DOI: 10.1111/j.1471-0528.2012.03323.

x
www.bjog.org

Amniotic fluid embolism: incidence, risk factors,

and impact on perinatal outcome
MS Kramer,a J Rouleau,b S Liu,b S Bartholomew,b KS Josephc for the Maternal Health Study Group
of the Canadian Perinatal Surveillance System
a
Departments of Pediatrics and of Epidemiology, Biostatistics, and Occupational Health, McGill University Faculty of Medicine, Montreal,
QC, Canada b Maternal and Infant Health Section, Health Surveillance and Epidemiology Division, Centre of Chronic Disease Prevention and
Control, Public Health Agency of Canada, Ottawa, ON, Canada c Department of Obstetrics and Gynaecology, Women’s Hospital and Health
Centre of British Columbia, and School of Population and Public Health, University of British Columbia, Vancouver, BC, Canada
Correspondence: MS Kramer, 2300 Tupper Street (Les Tourelles), Montreal, QC, Canada H3H 1P3. Email michael.kramer@mcgill.ca

Accepted 25 February 2012. Published Online 24 April 2012.

Objective To extend our previous work on AFE in Canada by seizure, nonimmune haemolytic or traumatic jaundice and length
including stricter criteria for case identification and by examining of hospital stay.
risks for stillbirth, neonatal mortality and serious maternal and
Results A total of 292 AFE cases were identified, of which only
neonatal morbidity.
120 (40%) were confirmed after applying our additional
Design Population-based cohort study. diagnostic criteria, yielding an AFE incidence of 2.5 per 100 000
deliveries. Of the 120 confirmed cases, 33 (27%) were fatal.
Setting Canada.
Significant modifiable risk factors included medical induction,
Population or sample In all, 4 508 462 hospital deliveries from caesarean delivery, instrumental vaginal delivery, and uterine or
fiscal year 1991/92 to 2008/09. cervical trauma. Amniotic fluid embolism was associated with
significantly increased risks of stillbirth and neonatal asphyxia,
Methods To reduce false-positive diagnoses, we restricted our
mechanical ventilation, sepsis, seizures and prolonged length of
analysis to AFE cases with cardiac arrest, shock or severe
hospital stay.
hypertension, respiratory distress, mechanical ventilation, coma,
seizure, or coagulation disorder. Linkage of maternal and neonatal Conclusions Amniotic fluid embolism remains a rare but serious
records, available since 2001/02, enabled us to examine the effects obstetric outcome, with several important modifiable risk
of AFE on neonatal outcomes. Detailed demographic and clinical factors and major implications for maternal, fetal and neonatal
data facilitated control for a broad array of potential confounding health.
variables.
Keywords Amniotic fluid embolism, maternal morbidity, maternal
Main outcome measures Amniotic fluid embolism, in-hospital mortality, neonatal morbidity, stillbirth.
neonatal death, asphyxia, mechanical ventilation, bacterial sepsis,

Please cite this paper as: Kramer M, Rouleau J, Liu S, Bartholomew S, Joseph K for the Maternal Health Study Group of the Canadian Perinatal Surveillance
System. Amniotic fluid embolism: incidence, risk factors, and impact on perinatal outcome. BJOG 2012; DOI: 10.1111/j.1471-0528.2012.03323.x.

Amniotic fluid embolism may arise from simultaneous

Introduction
tears in the fetal membranes and the uterine vessels, which
Amniotic fluid embolism (AFE) is a rare obstetric occur- permit amniotic fluid to enter the uterine vein and hence
rence, yet it is one of the leading causes of maternal mor- the maternal pulmonary arterial circulation,6,7 but its path-
tality in developed countries.1–3 In Canada, it was the cause ogenesis remains poorly understood.3 It is characterised by
of 13 of 99 direct maternal deaths in 1988–92 and 7 of 44 sudden dyspnoea, cardiopulmonary collapse and coagulop-
direct maternal deaths in 1997–2000, ranking third behind athy. The clinical picture resembles anaphylaxis, and evi-
cerebrovascular and hypertensive disorders and ahead of dence has been reported linking anaphylactic-type
postpartum haemorrhage and other pulmonary embolisms physiological mediators, including prostaglandins, leukotri-
as a cause of maternal deaths.4,5 enes, histamine and bradykinin. It is presumed, however,

