1

Blood Coagulation, Fibrinolysis and Cellular Haemostasis

Combined hormonal contraception and risk of venous

thromboembolism within the first year following pregnancy
Danish nationwide historical cohort 19952009
Jesper Friis Petersen1; Thomas Bergholt1; Anne Kristine Nielsen1; Michael J. Paidas2; Ellen Christine L. Lkkegaard1
1

Department of Gynecology and Obstetrics, Hillerd Hospital, Hillerd, University of Copenhagen, Denmark; 2Department of Obstetrics, Gynecology, and
Reproductive Sciences, Yale Women and Childrens Center for Blood Disorders, Yale University School of Medicine, New Haven, Conneticut, USA

not

use

combined

hormonal

contraceptives throughout the follow-

Summary

up year. After childbirth, individuals that

Estimating the risk of venous thromboembolism (VTE) associated with combined hormonal used

combined

hormonal

contraceptives following early terminated preg-nancies or birth, a Danish nationwide retrospective contraceptives were more likely than
cohort observing a one-year follow-up was defined using three unique registries. All Da-nish non-users to experience VTE de-picted
women with confirmed pregnancies aged 1549 during the peri-od of 19952009 were included. by crude incidence ratios; however, the
The main outcomes were relative and absolute risks of first time venous thromboembolism in difference was only sig-nificant after 14
users as well as non-users of combined hormonal contraceptives. In 985,569 per-son-years, 598 weeks. This implied that the numbers
venous thromboembolisms were recorded. After early terminated pregnancies and births, needed to harm were lower for those
respectively, 113 and 485 events occurred in 212,552 and 773,017 person-years. After early that used compared to those that did
terminated pregnancies, the crude VTE incidence ratios were similar, and the numbers needed to not use combined oral contraceptives
harm were equal between groups that did or did
in the initial 14 weeks postpartum. In
conclusion,

the

use

of

combined

hormonal contraceptives after early

terminated

pregnancies

was

not

detrimental, but during the puerperal

period, they should be used with
caution.

Keywords
Venous
thromboembolism,
early termination of pregnancy,
birth, puerperal period

Correspondence to:
Jesper Friis Petersen, MD
Department of Gynecology and Obstetrics
University of Copenhagen, Hillerd Hospital
Dyrehavevej 29, 3400 Hillerd, Denmark
Tel.: + 45 21 84 84 89
E-mail: jesperfriis@dadlnet.dk

Financial support:
The study was funded by the research foundation at Hillerd Hospital. The funding
organisation had no role in the study planning, data collection, analysis, writing of
the article, or in the decision to submit the article for publication. All authors held
positions independent from funding organisation.
Received: September 30, 2013
Accepted after major revision: January 25, 2014
Prepublished online:
http://dx.doi.org/10.1160/TH13-09-0797
Thromb Haemost 2014; 112:

from three to six

weeks
postpartum (5,
Introduction
6).
Venous thromboembolism (VTE) is the The aim of
leading cause of maternal mortality in the the present study
developed world (1-3). Both pregnancy andwas to estimate
the
use
of
combined
hormonalthe risk of VTE
contraceptives are established as inde-as-sociated with
pendent risk factors (4). An interactionthe use of any
between these separate risks has beentype
of
suggested; however, research addressingcombined
the matter is limited (5). Recently, thehormonal
World Health Organization (WHO)contra-ceptives
changed its recommendation on the time toover one year
initiate oral contra-ceptives following birth;
following
they postponed the commencement time

cardiovascular
pregnancy
in
disease, cancer, or
two
settings;
Materials and methods gynaecological
first, after early
terminated
All Danish women aged 15-49 in thesurgery that had
pregnancies
period of 1995-2009 were in-cluded inimpaired fertil-ity.
the
(terminated
a nationwide historical cohort. FromFrom
before 22 weeksthe Danish Central Person Registry,National Registry
Medical
of
gestation),women were identified by a uniqueof
Products
ten-digit
identification
and second, afterpersonal
number. This number was used to link Statistics, inforchildbirth.

