DOI: 10.1111/1471-0528.

13547
www.bjog.org

Abnormally invasive placenta—prevalence, risk

factors and antenatal suspicion: results from a
large population-based pregnancy cohort study
in the Nordic countries

ttir,g AM Tapper,h
L Thurn,a PG Lindqvist,b M Jakobsson,c LB Colmorn,d K Klungsoyr,e,f RI Bjarnado
PE Børdahl, K Gottvall, KB Petersen, L Krebs, M Gissler, J Langhoff-Roos,d K Ka
i j,k l m n,o
€llenj,p
a
Department of Obstetrics and Gynaecology, Blekinge Hospital, Karlskrona, Sweden b Department of Obstetrics and Gynaecology, Clintec,
Karolinska University Hospital, Stockholm, Sweden c Department of Obstetrics and Gynaecology, University Hospital, Helsinki, Finland
d
Department of Obstetrics, Rigshospitalet, Copenhagen, Denmark e Department of Global Public Health and Primary Care, University of Bergen,
Bergen, Norway f Medical Birth Registry of Norway, Norwegian Institute of Public Health, Bergen, Norway g Department of Obstetrics and
Gynaecology, Landspitali University Hospital, Reykjavik, Iceland h Department of Gynaecology and Paediatrics, University Hospital, Helsinki,
Finland i Department of Obstetrics and Gynaecology, Haukeland University Hospital, University of Bergen, Bergen, Norway j Department of
Evaluation and Analysis, Epidemiology and Methodological support unit, National Board of Health and Welfare, Stockholm, Sweden
k
Department of Public Health Sciences, Karolinska Institutet, Stockholm, Sweden l Fertility Clinic, Rigshospitalet, Copenhagen University
Hospital, Copenhagen, Denmark m Department of Obstetrics and Gynaecology, Holbæk Hospital, Holbæk and University of Copenhagen,
Copenhagen, Denmark n THL National Institute for Health and Welfare, Helsinki, Finland o Nordic School of Public Health, Gothenburg,
Sweden p Department of Reproduction Epidemiology, Tornblad Institute, Institution of Clinical Sciences, Lund University, Lund, Sweden
Correspondence: PG Lindqvist, Division of Obstetrics and Gynaecology, Department of Clinical Science Intervention and Technology (Clintec),
Karolinska Institutet, Karolinska University Hospital, Huddinge (K57), SE-14186 Stockholm, Sweden. Email [email protected]

Accepted 7 June 2015. Published Online 29 July 2015.

This article includes Author Insights, a video abstract available at https://vimeo.com/rcog/authorinsights13547

Objective The objective was to investigate prevalence, estimate risk after one prior caesarean section (CS) to 56-fold after three or more
factors, and antenatal suspicion of abnormally invasive placenta CS. Prior postpartum haemorrhage was associated with six-fold
(AIP) associated with laparotomy in women in the Nordic countries. increased risk of AIP (95% confidence interval 3.7–10.9).
Approximately 70% of all cases were not diagnosed antepartum. Of
Design Population-based cohort study.
these, 39% had prior CS and 33% had placenta praevia.
Setting and population A 3-year Nordic collaboration among
Conclusion Our findings indicate that a lower CS rate in the
obstetricians to identify and report on uterine rupture, peripartum
population may be the most effective way to lower the incidence
hysterectomy, excessive blood loss, and AIP from 2009 to 2012
of AIP. Focused ultrasound assessment of women at high risk will
The Nordic Obstetric Surveillance Study (NOSS).
likely strengthen antenatal suspicion. Prior PPH is a novel risk
Methods In the NOSS study, clinicians reported AIP cases from factor associated with an increased prevalence of AIP.
maternity wards and the data were validated against National
Keywords Incidence, placenta accreta, prenatal diagnosis, risk factors.
health registries.
Tweetable abstract An ultrasound assessment in women with
Main outcome measures Prevalence, risk factors, antenatal
placenta praevia or prior CS may double the awareness for AIP.
suspicion, birth complications, and risk estimations using
aggregated national data. Linked article This article is commented on by RM Silver, p. 1356
in this issue. To view this mini commentary visit http://
Results A total of 205 cases of AIP in association with laparotomy
dx.doi.org/10.1111/1471-0528.13583. The article has journal club
were identified, representing 3.4 per 10 000 deliveries. The single
questions by EYL Leung, p. 1357 in this issue. To view these visit
most important risk factor, which was reported in 49% of all cases
http://dx.doi.org/10.1111/1471-0528.13548.
of AIP, was placenta praevia. The risk of AIP increased seven-fold

Please cite this paper as: Thurn L, Lindqvist PG, Jakobsson M, Colmorn LB, Klungsoyr K, Bjarnad
ottir RI, Tapper AM, Børdahl PE, Gottvall K, Petersen KB,
Krebs L, Gissler M, Langhoff-Roos J, K€allen K. Abnormally invasive placenta—prevalence, risk factors and antenatal suspicion: results from a large
population-based pregnancy cohort study in the Nordic countries. BJOG 2016;123:1348–1355.

