Obstetrics

Risk Factors for Postpartum Hemorrhage:

Can We Explain the Recent Temporal Increase?
Michael S. Kramer, MD,1,2 Mourad Dahhou, MSc,1 Danielle Vallerand,3
Robert Liston, MD,4,5,6 K.S. Joseph, MD, PhD4,5,6
1

Department of Pediatrics, Faculty of Medicine, McGill University, Montreal QC

Department of Epidemiology, Biostatistics and Occupational Health, Faculty of Medicine, McGill University, Montreal QC

Department of Obstetrics and Gynecology, Faculty of Medicine, McGill University, Montreal QC

Department of Obstetrics and Gynaecology, University of British Columbia, Vancouver BC

School of Population and Public Health, University of British Columbia, Vancouver BC

Children and Womens Health Centre of British Columbia, Vancouver BC

Abstract

Rsum

Objective: To assess risk factors for postpartum hemorrhage (PPH)

and the extent to which changes in those risk factors may
explain the rising incidence of PPH recently reported from
industrialized countries.

Objectif: valuer les facteurs de risque dhmorragie postpartum

(HPP) et la mesure dans laquelle les changements quont
connus ces facteurs de risque peuvent expliquer la hausse
de lincidence dHPP rcemment constate au sein des pays
industrialiss.

Methods: We carried out a hospital-based cohort study of

103726 consecutive deliveries from January 1, 1978, to
January 31, 2007, from the computerized medical records of
a tertiary-care university maternity hospital in Montreal. We
examined adjusted odds ratios for any PPH (estimated blood
loss >500 mL for vaginal deliveries, >1000 mL for Caesarean
sections), severe PPH (estimated blood loss 1500 mL), and
PPH accompanied by blood transfusion and/or hysterectomy.
Results: Major independent risk factors for PPH included primiparity,
prior Caesarean section, placenta previa or low-lying placenta,
marginal umbilical cord insertion in the placenta, transverse
lie, labour induction and augmentation, uterine or cervical
trauma at delivery, gestational age <32 weeks, and birth weight
4500g. An overall increase in rate of PPH over the study
period (OR 1.029; 95% CI 1.024 to 1.034 per year) disappeared
(OR 0.995; 95% CI 0.988 to 1.001 per year) after inclusion of
maternal age, parity, prior Caesarean section, labour induction
and augmentation, placenta previa or low-lying placenta, and
abnormal placenta, with most of the reduction attributable to
rises in previous Caesarean section and labour augmentation.
Conclusion: Labour induction, augmentation of labour, and prior
Caesarean section are significantly associated with the risk of
PPH, and their increase over the study period largely explains
the observed rise in PPH.
J Obstet Gynaecol Can 2011;33(8):810819
Key Words: Postpartum hemorrhage, maternal morbidity, labour
induction, Caesarean
Competing Interests: None declared.
Received on February 8, 2011
Accepted on March 15, 2011

810 l AUGUST JOGC AOT 2011

Mthodes: Nous avons men une tude de cohorte en milieu

hospitalier ayant port sur 103726accouchements conscutifs
(du 1er janvier 1978 au 31janvier 2007) partir des dossiers
mdicaux informatiss dun hpital de maternit universitaire de
soins tertiaires de Montral. Nous avons examin les rapports
de cotes corrigs en ce qui concerne toute HPP (perte sanguine
estime >500ml dans les cas daccouchement vaginal,
>1 000ml dans les cas de csarienne), lHPP grave (perte
sanguine estime 1500ml) et lHPP saccompagnant dune
transfusion sanguine et/ou dune hystrectomie.
Rsultats: Parmi les principaux facteurs de risque indpendants
dHPP, on trouvait la primiparit, le fait davoir dj subi une
csarienne, le placenta praevia ou le placenta en prsentation
basse, une insertion marginale du cordon ombilical dans le
placenta, une prsentation transversale, le dclenchement et
lacclration du travail, un traumatisme utrin ou cervical au
moment de laccouchement, un ge gestationnel <32semaines
et un poids de naissance 4500g. Une hausse globale du
taux dHPP au cours de la priode dtude (RC, 1,029; IC
95%, 1,024 1,034 par anne) sest estompe (RC, 0,995; IC
95%, 0,988 1,001 par anne) la suite de linclusion de lge
maternel, de la parit, du fait davoir dj subi une csarienne,
du dclenchement et de lacclration du travail, du placenta
praevia ou du placenta en prsentation basse et du placenta
anormal, cette rectification la baisse tant en grande partie
attribuable des hausses des taux de csarienne pralable et
dacclration du travail.
Conclusion: Les taux de dclenchement du travail, dacclration
du travail et de csarienne pralable sont significativement
associs au risque dHPP et leur hausse au cours de la priode
dtude explique en grande partie la hausse constate en ce qui
concerne le taux dHPP.

Risk Factors for Postpartum Hemorrhage: Can We Explain the Recent Temporal Increase?

