DOI: 10.1111/1471-0528.

13300
www.bjog.org

Incidence, risk factors, management and

outcomes of amniotic-fluid embolism: a
population-based cohort and nested casecontrol
study
KE Fitzpatrick,a D Tuffnell,b JJ Kurinczuk,a M Knighta
a
National Perinatal Epidemiology Unit, University of Oxford, Oxford, UK b Bradford Teaching Hospitals NHS Foundation Trust, Bradford,
UK
Correspondence: KE Fitzpatrick, National Perinatal Epidemiology Unit, University of Oxford, Old Road Campus, Oxford OX3 7LF, UK. Email
Kate.fitzpatrick@npeu.ox.ac.uk

Accepted 7 December 2014. Published Online 12 February 2015.

Objective To describe the incidence, risk factors, management and

outcomes of amniotic-fluid embolism (AFE) over time.

Design A population-based cohort and nested casecontrol study

using the UK Obstetric Surveillance System (UKOSS).

Setting All UK hospitals with obstetrician-led maternity units.
Population All women diagnosed with AFE in the UK between

February 2005 and January 2014 (n = 120) and 3839 control

women.
Methods Prospective case and control identification through
UKOSS monthly mailing.
Main outcome measures Amniotic-fluid embolism, maternal

death or permanent neurological injury.

Results The total and fatal incidence of AFE, estimated as 1.7 and
0.3 per 100 000, respectively, showed no significant temporal
trend over the study period and there was no notable temporal
change in risk factors for AFE. Twenty-three women died (case

fatality 19%) and seven (7%) of the surviving women had

permanent neurological injury. Women who died or had
permanent neurological injury were more likely to present with
cardiac arrest (83% versus 33%, P < 0.001), be from ethnicminority groups (adjusted odds ratio [OR] 2.85, 95% confidence
interval [95% CI] 1.028.00), have had a hysterectomy
(unadjusted OR 2.49, 95% CI 1.026.06), had a shorter time
interval between the AFE event and when the hysterectomy was
performed (median interval 77 minutes versus 248 minutes,
P = 0.0315), and were less likely to receive cryoprecipitate
(unadjusted OR 0.30, 95% CI 0.110.80).
Conclusion There is no evidence of a temporal change in the

incidence of or risk factors for AFE. Further investigation is

needed to establish whether earlier treatments can reverse the
cascade of deterioration leading to severe outcomes.
Keywords Amniotic-fluid embolism, maternal morbidity,
maternal mortality.

Please cite this paper as: Fitzpatrick KE, Tuffnell D, Kurinczuk JJ, Knight M. Incidence, risk factors, management and outcomes of amniotic-fluid embolism:
a population-based cohort and nested casecontrol study. BJOG 2015; DOI: 10.1111/1471-0528.13300.

Introduction
Amniotic fluid embolism (AFE) although rare, remains
one of the leading causes of direct maternal mortality in
high-income countries,13 characterised by unexplained
sudden cardiovascular collapse, respiratory distress and
disseminated intravascular coagulation. The rarity of AFE
together with the fact that clinical diagnosis of the condition is one of exclusion makes it difficult to obtain
reliable information concerning incidence, risk factors,
management and outcomes. Previous reviews3,4 suggest

that the reported total incidence of AFE varies from 1.9

per 100 000 to 7.7 per 100 000 maternities; the reported
incidence of fatal AFE cases varies from 0.4 per 100 000
to 1.7 per 100 000 and the case fatality rate due to AFE
ranges from 11 to 43%; there is little consistency in the
factors reported to be associated with the occurrence of
AFE and very limited data are available regarding factors
associated with severe maternal outcomes.
The aim of this study was to extend our previous work,5
taking advantage of more recent data to not only describe
the incidence, risk factors, management and outcomes of

2015 The Authors. BJOG An International Journal of Obstetrics and Gynaecology published by John Wiley & Sons Ltd on behalf of
Royal College of Obstetricians and Gynaecologists.
This is an open access article under the terms of the Creative Commons Attribution-NonCommercial License, which permits use,
distribution and reproduction in any medium, provided the original work is properly cited and is not used for commercial purposes.

Fitzpatrick et al.

AFE, but also to further investigate temporal trends of the

condition and whether there are specific factors associated
with severe maternal outcomes.