1348 ª 2015 Royal College of Obstetricians and Gynaecologists

 Abnormally invasive placenta

who filled out a detailed questionnaire, either on paper or

Introduction through a web-based system. All countries used the same
Placenta accreta, or abnormally invasive placenta (AIP), is questionnaire. The information was validated using the
a severe obstetric complication with a high risk of major MBRs and Hospital Discharge Registers (HDR) and was
haemorrhage, peripartum hysterectomy, and potential com- searched for the presence of such a registration, i.e. inter-
plicated surgery.1 AIP is caused by an abnormal invasion of national classification of diseases (ICD-10) code
the chorionic villi into the myometrium that leads to pla- O43.2x.19,20 In Sweden, no additional cases were uncovered
centa accreta or increta. In severe cases, placenta percreta in the MBR and all prospectively identified cases were con-
may result if the penetration extends to other organs such firmed. In Finland, 80% of the cases reported came directly
as the bladder or intestines.2 The risk of AIP is known to from clinicians, and 14 more cases (20%) were found in
increase in conjunction with placenta praevia, after previ- the MBR or HDR. In Denmark, Norway and Iceland, the
ous caesarean section (CS), as well as with increasing majority of cases were identified using ICD-10 codes from
maternal age3–7 and parity.8 The average blood loss at local IT systems or from MBRs on a 6-month basis. The
delivery in women with AIP is reported to be over 3 l and responsible clinicians were then contacted and asked to fill
will require blood transfusion in 90% of the cases.9 AIP out the questionnaire, and the data were stored with the
has been suggested as a main cause of peripartum hysterec- designated NOSS member in each country. The final data
tomy on vital indication.10,11 from Denmark, Norway and Iceland were stored in a com-
Internationally the incidence of AIP has increased over mon database in Denmark.
the past 30 years and is now reported to occur in 2–90 per A case of AIP was defined as a woman with delivery by
10 000 births.3,4,6,8,12 A planned operative delivery in CS in current pregnancy and assessed by the obstetrician to
patients diagnosed with AIP has been reported to involve be AIP or a vaginal delivery assessed to be AIP where blood
fewer complications and decrease the need for blood trans- transfusion and a laparotomy were needed. We initially
fusions.13,14 To be able to plan for a delivery involving a planned to include women with vaginal deliveries involving
multidisciplinary surgical team, antenatal suspicion is cru- difficult and incomplete manual removal of the placentas,
cial. Both ultrasound and magnetic resonance imaging who had received blood transfusions within 48 hours, and
(MRI) are possible techniques to identify patients at risk who were assessed to be AIP. However, due to differences
for AIP,15,16 but in order to be feasible, a high-risk popula- in interpreting this definition and the risk of including
tion must be identified. common retained placenta, all vaginal births (n = 75) that
In the Nordic countries the average CS rate has almost did not lead to a laparotomy were excluded. Thus, the final
doubled over the last 30 years, from a range of 7–12% in study population compromised 205 women. The compar-
1980 to 14–21% in 2010.17 However, the prevalence, man- ison group for the study was the total background popula-
agement, antenatal suspicion, and outcome of AIP are not tion. In the case of some specific variables, the background
well known. To study a rare ‘near miss’ event such as AIP data were only available from Finland, Norway and Swe-
this study was planned as part of the Nordic Obstetric den. The comparison group was described by common
Surveillance Study (NOSS) in 2009.18 variables and included maternal age (<20, 20–29, 30–39,
The objective of this study is to elucidate and assess and ≥40 years of age); parity (categorised as 0, 1, 2 or ≥3);
prevalence, risk factors, risk groups and antenatal suspicion smoking (dichotomised into smokers or non-smokers in
of AIP. early pregnancy); body mass index (BMI) (categorised as
<18.5, 18.5–24, 25–29, and ≥30), and previous CS (cate-
gorised as 0, 1, 2 or ≥3 prior CSs). This information was
Methods used as summary statistics for comparative purposes.
The NOSS study group was initiated in 2009 as a collabo- The following data were registered from each country:
ration between the Nordic Obstetric and Gynaecology Soci- the number of previous terminations of pregnancy or other
eties (Denmark, Finland, Iceland, Norway, and Sweden) uterine surgery (yes/no), previous profuse postpartum
and the National Medical Birth Registries (MBR) in the haemorrhage (PPH) (yes/no), previous retained placenta
Nordic countries. An overview of the study’s design and (yes/no), major pregnancy complications and maternal
general findings has been published recently.18 The study morbidity. Diagnostic method of AIP and whether there
was planned as a prospective study. A national coordinator was an antenatal awareness of AIP were noted. Details on
appointed in each country sent regular reminders to clini- surgical and medical treatment were added. Estimations of
cians at all participating delivery units asking them to blood loss, need for transfusions, and use of haemostatic
report any AIPs that may have occurred. This was the task drugs were requested (factor VII, fibrinogen and tranex-
of a local obstetric representative at each maternity unit amic acid). In addition, gestational week and newborn

