hemorragia por atonia uterina

Postpartum Hemorrhage Resulting FromUterine Atony After Vaginal DeliveryFactors Associated With Severity

Marine Driessen, MD, MPH, Marie-Hèlène Bouvier-Colle, PhD, Corinne Dupont, PhD,Babak Khoshnood, MD, PhD, Renè-Charles Rudigoz, MD, and Catherine Deneux-Tharaux, MD, PhD forthe Pithagore6 Group*

OBJECTIVE: To identify factors associated with severityof postpartum hemorrhage among characteristics ofwomen and their delivery, the components of initialpostpartum hemorrhage management, and the organiza-tional characteristics of maternity units.

METHODS: This population-based cohort study includedwomen with postpartum hemorrhage due to uterine atonyafter vaginal delivery in 106 French hospitals betweenDecember 2004 and November 2006 (N�4,550). Severepostpartum hemorrhage was defined by a peripartumchange in hemoglobin of 4 g/dL or more. A multivariablelogistic model was used to identify factors independentlyassociated with postpartum hemorrhage severity.

RESULTS: Severe postpartum hemorrhage occurred in 952women (20.9%). In women with postpartum hemorrhage,factors independently associated with severity were: primi-parity; previous postpartum hemorrhage; previous cesarean

delivery; cervical ripening; prolonged labor; and episiot-omy; and delay in initial care for postpartum hemorrhage.Also associated with severity was 1) administration of oxy-tocin more than 10 minutes after postpartum hemorrhagediagnosis: 10–20 minutes after, proportion with severepostpartum hemorrhage 24.6% compared with 20.5%, ad-justed OR 1.38, 95% CI 1.03–1.85; more than 20 minutesafter, 31.8% compared with 20.5%, adjusted OR 1.86, CI1.45–2.38; 2) manual examination of the uterine cavity morethan 20 minutes after (proportion with severe postpartumhemorrhage 28.2% versus 20.7%, adjusted OR 1.83, 95% CI1.42–2.35); 3) call for additional assistance more than 10minutes after (proportion with severe postpartum hemor-rhage 29.8% versus 24.8%, adjusted OR 1.61, 95% CI1.23–2.12 for an obstetrician, and 35.1% compared with29.9%, adjusted OR 1.51, 95% CI 1.14–2.00 for an anesthesi-ologist); 4) and delivery in a public non-university hospital.Epidural analgesia was found to be a protective factor againstsevere blood loss in women with postpartum hemorrhage.

CONCLUSION: Aspects of labor, delivery, and theirmanagement; delay in initial care; and place of deliveryare independent risk factors for severe blood loss inwomen with postpartum hemorrhage caused by atony.(Obstet Gynecol 2011;117:21–31)DOI: 10.1097/AOG.0b013e318202c845

LEVEL OF EVIDENCE: II

Postpartum hemorrhage remains the leading causeof maternal mortality worldwide and the main

component of severe maternal morbidity in Westerncountries.1–4 Most postpartum hemorrhages are theresult of uterine atony. Although pharmacologic pre-vention of uterine atony in the third stage of laborsignificantly decreases the incidence of postpartumhemorrhage5 and is now recommended in interna-tional and national guidelines,6–11 reports from devel-oped countries indicate a recent rise in the postpar-tum hemorrhage rate.12–15 This increase is especially

See related editorial on page 3 and related article on page 14.

*For a list of Pithagore6 collaborators, see the Appendix online at http://links.lww.com/AOG/A211.

From INSERM U953 Epidemiological Research Unit on Perinatal Health andWomen’s and Children’s Health, UPMC Paris, Paris, France; and AurorePerinatal Network, Hopital de la Croix Rousse, Hospices Civils de Lyon, Lyon1 University, Lyon, France.

Funded by the French Ministry of Health under its Clinical Research HospitalProgram (contract no. 27-35). Dr. Driessen was supported by a student grantfrom the Fondation pour la Recherche Medicale.

The authors thank the staff from the participating maternity units for identifyingpostpartum hemorrhage cases, and the Fondation pour la Recherche Medicale forits financial support.

Corresponding author: Catherine Deneux-Tharaux, MD, PhD, INSERMU953, Batiment de recherche, Hopital Tenon, 4 rue de la Chine, 75020 Paris,France; e-mail: [email protected].

Financial DisclosureThe authors did not report any potential conflicts of interest.

© 2010 by The American College of Obstetricians and Gynecologists. Publishedby Lippincott Williams & Wilkins.ISSN: 0029-7844/10

VOL. 117, NO. 1, JANUARY 2011 OBSTETRICS & GYNECOLOGY 21

http://links.lww.com/AOG/A211

http://links.lww.com/AOG/A211

troubling because severe postpartum hemorrhage,even when not fatal, jeopardizes the woman’s fertility,exposes her to the risks of transfusion and intensivecare, and incurs costs. In this context, decreasing theprevalence of severe postpartum hemorrhage consti-tutes a major current obstetric challenge.

The likelihood of a continuum of morbidity be-tween simple and severe postpartum hemorrhagemakes the identification of factors that modulate thecourse of postpartum hemorrhage from excessivebleeding to severe hemorrhage an important ap-proach for increasing our understanding of thewomen and situations most at risk of severe postpar-tum hemorrhage.