Schattauer 2014

was
to other registries to acquire individualmation
retrieved
on
information. From the National
women
that
had
Registry of Patients, pregnancies were
identified by the discharge diagnosisfilled
encoded with ICD-10 criteria. Womenprescriptions for
were excluded when they had beencombined
diagnosed
with
prior
VTE,hormonal

Thrombosis and

Haemostasis

112.1/2014

contraceptives.
This
registry
records all prescriptions filled
by
Danish
citizens,
coded
according
to
anatomictherapeuticchemical
categories.
The
use of combined
hormonal
contraceptives
was considered a
time-varying
covariate

Petersen et al. Risk of VTE by contraceptives following

pregnancy

Smoking
Yes
Table 1: Characteristics of
No

The study year was

167,406
divided
533,470 into periods of
populations with different
0-6, 7-13, 14-26, 2772,141
pregnancy
out-comes: Unknown
39, and 40-52 weeks,
childbirth
or
early BMI, body-mass index; IR, incidence rate; PY, patient-years; VTE, venous
starting after either an
thromboembolism.
terminated
pregnancy.
early
terminated
Results
indicate
the
pregnancy
or
at
numbers of person years
childbirth.
The
risks
(PY)
and
venous
were estimated for
thromboembolisms (VTE);
each inter-val. Crude
the inci-dence rate (IR) was
incidence rates of VTE
calculated: VTE/(PY/1000).
were calculated per
Childbirth
1,000 per-son years.
From the National
PY
Registry of Patients,
Age (years)
information
was
<15
8,295
retrieved on potential
20
81,847
confounding variables,
25
258,401
including
maternal
30
285,667
age, calendar year,
35
118,027
education, body mass
40
20,023
index
(BMI)
and
45
757
smok-ing status. BMI
Calendar year
was not registered
until the year 2003,
1995
33,638
why only 23% of
1996
62,343
pregnancies had the
1997
60,482
information. Smoking
1998
57,856
was also not registered
1999
56,707
until the year 1997;
2000
55,581
consequently 10% of
2001
54,396
pregnancies miss this
2002
51,747
information.
2003
50,242
Education
was
2004
50,290
recorded
in
five
2005
49,514
categories: elementary
2006
48,931
school education only
(9-10
years
of
2007
48,045
schooling); on-going
2008
47,469
or completed high
2009
45,776
school education (2-3
years after elemen-tary
Education
school); high school
Elementary school 195,575
and
ongoing
or
High school
133,303
completed post-high
Middle education
228,916
school education (3-4
Long education
159,478
years
after
high
Unknown
55,746
school); high school
and
on-going
or
BMI (kg/m2)
completed higher-level
Underweight (<18.5) 11,253
education (5-6 years
Normal (18.525)
101,611
after high school); and
Overweight (2530) 37,214
no information on
Obese class 1 (3035) 14,165
education (typically,
Obese class 2 (3540) 5,152
the
youngest
Obese class 3 (>40) 2,474
individuals).
Unknown
601,148

hormonal
contraceptives was
The primary outcomeconsidered a timewas VTE during the firstvarying covariate in
year following pregnancy.multivariate PoissonThis outcome includedregression ana-lyses
deep venous thrombosiswith maternal age,
year,
of the lower extremities,calendar
education,
BMI
and
pulmonary
embolism,
smok-ing
status
cerebral venous thrombosis, and portal veinincluded as potential
thrombosis.
Diagnosesconfounders. These
pro-vided
were obtained from codesanalyses
used specifically foradjusted rate ratios
pregnant women andfor estimating the
from additional, generalrisk of combined
use
codes. Specific codescontraceptive
a
95%
from the ICD-10 werewith
either
in-cluded
orconfidence interval
excluded. ICD-8 (1977-(CI). Absolute risk
1995) was used towas determined by
the
exclude pre-vious venouscalculating
thromboembolism
(seenumbers needed to
Appendix). All recordsharm, which was an
of
the
were subdivided into twoestimate
groups for analysis: i)number of women
early terminated preg-that had to receive
nancies, which weretreatment for one
terminated before 22year to be able to
weeks of gestation (in-detect one venous
duced
abortions,throm-boembolism,
miscarriages, hydatiformcalculated as NNH =
mole, ectopic pregnan-cy,1/(incidence rate of
and unknown location ofcombined hormonal
the foetus) and ii)contraceptive users
incidence rate of
childbirths.
The use of combinedNON-users).
Thrombosis
and
Haemostasis
112.1/2014

Schattauer
2014

Statistics
were
conducted
with
SAS/STAT
(SAS
Inc., Cary, NC, USA)
software.

The study was

approved by The
Danish
Data
Protection Agen-cy
(No.:
2010-414778). No ethical
approval
was
required for registry-based studies in
Denmark.

Results
The
one-year
observation period
following
pregnancy
comprised 985,569
person years.
Early

terminated

pregnancies accounted
for

212,552

years.