ª 2012 The Authors BJOG An International Journal of Obstetrics and Gynaecology ª 2012 RCOG 1
 Kramer et al.

that entry of amniotic fluid into the maternal circulation Autopsy information was not available to confirm the diag-
must occur commonly, and the rarity of AFE strongly sug- nosis of AFE in fatal cases.
gests a genetic component to the disease. Linkage of the files of maternal and neonatal records and
In recent years, several population-based studies have data on parity in the Discharge Abstract Database have
examined its incidence, temporal trends, or risk factors been available since 2001/02. Based on this linked file, we
from Canada,8 Australia,9 the UK10 and the USA.11 Dif- were able to include the effects of AFE on neonatal death,
ferences in incidence and case fatality rate have been as well as the following categories of serious neonatal mor-
attributed largely to methodological differences in the bidity: asphyxia, mechanical ventilation, bacterial sepsis,
published studies, with no convincing evidence of signifi- seizure, jaundice due to trauma or haemolysis (other than
cant temporal trends, geographic differences, or effective haemolysis due to immune-mediated blood type incompat-
treatments.12 Moreover, although several previous studies ibility), as well as neonatal length of stay. These categories
have reported associations between AFE and fetal and were chosen because of their known relationship to the
neonatal death,9,10,13 none has reported on serious neona- hypoxia and hypotension characteristic of AFE. We also
tal morbidity. examined the mother’s receipt of any of the following
In this study, we have extended our previous popula- treatments and interventions: blood transfusion, plasma
tion-based study of AFE in Canada, not only to take transfusion, clotting factor transfusion, exchange transfu-
advantage of more recent data, but also to apply additional sion or hysterectomy.
criteria in an attempt to reduce false-positive diagnoses. In Statistical methods included overall descriptive statistics,
addition, since 2001/02, we have been able to link maternal highlighting the effect of the confirmatory maternal diag-
hospital records to those of their newborns, which allows nostic criteria on incidence, the proportion of confirmed
us to examine the impact of AFE on neonatal mortality cases with each of the criteria, the proportion receiving the
and serious morbidity, in addition to stillbirth and mater- above-noted treatments or interventions, and the case fatal-
nal morbidity. ity rate. We also examined risk factors for AFE using cross-
tabulations and logistic regression analyses to estimate
crude and adjusted odds ratios (OR) and their 95% confi-
Methods
dence intervals (95% CI), as well as associations of AFE
Ours is a population-based cohort study of 4 508 462 hos- with maternal mortality, hysterectomy and prolonged
pital deliveries (i.e. the hospitalisation in which delivery maternal length of hospital stay; stillbirth; and neonatal
occurred) in Canada from fiscal year 1991/92 to 2008/09. mortality, severe morbidity and prolonged neonatal length
These hospital records were collected by the Canadian of hospital stay. Covariates adjusted for in the multivariable
Institute for Health Information (CIHI) in its Discharge logistic regression models included maternal age (£19, 20–
Abstract Database. Quebec does not provide data to CIHI; 34, ‡35 years), elderly primigravidity (defined by CIHI as a
nor did Manitoba and Nova Scotia in earlier years, so all first pregnancy with age ‡35 years) on expected day of
three provinces were excluded from this study. Ascertain- delivery, grand multiparity (‡6 previous live births), medi-
ment of cases of AFE was based on International Classifica- cal and surgical induction of labour, previous caesarean
tion of Disease, ninth revision (ICD-9) code 673.1 for years delivery, presentation (cephalic versus noncephalic), pla-
up to and including 1999/2000 and ICD-10 code O88.1 centa praevia or placental abruption, hypertension (none,
since 2000/01, with some differences among the Canadian pre-eclampsia, eclampsia or other [pregestational hyperten-
provinces in the time of switching coding from ICD-9 to sion or gestational hypertension without proteinuria]), dia-
ICD-10. To improve the sensitivity of case identification, betes (gestational or pregestational), fetal macrosomia, fetal
we also included nondelivery postpartum admissions and growth restriction, postdates, prelabour rupture of mem-
hospital transfers with a discharge diagnosis of AFE using branes, prolonged first stage (‡18 hours for primiparae,
fifth-digit ICD codes. Procedures were coded according to ‡12 hours for multiparae) or second stage of labour
the Canadian Classification of Diagnostic, Therapeutic, and (‡2 hours for primiparae, ‡1 hour for multiparae, 1 addi-
Surgical Procedures from 1991 to 2001, and according to tional hour with receipt of epidural anaesthesia), cervical
the Canadian Classification of Interventions since 2001. laceration, uterine rupture, mode of delivery (spontaneous
To reduce false-positive diagnoses of AFE, we adapted vaginal, instrumental vaginal, caesarean), fetal distress
additional confirmatory criteria published by Roberts (ICD-9 codes denoting abnormal fetal heart rhythm, fetal
et al.9,13 A confirmed case required the presence of at least acidaemia, meconium staining of amniotic fluid, or cord
one of the following conditions or procedures: cardiac prolapse), dystocia and year of delivery.
arrest, shock or severe hypertension, respiratory distress, With the linked maternal–infant files available since
mechanical ventilation, coma, seizure or disseminated 2001/02, we repeated the above-mentioned multivariable
intravascular coagulation or other coagulation disorder. logistic regression analyses using actual parity (0, 1–2, 3–4,