Two categories of explanatory factors can be con-sidered: the individual characteristics of women anddeliveries and factors related to medical care, that is,both the content of care and the organization of health-care services. Various characteristics of women anddeliveries have been reported to be risk factors forpostpartum hemorrhage,16–18 but whether they are asso-ciated with an increased risk of severe postpartumhemorrhage once early postpartum hemorrhage hasoccurred is not known. On the other hand, focusing onprevention requires identifying the potential risk factorsassociated with medical care because they are mostamenable to change. Clinical guidelines for manage-ment of early postpartum hemorrhage are based mainlyon expert consensus, a low level of evidence. Datadocumenting the components of initial care that signif-icantly influence the course of postpartum hemorrhagewould be useful, making it possible to define the mostrelevant recommendations and thus perhaps increasetheir translation into practice.

The Pithagore6 trial, because it ascertained allcases of postpartum hemorrhage in 106 French ma-ternity units during 1 year and collected detailed dataon them, provides unique data for studying the vari-ous factors modulating the continuum of severity inpostpartum hemorrhage-related maternal morbidity.

The aim of this study was to identify factorsassociated with postpartum hemorrhage severityamong characteristics of women and deliveries, com-ponents of initial postpartum hemorrhage manage-ment, and organizational characteristics of maternityunits in women with postpartum hemorrhage result-ing from uterine atony after vaginal delivery.

MATERIALS AND METHODSThe study population was a cohort of women withpostpartum hemorrhage selected from the Pithagore6trial population.

The Pithagore6 trial was a cluster randomizedcontrolled trial in 106 French maternity units operat-ing as six perinatal networks. The main objective ofthis trial was to evaluate a multifaceted educationalintervention for reducing the rate of severe postpar-tum hemorrhage. No significant difference in the rateof severe postpartum hemorrhage was found betweenthe two groups of hospitals (details of this trial avail-able elsewhere19).

A 1998 French statute aimed at optimizing theorganization of obstetric care made it mandatory forall maternity units to belong to a perinatal network20

organized around one or more Level III units (refer-ence centers with an onsite neonatal intensive careunit) and including units rated as Level I (no facilitiesfor nonroutine neonatal care) and II (with a neonatalcare unit), both public and private. The six perinatalnetworks involved in the Pithagore6 trial were thePerinat Centre network around Tours (23 units), thePort-Royal St Vincent de Paul network in Paris (22units), and the four networks of the Rhone-Alpesregion: the Aurore network around Lyon (33 units),the Savoie network around Chambery (14 units), theGrenoble network (five units), and the St-Etiennenetwork (nine units). The 106 Pithagore6 maternityunits represented 17% of all French maternity unitsand covered 20% of deliveries nationwide. Data werecollected from December 2004 through November2005 in the Aurore network and from December 2005through November 2006 in the other five. Postpartumhemorrhage was clinically assessed by the caregiversif the estimated postpartum blood loss was greaterthan 500 mL or defined by a peripartum change inhemoglobin (Hb) greater than 2 g/dL (consideredequivalent to the loss of more than 500 mL of blood).Prepartum Hb was collected as part of routine prena-tal care during the last weeks of pregnancy; postpar-tum Hb was the lowest Hb level found in the 3 daysafter delivery. Birth attendants in each unit prospec-tively identified all deliveries with postpartum hem-orrhage and reported them to the research team. Inaddition, a research assistant reviewed the deliverysuite log book of each unit monthly as well ascomputerized patient charts when available. For ev-ery delivery with a mention of postpartum hemor-rhage, uterine cavity examination, or manual removalof the placenta, the patient’s obstetric file was furtherchecked to verify the postpartum hemorrhage diag-nosis. During the 1-year data collection period, 9,365cases of postpartum hemorrhage (defined either byestimated blood loss or drop in Hb) occurred among146,876 deliveries in the 106 Pithagore6 units for atotal PPH incidence of 6.4% of deliveries.

22 Driessen et al Postpartum Hemorrhage Resulting From Atony OBSTETRICS & GYNECOLOGY

For the present analysis, a specific definition ofpostpartum hemorrhage cases was used. We excludedcases of postpartum hemorrhage in which no exces-sive bleeding was clinically identified and that wereidentified only by a decreased Hb level, because, bydefinition, these cases did not receive any specificcare for postpartum hemorrhage, and one majorobjective was to study the association between com-ponents of initial care for postpartum hemorrhageand the risk of severe postpartum hemorrhage withinthis cohort of postpartum hemorrhage. The cohortwas further restricted to postpartum hemorrhage re-sulting from uterine atony after vaginal delivery, amore homogeneous situation that is the leading causeof postpartum hemorrhage and the main target ofclinical guidelines. Finally, the study population in-cluded 4,550 women. Figure 1 shows the process ofselection of the study population.

Characteristics of the patient, pregnancy, labor,delivery, and postpartum hemorrhage managementwere collected on a case report form from the chart ofevery delivery with confirmed postpartum hemor-rhage. The procedures for postpartum hemorrhagemanagement were considered to have been per-formed only if they were specifically mentioned in thechart.

The outcome was severe postpartum hemor-rhage, defined by a peripartum change in Hb of 4g/dL or more (considered equivalent to the loss of1,000 mL or more of blood).