Of

142,673

person
these,
were

attributed to induced
abortions; 59,634 were
miscarriages;
were

caused

hydatiform moles;

904
by

Petersen et al. Risk of VTE by contraceptives following pregnancy

The

7,498 were ectopic pregnancies;

influence of
and 1,843 had an unknown locombined
cation of the pregnancy. Over the
hormonal
first year after the early termincontraceptive
ated pregnancies, 113 cases
use on the risk
developed first-time occurrences
of VTE was
of VTE.
assessed with

Childbirths accounted for a total of

rate
ratios
773,017 person years. Among these, 485
(RR) adjusted
cases developed first-time occurrences
for ma-ternal
of VTE.

age, calendar
Maternal age and calendar year
year,
were similarly distributed across the
education,
two populations ( Table 1). However,BMI,
and
higher edu-cation, normal bodysmoking for
weight, and no smoking tended to beeach
time
associ-ated with reduced rates ofinterval. After
VTE.
early
During the initial seven weeks ofterminated
the observation period, 25% (n=28) ofpregnancies,
the total number of events werethe use of
diagnosed after early terminatedcom-bined
pregnancies, and 68% (n=329) were hormonal
diagnosed after childbirth. During thecontraceptives
subsequent time intervals (7-13, 14-conferred an
26, 27-39, and 40-52 weeks), 22increased risk
(19%), 24 (21%), 30 (27%), and nine after
seven
(8%) events, respectively, wereweeks
(RR:
diagnosed after early terminated2.3, 95% CI:
pregnancies and 39 (8%) 42 (9%), 411.0-5.4)
(
(8%), and 34 (7%), respect-ively,Table
2).
were diagnosed after childbirths.
Following
As illustrated by the incidencechildbirth, the
rates, for individuals that did or diduse
of
not
use
combined
hormonalcombined
contraceptives, the risk of VTE washormonal
high in the early postpartum period, contraceptives
asand it decreased later ( Table 2). Nowas
sociated
with
differences were found in the risk of
VTE associ-ated with use ofincreased risk
combined hormonal contraceptivesafter 14 weeks
between the two outcome groups(RR: 2.8, 95%
(early terminated pregnancies andCI: 1.5-5.3) (
Table 2).
birth) over the various time intervals (
Table 2).

The crude
incidence
rates
for
progesterone
only
formulations
during
the
entire

follow-up
were
2.3/10,000
person-years
and
progesterone
intrauterine
devices
at
2.1/10,000
person-years,
indicat-ing
low risk.
For
the
group
with
early
terminated
pregnancies,
the estimated
numbers
needed
to
harm indicated
no significant
change in risk
over the entire
year
of
observation. In
contrast, for
the group that
gave birth, the
number
needed
to
harm
was
much
lower
during
the
early
postpartum
period
(0-6
weeks) than
during
the
remain-ing
observation
period ( Table
3).

Discussi
on
This
showed

study
an

increased risk

of VTE in the postpartum period,10).

consistent with previous studies (5, 7- popuTable 2: Risk of venous
thromboembolism (VTE)
following pregnancy in
groups with different
pregnancy outcomes. The
incidence rates (per 1,000
person years) and rate ratios
are shown for individuals that
used combined hor-monal
contraceptives (CHC)
compared to those that did
not use CHC, with 95%
confidence inter-vals (CI).
Results were adjusted for the
covariates listed in Table 1.
Bold font indicates a
significant effect of CHC.

In

this

Childbirth

Weeks
Use of
following
CHC
pregnancy

Person- VTE
years

Incidence
rate
(95% CI)

06

No

94,270

327

3.5 (3.13.9) 1

Yes

405

4.9 (0.618)

No

101,112 33

0.3 (0.20.5) 1

Yes

8,321

0.7 (0.31.6) 2.2 (0.95.1)

No

160,433 25

0.2 (0.10.2) 1

Yes

37,813

0.5 (0.30.7) 2.8 (1.55.3)

No

141,453 16

0.1 (0.10.2) 1

Yes

48,071

0.5 (0.30.8) 4.6 (2.58.7)

No

124,915 14

0.1 (0.10.2) 1

Yes

56,223

20

0.4 (0.30.5) 3.3 (1.76.5)

No

28,620

23

0.8 (0.51.2) 1

Yes

3,378

1.5 (0.53.5) 2.0 (0.85.3)

No

27,329

14

0.5 (0.20.5) 1

Yes

7,475

1.1 (0.52.1) 2.3 (1.05.4)

No

40,113

0.2 (0.10.4) 1

Yes

15,521

15

1.0 (0.51.6) 4.8 (2.111)

No

33,478

16

0.5 (0.30.8) 1

Yes

14,053

14

1.0 (0.51.7) 2.4 (1.24.9)

No

29,715

0.1 (0.00.3) 1

12,871

0.5 (0.21.1) 9.3 (1.944)

713
1426
2739
4052
Early
terminated
pregnancy

06
713
1426
2739
4052

Yes

Schatt

auer
2014

Thr
omb

6
17
25

osis and sis 112.1/2014

Haemosta

Rate ratio
(95% CI)

1.3 (0.35.2)

Petersen et al. Risk of VTE by contraceptives following

pregnancy

Table 3: Estimated numbers needed to harm (NNH).

fects of combined hormonal contraceptives during the indicated time intervals (weeks) for groups with different pregnancy outcomes. NNH = 1/(incidence rate of combined hormonal contraceptive users incidence rate of
non-users).