Methods
A national, population-based cohort and nested casecontrol study was conducted using the UK Obstetric Surveillance System (UKOSS). UKOSS is a system that was set up
in 2005 to investigate rare and severe complications in
pregnancy and childbirth in the UK. Details of the UKOSS
methodology have been described elsewhere.6 Briefly, nominated clinicians in each obstetrician-led maternity unit in
the UK are sent a monthly case notification card with a list
of conditions under surveillance. Women may also deliver
in midwifery-led units or at home in the UK, but any
woman with a severe maternal morbidity in one of these
settings would be transferred to an obstetrician-led unit.
Between 1 February 2005 and 31 January 2014 clinicians
were asked to report cases of AFE, defined according to the
criteria outlined in Box 1. On reporting a case, clinicians
were sent a data collection form to complete from the
womans medical notes to confirm the diagnosis and ascertain further information concerning potential risk factors,
management and outcomes (data collection form available
at www.npeu.ox.ac.uk/downloads/files/ukoss/forms/UKOSSAmniotic-Fluid-Embolism.pdf). All data requested were
anonymous and up to five reminders were sent if completed forms were not returned. On receipt of data collection forms, cases were checked to confirm that they met
the case definition. All cases that had postpartum haemorrhage reported as the first symptom of AFE were reviewed
against the diagnostic criteria (Box 1) to determine
whether AFE was the most likely diagnosis. Information on
womans year of birth and expected date of delivery was
used to identify duplicate reports.
Information about maternal deaths from AFE during the
study period was obtained from the Centre for Maternal
and Child Enquires (CMACE) and the National Maternal,
Newborn and Infant Clinical Outcome Review Programme
run by MBBRACE-UK. This information was compared
with maternal deaths from AFE reported through UKOSS.
One additional case was identified through this route. In
this instance, the relevant UKOSS reporting clinician was
contacted and asked to complete a data collection form for
the case. There were also four cases reported to UKOSS
that were classified by the MBBRACE-UK confidential
enquiry process as not AFE cases; these four cases were
therefore excluded.
All statistical analysis was conducted using STATA 13
software (StataCorp, College Station, TX, USA). Incidence
rates for AFE with exact Poisson 95% confidence intervals
(95% CIs) were calculated using the available national birth

data (20052012)710 as the denominator for the estimated

number of maternities during the study period. Temporal
trends in incidence rates were investigated using Poisson
regression. The characteristics and management of women
with AFE were described, and putative risk factors for AFE,
identified from the literature, were evaluated by comparing
the cases with a control group of 3839 women using
unconditional logistic regression to estimate odds ratios
(ORs) with 95% CIs. The control women were identified
by UKOSS reporting clinicians as the two women delivering in the same hospital immediately before other UKOSS
study cases,1117 and are comparable in characteristics to
the available national data on women giving birth in the
UK. A full regression model was developed by including
both explanatory and potential confounding factors in a
core model if there was a pre-existing hypothesis or evidence to suggest that they were causally related to AFE.
Addition of interaction terms to the full model and subsequent likelihood ratio testing was used to assess plausible
interactions; to allow for multiple testing, a P-value of
<0.01 was deemed evidence of significant interaction. No
significant interactions were found. The analysis had 80%
power at the 5% level of statistical significance to detect
odds ratios of 1.7 or greater and 2.6 or greater, assuming
that putative risk factors have a prevalence of 40% and 5%,
respectively. Factors associated with severe maternal outcomes (fatality or permanent neurological injury) among
AFE cases were also investigated using unconditional logistic regression. Due to the high proportion of missing data
for some variables and on the basis of previous work,18 we
assumed that the data were not missing at random and
therefore in all analyses missing was included as an extra
category for variables that had any missing data.

Results
Incidence
All UK hospitals with obstetrician-led maternity units contributed data to UKOSS during the study period (100%
participation) and notified 186 cases of AFE, 23 of which
were subsequently reported by clinicians as not cases. Data
collection forms were received for 155 (95%) of the
remaining notified cases and the one additional case identified through the UK Confidential Enquires into Maternal
Deaths: 36 were subsequently excluded (11 because they
were duplicates, 21 because they did not meet the case definition and four because they were identified by MBBRACE
as not cases following confidential review), leaving a total
of 120 confirmed cases, 23 of which were fatal (case fatality
rate 19%, 95% CI 1229%), in an estimated 7 001 438
maternities.710 This represents a total incidence of 1.7 per
100 000 maternities (95% CI 1.42.1) and a fatal incidence
of 0.3 per 100 000 maternities (95% CI 0.20.5). There was

2015 The Authors. BJOG An International Journal of Obstetrics and Gynaecology published by John Wiley & Sons Ltd on behalf of
Royal College of Obstetricians and Gynaecologists.

Incidence, risk factors, management and outcomes of AFE

Box 1. Diagnostic criteria for amniotic-fluid embolism

In the absence of any other clear cause
EITHER
Acute maternal collapse with one or more of the following features
Acute fetal compromise
Cardiac arrest
Cardiac rhythm problems
Coagulopathy
Hypotension
Maternal haemorrhage
Premonitory symptoms, e.g. restlessness, numbness, agitation, tingling
Seizure
Shortness of breath
Excluding women with maternal haemorrhage as the first presenting feature in whom there was no evidence of early coagulopathy or cardiorespiratory compromise
OR
Women in whom the diagnosis was made at postmortem examination with the finding of fetal squames or hair in the lungs

no significant temporal trend in either the total or fatal

incidence of AFE during the study period (see Supporting
information, Figure S1; incidence rate ratio 0.97, 95% CI
0.921.03 and 0.90, 95% CI 0.791.03 for the total and
fatal incidences of AFE, respectively, for every 3-year
increase in time period).

Presentation of AFE
Amniotic-fluid embolism presented at or before delivery in
53% (62/117) of women, at a median gestation of 39 weeks
(range 2842 weeks); the remaining women (n = 55) presented with AFE a median of 19 minutes after delivery
(range 1 minute to 6 hours 27 minutes) having delivered
at a median gestation of 39 weeks (range 2842 weeks).
Diagnosis of AFE, including both antenatal and postnatal
cases, was first considered a median of 33 minutes (range
0 minutes to 2 days) after presentation. At presentation,
women had a median of four (range one to nine) of the
features of AFE (see Box 1), with women who died or who
had permanent neurological injury being significantly more
likely than those who survived and did not have permanent
neurological injury to present with cardiac arrest (83% versus 33%, P < 0.001) and exhibit this feature as the first
recognised symptom or sign of AFE (20% versus 6%,
P = 0.028). Women with AFE who died were also significantly more likely than those who survived to present with
cardiac arrest (87% versus 36%, P < 0.001) and to exhibit
this feature as the first recognised symptom or sign of AFE
(26% versus 5%, P = 0.006); no other significant differences in presentation were found between these groups (see
Supporting information, Table S1) The majority of women
(91%, 108/119) had ruptured membranes at or before AFE
presentation.