ª 2015 Royal College of Obstetricians and Gynaecologists 1349

 Thurn et al.

characteristics were included. However, due to a program- population. Among parous women with AIP, prior PPH
ming error in the electronic version of the detailed ques- was found six times more often than in the background
tionnaire, some fetal outcome variables were missing. population (OR 6.5, 95%CI 3.7–10.9). High BMI (≥30)
was found in 17% of the cases and in 11% of the back-
Statistical analysis ground population (P < 0.001).
Odds ratios (OR) were used as a proxy for relative risk. In Table 2 we illustrate the prevalence and risk estimates
With this rare outcome OR and relative risk are almost for different combinations of risk factors, based on data
identical. The prevalence and its OR were calculated with from Finland, Norway and Sweden, including 119 cases of
95% confidence intervals (CI). The OR ratios in Table 2 AIP and representing an overall prevalence of 2.8/10 000
were calculated by cross tabulation with 95%CIs, based on deliveries. The Table shows that placenta praevia and prior
number per 10 000 between the data of the respective CS are the main determinants of AIP. Pregnancies with
background population. Risk factor profiles were compared placenta praevia are at high risk of AIP, with a 170- to
with those of the background population described in the 640-fold increased risk compared with pregnancies without
MBRs. For the comparison of dichotomous data, a chi- placenta praevia. This resulted in an absolute risk of AIP of
square test or Fisher’s exact test were used, as appropriate. between 2 and 10%. Placenta previa was reported in 49%
All data were managed by IBM SPSS STATISTICS v22.0 of all cases in our study. OR increased seven-fold after one
(SPSS Inc., Chicago IL, USA). previous CS and up to 56-fold after three or more CSs.
Women with prior CS constitute half of all the AIP cases.
There were 64 women (31%) with AIP who had no pla-
Results
centa praevia or prior CS. The estimated prevalence of AIP
In all, 136 clinics (80%) participated; 605 362 (91%) of 666 in women without any previous CS was 1.6/10 000.
306 deliveries were included in the study (Supporting Nulliparous women younger than 35 years of age with-
Information Table S1). Thus, the data represents the major out placenta praevia had the lowest prevalence of AIP at
part of all deliveries and AIP cases in this Nordic cohort. 0.6/10 000 deliveries (Table 2). However, the proportion of
The 205 cases of AIP represent an overall prevalence of 3.4 nulliparous women without placenta praevia was surpris-
per 10 000 deliveries, with a variation of 1.0–5.0 per ingly large and consisted of 22% (45/205) of all cases of
10 000 between the Nordic countries (Figure 1, Table S1). AIP. In this subgroup, 13% (6/45) had prior surgical abor-
Figure 1 shows the total prevalence with the proportion of tions, 16% (7/45) anomalies with prior uterine surgery
AIP in conjunction with placenta praevia and prior CS, (two myoma uteri, two hysteroscopic septa, and three tra-
together with the mean rate of prior CS in the population chelectomies) and 33% (15/45) were 35 years of age or
of each country during the study period. older. Finally, of the 45 women in this group, 13% (6/45)
Demographic data for the study cohort and aggregated had IVF pregnancies. In 18% (8/45), peripartum hysterec-
data from background populations are shown in Table 1. tomy had to be performed.
Women with AIP are older, have greater parity, and have In Table 3 we summarise and compare outcome vari-
had more previous CSs compared with the background ables depending on antenatal suspicion of AIP. Less than
one-third (29%) of all 205 cases of AIP were diagnosed
antepartum, with a wide variation between countries
Total AIP per 10 000 AIP + prior CS
(Table S1). As expected, a higher proportion of antenatal
AIP + previa per 10 000 Prior CS in populaon (%) suspicion was found among women with placenta praevia
6 12%
or prior CS (P = 0.001 and P < 0.001, respectively). As
5 10% compared with no suspicion of AIP, antenatal suspicion
was related to significantly fewer emergency CSs (40 versus
per 10 000 deliveries