Pregnancy
outcome

Weeks following pregnancy

06

Childbirth

682

Early terminated 1478

Pregnancy

lation, the vast majority of cases occurred during the initial seven
weeks postpartum. The use of combined hormonal contraceptives
did not increase this relatively elevated risk observed in the procoagulative postpartum state (the first 14 weeks following birth).
Thereafter, the well-known association between combined hormonal contraceptives and VTE became evident (11, 12). Consequently, treatment of only 682 persons per year with combined
hormonal contraceptives (number needed to harm) was necessary
to detect one incident of VTE in the early interval following birth.
Later, the risk declined, as expected. Following early terminated
pregnancies, the number needed to harm was 1,478 per detectable
event in the early interval, with no changes in later intervals.
Jackson et al. reviewed the risk of VTE in the postpartum period (5) and identified 15 articles from 13 studies. In those studies,
the reference groups comprised women that were not pregnant or
postpartum. Compared to that group, the risk of VTE increased by
2.5- to 84-fold when combined hormonal contraceptives were
used during the initial six weeks after childbirth. Consequently, a
WHO group changed the recommendation regarding initiation of
combined hormonal contraceptives following childbirth (6). In the
present study, we examined the risk associated with combined hor-

What is known about this topic?

1 Both

pregnancy and
the use of combined
hormonal contracep-

VTE
postpartum
(14). In light of this
evidence, it would be
reason-tives
are
established
as
independent
risk
factors of venous
thromable
to

recommend
the
initiation
of
combined hormonal
contra-

boembolis
m (VTE).

An interaction
between
these
separate
risks has
been
suggested;
however,
research
addressing
the matter is
limited.

What does
this paper
add?

1 No

interaction
of
combined
hormonal
contracept
ives and
the
already
biologicall
y
increased
risk
of
VTE in the
early
postpartu
m period
was
observed.

2 Numbers

needed to
harm by
use
of
combined
hormonal
contraceptives
were low,
suggestin
g
reluctance
in
their
use in the
early
postpartu
m period.

3 After

early
terminated
pregnancie
s,
by
different
indications,
the use of

Thrombo
sis and
Haemost
asis
112.1/20
14

Schattauer
2014

combined
hormonalceptives
contraceptives did notdirectly
increase the absolute
following
risk of VTE.

early
terminated
pregnanci
es;
however,
we
observed
wide
confidenc
e limits,
due to the
limited
number of
cases, and
thus
clinicians
should
conduct
an
individual
assessmen
t in all
instances.
The
present
study had
some
limitations.
First,
the
low number
of
individuals
that
used
contraceptiv
es
early
after
pregnancy
might have
produced
misleading
results. The
conclusions
might differ
when tested
in a larger

cohort.
However,
this type of
study design
might be the
only way to
assess this
serious
complicatio
n
of
contraceptiv
e
use;
therefore,
the authors
recommend
that
additional
research
should
be
conducted in
other
countries
where these
registries are
available.
The validity
of
the
registry data
in
The
National
Registry of
Patients was
assessed by
Larsen et al.
They
reported a
positive
predictive
value
of
79.3% for
VTE
in
pregnancy
(15).
Problems
with validity
might cause
an overesti-

Petersen et al. Risk of VTE by contraceptives following pregnancy

Appendix: ICD-8 and -10 codes used (black) and excluded (grey) for identification of incident VTE.
ICD-8
ICD-10
VTE in non-pregnant women
450