INTRODUCTION

ostpartum hemorrhage (PPH) is a major cause of

maternal mortality and severe morbidity, particularly
in low-income countries.1,2 Recently, however, several
industrialized countries, including Australia, Canada, the
United Kingdom, and the United States, have reported an
increasing incidence of PPH.36
It is well recognized that appropriate obstetric management
(in particular, active management of the third stage of
labour) and access to blood transfusion and, if necessary,
hysterectomy, can prevent mortality and severe morbidity
once PPH occurs.7 Less is known, however, about
individual-level risk factors for the occurrence of PPH.8
Perhaps more importantly, the reasons for the recent
increase in its occurrence are unknown. An international
PPH collaborative group met and published a summary
of the recent evidence, as well as recommendations for
future surveillance and research.9 Reasons speculated for the
temporal increase include the rises in maternal obesity,10,11
previous Caesarean section,5,12,13 multiple pregnancy,5,8,12,14
and differences in the management of labour (including
induction and augmentation of labour and epidural
anaesthesia).8,9,15 Evidence supporting these possible
explanations is extremely limited, however.5,6 We took
advantage of a clinically rich hospital database comprising
deliveries over three decades to examine a number of
these hypothesized risk factors and their relationship to the
occurrence and severity of PPH, as well as to assess whether
changes in the prevalence of these risk factors could explain
the observed temporal changes.
METHODS

We carried out a hospital-based, historical cohort study

based on consecutive deliveries recorded in a perinatal
record system, the McGill Obstetrics and Neonatal
Database. This is a computerized archive that contains
maternal anthropometric data, detailed clinical and
sociodemographic information, obstetric history, and
pregnancy and neonatal outcomes for all births at
Montreals Royal Victoria Hospital since 1978. Maternal
and infant medical records are the raw data source for the
McGill Obstetrics and Neonatal Database and contain
the antenatal record and a range of neonatal and obstetric
forms that are completed during the birth hospitalization.
These data are routinely verified for accuracy and
completeness prior to computer entry.
The study sample was restricted to singleton deliveries that
occurred between January 1, 1978, and January 31, 2007.
Until March 2001, PPH was coded as yes or no, on the
basis of the delivering obstetricians estimate of blood loss
during and after delivery: >500mL for vaginal deliveries

and >1000mL for Caesarean section. A new version of the

McGill Obstetrics and Neonatal Database was initiated in
April 2001 and contains additional variables not previously
included in the database, including a quantitative estimate
of postpartum blood loss. Based on that estimate, we used
the same definition as previously for any PPH, but we also
defined severe postpartum hemorrhage as a postpartum
blood loss 1500mL. Because of the subjectivity and
variability inherent in estimating blood loss, we also analyzed
cases of PPH in which the delivering woman received a
blood transfusion and/or underwent hysterectomy during
her hospital stay.
Potential risk factors under study included maternal age,
parity, marital/cohabitation status, smoking during pregnancy,
history of infertility, previous history of spontaneous abortion,
previous Caesarean section, other uterine surgery, uterine
fibroids, pre-pregnancy BMI, pre-pregnancy hypertension,
pregnancy-induced hypertension (including preeclampsia),
pre-pregnancy or gestational diabetes or impaired glucose
tolerance, fetal presentation, abnormal placentation, amniotic
fluid volume, induction of labour, augmented labour, cervical
or vaginal trauma, delivery, placenta previa or low-lying
placenta (considered together), tocolysis, and infant birth
weight and gestational age.
Descriptive statistics for the potential risk factors in the
overall sample of deliveries over the study period were
analyzed, as were bivariate relationships between these
potential risk factors and both any PPH and severe PPH.
Multiple logistic regression analysis was used to assess the
independent associations between each of the risk factors
and the occurrence of any PPH and (for 2002 to 2007)
of severe PPH. To ensure that clustering of outcomes in
women with two or more deliveries at the study hospital
during the study period did not influence the results, we
also carried out a sensitivity analysis in which we randomly
selected one delivery for such women and repeated the
multiple logistic regression analysis.
Temporal trends were assessed graphically and by analyzing
the association between year of delivery and the occurrence
of any PPH. These trends were examined only for any PPH,
since data for severe PPH are available for only the last five
years of the study period. Data for 2007 were excluded from
the trend analysis, since only January deliveries had been
recorded in the database for that year. Potential risk factors
were also examined for temporal trends, using similar
methods. Those risk factors and covariates that varied over
time were then analyzed sequentially to examine their impact
on the overall temporal trend in the occurrence of PPH.
All statistical analyses were performed with SAS version 9.1
(SAS Institute, Cary NC).
AUGUST JOGC AOT 2011 l 811

Obstetrics

Table 1. Description of study cohort (n = 103 726)

Table 1. continued

Variable

Variable

Maternal age, years

< 20

Fetal presentation
2.3

20 to 34

80.1

35

17.6

Parity

47.1

35.4

2 to 4

16.7

0.8

Civil status

Married (legal or common-law)

83.5

Single, living alone

14.7

Separated, divorced, widowed

1.8

Cephalic

4.2

Transverse

0.4

Compound

0.3

Oblique

0.1

Other

0.1

Induced labour

23.2

Labour augmentation

30.4

Duration of labour >12 hours (n = 94 115)

24.6

Tocolytic therapy

12.2

Spinal

11 to 12

26.0

Epidural

13 to 15

26.3

Combined spinal/epidural

16

35.5
29.2

160 to 170

61.0
9.7

Pre-pregnancy BMI, kg/m2 (n = 52 387)