Risk factors for AFE

None of the multiparous women with AFE were reported
to have had AFE in a previous pregnancy. The characteristics of the women with AFE compared with the control
women are shown in Table 1. Women aged 35 years and
over had significantly raised odds of having AFE. The odds
of having AFE were also significantly increased in women
who had a multiple pregnancy, placenta praevia and induction of labour using any method. Information on the use
of prostaglandin labour induction and/or oxytocin use in
labour was available for women with AFE and a subset of
the control women; adjusted analysis suggested that compared with women who laboured without prostaglandin
induction or oxytocin in labour, the odds of AFE were
raised in women who had labour induced with prostaglandin and did not have oxytocin used in labour (adjusted
OR [aOR] 6.46, 95% CI 3.4712.01); were raised in women
who laboured without prostaglandin induction but had
oxytocin used in labour (aOR 2.80, 95% CI 1.485.29);
and were raised in women who had labour induced with
prostaglandin and oxytocin used in labour (aOR 2.43, 95%
CI 1.085.46). The type of prostaglandin used was only
known for women with AFE; 95% (40/42) had dinoprostone, 40% (16/40) of whom received a dose of more than
5 mg; only one woman had misoprostol and she was given
a dose of 800 lg.
Of the 55 women who presented with AFE postnatally,
11 (20%) had an instrumental vaginal delivery (ventouse
or forceps) and 39 (71%) delivered by caesarean. Both
instrumental vaginal delivery and caesarean delivery were
associated with significantly raised odds of having AFE
postnatally (aOR 9.51, 95% CI 3.1728.51 and aOR 16.15,
95% CI 6.2042.05, respectively). Of the women who pre-

2015 The Authors. BJOG An International Journal of Obstetrics and Gynaecology published by John Wiley & Sons Ltd on behalf of
Royal College of Obstetricians and Gynaecologists.

Fitzpatrick et al.

Table 1. Comparison of characteristics of cases and controls

Risk factor

No. (%) of cases No. (%) of controls

(n = 120)
(n = 3834)

Sociodemographic factors
Age (in years)
<35
35 or older
Missing
Ethnic group
White
Black or other minority ethnic group
Missing
Socio-economic group
Managerial and professional occupations
Other
Missing
Body mass index at booking (kg/m2)
<30
30 or more
Missing
Smoking status
Never/ex smoker
Smoked during pregnancy
Missing
Previous obstetric history
Parity
0
1 or more
Missing
Current pregnancy
Multiple pregnancy
No
Yes
Induction of labour using any method
No
Yes
Missing
Gestation age at delivery (weeks)**
Preterm (<37)
Term (37-41)
Post term (42 or more)
Missing
Macrosomia (birthweight 4500 g or more)**
No
Yes
Missing
Placenta praevia
No
Yes
Missing
Placental abruption
No
Yes
Missing

Unadjusted OR
(95% CI)

P-value

Adjusted OR
(95% CI)*

P-value

74 (62)
46 (38)
0 (0)

3062 (80)
766 (20)
6 (0)

1
2.48 (1.713.62)

1
<0.0001 2.15 (1.433.23)

0.0002

92 (77)
25 (21)
3 (3)

3064 (80)
699 (18)
71 (2)

1
1.19 (0.761.87)

1
0.4458 1.17 (0.721.91)

0.5183

42 (35)
60 (50)
18 (15)

977 (25)
2075 (54)
782 (20)

1
0.67 (0.451.01)

1
0.0529 0.81 (0.521.24)

0.3306

90 (75)
22 (18)
8 (7)

2797 (73)
674 (18)
363 (9)

1
1.01 (0.631.63)

1
0.9528 0.99 (0.601.62)

0.9559

99 (83)
20 (17)
1 (1)

3045 (79)
732 (19)
57 (1)

1
0.84 (0.521.37)

1
0.4842 1.12 (0.661.89)

0.6829

48 (40)
72 (60)
0 (0)

1669 (44)
2160 (56)
5 (0)

1
1.16 (0.801.68)

1
0.4353 1.07 (0.721.60)

0.7336

106 (88)
14 (12)

3786 (99)
48 (1)

71 (59)
49 (41)
0 (0)

2921 (76)
908 (24)
5 (0)

1
2.22 (1.533.22)

1
<0.0001 2.53 (1.703.75)

99
17
3
1

3414
276
125
10

2.12 (1.253.60)
1
3.45 (0.4427.20)

0.0053 1.34 (0.712.55)

1
0.2401 0.51 (0.151.71)

(83)
(14)
(3)
(1)

(89)
(7)
(3)
(0)

1
1
10.42 (5.5719.48) <0.0001 7.75 (3.6016.69) <0.0001

<0.0001

0.3653
0.2742

109 (91)
5 (4)
6 (5)

3742 (98)
77 (2)
6 (0)

1
2.23 (0.885.62)

1
0.0892 2.28 (0.885.92)

0.0892

116 (97)
3 (3)
1 (1)

3797 (99)
22 (1)
15 (0)

1
4.46 (1.3215.12)

1
0.0163 5.75 (1.6420.19)

0.0064

119 (99)
0 (0)
1 (1)

3806 (99)
6 (0)
22 (1)

1
3.93 (027.44)

*Adjusted for all factors in the table apart from placental abruption.
**Data do not necessarily add up to total number of cases/controls as this information is not applicable to miscarriages and terminations of
pregnancy.