4 8% 20%, P = 0.01) and more frequent use of embolisations

(33 versus 6%, P < 0.001). However, AIP pregnancies with
3 6%
antenatal suspicion had a significantly higher mean blood
loss, more massive blood transfusions (≥6 units of red
2 4%
blood cells, RBC), more frequent use of haemostatic drugs,
1 2% more organ damage, and more hysterectomies compared
with the group without suspicion of AIP. The group with
0 0% antenatal suspicion also had a higher rate of preterm deliv-
Denmark Finland Sweden Norway Iceland Total eries (73 versus 31%, P < 0.01) but no excess risk was
Figure 1. Prevalence of total AIP, AIP with placenta praevia, prior observed for extreme prematurity (<32 weeks). In 50% of
caesarean in relation to rate of prior caesarean in study population. the cases with antenatal suspicion, vaginal bleeding was

1350 ª 2015 Royal College of Obstetricians and Gynaecologists

 Abnormally invasive placenta

reported, as compared with 28% in the group with no sus-

Table 1. Demographics of study and background populations
picion of AIP. All of the diagnoses were made using ultra-
Abnormally Background Significance sound, but 29% (17 of 59) of the examinations were
invasive population of
complemented by MRI.
placenta difference
(AIP) (P)
The mean blood loss was 4.6 l and the mean number of
blood transfusions was 7 units (RBC). Radiologic interven-
n = 205 % n = 605 362 % tion with embolisation, coiling or balloon occlusion was
performed in 14% of the cases. Hysterectomies were per-
Maternal age formed in 44% of all AIP cases, ranging between 19% in
<20 0 0 10 729 2 <0.001
Denmark and 100% in Iceland. There were no cases of
20–34 112 55 471 765 78
≥35 93 46 122 840 20
maternal death; however, one stillbirth and one perinatal
Missing 0 28 death were identified in the group with no antenatal suspi-
Smoking 12 6 61 877 10 =0.04 cion.
Body mass index (kg/m2)
In our series we included only deliveries involving a
≤24 113 58 333 631 64 <0.001
25–29 49 25 118 888 23
laparotomy, which is why we present a lower prevalence
30–34 24 12 44 187 9 (3.4 versus 4.6/100 000) than in the overview article.18
≥35 10 5 21 205 4
Missing 9 87 451
Previous conditions Discussion
Deliveries
0 60 30 262 373 44 0.001 Main findings
1 66 33 216 739 36 The prevalence of AIP associated with laparotomy in the
2 31 16 84 347 14
Nordic countries was 3.4/10 000 deliveries and, as expected,
≥3 42 21 38 820 6
Missing 6 3083
placenta praevia and prior CS were the major determinants
Surgical abortions 50 24 na for AIP. Previous PPH was identified as a novel six-fold
Cesarean deliveries* increased risk factor for AIP. Possible causes for this rela-
0 101 49 544 732 90 <0.001
tionship include abnormal placentation in a previous preg-
Yes 104 51 60 630 10
1 72 35 35 069 8
nancy, resulting in a PPH. Another possibility is that some
2 19 9 4130 1 surgical procedure was done, giving rise to surgical trauma
≥3 13 6 850 0 of the uterine endometrium (exploration or exeresis). High
Other uterine surgery 16 8 na
BMI was also associated with an increased risk for AIP. As
PPH** 28 14 10 147 2 <0.001
Complications in this pregnancy
high BMI is related to an increased risk of CS, this may
ap vaginal Bleeding*** 42 27 na explain this relationship.21
Placenta previa**** 100 49 1389 0.3 <0.001 We found a dose-dependent relationship with prior CS
Preeclampsia 13 6 na
ranging from a seven-fold increased risk after one prior CS
Mode of delivery
Vaginal not instr 15 7 432 955 74 <0.001
to a 56-fold increased risk after three or more CSs. This is
Vaginal instrumental 5 2 46 035 8 in agreement with several studies3,5,12 but contrasts with a
Planned CS 114 56 43 606 7 recent study from the UK.6 It has been estimated that AIP
Emergency CS 71 35 61 879 11
will occur in 1/800 to 1/2000 women with prior CS, which
Missing 0 18 469
Gestational age 36
is in line with our findings (1/720).22 There were large vari-
(mean weeks) ations in the absolute risk of AIP (Table 2).
Preterm Antenatal detection has been reported to lower the risk
<37+0 82 40 30 834 5 <0.001
of surgical complications and the need for blood transfu-
<32+0 19 9 5416 1
Multiple pregnancies 12 6 10 530 2 <0.001
sion;14,16,23 however, this could not be confirmed in our
IVF***** 10 6 12 143 3 study (Table 3). We believe this heterogeneity reflects that
the most severe cases of AIP were predominantly diagnosed
*Caesarean 1, 2, >2 Background data from Finland, Norway,
antenatally. When diagnoses of AIP were made before
Sweden, Iceland.
**Background data from Finland, Norway, Sweden/428 044 delivery, hysterectomies were performed more often (65%)
deliveries. na = non available. than in the group with no antenatal suspicion of AIP
***Case data from Denmark, Norway, Sweden / 153 cases, (40%), supporting the above assumption. It is also likely
ap = antepartum. that in some cases there was an the woman with suspected
****Background data from Finland, Norway, Sweden/428 044
AIP agreed to have a hysterectomy, as opposed to the
deliveries.
*****Case data from Finland, Denmark, Norway/163 cases of AIP. reluctance to do so in an acute situation in which AIP was
not suspected.