9.4

18.5 to < 25

68.2

25 to < 30

15.5

30

Other

12.7
3.0
12.4
1.1
70.8

Labour analgesia

< 160

< 18.5

6.6

Labour anaesthesia

< 11

Height, cm (n = 60 317)

94.9

Breech

None

Maternal education, years (n = 81849)

> 170

6.9

Smoking in pregnancy

16.6

Prior spontaneous abortion

18.4

None

97.6

Narcotic

2.2

Other

0.2

Delivery mode

Spontaneous vaginal

64.9

Forceps or vacuum

13.8

Caesarean section

21.3

Trauma at delivery

None

63.9

3.3

Uterine

0.2

Prior Caesarean section

10.8

Cervical

0.3

History of uterine surgery

0.4

Fibroids

1.9

Pre-pregnancy hypertension

1.3

< 32

1.9

Pregnancy-induced hypertension/preeclampsia

7.4

32 to 36

7.0

Pre-pregnancy or gestational diabetes, or

impaired glucose tolerance

5.4

History of infertility (n = 52 431)

93.2

Polyhydramnios

1.7

Oligohydramnios

5.0

Placenta previa or low-lying placenta

0.8

Gross placental pathology

Normal

87.9

Infarction

0.9

Marginal cord insertion

4.4

Circumvallate placenta

0.7

Chorangioma

2.7

Other abnormality

3.4

continued

812 l AUGUST JOGC AOT 2011

35.6

37

91.1

Birth weight, g

Amniotic fluid volume

Normal

Other (vaginal, perineal)

Gestational age, weeks

< 2500

6.3

2500 to 3999

82.8

4000 to 4499

9.5

4500

1.5

Risk Factors for Postpartum Hemorrhage: Can We Explain the Recent Temporal Increase?

RESULTS

A total of 103726 deliveries occurred at the study hospital

during the study period, among which 2346 (2.3%) had a
recorded PPH and 157 (0.15%) had a PPH accompanied
by a blood transfusion and/or hysterectomy. Of the
20445 deliveries since April 2001, 39 (0.19%) had severe
PPH (estimated blood loss 1500mL).
The study sample is summarized in Table 1 according to
the risk factors and covariates under study. The bivariate
(crude) relationship between each of the potential
risk factors and covariates and the occurrence of any
PPH and severe PPH is shown in Table 2. PPH was
significantly associated with older maternal age, unmarried
status, nulliparity or multiparity, maternal overweight/
obesity, noncephalic presentation (other than breech),
abnormal placenta, history of infertility, uterine surgery,
prior Caesarean section, pre-pregnancy hypertension,
pregnancy-induced hypertension, diabetes or impaired
glucose tolerance, polyhydramnios, placenta previa or
low-lying placenta, uterine or cervical trauma at delivery,
Caesarean section in the index birth, labour induction,
labour augmentation, tocolytic therapy, preterm birth, and
both low and high birth weight. Neither anaesthesia nor
analgesia type was associated with the risk of PPH, and
smoking was significantly (albeit very modestly) inversely
related to the risk of PPH. A graded increase in risk of any
PPH was observed with increasing maternal pre-pregnancy
BMI, although data on the latter variable were available for
only half the overall study sample. Most of the factors
associated with the risk of any PPH were also associated
with severe PPH, although the much smaller sample size
and low rate of occurrence (<10% of the rate for total
PPH) resulted in higher P values for severe PPH.
The results of the multiple logistic regression analysis for
PPH and severe PPH are shown in Table 3. Pre-pregnancy
and pregnancy-induced hypertension, transverse or
compound presentation, abnormal placenta, prior Caesarean
section, preterm birth, high birth weight, nulliparity,
multiparity, uterine or cervical trauma during delivery, and
placenta previa or low-lying placenta remained significantly
associated with the risk of PPH. Of note, labour induction,
augmentation of labour, and tocolytic therapy remained
significantly associated with the risk of PPH. Both Caesarean
section and operative vaginal delivery (forceps or vacuum) in
the index birth were also significantly associated with PPH
risk. The largest odds ratios were associated with transverse
lie, previous Caesarean section, early preterm birth, and
uterine and cervical trauma during delivery. Odds ratios
for severe PPH were highest for older maternal age, the
extremes of parity, pre-pregnancy hypertension, fibroids,