Three groups of potential risk factors for severepostpartum hemorrhage were examined: characteris-tics of the women and aspects of labor and deliverybefore postpartum hemorrhage; components of initialpostpartum hemorrhage management; and organiza-tional characteristics of the units. The individualpreexisting characteristics were as follows: age inyears at delivery in three categories: less than 25,25–35, greater than 35; body mass index (calculatedas weight (kg)/[height (m)]2) at conception in fourcategories: 18 or less, greater than 18–25, greater than25–30, and greater than 30; and parity and previouscesarean delivery, categorized as: primiparous, mul-tiparous without previous cesarean delivery, and mul-tiparous with previous cesarean delivery (one ormore). History of postpartum hemorrhage, multiplepregnancy, hydramnios, epidural analgesia, pro-longed labor (defined as an active phase of labor morethan 6 hours without expulsive efforts), prolongedexpulsive efforts (defined as a duration of pushingmore than 30 minutes), and prophylactic oxytocinafter delivery were analyzed as dummy variables.Onset of labor was categorized as spontaneous, in-duced by oxytocin, and induced cervical ripening.Gestational age at delivery in weeks was categorizedas preterm (less than 37), term (37–41), or posttermdelivery (more than 41). Episiotomy and deliverywere categorized as spontaneous delivery withoutepisiotomy, spontaneous delivery with episiotomy,operative delivery without episiotomy, and operativedelivery with episiotomy. Birth weight in grams wasstudied in three categories: less than 2,500, 2,501–3,999, and 4,000 or more. Data were missing for nomore than 3% of women for all variables, except bodymass index (13.2%) and prolonged expulsive efforts(14.3%); a specific missing data class was added forthem.

Four components of initial care for postpartumhemorrhage were studied. For all of them, the refer-ence category was the performance within 10 minutesof postpartum hemorrhage diagnosis as recom-mended by the 2004 French national guidelines.11

Oxytocin administration and manual examination ofthe uterine cavity were both classified as performed in10 minutes or less, more than 10–20 minutes, morethan 20 minutes, done but delay unknown, and notdone. The calls for assistance from a senior obstetri-cian and an anesthesiologist were classified as: present

DeliveriesN=146,781

Women with postpartumhemorrhage (estimated

blood loss, drop in hemoglobin, or both)

n=9,365 Women with postpartumhemorrhage assessed

only by hemoglobin measurement

n=2,705Women with clinicallyassessed postpartum

hemorrhagen=6,660

Women with postpartum hemorrhage due to uterine atony

n=5,228

Women with postpartum hemorrhage not due

to uterine atonyn=1,432

Women with postpartum hemorrhage after cesarean delivery

n=678Women with clinicallyassessed postpartum

hemorrhage due to uterineatony after vaginal delivery

n=4,550

Fig. 1. Study population.Driessen. Postpartum Hemorrhage Resulting From Atony. ObstetGynecol 2011.

VOL. 117, NO. 1, JANUARY 2011 Driessen et al Postpartum Hemorrhage Resulting From Atony 23

or called in 10 minutes or less, more than 10 minutes,called but delay unknown, and not called.

The organizational characteristics of the unitsincluded: status, classified as university public, otherpublic, or private; number of annual deliveries, cate-gorized as fewer than 1,500, 1,500–2,500, or morethan 2,500; level of neonatal care, categorized into 1,2, or 3; and 24-hour-a-day onsite presence of anobstetrician, and of an anesthesiologist, studied asdummy variables.

The characteristics of women, labor, delivery,and initial postpartum hemorrhage management weredescribed as proportions in all postpartum hemor-rhage deliveries meeting the study case definition.The percentage of postpartum hemorrhage deliveriesmeeting severity criteria was calculated overall and bypregnancy characteristics. The crude associations ofsevere postpartum hemorrhage with these variableswere tested with chi square statistics and quantifiedwith unadjusted odds ratios and their 95% confidenceintervals. Multivariable logistic regression modelingwas used to assess the independent effect of eachvariable. Given the hierarchical structure of our data,level 1: women, level 2: centers (“clusters”), we tookinto account the intraclass (or intracluster) correlationfor outcomes of women cared for at a given center byusing random-intercept hierarchical logistic regres-sion models.21 Such modeling provides a more accu-rate estimation of associations and makes it possible tostudy explanatory variables at both levels. In a firststep, a logistic regression analysis including all rele-vant characteristics of women, labor, and deliverybefore postpartum hemorrhage was performed todetermine whether these characteristics were inde-pendently associated with postpartum hemorrhageseverity. Then, separate multilevel models tested theassociation of each component of initial postpartumhemorrhage care with postpartum hemorrhage sever-ity after adjustment for the significant characteristicsof women, labor, and delivery. Finally, the associationof each organizational characteristic with postpartumhemorrhage severity was examined after adjustmentfor characteristics of women, labor, delivery, andcomponents of initial care.

Cases with one or more missing value among thecharacteristics of women, labor, and delivery werenot included in the multivariate analyses (n�151women, 3.3% of total). Cases with missing data for thetiming of procedures were included in a specificcategory “done but unknown delay,” Organizationalcharacteristics were available for all units.

Based on a sample size of 4,500 women withpostpartum hemorrhage delivery that met the study

definition and an expected 20% prevalence of severepostpartum hemorrhage in this group, we estimatedthat the power of the study would be more than 80%to detect a relative risk of 2.0 between exposed andunexposed women for variables with a prevalence of2% or more of deliveries and to detect a relative riskof 1.5 for variables with a prevalence of 6% or more ofdeliveries. Statistical analysis was performed usingStata 10 software. Approval for the study was ob-tained from the Sud Est III institutional review boardand from the French Data Protection Agency.

RESULTSAmong 4,550 women with postpartum hemorrhage inthe study population, 952 (20.9%) had severe postpar-tum hemorrhage. Table 1 reports the distributions ofthe characteristics of women, labor, and delivery inthe cohort of women with postpartum hemorrhageand their association with severe postpartum hemor-rhage. After adjustment for other individual potentialrisk factors, the risk of severe postpartum hemorrhagefor women with postpartum hemorrhage was signifi-cantly higher in primiparas, multiparas with previouscesarean delivery, women with previous postpartumhemorrhage, women who had induced cervical rip-ening, prolonged labor, episiotomy (for both sponta-neous and instrumental delivery), and women whoreceived prophylactic uterotonics. Epidural analgesiawas associated with a significantly reduced risk ofsevere postpartum hemorrhage.