ª 2015 Royal College of Obstetricians and Gynaecologists 1351

 Thurn et al.

Table 2. Absolute and relative risk of abnormally invasive placenta (AIP) in women categorized by different combinations of risk factors*

Category (not exclusive) Number of AIP Absolute risk of Total deliveries Odds ratio (OR) 95%CI
of pregnant women AIP/10 000 in category

All women 119 2.8 42 7652

Parity
Multiparous 91 3.7 243 495 1.0 Reference
0 Para 28 1.5 184 157 0.4 0.3–0.6
0-para not praevia 22 1.3 171 836 0.3 0.2–0.5
0-para not praevia <35 10 0.6 156 582 0.2 0.0–0.3
Multiparous, no previous CS 33 1.6 201 667 1.0 Reference
Multipara, any previous CS 58 13.9 41 828 8.5 5.5–13.1
0 Para 28 1.5 184 157 0.9 0.6–1.5
Caesarean section (CS)
No previous CS 61 1.6 385 824 1.0 Reference
Any previos CS 58 13.9 41 828 8.8 6.1–12.6
1 prior CS 38 10.4 36 561 6.6 4.4–9.8
2 prior CS 12 27.5 4358 17.4 9.0–31.4
≥3 prior CS 8 88.0 909 55.9 25.0–110.3
Prior CS not praevia 24 5.8 41 406 3.7 2.3–5.8
Placenta praevia
No praevia 61 1.4 426 263 1.0 Reference
Placenta praevia 58 417.6 1389 292 196–400
Praevia and previous CS 34 878.6 387 614 372–844
Praevia, not previous CS 24 239.5 1002 168 101–258
Praevia, prior CS, age >35 15 1013.5 148 643 362–1000
Other
Multiparous, no prior PPH** 56 3.1 174 115 1.0 Reference
Prior PPH in multiparous** 18 21.0 8580 6.5 3.7–10.9
Maternal age <35 years 55 1.6 341 853 1.0 Reference
Maternal age ≥35 years 64 7.5 85 799 4.6 3.2–6.7
No IVF 109 2.6 415 509 1.0 Reference
IVF 10 8.2 12 143 3.1 1.6–5.8

CI, confidence interval; CS, caesarean section; PPH, previous postpartum haemorrhage.
*Specific denominator data on subcategories only available from Finland, Norway and Sweden; 119 cases of AIP.
**Data available for Sweden and Finland.

It is well known that parous women are at an increased high quality health register data for identification of non-
risk of AIP, mainly due to an increased prevalence of prior reported cases. This ensures a minimum number of miss-
CS and placenta previa. However, 29% of AIP cases in our ing cases and false positive cases, both of which might
study (n = 60) were primiparous and the majority had no occur in studies relying entirely on retrospective data
placenta praevia (n = 45). Among these 45 women, almost from medical registries or journals. During the 3-year
one-third (13/45) had had previous uterine surgery (includ- study period, medical strategies, diagnostics and reporting
ing surgical exereses) and 15 had no known risk factor at all. clinicians were likely to remain the same, ensuring similar
Although IVF treatment was not recognised as a risk factor conditions, in comparison with studies collecting data
for AIP and was not included in the study protocol, six of the over a much longer time period.
primiparous women (10%) had become pregnant through The major limitation of the study is that the participat-
IVF.6,24 IVF pregnancies have known increased risk of low ing countries used different approaches to data collection.
placentation, which may be due to thick endometrium.25,26 In addition, we were unable to achieve full coverage in
Sweden and Norway. However, in Sweden we ascertained a
Strengths and limitations similar prevalence in centres that did not participate.
The major strengths of our investigation are the combi- Moreover, we had no economic incentive as in the UKOSS
nation of a prospective cohort study design with local study, and participation was only on a voluntarily basis.
clinicians reporting cases of AIP on a regular basis and Another shortcoming was differences in the definition of

