday practice, fear of transfusion complications, traditions in transfusion policy, or inappropriate transfusion guidelines. We need to evaluate the impact of deviations from transfusion guidelines and compare RBC transfusion with alternatives such as intravenous iron supplementation. Transfusion practices vary by mode of delivery and cause of PPH. Operative vaginal delivery, coagulation disorders, and abnormal placentation were characterized by higher blood loss and more transfusions.
The authors thank staff from the participating maternity units. We are also grateful to Dr Antoinette Wolfe MD for helping to prepare the manuscript.
The Pithagore6 project was funded by the French Ministry of Health under its Clinical Research Hospital Program (contract no. 27–35). This study was supported by a doctoral grant from AXA Research Funds.