The distribution of the components of initialpostpartum hemorrhage management in the cohort isshown in Table 2 as well as their crude associationswith severe postpartum hemorrhage.

Oxytocin was administered late or not at all to24.5% of women with postpartum hemorrhage, whotherefore did not receive the recommended care.Manual examination of the uterine cavity was inap-propriate (late or not done) for 33.2%. In this cohort,40.6% of women with postpartum hemorrhage weremanaged with no senior obstetrician called or presentand 63.2% with no anesthesiologist called or present.

Delayed care, compared with the recommendedmanagement, was associated with an increased risk ofsevere postpartum hemorrhage (Table 2), and theassociations remained significant when controlling forcharacteristics of women, labor, and delivery beforepostpartum hemorrhage (Table 3). After adjustmentfor preexisting factors, the risk of severe postpartumhemorrhage was 1.4 times higher in women whoreceived oxytocin between 10 and 20 minutes afterpostpartum hemorrhage diagnosis and 1.9 timeshigher when it was administered more than 20 min-


utes after diagnosis compared with those who re-ceived it within the first 10 minutes (Table 3, model1), and 1.8 times higher in women who had a manualexamination of the uterine cavity more than 20minutes after diagnosis compared with the first 10

minutes (Table 3, model 2). Similarly, a delayed callfor obstetric assistance was associated with a 1.6 timeshigher risk of severe postpartum hemorrhage com-pared with cases in which a senior obstetrician waspresent or called within 10 minutes (Table 3, model

Table 1. Characteristics of Women, Labor, and Delivery: Distribution in the Cohort of PostpartumHemorrhage and Risk of Severe Postpartum Hemorrhage, Univariable and MultivariableAnalyses

n %Proportion WithSevere PPH (%) OR* 95% CI aOR† 95% CI

Women and pregnancyAge (y)

Younger than 25 876 19.3 22.0 1.05 0.87–1.26 0.95 0.77–1.1725–35 2,970 65.3 21.3 1.00 — 1.00 —Older than 35 700 15.4 18.3 0.83 0.67–1.02 0.98 0.77–1.24

BMI (kg/m2)18 or less 215 5.4 24.2 1.18 0.86–1.64 1.10 0.78–1.55More than 18–25 2,864 72.5 21.1 1.00 — 1.00 —More than 25–30 595 15.1 18.7 0.85 0.68–1.06 0.85 0.67–1.08More than 30 275 7.0 19.3 0.89 0.65–1.20 0.89 0.63–1.24

Previous PPH 249 5.5 20.9 0.99 0.73–1.37 1.47 1.02–2.13Fibroma 33 0.7 21.2 1.02 0.44–2.35 0.74 0.29–1.98Hydramnios 44 1.0 25.0 1.26 0.64–2.51 0.94 0.42–2.11Parity

Primiparous 2,268 49.9 26.2 1.99 1.70–2.32 1.88 1.51–2.33Multiparous with no previous

cesarean delivery2,036 44.8 15.1 1.00 — 1.00 —

Multiparous with previouscesarean delivery

245 5.4 20.7 1.47 1.05–2.04 1.66 1.15–2.41

Multiple pregnancy 119 2.6 26.9 1.40 0.93–2.12 1.17 0.70–1.96Labor

Onset of laborSpontaneous 3,457 76.0 20.3 1.00 — 1.00 —Induction 571 12.5 21.2 1.06 0.86–1.32 1.20 0.93–1.55Induced cervical ripening 522 11.5 25.7 1.37 1.10–1.69 1.45 1.13–1.85

Epidural analgesia 3,552 78.2 20.2 0.83 0.70–0.98 0.53 0.43–0.67Prolonged labor 1,376 31.2 24.5 1.38 1.19–1.61 1.27 1.06–1.53Oxytocin during labor 3,029 66.7 21.7 1.15 0.98–1.34 1.04 0.85–1.28Prolonged expulsive efforts 534 13.7 27.0 1.45 1.18–1.79 0.97 0.77–1.24

DeliveryGestational age (wk)

Less than 37 218 4.8 23.9 1.25 0.90–1.72 1.14 0.70–1.8537–41 3,565 78.5 20.1 1.00 — 1.00 —More than 41 759 16.7 24.1 1.26 1.05–1.52 1.10 0.89–1.37

DeliverySpontaneous without episiotomy 2,444 53.7 16.3 1.00 — 1.00 —Spontaneous with episiotomy 1,230 27.0 25.2 1.73 1.46–2.05 1.55 1.27–2.87Operative without episiotomy 176 3.9 19.3 1.23 0.83–1.81 1.05 0.69–1.62Operative with episiotomy 698 15.4 30.1 2.21 1.82–2.69 1.70 1.33–2.18

Prophylactic uterotonics 2,486 54.6 21.7 1.11 0.96–1.28 1.22 1.03–1.43Birth weight (g)

2,500 or less 179 3.9 23.5 1.18 0.83–1.68 0.97 0.56–1.692,501–3,999 3,880 85.3 20.6 1.00 — 1.00 —4,000 or greater 486 10.7 22.4 1.11 0.89–1.40 1.21 0.95–1.58

Total 4,550 100.0 20.9 NR

PPH, postpartum hemorrhage; OR, odds ratio; CI, confidence interval; aOR, adjusted odds ratio; BMI, body mass index; NR, notrelevant.

* Simple logistic regression.† Multivariable logistic regression including all variables.