prior Caesarean section, prior uterine surgery, transverse lie,

early preterm birth, placenta previa and low-lying placenta,
and Caesarean section for the index birth. The results shown
in Table 3 do not include maternal pre-pregnancy BMI or
history of infertility, because a logistic model that included
those variables in addition to those shown in this Table
showed no significant association between PPH and either
of these variables, and the sample size was severely reduced
(n = 19 929) because of the large proportion of subjects
with missing values for those variables (Tables 1 and 2).
Logistic regression analysis for PPH accompanied by blood
transfusion and/or hysterectomy yielded similar results to
those for severe PPH, except for higher adjusted ORs for
trauma at delivery: 6.5 (95% CI 2.2 to 1.3) for uterine trauma
and 20.0 (95% CI 6.0 to 66.5) for cervical trauma.
The sensitivity analysis (inclusion of only one randomly
selected birth for mothers with two or more births
recorded in the database during the study period)
yielded very similar results to those presented above
for the entire sample of births. The sample size was
reduced from 103726 to 69121 in the sensitivity
analysis, and thus confidence intervals around the point
estimates were wider. Statistically significant risk factors
in the multiple logistic regression analysis included
primiparity (aOR 1.4; 95% CI 1.2 to 1.7), parity 2 to 4
(aOR 1.3; 95% CI 1.1 to 1.6), prior Caesarean section
(aOR 1.3; 95% CI 1.04 to 1.6), pregnancy-induced
hypertension (aOR 1.4; 95% CI 1.2 to 1.7), transverse
lie (aOR 3.1; 95% CI 1.9 to 5.2), marginal cord insertion
(aOR 1.4; 95% CI 1.1 to 1.7), other abnormal placentation
(aOR 1.8; 95% CI 1.4 to 2.2), birth weight 4000 to
4499g (aOR 1.6; 95% CI 1.5 to 2.1), birth weight
4500 g (aOR 2.6; 95% CI 1.9 to 3.5), gestational age <32
weeks (aOR 2.9; 95% CI 1.7 to 4.7), induced labour (aOR
1.2; 95% CI 1.1 to 1.4), uterine (aOR 4.6; 95% CI 2.6 to 8.1)
or cervical (aOR 3.1; 95% CI 1.4 to 6.7) trauma at delivery,
placenta previa or low-lying placenta (aOR 3.6; 95%
CI 2.6 to 5.0), forceps or vacuum delivery (aOR 1.6; 95%
CI 1.3 to 1.9), and Caesarean section for the index birth
(aOR 1.3; 95% CI 1.1 to 1.6). Risk factors that were
no longer significant in the sensitivity analysis were
augmentation of labour (aOR 1.1; 95% CI 1.0 to 1.3) and
tocolytic therapy (aOR 1.2; 95% CI 1.0 to 1.6).
The yearly rate (proportion of all deliveries) of PPH
from 1978 to 2006 is shown in the Figure. The rate rose,
albeit irregularly, from 1978 to 1994, declined from 1994
to 1997, and then remained stable until 2006. The rates
of several of the risk factors significantly associated with
PPH also rose during the study period, including induction
of labour, augmentation of labour, previous Caesarean
AUGUST JOGC AOT 2011 l 813

Obstetrics

Table 2. Variations in rates (%) of PPH and severe

PPH by risk factors and covariates

Table 2. continued
Variable

PPH
(n = 103 726)

Severe PPH
(n = 20 445)

PPH
(n = 103 726)

Severe PPH
(n = 20 445)

***

***

Yes

2.8

0.68

< 20

1.7

0.37

No

2.3

0.17

20 to 34

2.2

0.10

35

2.7

0.41

Variable
Maternal age, years

Parity

***

Fibroids

Pre-pregnancy hypertension

**

***

***

Yes

4.1

1.14

No

2.2

0.18

2.6

0.18

1.8

0.15

2 to 4

2.3

0.32

Yes

3.5

0.11

0.43

No

2.2

0.19

3.1

Civil status

**

Married (legal or common-law)

2.2

0.17

Single, living alone

2.5

0.29

Separated, divorced, widowed

3.0

0.30

Maternal education, years

(n = 81 849)
2.4

0.31

11 to 12

2.5

0.19

13 to 15

2.6

0.20

16

2.5

0.14

< 160

2.1

0.20

160 to 170

2.2

0.19

> 170

2.3

0.08

Height, cm

***

< 18.5

1.7

0.00

18.5 to < 25

2.1

0.27

25 to < 30

2.8

0.06

30

3.0

0.09

Smoking in pregnancy

Yes

2.0

0.31

No

2.3

0.18

Prior spontaneous abortion

Yes

3.0

0.29

No

2.7

0.15

History of infertility (n = 52 431)

Pre-pregnancy or gestational
diabetes, or impaired glucose
tolerance

***

***

***

Yes

3.2

0.20

No

2.2

0.19

Amniotic fluid volume

< 11

Pre-pregnancy BMI, kg/m2

(n = 52 387)

Pregnancy-induced hypertension/
preeclampsia

***

Normal

2.2

0.19

Polyhydramnios

3.6

0.31

Oligohydramnios

2.2

0.15

***

***

Yes

9.3

3.81

No

2.2

0.15

Placenta previa or low-lying

placenta

Gross placental pathology

***

Normal

2.1

0.16

Infarction

3.3

0.75

Marginal cord insertion

3.7

0.24

Circumvallate placenta

3.4

0.23

Chorangioma

2.5

0.26

Other abnormality

4.1

0.73

Fetal presentation

Cephalic

***

***

2.2

0.17

Breech

1.7

0.42

Transverse

10.2

3.75

Compound

4.7

0.00

Oblique

3.8

0.00

Other

1.7

0.00

Induced labour

***

Yes

4.8

0.24

Yes

2.6

0.10

No

2.6

0.19

No

2.1

0.23

**

***

Yes

3.0

0.56

Yes

2.8

0.11

No

2.5

0.14

No

2.5

0.23

**

***

Yes

4.2

2.01

12

2.1

No

2.3

0.18

> 12

2.8

Prior Caesarean section

History of uterine surgery

continued

814 l AUGUST JOGC AOT 2011

Labour augmentation

Duration of labour, hours

***
0.20
0.15
continued

Risk Factors for Postpartum Hemorrhage: Can We Explain the Recent Temporal Increase?