1352 ª 2015 Royal College of Obstetricians and Gynaecologists

 Abnormally invasive placenta

Table 3. Difference in characteristics and outcome by antenatal suspicion of abnormally invasive placenta (AIP), or not, in the Nordic countries

Proportion of AIP Antenatal suspicion No antenatal suspicion Significance of

n = 60 n = 145 difference (P)

29% 71%

Maternal charecteristics
Age ≥ 35 (years) 23 38 57 39 0.6
BMI ≥ 30 (kg/m2) 12 18 20 14 0.3
Previous caesarean delivery 46 77 58 40 <0.001
Pregnancy conditions
Placenta praevia 50 83 48 33 <0.001
Twins 1 2 11 8 0.1
Antepartum bleeding 30 50 41 28 0.003
Pre-eclampsia 1 2 11 8 0.1
Percrete 9 15 5 3 0.003
Mode of delivery
Vaginal delivery 0 0 21 14 0.002
Emergency CS 12 20 58 40 0.006
Blood loss and measurements against blood loss
Hysterectomy 39 65 58 40 0.001
Hysterectomy + transfusion ≥6 RBC 22 37 19 13 <0.001
Tamponade, intrauterine 10 17 16 11 0.3
Embolisation/occlusion 20 33 9 6 <0.001
Blood mean (l) 5.6 4.2
Blood loss > 5 l (%) 18 30 40 28 0.7
RBC transfusion ≥ 6 33 55 57 39 0.04
Haemostatic drugs 31 52 56 39 0.09
Major complications
Bladder injury 5 8 3 2 0.03
ICU 13 22 43 30 0.2
Relaparotomy 5 8 23 16 0.2
Major morbidity* 5 8 8 6 0.5
Newborn outcome
Birthweight Mean (g) 2498 3007
Preterm < 37 + 0 (weeks) 44 73 45 31 <0.001
Preterm < 32 + 0 (weeks) 6 10 13 9 0.8

*Severe ischaemic pain (n = 1), ureter damage (n = 1), postop retroperitoneal haemorrhage (n = 1), Ileus (n = 2), necrosis of bladder wall (n = 1),
lesion of bladder (n = 4), resuscitation (n = 1), na (n = 2).
ICU, Intensive care unit; RBC, red blood concentrate.

PPH among the countries. Finally, we could only use back- We attempted to investigate cases that reflect the clinical sit-
ground data from Finland and Sweden when calculating uation of AIP and decided not to rely solely on histopatho-
absolute prevalence estimations in this particular subgroup logical results, which usually result in hysterectomy.
(Table 2). Because of the data access and the ethics permis- The differences in prior CS rates in a population will have
sion received, we employed stratified analysis. a substantial influence on the risk of AIP, both directly and
indirectly, through an increased risk of placenta praevia in
Interpretation the future. We had a lower prevalence of prior CS (10%)
The prevalence of AIP in our study was in the lower range of (Figure 1) in the Nordic sample than in the UK (15%),6 and
the reported data. However, the UKOSS study recently presumably also lower than in the US, where the incidence
reported an incidence of 1.7/10 000 deliveries.6 They did not of AIP has increased from 1/30 000 deliveries around 1930
have the opportunity of using register data to identify miss- to 1/2000–3000 deliveries in the last decade.22
ing cases, which might partly explain the lower incidence. Our data include absolute risks, group size, and the
Until there is an international agreement on the definition of ORs associated with both individual and risk factors. This
AIP, it will be difficult to compare results between studies. information might be helpful when deciding whom to assess

ª 2015 Royal College of Obstetricians and Gynaecologists 1353

 Thurn et al.