2015 The Authors. BJOG An International Journal of Obstetrics and Gynaecology published by John Wiley & Sons Ltd on behalf of
Royal College of Obstetricians and Gynaecologists.

Incidence, risk factors, management and outcomes of AFE

sented with AFE after caesarean delivery, just over half (21/
39, 54%) did not labour. Caesarean delivery was performed
as an emergency (grade 1 or 2 urgency19) in four of these
women (two for maternal compromise and two for fetal
compromise) and as an elective procedure (grade 3 or 4
urgency19) in 17 of these women (three for maternal
request, six for previous caesarean delivery, eight for other
obstetric indications).
Further analyses were performed in which the characteristics of the cases and controls reported between 1 February
2005 and 31 July 2009 were compared and the characteristics of the cases and controls reported between 1 August
2009 and 31 January 2014 were compared; no notable
change was evident in the risk factors for AFE between the
two time periods (data not shown).

Management of AFE
Table 2 shows the factors used to manage coagulation following AFE presentation. Forty-four (37%) of the women
did not receive any coagulopathy management. The variety
of other management strategies used following AFE presentation is illustrated in Table 3. Thirty-one (26%) of the
women had one and 21 (18%) of the women had more
than one of these other management strategies used. Hysterectomy was known to have been performed in a total of
31 of the women a median of 200 minutes (range 0 minutes to 12 days) after AFE presentation. An obstetrician
was present a median of 0.5 minutes (range 0166 minutes) after AFE presentation and an anaesthetist was present a median of 1 minute (range 0198 minutes) after AFE
presentation.

Outcomes of AFE
Twenty of the 23 women with AFE who died had information on when AFE presented and when death occurred;
death occurred a median of 1 hour and 42 minutes after
AFE presentation (interquartile range 50 minutes to
10 hours 58 minutes). Three (15%) of the women died
more than 1 day after the initial event. Seven (7%) of the
women who survived had permanent neurological injury,
including persistent vegetative state/anoxic/hypoxic brain
injury or cerebrovascular accident. Women with AFE who
died or had permanent neurological injury were less likely
than those who survived and did not have permanent neurological injury to have cryoprecipitate given, were more
likely to have had a hysterectomy, had a shorter time interval between the AFE event and when the hysterectomy was
performed (median interval 77 minutes, range 0360 minutes versus median interval 248 minutes, range 23 minutes
to 12 days, P = 0.0315) and were more likely to be from
black or other minority ethnic groups (Tables 2, 3 and 4).
When the analysis was limited to comparing women with
AFE who died with those who survived, no significant differences were found (data not shown), noting the limited
power of the analysis. Of the women with AFE who survived and did not have permanent neurological injury, 15
(17%) had other major morbidities: five had a thrombotic
event (none of these women were treated with factor VIIa),
three had septicaemia, five had renal failure/impairment
and three had pulmonary oedema. The majority of women
who survived (87/97, 90%) were admitted to an intensive
therapy unit/high-dependency unit (ITU/HDU). Only 35%
(8/23) of those who died were admitted to ITU/HDU with

Table 2. Factors used in coagulation management of AFE cases that had severe outcome compared with those that did not have severe outcome
Number (%) of cases that
had severe outcome* (n = 30)
Cryoprecipitate
No
Yes
FFP
Non
Yes
Platelets
No
Yes
Factor VIIa
No
Yes
Fibrinogen
No
Yes

Number (%) of cases that did

not have severe outcome* (n = 90)

Unadjusted OR
(95% CI)

P-value

24 (80)
6 (20)

49 (54)
41 (46)

1
0.3 (0.110.80)

0.0163

13 (43)
17 (57)

38 (42)
52 (58)

1
0.96 (0.412.20)

0.9151

22 (73)
8 (27)

54 (60)
36 (40)

1
0.55 (0.221.36)

0.1929

24 (80)
6 (20)

70 (78)
20 (22)

1
0.88 (0.312.43)

0.7981

29 (97)
1 (3)

89 (99)
1 (1)

1
3.07 (0.1950.64)

0.433

*Died or had permanent neurological injury.

2015 The Authors. BJOG An International Journal of Obstetrics and Gynaecology published by John Wiley & Sons Ltd on behalf of
Royal College of Obstetricians and Gynaecologists.

Fitzpatrick et al.