Table 2. continued
Variable
Tocolytic therapy

PPH
(n = 103 726)

Severe PPH
(n = 20 445)

**

Yes

2.8

0.24

No

2.2

0.19

None

2.5

0.39

Spinal

2.5

0.62

Epidural

2.1

0.05

Combined spinal/epidural

2.1

0.23

Other

2.2

0.00

None

2.3

0.20

Narcotic

2.1

0.00

Other

1.7

0.00

***

***

Spontaneous vaginal

2.0

0.05

Forceps or vacuum

2.4

0.08

Caesarean section

3.0

0.61

Labour anaesthesia

***

Labour analgesia

Delivery mode

Trauma at delivery

***

**

None

2.4

0.26

Uterine

9.6

1.08

Cervical

4.7

0.00

Other (vaginal, perineal)

2.4

0.04

***

***

< 32

3.9

1.04

32 to 36

2.5

0.36

37

2.2

0.16

Gestational age, weeks

Birth weight, g

***

**

< 2500

2.4

0.59

2500 to 3999

2.1

0.16

4000 to 4499

3.7

0.26

4500

4.2

0.00

* P < 0.05

lying placenta, and abnormal placenta. As shown in Table

4, the OR for the yearly term diminished at each step, and
after all of these seven risk factors were entered into the
model, it became non-significant (OR 0.995; 95% CI 0.988
to 1.001). This suggests that in the absence of temporal
trends in these seven risk factors, no overall rise in PPH
incidence would have been observed over the study period.
The largest decreases in the yearly effect were observed
with prior Caesarean section and labour augmentation,
suggesting that temporal increases in the rates of these two
obstetric interventions were responsible for most of the
crude rise in PPH.
When we restricted the trend analysis to the period of
observed increase in PPH incidence (see Figure), i.e.,
from 1978 to 1994, the yearly crude OR was 1.152 (95%
CI 1.138 to 1.167), but the inclusion of the abovementioned risk factors reduced the yearly effect to 1.108
(95% CI 1.089 to 1.128). In other words, the temporal
trend in these seven risk factors explained only 0.044/0.152
(29%) of the observed yearly increase from 1978 to
1994. When we examined the temporal decline in PPH
incidence observed from 1994 to 1997, the yearly decrease
(crude OR 0.741; 95% CI 0.678 to 0.810) was unaffected
by changes in the seven risk factors (aOR 0.737; 95%
CI 0.674 to 0.807).
A crude rise in severe PPH incidence was observed over
the short interval (2001 to 2006) in which this variable
was classifiable in the database (crude yearly OR 1.145;
95% CI 0.940 to 1.395), but the low overall rate of
occurrence (0.19% of deliveries) led to a highly imprecise
and non-significant estimate. Similarly, the low rate of
occurrence of PPH accompanied by blood transfusion
and/or hysterectomy (0.15%) showed no significant
increase over the entire study period (crude yearly
OR 1.009; 95% CI 0.990 to 1.029).

** P < 0.01
*** P < 0.001

section, placenta previa or low-lying placenta, abnormal

placentation, older maternal age, and parity 5.
Over the entire study period, we estimated the average
yearly effect (OR; 95% CI) estimated from univariable
logistic regression analysis as 1.029 (1.024 to 1.034). We
then sequentially entered the following additional variables
that were significantly positively associated with PPH in
the logistic regression analysis and that showed graphical
evidence of a temporal increase during the study period:
maternal age, parity, prior Caesarean section, labour
induction, augmentation of labour, placenta previa or low-

DISCUSSION

We studied a large number of potential risk factors for

postpartum hemorrhage, including many that are not
available in population-based birth registries or databases
of hospital discharges. These potentially important risk
factors include placenta previa, abnormal gross placental
pathology, fetal presentation, duration of labour, uterine
and cervical trauma at delivery, and uterine fibroids. We
confirmed previously reported associations with prior
Caesarean section, labour induction and augmentation, and
fetal macrosomia.8,15 In addition, we observed increased
risks associated with transverse lie, marginal umbilical
cord insertion in the placenta, placenta previa or lowAUGUST JOGC AOT 2011 l 815

Obstetrics

Table 3. Results of multiple logistic regression analyses

PPH
aOR (95% CI)

Severe PPH
aOR (95% CI)

< 20

0.8 (0.5 to 1.2)

4.7 (0.5 to 42.2)

20 to 34

1.0 (reference)

1.0 (reference)

35

1.0 (0.9 to 1.2)

2.9 (1.3 to 6.6)

1.5 (1.4 to 1.7)

2.4 (0.7 to 8.2)

1.0 (reference)

1.0 (reference)

2 to 4

1.3 (1.1 to 1.5)

1.9 (0.7 to 5.0)

1.4 (0.9 to 2.2)

4.5 (0.5 to 40.3)

Married (legal or common-law)

1.0 (reference)

1.0 (reference)