for AIP. We were surprised to find the low degree of antena- hospital for validation of number of cases (O € 8-2011). In
tal suspicion among AIP cases (29%) in our study popula- Finland, the Ministry of Social Affairs and Health approved
tion. As many cases with no suspicion had placenta praevia the collection of primary data and the data linkage of
(33%) or prior CS (39%), we need to focus on these factors nationwide health register to these data (STM/1373/2009).
to increase the antepartum detection rate. In the subgroup In Norway, ethical permission was granted from MBR Nor-
of women with placenta praevia (0.3% of the background way and the ethical committee of Norway. In Denmark, the
population) the risk is so high (OR  290) that it is recom- Danish Data Protection Agency approved the study. In Ice-
mendable to screen routinely for AIP with ultrasound.16 An land, the Directorate of Health granted permission to collect
ultrasound assessment might also be indicated in gravida primary case data and data linkage to MBR.
with prior CS (10% in our background population) where
there is a low-lying placenta over the uterine scar (0.5% of Funding
population).27 With the high reported sensitivity (73–93%) This study was financially supported by the NFOG (Nordic
and positive predictive value (65–93%) in diagnosing AIP Federation of Societies of Obstetrics and Gynaecology)
with ultrasound,15,28 the implementation of the above sug- foundation and by Clintec, Karolinska Institutet, Stock-
gestions should increase the antepartum suspicion of AIP, holm, Sweden.
and reduce the risk of this near-miss event by planning for
multidisciplinary care at delivery. Acknowledgements
A theoretical calculation of assessing AIP by ultrasound We acknowledge all clinicians in the NOSS teams for their
using 80% sensitivity and 90% specificity (above) on the work. See Supporting Information Appendix S1.
small subgroups of women with placenta praevia (preva-
lence 3/1000) and prior CS with placenta overlying the scar
( 5/1000) would yield antepartum identification of 113
Supporting Information
cases (141*0.8) of 4800 assessments (600 000*8 per thou- Additional Supporting Information may be found in the
sand) with our study population, thus a doubled antepar- online version of this article:
tum rate of suspicion. The efficacy of such a screening Table S1. Background data from the Nordic countries
programme remains to be determined. and summary of outcome.
Appendix S1. Acknowledgements.
Data S1. Powerpoint slides summarising the study. &
Conclusion
Our findings indicate that a lower CS rate in the population
may be the most effective way to lower the incidence of AIP.
Focused ultrasound assessment of women at high risk will
References
likely increase antenatal suspicion. Prior PPH is a novel risk 1 Chantraine F, Langhoff-Roos J. Abnormally Invasive Placenta–AIP.
factor associated with an increased prevalence of AIP. Awareness and pro-active management is necessary. Acta Obstet
Gynecol Scand 2013;92:369–71.
2 Tantbirojn P, Crum CP, Parast MM. Pathophysiology of placenta
Disclosure of interests creta: the role of decidua and extravillous trophoblast. Placenta
Full disclosure of interests available to view online as sup- 2008;29:639–45.
porting information. 3 Miller DA, Chollet JA, Goodwin TM. Clinical risk factors for placenta
previa-placenta accreta. Am J Obstet Gynecol 1997;177:210–4.
4 Wu S, Kocherginsky M, Hibbard JU. Abnormal placentation: twenty-
Contribution to authorship year analysis. Am J Obstet Gynecol 2005;192:1458–61.
LT wrote the first draft of the paper and carried out the 5 Silver RM, Landon MB, Rouse DJ, Leveno KJ, Spong CY, Thom EA,
initial analysis supervised by PL. PL, MJ, A-MT, LK, PB, et al. Maternal morbidity associated with multiple repeat cesarean
KK, KG, KBP, LBC, KK, RIB, JL-RMG designed the study deliveries. Obstet Gynecol 2006;107:1226–32.
and KK, RIB and MG were responsible for the background 6 Fitzpatrick KE, Sellers S, Spark P, Kurinczuk JJ, Brocklehurst P,
Knight M. Incidence and risk factors for placenta accreta/increta/
data from the MBRs. All authors contributed to the revis-
percreta in the UK: a national case-control study. PLoS ONE 2012;7:
ing of the manuscript and approved the final submission. e52893.
PL is the guarantor for the data. 7 Eshkoli T, Weintraub AY, Sergienko R, Sheiner E. Placenta accreta:
risk factors, perinatal outcomes, and consequences for subsequent
Details of ethics approval births. Am J Obstet Gynecol 2013;208:219 e1–7.
8 Gielchinsky Y, Rojansky N, Fasouliotis SJ, Ezra Y. Placenta accreta—
Ethical approval was obtained separately from each country.
summary of 10 years: a survey of 310 cases. Placenta 2002;23:
In Sweden, permission was obtained from the Central Ethi- 210–4.
cal Board for medical research to register the cases anon- 9 Hudon L, Belfort MA, Broome DR. Diagnosis and management of
ymised with 4 digitals (year and month) and from the placenta percreta: a review. Obstet Gynecol Surv 1998;53:509–17.