Table 3. Other management strategies for AFE cases that had severe outcome compared with those that did not have severe outcome
Number (%) of cases that
had severe outcome* (n = 30)

Number (%) of cases that did not

have severe outcome* (n = 90)

Hysterectomy
No
18 (60)
Yes
12 (40)
Other surgery for haemorrhage**
No
25 (83)
Yes
5 (17)
Exchange transfusion
No
30 (100)
Yes
0 (0)
Plasma exchange
No
30 (100)
Yes
0 (0)
Misoprostol
No
28 (93)
Yes
2 (7)
Hemabate
No
28 (93)
Yes
2 (7)
Intrauterine balloons
No
29 (97)
Yes
1 (3)
Embolisation
No
30 (100)
Yes
0 (0)
Tranexamic acid
No
29 (97)
Yes
1 (3)
Obstetrician present at time of event or arrived within 5 minutes
No
10 (33)
Yes
17 (57)
Missing
3 (10)
Anaesthetist present at time of event or arrived within 5 minutes
No
9 (30)
Yes
18 (60)
Missing
3 (10)
Obstetrician and/or anaesthetist present at time of event
No
11 (37)
Yes
16 (53)
Missing
3 (10)

Unadjusted OR
(95% CI)

P-value

71 (79)
19 (21)

1
2.49 (1.026.06)

0.0441

81 (90)
9 (10)

1
1.8 (0.555.87)

0.3296

87 (97)
3 (3)

1
0.77 (07.33)

0.8366

86 (96)
4 (4)

1
0.55 (04.58)

0.6221

86 (96)
4 (4)

1
1.54 (0.278.84)

0.6309

84 (93)
6 (7)

1
1 (0.195.24)

84 (93)
6 (7)

1
0.48 (0.064.18)

0.5085

88 (98)
2 (2)

1
1.24 (016.10)

85 (94)
5 (6)

1
0.59 (0.075.23)

0.6324

28 (31)
55 (61)
7 (8)

1
0.87 (0.352.14)

0.7541

26 (29)
57 (63)
7 (8)

1
0.91 (0.362.30)

0.8457

35 (39)
47 (52)
8 (9)

1
1.08 (0.452.62)

0.8593

*Died or had permanent neurological injury.

**Inlcudes B-Lynch suture, artery ligation.

the remainder dying before they could be admitted. The

median length of stay in ITU/HDU was 3 days (range
144 days).
Women with AFE gave birth to a total of 134 infants
(106 singletons and 14 twin pregnancies). There were four
stillbirths (one antepartum and three intrapartum) and five
early neonatal deaths among these infants, equating to a
perinatal mortality rate of 67 per 1000 (95% CI 3112),
significantly higher than the national rate of 7.5 per 100020
(relative risk 8.91, 95% CI 4.7416.75). Mother and infant

did not survive in four cases. Major complications were

reported in 17 of the surviving infants, including ten diagnosed with neonatal encephalopathy and three requiring
phototherapy for severe jaundice. A total of 75/127 (59%)
of the infants were admitted to a neonatal unit for a median duration of 7 days (range 167 days). The perinatal
mortality rate was higher in infants born to women diagnosed with AFE at or before delivery compared with after
delivery (121 per 1000, 95% CI 54225 versus 16 per 1000,
95% CI 0.485, P = 0.033), as was the proportion of

2015 The Authors. BJOG An International Journal of Obstetrics and Gynaecology published by John Wiley & Sons Ltd on behalf of
Royal College of Obstetricians and Gynaecologists.

Incidence, risk factors, management and outcomes of AFE

Table 4. Sociodemographic, previous obstetric and medical history and current pregnancy characteristics of AFE cases that had severe outcome
compared with those that did not have severe outcome

Sociodemographic factors
Age
<35
35 or older
Ethnic group
White
Black or other minority ethnic group
Missing
Socio-economic group
Managerial and professional occupations
Other
Missing
Body mass index at booking (kg/m2)
<30
30 or more
Missing
Smoking status
Never/ex smoker
Smoked during pregnancy
Missing
Previous obstetric and medical history
Parity
0
1 or more
History of atopy
No
Yes
History of allergy
No
Yes
Current pregnancy
Multiple pregnancy
No
Yes
Induction of labour using any method
No
Yes
Mode of delivery
Spontaneous vaginal
Instrumental vaginal
Caesarean
Gestation age at delivery (weeks)
Preterm (<37)
Term (3741)
Post term (42 or more)
Missing
When AFE occurred
Before or at delivery
Post delivery
Missing

P-value

No. (%) of cases

that did not have
severe outcome*
(n = 90)

Unadjusted
OR
(95% CI)

16 (53)
14 (47)

58 (64)
32 (36)

1
1.59 (0.693.66)

0.2802

1
1.31 (0.523.30)

0.5645

19 (63)
10 (33)
1 (3)

73 (81)
15 (17)
2 (2)

1
2.56 (0.996.60)

0.0514

1
2.85 (1.028.00)

0.0462

10 (33)
17 (57)
3 (10)

32 (36)
43 (48)
15 (17)

1
1.27 (0.513.13)

0.6106

1
1.22 (0.473.16)

0.6841

22 (73)
6 (20)
2 (7)

68 (76)
16 (18)
6 (7)

1
1.16 (0.403.33)

0.7837

1
1.43 (0.474.40)

0.5298

26 (87)
4 (13)
0 (0)

73 (81)
16 (18)
1 (1)

1
0.7 (0.212.29)

0.5578

11 (37)
19 (63)

37 (41)
53 (59)

1
1.21 (0.512.83)

0.6672

1
1.12 (0.442.84)

0.8158

26 (87)
4 (13)

69 (77)
21 (23)

1
0.51 (0.161.61)

0.2492

25 (83)
5 (17)

70 (78)
20 (22)