Single, living alone

1.0 (0.8 to 1.1)

2.1 (0.9 to 4.8)

Separated, divorced, widowed

Variable
Maternal age, years

Parity

Civil status

1.2 (0.9 to 1.7)

1.5 (0.2 to 12.1)

Smoking in pregnancy

1.0 (0.9 to 1.1)

1.7 (0.6 to 5.0)

Prior spontaneous abortion

1.1 (1.0 to 1.2)

1.3 (0.6 to 2.7)

Prior Caesarean section

1.4 (1.2 to 1.6)

3.1 (1.0 to 10.1)

History of uterine surgery

1.3 (0.8 to 2.3)

4.6 (1.2 to 17.7)

Fibroids

0.8 (0.6 to 1.1)

4.0 (1.4 to 11.4)

Pre-pregnancy hypertension

1.3 (0.9 to 1.8)

4.8 (1.2 to 19.5)

Pregnancy-induced hypertension/
preeclampsia

1.4 (1.2 to 1.6)

0.2 (0.02 to 1.7)

Pre-pregnancy or gestational diabetes, or

impaired glucose tolerance

1.1 (0.9 to 1.3)

0.3 (0.1 to 1.4)

Normal

1.0 (reference)

1.0 (reference)

Polyhydramnios

1.3 (1.0 to 1.7)

1.2 (0.2 to 5.9)

Oligohydramnios

0.8 (0.7 to 1.1)

0.2 (0.02 to 1.3)

3.5 (2.7 to 4.5)

17.1 (6.2 to 46.7)

Normal

1.0 (reference)

1.0 (reference)

Infarction

1.2 (0.8 to 1.8)

3.9 (0.4 to 34.0)

Marginal cord insertion

1.5 (1.2 to 1.8)

1.6 (0.5 to 5.0)

Circumvallate placenta

1.5 (1.0 to 2.3)

2.0 (0.3 to 16.1)

Chorangioma

0.9 (0.7 to 1.2)

1.1 (0.1 to 8.4)

Other abnormality

1.5 (1.3 to 1.9)

2.8 (0.8 to 9.0)

Cephalic

1.0 (reference)

reference

Breech

0.7 (0.5 to 0.9)

reference

Transverse

3.4 (2.3 to 5.1)

7.3 (1.8 to 30.0)*

Compound

1.9 (1.03 to 3.3)

reference

Oblique

1.2 (0.4 to 4.0)

reference

Other

Amniotic fluid volume

Placenta previa or low-lying placenta

Gross placental pathology

Fetal presentation

0.6 (0.1 to 2.5)

reference

Induced labour

1.2 (1.1 to 1.4)

0.8 (0.2 to 2.7)

Labour augmentation

1.2 (1.1 to 1.4)

0.7 (0.2 to 2.6)

Tocolytic therapy

1.3 (1.1 to 1.5)

0.4 (0.1 to 2.2)

continued

816 l AUGUST JOGC AOT 2011

Risk Factors for Postpartum Hemorrhage: Can We Explain the Recent Temporal Increase?

Table 3. continued
Variable

PPH
aOR (95% CI)

Severe PPH
aOR (95% CI)

Duration of labour > 12 hours

1.0 (0.9 to 1.2)

1.9 (0.7 to 5.6)

Spontaneous vaginal

1.0 (reference)

1.0 (reference)

Forceps or vacuum

1.5 (1.3 to 1.7)

1.9 (0.2 to 16.1)

Caesarean section

1.3 (1.1 to 1.5)

4.8 (1.4 to 16.0)

Delivery mode

Trauma at delivery

None

1.0 (reference)

1.0 (reference)

Uterine

3.8 (2.4 to 6.2)

0.7 (0.2 to 2.8)

Cervical

4.0 (2.3 to 7.0)

Other (vaginal, perineal)

1.1 (1.0 to 1.2)

Gestational age, weeks

< 32

2.9 (1.9 to 4.4)

32 to 36

1.3 (1.04 to 1.5)

1.7 (0.4 to 6.3)

37

1.0 (reference)

1.0 (reference)

7.3 (1.0 to 56.6)

Birth weight, g

< 2500

0.6 (0.4 to 0.8)

1.0 (0.2 to 5.4)

2500 to 3999

1.0 (reference)

1.0 (reference)

4000 to 4499

1.8 (1.6 to 2.1)

1.2 (0.4 to 3.6)

4500

2.2 (1.6 to 2.8)

* Because of insufficient numbers, transverse lie was compared to all other presentations.
Because of insufficient numbers, uterine, cervical, vaginal, and perineal trauma were combined into a
single category.
Because of insufficient numbers, birth weights 4000g were combined into a single category.