3). The same was true for a delayed call for ananesthesiologist (Table 3, model 4). Associations be-tween delayed management and severe postpartumhemorrhage remained when several components ofcare were included in the same model (Table 3,models 5 and 6), except for the obstetrician call.When all four components of care were included,only delayed administration of oxytocin remainedsignificantly associated with severe postpartum hem-orrhage (Table 3, model 7).

Women in the category “done but delay un-known” were at lower risk of severe postpartumhemorrhage than those for whom the procedure wasperformed within 10 minutes for oxytocin adminis-tration and manual examination (Table 3, models 1and 2).

The risk of severe postpartum hemorrhage waslower when the obstetrician or the anesthesiologistwas absent and not called than when they were calledpromptly (Table 3, models 3 and 4). We performed a

Table 2. Initial Management of Postpartum Hemorrhage and Characteristics of the Units: Distributionin the Cohort of Postpartum Hemorrhage and Risk of Severe Postpartum Hemorrhage,Univariable Analysis

n %Proportion WithSevere PPH (%) OR 95% CI

Initial management of PPH*Oxytocin administration

10 min or less 2,208 48.5 20.5 1.00 —More than 10–20 min 329 7.2 24.6 1.27 0.97–1.66More than 20 min 447 9.8 31.8 1.81 1.45–2.26Done but delay unknown 1,224 27.0 17.7 0.83 0.63–1.13Not done 342 7.5 17.8 0.8 0.70–1.00

Manual examination of the uterine cavity10 min or less 2,114 46.5 20.7 1.00 —More than 10–20 min 326 7.2 23.9 1.21 0.92–1.59More than 20 min 490 10.8 28.2 1.50 1.20–1.88Done but delay unknown 929 20.4 15.6 0.71 0.58–0.87Not done 691 15.2 22.3 1.10 0.89–1.35

Call for obstetricianPresent or call 10 min or less 2,050 45.1 24.8 1.00 —Call more than 10 min 362 8.0 29.8 1.29 1.01–1.65Called but delay unknown 294 6.5 24.8 1.00 0.76–1.33Not called, not present 1,844 40.6 14.3 0.50 0.43–0.6

Call for anesthesiologistPresent or call 10 min or less 999 22.0 29.9 1.00 —Call more than 10 min 356 7.8 35.1 1.27 0.98–1.64Called but delay unknown 318 7.0 28.3 0.92 0.70–1.22Not called, not present 2,877 63.2 15.2 0.42 0.35–0.50

Characteristics of the units†

StatusUniversity public 1,423 31.3 17.6 1.00 —Other public 2,219 48.8 23.8 1.45 1.09–1.92Private 908 19.9 19.2 1.07 0.77–1.50

Level of care1 1,369 30.1 21.4 0.97 0.76–1.232 2,219 48.8 21.1 1.00 —3 962 21.1 19.8 0.92 0.64–1.33

Number of deliveries annuallyFewer than 1,500 1,483 32.6 23.3 1.18 0.93–1.511,500–2,500 1,922 42.2 19.6 1.00 —More than 2,500 1,145 25.2 20.2 1.02 0.73–1.43

24-h presence of obstetrician 3,318 73.0 19.7 0.80 0.64–1.0024-h presence of anesthesiologist 4,084 89.8 20.8 1.07 0.78–1.47

Total 4,550 100.0 20.9 NR

PPH, postpartum hemorrhage; OR, odds ratio; CI, confidence interval; NR, not relevant.* Logistic regression.† Multilevel logistic regression.


secondary analysis restricted to the population ofpostpartum hemorrhage women who received sulpr-ostone (second-line pharmacologic treatment) to testthe hypothesis that an indication bias might explainthese associations given that midwives and juniordoctors manage the most minor cases of postpartumbleeding alone. In this population, the obstetrician“not called, not present” category was associated withan increased risk of severe postpartum hemorrhage(adjusted odds ratio 2.01; 95% confidence interval1.44–2.84); no significant association was found be-tween the presence of an anesthesiologist and the riskof severe postpartum hemorrhage (data not shown).

The distribution of hospital-of-birth characteris-tics among the postpartum hemorrhage deliveries isshown in Table 2. The risk of severe postpartumhemorrhage was 1.5 times higher for postpartumhemorrhage in nonteaching public hospitals com-pared with university hospitals, and this significantassociation remained after adjustment for the charac-teristics of the women, labor, delivery, and compo-nents of early postpartum hemorrhage management(Table 4). When we controlled for the characteristicsof women, labor, and delivery, severe postpartumhemorrhage was not significantly associated with theannual number of deliveries, the level of care, or theon-site presence of an obstetrician or an anesthesiol-ogist (Table 4).

DISCUSSIONThis study documents the factors that modulate thecourse from simple to severe postpartum hemor-rhage. Our results suggest that various specific char-acteristics are independent determinants of postpar-tum hemorrhage aggravation. These include thewoman’s obstetric history, aspects of delivery beforepostpartum hemorrhage, delay in initial care forpostpartum hemorrhage, and hospital status.

We found that several characteristics of thewoman and her pregnancy, previously described asrisk factors for postpartum hemorrhage,16–18 also areassociated with a higher risk of severity once postpar-tum hemorrhage has occurred. Although any postpar-tum hemorrhage requires immediate management,for women with a history of postpartum hemorrhageor cesarean delivery, those having a first newborn,those whose labor was managed with induced cervicalripening or was prolonged, and those who had anepisiotomy, excessive but not severe postpartumbleeding requires even more careful attention becausethey are at higher risk of severe hemorrhage. Inter-estingly, several of these characteristics—previous ce-sarean delivery, cervical ripening, and episiotomy—

are related to the management of labor and delivery,and the associations found here provide additionalevidence to be considered in balancing the risks andbenefits of those procedures. In our cohort of 4,550women with postpartum hemorrhage, 2,314 (51%)had at least one of these three characteristics; effortsto decrease the rate of these procedures may actuallyreduce the importance of this group and possibly theincidence of severe postpartum hemorrhage.