1
0.7 (0.242.06)

0.5179

27 (90)
3 (10)

79 (88)
11 (12)

1
0.8 (0.213.08)

0.7431

14 (47)
16 (53)

57 (63)
33 (37)

1
1.97 (0.864.55)

0.1107

1
2.29 (0.935.63)

0.0726

1 (3)
3 (10)
26 (87)

8 (9)
16 (18)
66 (73)

1
1.5 (0.1316.82)
3.15 (0.3826.46)

0.7423
0.2904

4
26
0
0

13
73
3
1

(14)
(81)
(3)
(1)

0.86 (0.193.14)
1
0.75 (07.15)

0.8189

45 (50)
43 (48)
2 (2)

1
0.74 (0.321.73)

0.4845

(13)
(87)
(0)
(0)

17 (57)
12 (40)
1 (3)

Adjusted aOR
(95% CI)**

P-value

No. (%) of cases

that had severe
outcome*
(n = 30)

*Died or had permanent neurological injury.

**Adjusted for maternal age, ethnicity, socio-economic group, body mass index, parity and induction of labour.

2015 The Authors. BJOG An International Journal of Obstetrics and Gynaecology published by John Wiley & Sons Ltd on behalf of
Royal College of Obstetricians and Gynaecologists.

Fitzpatrick et al.

surviving infants that had major complications (22% versus

7%, P = 0.013) and the proportion of infants admitted to
a neonatal unit (71% versus 45%, P = 0.003).

Discussion

reducing the potential for information bias. Although our

study is one of the largest conducted of AFE using validated case criteria, another limitation of our study is that
we had limited power to examine factors associated with
severe outcomes due to the relatively small number of
women with severe outcomes.

Main findings
The total incidence of AFE as estimated by this 9-year prospective population-based study is 1.7 per 100 000 maternities and the estimated fatal incidence is 0.3 per 100 000
maternities, with a case fatality rate of 19%. No significant
temporal trend in either the total or fatal incidence during
the study period was found. Older maternal age, multiple
pregnancy, placenta praevia and induction of labour were
associated with the occurrence of AFE. Instrumental vaginal
and caesarean deliveries were associated with the occurrence of AFE postnatally. No notable change was evident in
the risk factors for AFE during the study period. Among
women with AFE, those who died or had permanent neurological injury were more likely to present with cardiac
arrest, be from black or other minority ethnic groups, were
more likely to have had a hysterectomy, had a shorter time
interval between the AFE event and when the hysterectomy
was performed and were less likely to have received cryoprecipitate.

Strengths and limitations

A major strength of our study is its prospective population-based design, not relying on coded data from routine
hospital administrative systems to ascertain cases. The
design of our study minimises the possibility of selection
bias by recruiting cases of AFE from all obstetrician-led
maternity units in the UK. Furthermore the controls in our
study, identified by UKOSS reporting clinicians as the two
women delivering immediately before other UKOSS study
cases, are comparable in characteristics to the available
national data on women giving birth in the UK. Despite
the active nature of the UKOSS data collection system and
the presence of several reporting clinicians in each hospital
we do, however, recognise that there is the possibility that
we have under-ascertained cases or ascertained only the
more severe cases. However, all apart from one of the fatal
cases of AFE reported to the UK Confidential Enquiries
into Maternal Deaths during the study period were
reported to UKOSS. This suggests that significant underreporting is unlikely to have occurred. Furthermore, we have
no evidence of systematic underreporting which may have
affected the validity of our results. We were only able to
examine information adequately recorded in the hospital
case notes, but this still allowed a detailed analysis of the
risk factors, management and outcomes of AFE and had
the advantage that much of the information would have
been documented prospectively before the AFE or delivery,

Interpretation
The observed variation in the reported incidence estimates of
AFE has largely been attributable to methodological differences between studies; rates obtained through the analysis of
population databases without additional criteria to exclude
false-positive cases are significantly higher than those
obtained through validated case criteria.4,21 Our incidence
estimate is comparable to other studies that have used validated case criteria, and again demonstrates the rarity of the
condition. Advanced maternal age and induction of labour
are the only factors that have been reported to be consistently
associated with AFE across five high-income countries,4
although the associations were not always statistically significant and the induction method investigated varied. Our
study investigated and found a significant association
between the occurrence of AFE and advanced maternal age
as well as induction of labour, in keeping with the literature.
We noted six times the odds of AFE in women who had
labour induced with prostaglandin, and more than double
the odds of AFE among women who received oxytocin in
labour or had both oxytocin and prostaglandin. Several studies have also reported an increased risk of AFE associated
with instrumental vaginal and caesarean delivery,2225
although these studies lacked information on timing of delivery in relation to when the AFE occurred making it uncertain
whether particular modes of delivery are a cause or a consequence of AFE. Our study did have information on timing,
and found that instrumental vaginal and caesarean delivery
were both significantly associated with postnatal occurrence
of AFE, suggesting that there may be a causal relationship.
Of note, three women had AFE after caesarean delivery at
maternal request.
Despite temporal increases in some of the identified risk
factors for AFE such as maternal age and caesarean delivery,26,27 similar to a Canadian study22 we did not observe a
significant temporal trend in either the total or fatal incidence of AFE over the study period. This may reflect the
low absolute risk of having the condition or it could be
that our study lacked sufficient power to detect a significant trend.
Although rare, AFE is associated with significant morbidity and mortality. Despite this, there has been limited
investigation of factors associated with severe outcomes
among women with AFE: an Australian study of 20 cases
(six fatal) and a US study of 227 cases (49 fatal) found no
significant association between common demographic or

2015 The Authors. BJOG An International Journal of Obstetrics and Gynaecology published by John Wiley & Sons Ltd on behalf of
Royal College of Obstetricians and Gynaecologists.