lying placenta, uterine or cervical trauma at delivery, and

gestational age <32 weeks. For the last six years of the
study (2001 to 2006), we were also able to study risk factors
associated with severe PPH, defined as postpartum blood
loss 1500 mL. Strong associations with severe PPH were
observed for maternal age 35 years, prior Caesarean
section, history of uterine surgery, uterine fibroids, prepregnancy hypertension, placenta previa or low-lying
placenta, transverse presentation, and gestational age <32
weeks. Similar associations were observed over the entire
study period for PPH accompanied by blood transfusion
and/or hysterectomy and were extremely strong with
uterine or cervical trauma at delivery.
We observed a temporal increase in the incidence of
PPH at the study hospital over the study period. The rise
was not continuous, however, decreasing between 1994
and 1997 and stabilizing (at a rate of approximately 2%
of deliveries) from 1997 to 2006. We observed temporal
increases in the rates of several of the important risk
factors for PPH, including prior Caesarean section, labour
induction, augmentation of labour, placenta previa or
low-lying placenta, and abnormal placenta. Once these
five risk factors were entered into a multiple logistic

regression model, along with maternal age and parity, the

crude increase in occurrence of PPH over the study period
disappeared, suggesting that in the absence of temporal
changes in these risk factors, the hospital would not have
experienced any rise in the occurrence of PPH. Previous
population-based studies from Canada5 and Australia6
have not succeeded in explaining temporal trends in PPH
incidence by concurrent trends in the incidence of risk
factors, perhaps because of their reliance on ICD-based
diagnoses and procedure codes and limited information
Table 4. Sequential logistic regression analysis of
temporal (yearly) trends in PPH
OR (95%)
Year only

1.029 (1.024 to 1.034)

Plus maternal age

1.028 (1.023 to 1.033)

Plus parity

1.027 (1.022 to 1.033)

Plus prior Caesarean section

1.006 (1.000 to 1.012)

Plus labour induction

1.005 (0.999 to 1.011)

Plus augmentation

0.998 (0.992 to 1.004)

Plus placenta previa or low-lying placenta

0.997 (0.991 to 1.004)

Plus abnormal placenta

0.995 (0.988 to 1.001)

AUGUST JOGC AOT 2011 l 817

Obstetrics

Temporal (yearly) trend in occurrence of any PPH, Royal Victoria Hospital,

19782006
6
5

Rate (%)

4
3
2
1
0
1978

Birth year

1982

1986

1990

on prior clinical history, details of labour and delivery, and

placental examination.
Although the benefit of routine post-term induction of
labour has been convincingly demonstrated,16 recent reports
have documented substantial increases in rates of induction
even at early term and late preterm gestations.1722 When
combined with reports of increased risk of amniotic fluid
embolism associated with labour induction,23,24 our findings
suggest that recent interventionist policies regarding
labour and delivery may have adverse consequences for
the mother in addition to those recently reported for the
newborn infant and for subsequent child development.2527
Several limitations of this study should be acknowledged.
Primary among these is the fact that the study is based
on deliveries from a single hospital. This is a tertiarycare referral hospital, and although the referral base did
not change over the study period, we have no way of
assessing whether more or different kinds of women at
high risk for PPH were delivered at the study hospital over
this period. Other selection factors might be operating
that could bias associations between risk factors and
PPH upwards or downwards, although many of the risk
factors we identified are consistent with those reported in
previously published studies.8 We have no data on placenta
accreta or percreta, which could have been used to better
understand the relationship between prior Caesarean
section or other uterine surgery and the occurrence of
PPH in subsequent pregnancies. Nor can we ensure the
validity of obstetricians estimates of blood loss during
or after delivery, or their uniformity over the prolonged
818 l AUGUST JOGC AOT 2011

1994

1998

2002

2006

study period. The similar results observed in our analysis

of PPH accompanied by blood transfusion and/or
hysterectomy, however, strengthen the validity of our
overall findings. Our inability to determine the temporal
sequence of risk factors and PPH is another limitation of
our data source. Thus the positive association of PPH with
Caesarean section for the index birth may reflect the use
of Caesarean section to control excessive bleeding, rather
than it being a cause of subsequent bleeding. We also have
no assurance that blood transfusions or hysterectomies
occurred after (rather than before) the PPH in all cases.
The temporal increase in some risk factors could be partly
explained by closer diagnostic surveillance, especially for
those diagnosed by ultrasound: placenta previa or lowlying placenta, fibroids, and polyhydramnios. As shown in
Table 4, however, placenta previa or low-lying placenta was
the only one of these factors whose rise had a perceptible
impact on the temporal increase in PPH, and that impact
was small. Finally, no observational study (including ours)
can adequately control for confounding by indication
for such obstetric interventions as labour induction and
augmentation. Thus uterine hypotonicity (of any or
unknown cause) may be an underlying cause of labour
induction and/or augmentation and may lead to PPH
independently of these obstetric interventions.
Strengths of our study include the large number of deliveries,
consistently extracted and coded risk factors and pregnancy
complications, and a study period that was long enough to
observe and assess risk factors for temporal trends.

Risk Factors for Postpartum Hemorrhage: Can We Explain the Recent Temporal Increase?