2 ª 2012 The Authors BJOG An International Journal of Obstetrics and Gynaecology ª 2012 RCOG
 Amniotic fluid embolism

‡5) instead of ICD-based diagnoses of elderly primigravidi- ments and interventions under study, length of maternal
ty and grand multiparity; gestational age (<32, 32–36, 37– hospital stay and hysterectomy rate.
41, ‡42 completed weeks) instead of postdates; and birth- Table 2 shows the distribution of potential risk factors
weight-for-gestational age z-score (based on a published and covariates, the number and rate of confirmed AFE
Canadian population-based reference14) of <)1, )1 to +1, cases with or without the risk factors, and the crude and
or >+1 instead of fetal macrosomia and fetal growth adjusted odds ratios and their 95% confidence intervals.
restriction. All statistical analyses were carried out using Significant associations were found with older maternal
SAS version 9.2 (SAS Institute, Cary, NC, USA). age, grand multiparity, multiple birth, polyhydramnios,
eclampsia, pre-eclampsia and placenta praevia or placental
abruption. A previous caesarean section was not signifi-
Results
cantly protective against AFE (adjusted OR 0.5; 95% CI
Of the 4 058 462 total deliveries in the CIHI file, 292 were 0.3–1.0). Significant labour and delivery factors that
coded as having experienced AFE based on the ICD-9 or increased the risk of AFE included premature rupture of
ICD-10 codes. The overall crude incidence including all membranes, medical induction, fetal distress, caesarean
292 cases was 6.5 per 100 000 deliveries. Of the 292 total, delivery, instrumental vaginal delivery and cervical lacera-
however, only 120 (40%) had one or more of the required tion or uterine rupture. Noncephalic presentation was pro-
confirmatory maternal conditions, yielding a confirmed tective for AFE (adjusted OR 0.4; 95% CI 0.2–0.9).
(corrected) AFE incidence of 2.5 per 100 000. Of these 120, The sensitivity analysis based on the linked maternal and
113 were noted on the delivery record, and the remaining neonatal files confirmed the association of all the above-
seven were noted only after postpartum transfer to another noted risk factors. With the data available on actual parity
hospital. Of the 42 fatal cases, 33 were confirmed, for a in the more recent maternal and linked files, both parity of
confirmed fatal AFE rate of 0.8 per 100 000 and a con- 3 or 4 (adjusted OR 1.1; 95% CI 1.03–1.1) and 5 or more
firmed case fatality rate of 27%. Table 1 describes the 120 (adjusted OR 1.1; 95% CI 1.1–1.2) were associated with a
women with confirmed AFE in greater detail, showing the significantly increased risk of AFE. In addition, preterm
