Between 2004 and 2006, the incidence of postpartum hemorrhage treated with massive transfusion was notably high in the Netherlands (91 per 100,000 deliveries). This is four times the incidence reported for the United Kingdom between 2012 and 2013 (23 per 100,000 deliveries), and one-and-a-half times the incidence reported for the state of New York between 1998 and 2007 (60 per 100,000 deliveries) [12, 18]. We found that the leading cause of PPH with massive transfusion was uterine atony. One quarter of all women receiving massive transfusion underwent hysterectomy to control bleeding.
The difference in the incidence of massive transfusion due to PPH between the Netherlands and the UK is remarkable. Whereas incidence of major obstetric hemorrhage has differed between various countries as a result of varying inclusion criteria [15, 19, 20], our study applied the same inclusion criteria for massive transfusion described by Green et al.  The difference in incidence between the Netherlands and the state of New York is also of note, particularly since Mhyre et al.  used a higher threshold for the number of red blood cell concentrates transfused to define massive transfusion (≥10 units) and included both antepartum and postpartum hemorrhage.
A distinct difference between the national guidelines for the management of PPH between the Netherlands (Dutch Society of Obstetrics and Gynaecology (NVOG)) and the United Kingdom (Royal College of Obstetricians and Gynaecology (RCOG)) is that the RCOG specifically recommends that ‘surgical interventions should be initiated sooner rather than later’. Both guidelines are inconclusive concerning the administration of blood products; NVOG (see Additional file 2 for a summary of the NVOG PPH guideline represented as a chart ) recommends not to deviate from the local guidelines of the hospital, while RCOG states that the decision to provide blood transfusion ‘should be based on both clinical and hematological assessment’ [9, 10, 21]. Furthermore, it is noteworthy that the median estimated blood loss in the study from Green et al. was 6 L (4.5–8.0 L) versus 4.5 L (3.3–6 L) in our cohort, whilst the massive transfusion rate was four times higher in the Netherlands . This may suggest that the difference in transfusion rate is due to differences in transfusion policy, which would emphasize the need for uniform guidelines .
During the study period, there were 358,874 deliveries in the Netherlands and 145,703 deliveries (40.6%) were under the responsibility of a primary care giver, making the risk of massive transfusion due to PPH 13 per 100.000 deliveries in midwifery care. Comparison of women requiring massive transfusion due to PPH with the general pregnant population in the Netherlands showed that women requiring massive transfusion had a multiple pregnancy in 11% of all cases versus 1.7% in the general population , suffered from preeclampsia in 17% of all cases versus 4% in the general population , had labour induced in 31% of all cases versus 12.5% in the general population , had a preterm delivery in 26% of all cases versus 5.8% in the general population  and delivered by cesarean section in 46% in all cases versus 13% in the general population . These characteristics are known risk factors of PPH and highlight that the management of postpartum hemorrhage should not only be focused on treatment, but on prevention as well [24, 25].
Uterine atony was the most frequent cause of postpartum hemorrhage as is consistent with the literature [12, 26, 27]. Atony was also the commonest cause of PPH in home deliveries . In elective cesarean sections the leading cause of massive transfusion due to PPH was placenta previa. Green et al. reported placenta accreta as the most frequent cause of PPH in women delivering by elective cesarean section, while Dupont et al. in France found that uterine atony remained the main cause of PPH regardless of mode of birth [12, 27]. The higher percentage of laparotomies performed after cesarean section is consistent with previous findings from the LEMMoN-cohort that the risk of postpartum laparotomy was more than 16 times higher in women who delivered by cesarean section compared to those who delivered vaginally .
As a last resort to arrest heavy bleeding, a quarter of all women underwent hysterectomy. This percentage is considerably lower than reported by Green et al. for the UK, where the overall rate of hysterectomy was 45%. A possible explanation for this difference could be the lower rates of previous cesarean deliveries; 66/327 (20%) in our study versus 73/181 (40%) in the . Two studies showed that the risk of peripartum hysterectomy increased with the number of previous cesarean deliveries [30, 31]. Another contributing factor could be the higher rate of embolization in our study, 72/327 (22%) versus 29/181 (16%) in the UK, and thereby preventing the need for hysterectomy. Furthermore, uterine rupture or an abnormally invasive placenta had the highest rates of hysterectomy compared to other causes. This is coherent with the recommendation of the Dutch Society of Obstetrics and Gynaecology guideline that states that hysterectomy should not be postponed if the cause of hemorrhage is related to a placenta accreta or ruptured uterus .
The maternal mortality rate of massive transfusion due to PPH in our study was low with 0.84 deaths per 100.000 maternities. This is comparable with the maternal mortality rate of PPH in the Netherlands reported by Schutte et al. between 1993 and 2005 (0.7 deaths per 100.000 maternities) . Nearly three-quarters of women who received massive transfusion were admitted to an ICU, and about one-third experienced morbidity. This high rate of morbidity is consistent with other studies [12, 26]. The rate of morbidity may be higher in low-income settings where not all treatment modalities are available or for Jehovah’s witnesses who refuse blood products [3, 15, 33].
A key strength is that our results were based on a nationwide cohort compromising all hospitals in the Netherlands with a maternity unit. Considering that PPH cases requiring massive transfusion must have been managed in one of these units, our results are population-based. Furthermore, our results are directly comparable to those of Green et al. who used the same definition for massive transfusion in their analysis of a UK cohort .
However, number of red blood cell concentrates as definition for massive transfusion remains an indicator with shortcomings as well, since it can be influenced by other factors, such as obstetrician’s decision-making. We also acknowledge that our data are from 2004 to 2006 and may not reflect the current situation. Since the incidence of PPH increased significantly throughout the years in many countries, but the incidence of obstetric blood transfusion in the Netherlands decreased , it is possible that the incidence of massive transfusion due to PPH may have reduced in recent years, but this is subject of further assessment. There may have been inclusion bias, since identification and management of cases may differ between obstetricians and hospitals. Underreporting is a concern, however, we have previously observed that there is a negative correlation between the rate of underreporting and the number of red blood cell concentrates transfused . Therefore, we would expect a low rate of underreporting. The considerable number of women without a known Hb-level at discharge is likely due to missing data, as a result of the design of the LEMMoN-database that did not specifically include Hb-level at discharge.
Nevertheless, this study makes clear that the incidence of PPH requiring massive transfusion was high in the Netherlands at that time compared to other countries and further research of contemporary obstetric cohorts is needed to allow for more up to date international comparisons of rates of transfusion and hemorrhage-related morbidity. Networks such as the International Network of Obstetric Surveillance Systems (INOSS) could facilitate such studies .