Interestingly, episiotomy is associated with ahigher risk of severe postpartum hemorrhage, al-though the study population was restricted to postpar-tum hemorrhage resulting from uterine atony withpostpartum hemorrhage resulting from bleeding epi-siotomy excluded. This finding suggests that the exis-tence of multiple sources of blood loss, even in aphysiological range, increases the risk of severe post-partum hemorrhage and reinforces the relevance ofpolicies for limited use of episiotomy at vaginal birth.

An unexpected result was the increased risk ofsevere postpartum hemorrhage in women who re-ceived preventive oxytocin as compared with womenwith postpartum hemorrhage who had no prophy-laxis. This may reflect an indication bias, prophylaxisbeing more likely in women with risk factors, al-though these risk factors were taken into account inour analysis. Alternatively, excessive postpartumbleeding occurring after and despite prevention maybe more likely to be severe, because, by definition,prophylactic oxytocin was not able to prevent it.Another possible explanation is that the surveillanceof postpartum blood loss may be less intense whenprophylaxis has been done, leading to a delayeddiagnosis.

Epidural anesthesia had a protective effect here.It has previously been discussed as a risk factor forpostpartum hemorrhage,17,22 presumably by lengthen-ing labor or negatively affecting the endogenousoxytocin level or both, but evidence for such an effectis weak so far. Our results suggest that regardless ofthe effect of the epidural on the occurrence of post-partum hemorrhage, women diagnosed with postpar-tum hemorrhage who already have an epidural are ata smaller risk of severe bleeding. This unexpectedresult illustrates the importance of analyzing the roleof risk factors at different levels of the continuum ofseverity. The presence of the epidural catheter likelyfacilitates immediate management of postpartumhemorrhage because some procedures such as exam-ination of the uterine cavity, manual removal of theplacenta, or instrumental examination of the vaginaand cervix are usually done under anesthesia. In-versely, the need for anesthesia may delay initial care


for postpartum hemorrhage and thus increase the riskof severe postpartum hemorrhage in women whodelivered without an epidural; in our study popula-tion, this group had a significantly higher proportionof women with no or delayed examination of theuterine cavity than did the women with epidurals.

Delay in initial postpartum hemorrhage care(manual examination of the uterine cavity, oxytocinadministration, and call for extra help) was associatedwith an increased risk of severe postpartum hemor-rhage. These results might appear expected or evenobvious. However, gathering evidence to supportclinical practice recommendations is the principle ofevidence-based medicine and an essential task, be-cause it increases their level of proof and thus theirlegitimacy, both factors that may improve their trans-lation into practice.23 The content of postpartumhemorrhage-related guidelines for the initial steps isvery similar in all countries. This study providesevidence to support the recommendations for imme-diate management of excessive bleeding.

The risk of severe blood loss is higher for womenwith a postpartum hemorrhage after vaginal birth inpublic nonuniversity hospitals compared with otherpublic university or private hospitals, and this excess

risk is not explained by characteristics of the women,their delivery, or the initial postpartum hemorrhagemanagement. We hypothesize that second-line treat-ment for postpartum hemorrhage may be inappropri-ate or delayed in these hospitals because of limitedhuman (eg, available staff, surgical skills of obstetri-cians) or material (eg, interventional radiology) re-sources. However, these further steps of postpartumhemorrhage management are less standardized andtheir appropriateness is more difficult to assess be-cause the corresponding content of guidelines is quiteimprecise. That is why they were not considered inthis study.

The design of the present study had severalstrengths. It was population-based, covering all ma-ternity units and consequently all deliveries in a givenarea. This feature ensures the external validity of itsresults. The prospective identification of deliverieswith postpartum hemorrhage and the characterizationof severe postpartum hemorrhage within the cohort ofwomen with identified postpartum hemorrhage pro-vided unbiased comparison groups with regard to thestudy objectives. The large number of units anddeliveries provided good power for studying theindependent role of multiple factors and allowed an

Table 3. Initial Management of Postpartum Hemorrhage and Risk of Severe Postpartum Hemorrhage,Multivariable Analysis*

Model 1 Model 2 Model 3

Oxytocin administration10 min or less 1.00More than 10–20 min 1.38 (1.03–1.85)More than 20 min 1.86 (1.45–2.38)Done but delay unknown 0.75 (0.61–0.91)Not done 0.82 (0.60–1.13)

Manual examination of uterine cavity10 min or less 1.00More than 10–20 min 1.30 (0.97–1.76)More than 20 min 1.83 (1.42–2.35)Done but delay unknown 0.62 (0.50–0.78)Not done 0.96 (0.76–1.21)

Call for obstetricianPresent, call 10 min or less 1.00Call more than 10 min 1.61 (1.23–2.12)Called but unknown delay 1.14 (0.84–1.56)Not called, not present 0.54 (0.44–0.65)

Call for anesthesiologistPresent, call 10 min or lessCall more than 10 minCalled but delay unknownNot called, not present

PPH, postpartum hemorrhage.Data are adjusted odds ratio (95% confidence interval).* Models 1–7: multilevel logistic regression models adjusted for characteristics of women before PPH: previous PPH, parity and

previous cesarean delivery, multiple pregnancy, onset of labor, epidural analgesia, prolonged labor, gestational age, prophylacticuterotonics, delivery and episiotomy, birth weight (n�4,399 PPH).


analysis that could take the hierarchical structure ofthe data into account and explore the role of factorsat the levels of both the women and the units. Finally,the definition of severe postpartum hemorrhage wasbased on change in Hb, a more objective criterionthan the clinically assessed volume of blood loss, orthe need for surgery, embolization, or transfusion, alldependent on practices likely to vary widely amongclinicians and centers.