proportion with each of the confirmatory criteria, as well birth (particularly preterm birth <32 weeks, adjusted OR
as the number and percent who received each of the treat- 10.2; 95% CI 9.8–10.6) and low birthweight for gestational
age (z-score < )1) (adjusted OR 1.8; 95% CI 1.8–1.8) were
also associated with increased risks of AFE.
Table 3 shows the numbers and rates (%) for stillbirth
Table 1. Description of confirmed AFE cases (n = 120) (for the entire file) and for the studied neonatal outcomes
Characteristic Number %
associated with AFE, the latter based on the maternal–
infant linked file available since 2001/02. No cases of in-
hospital neonatal death occurred among infants born to
Confirmatory diagnostic criteria
Cardiac arrest 56 46.7 mothers with AFE. Highly significant increases in risk,
Shock or severe hypotension 39 32.5 however, were observed for stillbirth, asphyxia, mechanical
Respiratory distress 17 14.2 ventilation, bacterial sepsis, seizures and prolonged length
Mechanical ventilation 27 22.5 of neonatal hospital stay, but not for jaundice. These rela-
Coma 2 1.7 tionships were even stronger when the analysis was
Seizure 2 1.7
restricted to births ‡37 completed weeks of gestation.
DIC or other coagulation disorder 36 30.0
Treatments/interventions
Blood transfusion 68 56.7 Discussion
Plasma transfusion 53 44.2
Clotting factor transfusion 37 30.8 As we hypothesised, the use of the additional maternal
Exchange transfusion 0 0.0 morbidity confirmation criteria sharply reduced the num-
Hysterectomy 20 16.7 ber of cases and incidence of AFE. Our inference from
Length of hospital stay (days)
these findings is that many of the AFE diagnoses without
0–3* 39 32.5
these criteria were false positives. In addition, the effect of
4–7* 36 30.0
‡8* 45 37.5 including these confirmation criteria reduced the number
of nonfatal cases to a greater degree than for fatal cases,
DIC, disseminated intravascular coagulation. hence leading to a sharp increase in the case fatality rate.
*For the 87 nonfatal cases, the numbers and percentages are 10
This finding underscores the findings from our previous
(11.5%) for 0–3 days, 36 (41.4%) for 4–7 days, and 41 (47.1%) for
‡8 days.
study, where we used analysis of fatal AFE to reduce false-
positive diagnoses.8

ª 2012 The Authors BJOG An International Journal of Obstetrics and Gynaecology ª 2012 RCOG 3
 Kramer et al.

Table 2. Risk factor analysis for confirmed cases of AFE (n = 120)

Characteristic Proportion (%) of Confirmed cases, Crude OR (95% CI) Adjusted OR

study cohort n (rate per 100 000) (95% CI)

Maternal age (years)