1354 ª 2015 Royal College of Obstetricians and Gynaecologists

 Abnormally invasive placenta

10 Kwee A, Bots ML, Visser GH, Bruinse HW. Emergency peripartum 20 Ludvigsson JF, Andersson E, Ekbom A, Feychting M, Kim JL,
hysterectomy: a prospective study in The Netherlands. Eur J Obstet Reuterwall C, et al. External review and validation of the
Gynecol Reprod Biol 2006;124:187–92. Swedish national inpatient register. BMC Public Health
11 Machado LS. Emergency peripartum hysterectomy: incidence, 2011;11:450.
indications, risk factors and outcome. North Am J Med Sci 21 Morken NH, Klungsoyr K, Magnus P, Skjaerven R. Pre-pregnant
2011;3:358–61. body mass index, gestational weight gain and the risk of operative
12 Bowman ZS, Eller AG, Bardsley TR, Greene T, Varner MW, Silver delivery. Acta Obstet Gynecol Scand 2013;92:809–15.
RM. Risk factors for placenta accreta: a large prospective cohort. 22 Timor-Tritsch IE, Monteagudo A, Cali G, Palacios-Jaraquemada JM,
Am J Perinatol 2014;31:799–804. Maymon R, Arslan AA, et al. Cesarean scar pregnancy and early
13 Robinson BK, Grobman WA. Effectiveness of timing strategies for placenta accreta share common histology. Ultrasound Obstet
delivery of individuals with placenta previa and accreta. Obstet Gynecol 2014;43:383–95.
Gynecol 2010;116:835–42. 23 Fitzpatrick K, Sellers S, Spark P, Kurinczuk J, Brocklehurst P, Knight
14 Tikkanen M, Paavonen J, Loukovaara M, Stefanovic V. Antenatal M. The management and outcomes of placenta accreta, increta,
diagnosis of placenta accreta leads to reduced blood loss. Acta and percreta in the UK: a population-based descriptive study. BJOG
Obstet Gynecol Scand 2011;90:1140–6. 2014;121:62–71.
15 Comstock CH, Bronsteen RA. The antenatal diagnosis of placenta 24 Esh-Broder E, Ariel I, Abas-Bashir N, Bdolah Y, Celnikier DH.
accreta. BJOG 2014;121:171–81; discussion 81-2. Placenta accreta is associated with IVF pregnancies: a retrospective
16 Warshak CR, Ramos GA, Eskander R, Benirschke K, Saenz CC, Kelly chart review. BJOG 2011;118:1084–9.
TF, et al. Effect of predelivery diagnosis in 99 consecutive cases of 25 Grady R, Alavi N, Vale R, Khandwala M, McDonald SD. Elective
placenta accreta. Obstet Gynecol 2010;115:65–9. single embryo transfer and perinatal outcomes: a systematic review
17 National Institute for Health and Welfare F. Perinatal statistics in the and meta-analysis. Fert Steril 2012;97:324–31.
Nordic countries. 2012 Contract No.: Statistical report 7/2012, 26 Rombauts L, Motteram C, Berkowitz E, Fernando S. Risk of placenta
27.3.2012.THL. praevia is linked to endometrial thickness in a retrospective cohort
18 Colmorn LB, Petersen KB, Jakobsson M, Lindqvist PG, Klungsoyr K, study of 4537 singleton assisted reproduction technology births.
K€all

en K, et al. The Nordic Obstetric Surveillance study: a study of Human Reprod 2014;29:2787–93.
complete uterine rupture, abnormally invasive placenta, peripartum 27 Committee on Obstetric P. Committee opinion no. 529: placenta
hysterectomy and severe blood loss at delivery. Acta Obstet Gynecol accreta. Obstet Gynecol 2012;120:207–11.
Scand 2015;94:734–44. 28 Warshak CR, Eskander R, Hull AD, Scioscia AL, Mattrey RF,
19 Langhoff-Roos J, Krebs L, Klungsoyr K, Bjarnadottir RI, Kallen K, Tapper Benirschke K, et al. Accuracy of ultrasonography and magnetic
AM, et al. The Nordic medical birth registers: a potential goldmine for resonance imaging in the diagnosis of placenta accreta. Obstet
clinical research. Acta Obstet Gynecol Scand 2014;93:132–7. Gynecol 2006;108:573–81.

ª 2015 Royal College of Obstetricians and Gynaecologists 1355