Our definition of severe postpartum hemorrhagealso has some limitations. Peripartum change in Hbmay not always accurately reflect postpartum bloodloss. It may overestimate blood loss in women whoreceived large amounts of fluids, who could then bewrongly classified as severe postpartum hemorrhage;it may underestimate the total blood loss if notperformed after 48 hours of delivery. This is howeverunlikely to bias our conclusions, because the conse-quence would actually be an underestimation of thestrength of the associations we found with severepostpartum hemorrhage. Given our study objectivesand the constraints of our data, the definition ofsevere postpartum hemorrhage by a maximumchange in Hb appeared as the least biased option.

Selection by indication bias is common in obser-vational studies assessing the role of procedures or

treatments on health outcomes. In the present study,some women with postpartum hemorrhage may havereceived more appropriate immediate managementbecause their bleeding was considered at high risk ofimmediate aggravation. Conversely, in women withexcessive bleeding after delivery but not consideredto be at risk for heavy blood loss, delay in adequatemanagement is more likely. The effect of this bias is toattenuate the negative effect of inadequate care. Theactual associations between delayed initial care andsevere postpartum hemorrhage may therefore bestronger than we found here. As mentioned, thisbias probably also explains the apparent lower riskof severe postpartum hemorrhage in cases in whichan obstetrician or anesthesiologist was not calledpromptly.

We cannot exclude the possibility that someprocedures were performed but not recorded in themedical files, although this seems unlikely for phar-macologic treatments such as oxytocin or invasiveexaminations such as examination of the uterus. Therelatively high proportion of missing data for thetiming of oxytocin administration and manual exam-ination shows that the quality of data recording inobstetrics files needs to improve. We found that thewomen with postpartum hemorrhage for whom these

Model 4 Model 5 Model 6 Model 7

1.00 1.00 1.001.39 (1.02–1.89) 1.38 (1.01–1.88) 1.33 (0.97–1.83)1.63 (1.26–2.11) 1.60 (1.23–2.08) 1.49 (1.14–1.94)0.95 (0.74–1.21) 0.97 (0.76–1.25) 1.00 (0.78–1.30)0.96 (0.68–1.33) 0.97 (0.69–1.36) 1.02 (0.72–1.43)

1.00 1.00 1.001.12 (0.82–1.55) 1.04 (0.75–1.43) 0.87 (0.62–1.20)1.60 (1.23–2.10) 1.35 (1.01–1.80) 1.05 (0.78–1.41)0.69 (0.52–0.92) 0.67 (0.50–0.89) 0.68 (0.50–0.91)1.01 (0.79–1.27) 1.10 (0.86–1.40) 1.30 (1.01–1.66)

1.00 1.001.29 (0.96–1.73) 1.10 (0.80–1.52)1.32 (0.96–1.82) 1.24 (0.87–1.76)0.56 (0.46–0.68) 0.70 (0.57–0.85)

1.00 1.001.51 (1.14–2.00) 1.33 (0.96–1.84)0.99 (0.73–1.35) 1.06 (0.75–1.50)0.37 (0.31–0.46) 0.43 (0.35–0.54)


two procedures were done, but at an unknown time,were at lower risk of severe postpartum hemorrhage.One possible explanation for this finding is that theprocedures were performed immediately after post-partum hemorrhage diagnosis and that the specifictime was not recorded because implicitly consideredsimultaneous with the diagnosis.

Identifying factors that influence the course ofpostpartum hemorrhage from simple to severe hasdirect potential implications for clinicians, especiallyfor factors related to care, which have been rarelyexplored so far. Our study shows that some aspects ofthe management of labor and delivery as well asdelayed initial care for postpartum hemorrhage andplace of delivery increase the risk of heavy postpar-tum bleeding caused by atony. More specifically, itprovides evidence suggesting that reducing the usecervical ripening, episiotomy, or cesarean delivery, inparticular in situations in which these interventions donot provide clear benefits as well improving therapidity of first care once postpartum hemorrhage hasoccurred, may reduce the incidence of severe post-partum hemorrhage.

REFERENCES1. Brace V, Penney G, Hall M. Quantifying severe maternal

morbidity: a Scottish population study. BJOG 2004;111:481–4.

2. Callaghan WM, Mackay AP, Berg CJ. Identification of severematernal morbidity during delivery hospitalizations, UnitedStates, 1991–2003. Am J Obstet Gynecol 2008;199:133.e1–8.

3. Zhang WH, Alexander S, Bouvier-Colle MH, Macfarlane A.Incidence of severe pre-eclampsia, postpartum haemorrhageand sepsis as a surrogate marker for severe maternal morbidityin a European population-based study: the MOMS-B survey.BJOG 2005;112:89–96.

4. Zwart JJ, Richters JM, Ory F, de Vries JI, Bloemenkamp KW,van Roosmalen J. Severe maternal morbidity during preg-nancy, delivery and puerperium in the Netherlands: a nation-wide population-based study of 371,000 pregnancies. BJOG2008;115:842–50.

5. Elbourne DR, Prendiville WJ, Carroli G, Wood J, McDonaldS. Prophylactic use of oxytocin in the third stage of labour. TheCochrane Database of Systematic Reviews 2001, Issue 4. Art.No.: CD001808. DOI: 10.1002/14651858.CD001808.