£19 5.5 2 (0.8) 0.4 (0.1–1.5) 0.4 (0.1–1.7)
20–34 79.0 78 (2.2) 1.0 (reference) 1.0 (reference)
‡35 15.5 40 (5.7) 2.6 (1.8–3.8) 2.3 (1.5–3.4)
Elderly primigravidity 1.0 2 (4.4) 1.7 (0.4–6.8) 0.4 (0.1–1.7)
Grand multiparity 0.3 1 (7.7) 2.9 (0.4–20.8) 2.5 (0.3–17.9)
Previous caesarean delivery 10.9 17 (3.5) 1.3 (0.8–2.3) 0.5 (0.3–1.0)
Hypertension
None 95.6 106 (2.5) 1.0 (reference) 1.0 (reference)
Pre-eclampsia 3.8 12 (7.0) 2.8 (1.6–5.2) 1.7 (0.9–3.2)
Eclampsia 0.1 2 (75.5) 30.7 (7.6–124.5) 16.3 (4.0–67.2)
Other 5.6 0 (0.0) Undefined Undefined
Diabetes 1.3 4 (6.8) 2.6 (1.0–7.1) 1.4 (0.5–3.9)
Polyhydramnios 0.5 4 (19.0) 7.3 (2.7–19.9) 3.8 (1.4–10.5)
Amnionitis 1.1 2 (4.0) 1.5 (0.4–6.2) 0.6 (0.2–2.6)
Placenta praevia or abruption 2.1 20 (21.6) 9.5 (5.9–15.4) 5.0 (3.0–8.3)
Multiple pregnancy 1.2 5 (9.3) 3.6 (1.5–8.8) 2.5 (1.0–6.2)
Fetal macrosomia 2.5 7 (6.2) 2.4 (1.1–5.2) 1.6 (0.7–3.5)
Fetal growth restriction 2.3 2 (1.9) 0.7 (0.2–2.9) 0.4 (0.1–1.7)
Postdates 9.1 21 (5.1) 2.1 (1.3–3.4) 1.8 (1.1–3.1)
Noncephalic presentation 7.4 10 (3.0) 1.1 (0.6–2.2) 0.5 (0.3–1.0)
Prelabour rupture of membranes 8.7 17 (4.4) 1.7 (1.0–2.9) 1.6 (0.9–2.7)
Induction of labour
Medical 16.2 36 (4.9) 2.2 (1.5–3.3) 1.9 (1.2–3.0)
Surgical 7.6 6 (1.8) 0.6 (0.3–1.5) 0.4 (0.2–1.0)
Prolonged labour
Prolonged first stage 1.5 1 (1.4) 0.5 (0.1–3.8) 0.5 (0.1–3.7)
Prolonged second stage 4.8 5 (2.3) 0.9 (0.4–2.1) 0.6 (0.2–1.7)
Long labour, unspecified 0.3 0 (0.0) Undefined Undefined
Fetal distress 16.0 47 (6.5) 3.4 (2.3–4.9) 1.5 (1.0–2.3)
Mode of delivery
Spontaneous vaginal 64.7 17 (0.6) 1.0 (reference) 1.0 (reference)
Instrumental vaginal 13.0 28 (4.8) 8.2 (4.5–15.0) 7.6 (4.0–14.4)
Caesarean 22.4 75 (7.4) 12.8 (7.5–21.6) 11.7 (6.4–21.5)
Dystocia 25.0 47 (4.2) 1.9 (1.3–2.8) 0.8 (0.5–1.3)
Cervical laceration or uterine rupture 0.2 6 (56.2) 22.2 (9.8–50.5) 12.7 (5.5–29.1)

Table 3. Fetal/infant outcomes among confirmed AFE cases Nonetheless, our updated and improved study confirmed
(n = 120 for stillbirths, n = 54 for neonatal outcomes*) our previously reported association of AFE with medical
induction of labour, as well as with all forms of operative
Outcome n (%) Adjusted OR (95% CI)
delivery, including caesarean, forceps and vacuum deliver-
ies.8 Many of the confirmed cases of AFE were associated
Stillbirth 5 (4.2) 5.9 (2.0–17.4)
with multiple categories of serious maternal morbidity, as
Asphyxia 15 (27.8) 36.0 (18.6–69.7)
well as with hysterectomy and prolonged length of stay. As
Mechanical ventilation 17 (31.5) 11.8 (6.0–23.4)
Bacterial sepsis 5 (9.3) 5.0 (1.8–14.0) in our previous study, we found no significant temporal
Seizure 8 (14.8) 22.8 (9.7–53.3) increase of AFE during the 18-year study. Finally, we found
Jaundice (nonimmune 9 (16.7) 1.4 (0.6–3.2) major increased risks of stillbirth and serious neonatal
haemolytic/traumatic) morbidity associated with AFE.
Length of stay >7 days 18 (33.3) 18.5 (8.5–40.2) The lower incidence rate and higher case fatality rate
*Based on linked maternal and infant file available since 2001/02. (compared with our previous report8) bring our data closer
in line with those of more recent studies based on similar