CONCLUSION

Our results suggest that the temporal increase in the

incidence of PPH, at least at the study hospital, was
attributable to rises in the incidence of several key risk
factors, including previous Caesarean section, labour
induction and augmentation, placenta previa, and abnormal
placentation. These results raise some troubling questions
about the consequences of current trends in obstetric
intervention, particularly the rises in rates of Caesarean
section, labour induction, and augmentation of labour.
The evidence suggests an urgent need for randomized
trials of these interventions to ensure that the benefits of
their increased use outweigh their harmful effects.
ACKNOWLEDGEMENTS

Our study was funded by a grant from the Canadian

Institutes of Health Research.
REFERENCES
1. World Health Organization. The world health report: 2005: make every
mother and child count. Geneva: WHO; 2005.
2. Minio AM, Heron MP, Murphy SL, Kochanek KD; Centers for Disease
Control and Prevention National Center for Health Statistics National
Vital Statistics System. Deaths: final data for 2004. Natl Vital Stat Rep
2007;55:1119.
3. Cameron CA, Roberts CL, Olive EC, Ford JB, Fischer WE. Trends in
postpartum haemorrhage. Aust N Z J Public Health 2006;30:1516.
4. Haynes K, Stone C, King J. Major morbidities associated with childbirth
in Victoria: obstetric haemorrhage and associated hysterectomy.
Melbourne: Public Health Group, Department of Human Services; 2004.
5. Joseph KS, Rouleau J, Kramer MS, Young DC, Liston RM, Baskett TF.
Investigation of an increase in postpartum haemorrhage in Canada.
BJOG 2007;114:75159.
6. Ford JB, Roberts CL, Simpson JM, Vaughan J, Cameron CA. Increased
postpartum hemorrhage rates in Australia. Int J Gynaecol Obstet
2007;98:23743.
7. Cabero Roura L, Keith LG. Post-partum haemorrhage: diagnosis,
prevention and management. J Matern Fetal Neonat Med 2009;22:3845.

Postpartum Hemorrhage Collaborative Group. BMC Pregnancy

Childbirth 2009;9:55.
10. Robinson HE, OConnell CM, Joseph KS, McLeod NL. Maternal
outcomes in pregnancies complicated by obesity. Obstet Gynecol
2005;106:135764.
11. Sebire NJ, Jolly M, Harris JP, Wadsworth J, Joffe M, Beard RW, et al.
Maternal obesity and pregnancy outcome: a study of 287,213 pregnancies
in London. Int J Obes 2001;25:117582.
12. Knight M, Kurinczuk JJ, Spark P, Brocklehurst P. Cesarean delivery and
peripartum hysterectomy. Obstet Gynecol 2008;111:97105.
13. Daltveit AK, Tollnes MC, Pihlstrm H, Irgens LM. Cesarean delivery
and subsequent pregnancies. Obstet Gynecol 2008;111:132734.
14. Cruikshank DP. Intrapartum management of twin gestations. Obstet
Gynecol 2007;109:116776.
15. Guerra GV, Cecatti JG, Souza JP, Faundes A, Morais SS, Gulmezoglu AM,
et al. Factors and outcomes associated with the induction of labour in
Latin America. BJOG 2009;116:176272.
16. Sanchez-Ramos L, Olivier F, Delke I, Kaunitz AM. Labor induction
versus expectant management for postterm pregnancies: a systematic
review with meta-analysis. Obstet Gynecol 2003;101:13128.
17. Zhang J, Yancey MK, Henderson CE. U.S. national trends in labor
induction, 19891998. J Reprod Med 2002;47:1204.
18. MacDorman MF, Mathews TJ, Martin JA, Malloy MH. Trends and
characteristics of induced labour in the United States, 198998.
Paediatr Perinat Epidemiol 2002;16:26373.
19. Russell SK. Trends in labor induction in the United States: is it true that
what goes up must come down? Birth 2004;31:14851.
20. Wing DA. Elective induction of labor in the USA. Curr Opin Obstet
Gynecol 2000;12:45762.
21. Rayburn WF, Zhang J. Rising rates of labor induction: present concerns
and future strategies. Obstet Gynecol 2002;100:1647.
22. Grobman WA. Elective induction: when? ever? Clin Obstet Gynecol
2007;50:53746.
23. Kramer MS, Rouleau J, Baskett TF, Joseph KS, for the Maternal Health
Study Group of the Canadian Perinatal Surveillance System. Amnioticfluid embolism and medical induction of labour: a retrospective,
population-based cohort study. Lancet 2006;368:14448.
24. Knight M, Tuffnell D, Brocklehurst P, Spark P, Kurinczuk JJ, on behalf
of the UK Obstetric Surveillance System. Incidence and risk factors for
amniotic-fluid embolism. Obstet Gynecol 2010;115:9107.
25. Zhang X, Kramer MS. Variations in mortality and morbidity by gestational
age among infants born at term. J Pediatr 2009;154:35862.

8. Al-Zirqi I, Vangen S, Forsen L, Stray-Pedersen B. Prevalence and risk

factors of severe obstetric haemorrhage. BJOG 2008;115:126572.

26. Yang S, Platt RW, Kramer MS. Variation in child cognitive ability
by week of gestation among healthy term births. Am J Epidemiol
2010;171:399406.

9. Knight M, Callaghan WM, Berg C, Alexander S, Bouvier-Colle M-H,

Ford JB, et al. Trends in postpartum hemorrhage in high resource
countries: a review and recommendations from the International

27. Zhang X, Joseph KS, Kramer MS. Decreased term and postterm
birthweight in the United States: impact of labor induction.
Am J Obstet Gynecol 2010;203:124.e1124.e7.

AUGUST JOGC AOT 2011 l 819