6. Department of health New South Wales Government Austra-lia. Postpartum haemorrhage—framework for prevention, earlyrecognition and management. 2005. Available at: www.health.nsw.gov.au/policies/PD/2005/PD_264.html. Accessedon May 31, 2010.

7. National Guideline Clearinghouse. Postpartum haemorrhage.American College of Obstetricians and Gynecologists. 2006.Available at: www.guideline.gov/summary/pdf.aspx?doc_id�10922&stat�1&string�. Accessed on May 31, 2010.

8. The Society of Obstetricians and Gynecologists of Canada(SOGC). Prevention and management of postpartum haemor-rhage. 2000. Available at: www.sogc.org/guidelines/public/88E-CPG-April2000.pdf. Accessed on May 31, 2010.

9. World Health Organization. WHO recommendations for theprevention of postpartum haemorrhage. Geneva (Switzerland):World Health Organization Department of Making PregnancySafer; 2007. Available at: www.who.int/making_pregnancy_safer/publications/WHORecommendationsforPPHaemorrhage.pdf. Accessed on May 31, 2010.

10. Lalonde A, Daviss BA, Acosta A, Herschderfer K. Postpartumhemorrhage today: ICM/FIGO initiative 2004–2006. Int JGynaecol Obstet 2006;94:243–53.

11. Goffinet F, Mercier F, Teyssier V, Pierre F, Dreyfus M,Mignon A, et al. Postpartum haemorrhage: recommendationsfor clinical practice by the CNGOF (December 2004) [inFrench]. Gynecol Obstet Fertil 2005;33:268–74.

12. Callaghan WM, Kuklina EV, Berg CJ. Trends in postpartumhemorrhage: United States, 1994–2006. Am J Obstet Gynecol2010;202:353.e1–6.

13. Joseph KS, Rouleau J, Kramer MS, Young DC, Liston RM,Baskett TF. Investigation of an increase in postpartum haem-orrhage in Canada. BJOG 2007;114:751–9.

14. Knight M, Callaghan WM, Berg C, Alexander S, Bouvier-Colle MH, Ford JB, et al. Trends in postpartum hemorrhage inhigh resource countries: a review and recommendations fromthe International Postpartum Hemorrhage CollaborativeGroup. BMC Pregnancy Childbirth 2009;9:55.

15. Roberts CL, Ford JB, Algert CS, Bell JC, Simpson JM, MorrisJM. Trends in adverse maternal outcomes during childbirth: apopulation-based study of severe maternal morbidity. BMCPregnancy Childbirth 2009;9:7.

16. Al-Zirqi I, Vangen S, Forsen L, Stray-Pedersen B. Prevalenceand risk factors of severe obstetric haemorrhage. BJOG 2008;115:1265–72.

17. Combs CA, Murphy EL, Laros RK Jr. Factors associated withpostpartum hemorrhage with vaginal birth. Obstet Gynecol1991;77:69–76.

Table 4. Characteristics of the Units and Risk ofSevere Postpartum Hemorrhage,Multivariable Analysis*

VariableAdjusted

OR 95% CI

StatusUniversity public 1.00 —Other public 1.46 1.10–1.87Private 1.00 0.73–1.37

Level of care1 0.92 0.72–1.172 1.00 —3 0.94 0.66–1.34

Number of deliveries annuallyFewer than 1,500 1.11 0.81–1.531,500–2,500 1.00 —More than 2,500 1.06 0.76–1.46

24-h presence of obstetrician 0.83 0.66–1.0424-h presence of anesthesiologist 1.05 0.78–1.42

OR, odds ratio; CI, confidence interval.* Multilevel logistic regression adjusted for characteristics of

women before postpartum hemorrhage (PPH) (previous PPH,parity and previous cesarean delivery, multiple pregnancy,onset of labor, epidural analgesia, prolonged labor,gestational age, prophylactic uterotonics, delivery andepisiotomy, birth weight) and early management of PPH(oxytocin administration, manual examination of the uterinecavity) (n� 4,399 PPH).


18. Sosa CG, Althabe F, Belizan JM, Buekens P. Risk factors forpostpartum hemorrhage in vaginal deliveries in a Latin-Amer-ican population. Obstet Gynecol 2009;113:1313–9.

19. Deneux-Tharaux C, Dupont C, Colin C, Rabilloud M, TouzetS, Lansac J, et al. Multifaceted intervention to decrease the rateof severe postpartum haemorrhage—the PITHAGORE6 clus-ter-randomized controlled trial. BJOG 2010;117:1278–87.

20. French Ministry of Health. Decret 98-899. 1988. Available at:http://www.sante.gouv.fr/adm/dagpb/bo/1999/99 –31/a0312082.htm. Accessed on May 31, 2010.

21. Raudenbush SW, Bryk AS. Hierarchical linear models: appli-cations and data analysis methods. 2nd ed. Thousand Oaks(CA): Sage Publications; 2002.

22. Saunders NS, Paterson CM, Wadsworth J. Neonatal andmaternal morbidity in relation to the length of the second stageof labour. Br J Obstet Gynaecol 1992;99:381–5.

23. Davis DA, Taylor-Vaisey A. Translating guidelines into prac-tice. A systematic review of theoretic concepts, practical expe-rience and research evidence in the adoption of clinicalpractice guidelines. CMAJ 1997;157:408–16.


http://www.sante.gouv.fr/adm/dagpb/bo/1999/99�31/a0312082.htm

http://www.sante.gouv.fr/adm/dagpb/bo/1999/99�31/a0312082.htm

hemorragia por atonia uterina

Documents