4 ª 2012 The Authors BJOG An International Journal of Obstetrics and Gynaecology ª 2012 RCOG
 Amniotic fluid embolism

population-based hospitalisation databases (Australia)9 and our previous study, as well as in a hospital discharge-based
tailor-made surveillance studies from the UK.10 All three of study from the USA.11 It is unclear, however, whether the
these studies found a significant association with labour seizures leading to the diagnosis of eclampsia preceded the
induction. One study from the USA found a 50% increase signs and symptoms of AFE or were consequences of it, i.e.
in the odds of induction that was not statistically signifi- occurred after its onset. Temporality is also an issue for
cant.11 The US study, however, did not include confirma- neonatal outcomes. Neonatal morbidity can be a conse-
tory maternal morbidity criteria to reduce false-positive quence of AFE only for those cases occurring before birth,
diagnoses of AFE, hence probably diluting the true effect of and so the strong associations we observed may be diluted
induction on AFE occurrence. by our inability to identify and exclude cases of postnatal
Our study has both strengths and limitations. Strengths AFE.
include a population-based cohort of all hospital births Several other issues should be considered. We did not
(over 98% of all Canadian births occur in hospital) in observe any significant temporal trend in the occurrence of
Canadian provinces and territories except Quebec, Mani- AFE, despite strong temporal trends in medical induction,
toba and Nova Scotia, comprising over 70% of Canadian caesarean delivery, advanced maternal age and other identi-
births over the study period. This enabled a large number fied risk factors. It is possible that increases in induction
of AFE cases to be confirmed, despite the rarity of the con- and caesarean were counter-balanced by other trends,
dition, as well as inclusion of a large number of potential including reduction in multiparity and in use of operative
risk factors and confounding variables. Finally, the linked vaginal delivery. Nonetheless, the lack of temporal trend
files of mothers and infants available since 2001 enabled a in AFE confirms findings from other recently published
sensitivity analysis for a number of risk factors (parity, studies.12
birthweight and gestational age) not available in the Future research in AFE should aim to better understand
unlinked maternal file, as well as a heretofore unavailable the biological mechanisms underlying the disease, based on
assessment of adverse neonatal outcomes. animal models, and to identify specific genes that may
One important limitation of our study is the uncertain increase the risk when amniotic fluid enters the maternal
temporal relationship between caesarean delivery and the circulation. From the technical and public health perspec-
first signs and symptoms of AFE, which prevents us from tives, the strong associations of labour induction and (espe-
knowing whether caesarean delivery is a true risk factor cially) caesarean delivery with AFE should be kept in mind
(and potential cause) of AFE, or a consequence of an when balancing the risks and benefits of these procedures,
attempt to rescue the fetus or the mother with rapid deliv- given their increasing use worldwide.
ery. The same can be said for forceps and vacuum delivery
in cases where labour was so advanced that delivery would Disclosure of interests
be more expeditious with instrumental vaginal delivery None.
than with a caesarean. In the UK study,10 however, caesar-
ean delivery was significantly associated with the risk of Contribution to authorship
AFE even when restricted to cases occurring postnatally MSK designed the study, supervised the analysis, and wrote
(after delivery). the manuscript. JR carried out the analysis and contributed
Finally, administrative databases like CIHI’s Discharge to the study design and the first draft of the manuscript.
Abstract Database inevitably imply a risk of coding errors, SL, SB and KSJ contributed to the study design and
not only for AFE, but also for confirmatory maternal reviewed and provided feedback on earlier drafts of the
diagnoses, neonatal diagnoses and demographic and clini- manuscript. Several other members of the Maternal Health
cal covariates. In addition, the absence of data on parity Study Group provided feedback on the penultimate draft
(which is considered neither a diagnosis nor a procedure of the manuscript.
under ICD coding) forced us to rely on the ICD ‘diagno-
ses’ of elderly primigravidity and grand multiparity, both Details of ethics approval
of which are likely to be under-reported. Associations Not applicable.
with parity ‡3 were confirmed, however, using the parity
variable contained in the file since 2001/02. Moreover, Funding
most coding errors would be expected to be nondifferen- This study was supported by the Canadian Institutes of
tial with respect to the occurrence of AFE and thereby Health Research (MSK) and the Child and Family Research
would reduce associations with AFE, rather than bias Institute (KSJ).
them upwards.
The extremely high odds ratio associated with eclampsia Acknowledgements
also merits further comment. This result was reported in None. j