Incidence, risk factors, management and outcomes of AFE

obstetric factors and fatality;4,23 a Canadian study of 180

cases (24 fatal) reported a crude association between medical induction of labour and fatality;22 based on an analysis
of the first 60 cases (12 fatal) in the UK, included within
the cases in this study, we previously reported an increased
risk of fatality in ethnic minority women;5 and based on a
review of 44 case reports of AFE, Leighton et al.28 reported
an increased risk of death or permanent disability associated with factor VIIa treatment. Our study investigated the
relationship between a range of factors, including sociodemographic, obstetric and management strategies, and fatality or permanent neurological injury. We found that
women who died or had permanent neurological injury
were more likely to be from black or other minority ethnic
groups, an association that could not be explained by differences in age, socio-economic status, body mass index,
parity or induction of labour. We also found that women
who died or had permanent neurological injury were more
likely to present with cardiac arrest, have had a hysterectomy, had a shorter time interval between the AFE event
and when the hysterectomy was performed, and were less
likely to have had cryoprecipitate given. These data suggest
that hysterectomy is being performed in the sickest of
women and may either reflect a lack of time to give cryoprecipitate in the sickest of women or could indicate a benefit of earlier correction of coagulation abnormality. We
did not have information on the amount or timing of
coagulation products given, so were unable to investigate
this in relation to outcomes. No other significant associations were found including in relation to any delay in the
involvement of medical personnel, acknowledging the limited power of our study to examine associations due to the
relatively small number of women with severe outcomes.
Previous studies have found a higher risk of severe maternal morbidity and mortality among non-white ethnic
groups in the UK.2,29 The increased risk in ethnic minority
women may reflect differences in underlying medical conditions or factors related to maternity care.29

Conclusion
For every one woman with AFE who dies, four survive. Of
the women with AFE who survive, 7% have permanent neurological injury and of the women with AFE who survived
and do not have permanent neurological injury, 17% have
other major morbidity. There is no evidence of a temporal
change in the incidence of or risk factors for AFE. Further
investigation is needed to establish whether better and more
rapid correction of coagulopathy, through the use of cryoprecipitate, fresh frozen plasma, platelets and fibrinogen is
associated with improved outcomes. Future studies should
collect information on the amount and timing of coagulation
products given. Outcome appears largely related to the sever-

ity of the clinical presentation. Women who die or have permanent neurological disability are more likely to have
cardiac arrest at presentation and death most commonly
occurs on the same day as the event. Women who die or have
permanent neurological disability are more likely to have a
hysterectomy, suggesting more resistant haemorrhage. Consideration needs to be given to whether earlier treatments,
including correction of coagulopathy, can reverse the cascade
of deterioration that seems to be present with AFE, and so
improve survival in the most serious cases.

Disclosure of interests
The authors declare that no competing interests exist.

Contribution to authorship
KF carried out the analysis and wrote the manuscript. DT,
JJK and MK designed the study and contributed to the
writing of the manuscript.

Details of ethics approval

The London Research ethics committee approved the
UKOSS general methodology (04/MRE02/45) and this
study (04/MRE02/46).

Funding
This article presents independent research funded by the
National Institute for Health Research (NIHR) under the
Beyond maternal death: improving the quality of maternity
care through national studies of near-miss maternal morbidity programme (programme grant RP-PG-0608-10038).
The views expressed in this publication are those of the
author(s), and not necessarily those of the NHS, the NHIR,
or the Department of Health. The funder had no role in
the study design, data collection and analysis, decision to
publish, or preparation of the article.

Acknowledgements
The authors would like to thank the UK Obstetric Surveillance System (UKOSS) reporting clinicians who notified
cases and completed the data-collection forms, without
whose contribution it would not have been possible to
carry out this study. We would also like to acknowledge Dr
Rowan Wilson for helpful comments on the article and the
help of staff from CMACE and MBRRACE-UK in ascertaining cases of fatal AFE.

Supporting Information
Additional Supporting Information may be found in the
online version of this article:
Figure S1. Total and fatal incidence of AFE in the UK, 1
February 2005 to 31 January 2014.
Table S1. Features of AFE at presentation. &

2015 The Authors. BJOG An International Journal of Obstetrics and Gynaecology published by John Wiley & Sons Ltd on behalf of
Royal College of Obstetricians and Gynaecologists.

Fitzpatrick et al.