ª 2012 The Authors BJOG An International Journal of Obstetrics and Gynaecology ª 2012 RCOG 5
 Kramer et al.

9 Roberts CL, Algert CS, Knight M, Morris JM. Amniotic fluid embo-
References
lism in an Australian population-based cohort. BJOG 2010;
1 Hogberg U, Joelsson I. Amniotic fluid embolism in Sweden, 1951– 117:1417–21.
1980. Gynecol Obstet Invest 1985;20:130–7. 10 Knight M, Tuffnell D, Brocklehurst P, Spark P, Kurinczuk JJ, on
2 Atrash HK, Koonin LM, Lawson HW, Franks AL, Smith JC. Maternal behalf of the UK Obstetric Surveillance System. Incidence and
mortality in the United States, 1979–1986. Obstet Gynecol 1990; risk factors for amniotic-fluid embolism. Obstet Gynecol 2010;115:
76:1055–60. 910–7.
3 Conde-Agudelo A, Romero R. Amniotic fluid embolism: an evi- 11 Abenhaim HA, Azoulay L, Kramer MS, Leduc L. Incidence and risk
dence-based review. Am J Obstet Gynecol 2009;201:445.e1–e13. factors of amniotic fluid embolisms: a population-based study on 3
4 Turner LA, Kramer MS, Liu S, for the Maternal Mortality and Mor- million births in the United States. Am J Obstet Gynecol 2008;
bidity Study Group of the Canadian Perinatal Surveillance System. 199:e1–8.
Cause-specific mortality during and after pregnancy and the defini- 12 Knight M, Berg C, Brocklehurst P, Kramer M, Lewis G, Oats J, et al.
tion of maternal death. Chronic Dis Can 2002;23:31–6. Amniotic fluid embolism incidence, risk factors and outcomes: a
5 Health Canada. Special Report on Maternal Mortality and Severe review and recommendations. BMC Pregnancy Childbirth 2012;12:7.
Morbidity in Canada – Enhanced Surveillance: The Path to Preven- 13 Roberts CL, Cameron CA, Bell JC, Algert CS, Morris JM. Measuring
tion. Ottawa: Minister of Public Works and Government Services maternal morbidity in routinely collected health data: development
Canada, 2004. and validation of a maternal morbidity outcome indicator. Med Care
6 Morgan M. Amniotic fluid embolism. Anaesthesia 1979;34:20–32. 2008;46:786–94.
7 Clark SL. New concepts of amniotic fluid embolism: a review. Obstet 14 Kramer MS, Platt RW, Wen SW, Abrahamowicz M, Joseph KS, Allen
Gynecol Surv 1990;45:360–8. A, Blondel B, Bréart G, for the Fetal/Infant Health Study Group of
8 Kramer MS, Rouleau J, Baskett TF, Joseph KS, for the Maternal the Canadian Perinatal Surveillance System. A new and improved
Health Study Group of the Canadian Perinatal Surveillance System. population-based Canadian reference for birth weight for gesta-
Amniotic-fluid embolism and medical induction of labour: a retro- tional age. Pediatrics 2001;108:e35.
spective, population-based cohort study. Lancet 2006;368:1444–8.

6 ª 2012 The Authors BJOG An International Journal of Obstetrics and Gynaecology ª 2012 RCOG