References
1 Berg CJ, Callaghan WM, Syverson C, Henderson Z. Pregnancyrelated mortality in the United States, 1998 to 2005. Obstet
Gynecol 2010;116:13029.
2 Cantwell R, Clutton-Brock T, Cooper G, Dawson A, Drife J, Garrod
D, et al. Saving mothers lives: reviewing maternal deaths to make
motherhood safer: 20062008. The eighth report of the
confidential enquiries into maternal deaths in the United Kingdom.
BJOG 2011;118 (Suppl 1):1203.
3 Conde-Agudelo A, Romero R. Amniotic fluid embolism: an
evidence-based review. Am J Obstet Gynecol 2009;201:445.e113.
4 Knight M, Berg C, Brocklehurst P, Kramer M, Lewis G, Oats J, et al.
Amniotic fluid embolism incidence, risk factors and outcomes: a
review and recommendations. BMC Pregnancy Childbirth 2012;12:7.
5 Knight M, Tuffnell D, Brocklehurst P, Spark P, Kurinczuk JJ.
Incidence and risk factors for amniotic-fluid embolism. Obstet
Gynecol 2010;115:9107.
6 Knight M, Kurinczuk JJ, Tuffnell D, Brocklehurst P. The UK Obstetric
Surveillance System for rare disorders of pregnancy. BJOG
2005;112:2635.
7 Office for National Statistics. Key Population and Vital Statistics
(2005, 2006 and 2007). Newport, UK: Office for National Statistics,
2007, 2008 and 2009.
8 General Register Office for Scotland. Vital Events Reference Tables
(2008, 2009, 2010, 2011 and 2012). Section 3: Births. Edinburgh,
UK: General Register Office for Scotland, 2009, 2010, 2011, 2012
and 2013.
9 Northern Ireland Statistics and Research Agency. Registrar General
Annual Reports (2008, 2009, 2010, 2011 and 2012) Section 3:
Births. Belfast, UK: Northern Ireland Statistics and Research Agency,
2009, 2010, 2011, 2012 and 2013.
10 Office for National Statistics. Birth Summary Tables, England and
Wales 2012. Newport, UK: Office for National Statistics, 2013.
11 Knight M. Antenatal pulmonary embolism: risk factors, management
and outcomes. BJOG 2008;115:45361.
12 Knight M, Kurinczuk JJ, Spark P, Brocklehurst P. Cesarean delivery
and peripartum hysterectomy. Obstet Gynecol 2008;111:97105.
13 Scott CA, Bewley S, Rudd A, Spark P, Kurinczuk JJ, Brocklehurst P,
et al. Incidence, risk factors, management, and outcomes of stroke
in pregnancy. Obstet Gynecol 2012;120:31824.
14 Fitzpatrick KE, Sellers S, Spark P, Kurinczuk JJ, Brocklehurst P,
Knight M. Incidence and risk factors for placenta accreta/increta/
percreta in the UK: a national casecontrol study. PLoS One 2012;7:
e52893.

10

15 Fitzpatrick KE, Hinshaw K, Kurinczuk JJ, Knight M. Risk factors,

management, and outcomes of hemolysis, elevated liver enzymes,
and low platelets syndrome and elevated liver enzymes, low
platelets syndrome. Obstet Gynecol 2014;123:61827.
16 Knight M, Kurinczuk JJ, Spark P, Brocklehurst P. Extreme obesity in
pregnancy in the United Kingdom. Obstet Gynecol 2010;115:989
97.
17 Knight M. Eclampsia in the United Kingdom 2005. BJOG
2007;114:10728.
18 Lindquist A, Knight M, Kurinczuk JJ. Variation in severe maternal
morbidity according to socioeconomic position: a UK national case
control study. BMJ Open 2013;3:e002742.
19 Lucas DN, Yentis SM, Kinsella SM, Holdcroft A, May AE, Wee M,
et al. Urgency of caesarean section: a new classification. J R Soc
Med 2000;93:34650.
20 Centre for Maternal and Child Enquiries. Perinatal Mortality 2008.
London: Centre for Maternal and Child Enquiries, 2010.
21 Clark SL. Amniotic fluid embolism. Obstet Gynecol 2014;123:337
48.
22 Kramer MS, Rouleau J, Baskett TF, Joseph KS. Amniotic-fluid
embolism and medical induction of labour: a retrospective,
population-based cohort study. Lancet 2006;368:14448.
23 Abenhaim HA, Azoulay L, Kramer MS, Leduc L. Incidence and risk
factors of amniotic fluid embolisms: a population-based study on 3
million births in the United States. Am J Obstet Gynecol
2008;199:49.e18.
24 Roberts CL, Algert CS, Knight M, Morris JM. Amniotic fluid
embolism in an Australian population-based cohort. BJOG
2010;117:141721.
25 Kramer MS, Rouleau J, Liu S, Bartholomew S, Joseph KS. Amniotic
fluid embolism: incidence, risk factors, and impact on perinatal
outcome. BJOG 2012;119:8749.
26 RCOG Clinical Effectiveness Support Unit. The National Sentinel
Caesarean Section Audit Report. London: RCOG, 2001.
27 Office for National Statistics. Birth Statistics: Review of the Registrar
General on Births and Patterns of Family Building in England and
Wales. Newport: Office for National Statistics, 2007.
28 Leighton BL, Wall MH, Lockhart EM, Phillips LE, Zatta AJ. Use of
recombinant factor VIIa in patients with amniotic fluid embolism: a
systematic review of case reports. Anesthesiology 2011;115:12018.
29 Knight M, Kurinczuk JJ, Spark P, Brocklehurst P. Inequalities in
maternal health: national cohort study of ethnic variation in severe
maternal morbidities. BMJ 2009;338:b542.

2015 The Authors. BJOG An International Journal of Obstetrics and Gynaecology published by John Wiley & Sons Ltd on behalf of
Royal College of Obstetricians and Gynaecologists.