DI26

Pulmonary embolism

DI801

Phlebitis and thrombophlebitis of femoral vein

451,08

DI802

Phlebitis and thrombophlebitis of other deep vessels of lower extremities

451

DI803

Phlebitis and thrombophlebitis of lower extremities, unspecified

452

DI81

Portal vein thrombosis

453,02

DI822

Embolism and thrombosis of vena cava

DI823

Embolism and thrombosis of renal vein

451,99

DI828

Embolism and thrombosis of other specified veins

451,99

DI829

Embolism and thrombosis of unspecified vein

VTE in pregnancy or puerperium

DO082

Embolism following abortion and ectopic and molar pregnancy

DO087

Other venous complications following abortion and ectopic and molar pregnancy

DO22

Venous complications in pregnancy

DO220

Varicose veins of lower extremity in pregnancy

DO221

Genital varices in pregnancy

DO221A

Perineal varices in pregnancy

DO221B

Vaginal varices in pregnancy

DO221C

Vulval varices in pregnancy

DO222

Superficial thrombophlebitis in pregnancy

DO223

Deep phlebothrombosis in pregnancy

DO224

Haemorrhoids in pregnancy

DO225

Cerebral venous thrombosis in pregnancy

DO225A

Cerebrovenous sinus thrombosis in pregnancy

DO228

Other venous complications in pregnancy

DO229

Venous complication in pregnancy, unspecified

DO87

Venous complications in the puerperium

DO870

Superficial thrombophlebitis in the puerperium

DO871

Deep phlebothrombosis in the puerperium

DO871A

Deep-vein thrombosis, postpartum

DO872

Haemorrhoids in the puerperium

DO873

Cerebral venous thrombosis in the puerperium

DO873A

Cerebrovenous sinus thrombosis in the puerperium

671,08

DO878

Other venous complications in the puerperium

671,09

DO879

Venous complication in the puerperium, unspecified

DO88

Obstetric embolism

DO880

Obstetric air embolism

DO880A

Obstetric air embolism in the puerperium

631

634,99

671,01

673

Schattauer 2014
Thrombosis and

Haemostasis 112.1/2014

Petersen et al. Risk of VTE by contraceptives following

pregnancy

increased risk of
VTE during the
mation or underestimationinitial seven weeks
of the number of cases, postpartum.
depending on whether aAttempting to exsuperficial event wasplain this, our study
coded as a VTE (leadingdistinguished
the
to overes-timation) or as abetween
insubclinical event (leadingbiologically
to
underestimation).creased risk due to
Finally, the results werepregnancy and the
based on the assumptionrisk imposed by the
that filled pre-scriptionsuse of combined
implied actual use ofhormonal
No
combined
hormonalcontraceptives.
addition
to
the
contracep-tives.
already
biologically
Overestimation of the use
of combined hormonalincreased risk was
when
contracep-tives
couldobserved
adding
the
use
of
occur
when
women
combined
hormonal
decided not to take the
pills, des-pite the filledcontraceptives.
prescription. On the otherHowever, because of
hand,
underestimationthe low numbers
could occur when womenneeded to harm in the
postpartum
used an old package ofearly
period,
our
re-sults
combined
hor-monal
supported
the
contraceptives
before
altered
filling a new prescription. recently
The major strength ofrecommendations
WHO
the study was the largefrom
suggesting
a
pause
of
number of pa-tients.
at
least
six
weeks
This was due to the
nationwide databases,before commencing
which
included
allhor-monal
pregnancies
thatcontraception.
Following early
occurred in women aged
terminated
15-49 throughout a 15the
year
period.pregnancies,
numbers
needed
to
Additionally, the register
approach eliminated po-harm showed no
changes
tential recall or non-major
throughout
the
responder
biases
follow-up
peri-od;
compared to different
meth-ods. This studythis result suggested
combined
was one of the first tothat
hormonal
examine
the
risk
associated with the usecontraceptives were
of combined hormonalrelatively safe to use
these
contraceptives
duringunder
the postpartum period. conditions supporting
the clinical option of
immediate
commencement.

Conclusion

In our national setting,

this study illustrated an

Conflicts of
interest

None declared.

References
1. Bodker

B, Hvidman
L, Weber T, et al.
Maternal deaths in
Denmark 2002-2006.
Acta Obstet Gynecol
Scand 2009; 88: 556562.

2. Chang

J, Elam-Evans
LD, Berg CJ, et al.
Pregnancy-related
mortality
surveillance--United States,
1991--1999. MMWR
Surveill Summ 2003;
52: 1-8.

3. Cantwell

R, CluttonBrock T, Cooper G, et
al. Saving Mothers'
Lives:
Reviewing
maternal deaths to
make
motherhood
safer: 2006-2008. The
Eighth Report of the
Confidential
Enquiries
into
Maternal Deaths in
the United Kingdom.
Br J Obstet Gynaecol
2011; 118 (Suppl 1):
1-203.

4. Manzoli

L, De Vito
C, Marzuillo C, et al.
Oral contraceptives
and venous thromboembolism:
a
systematic review and
meta-analysis. Drug
Saf 2012; 35: 191205.

5. Jackson

E, Curtis KM,
Gaffield ME. Risk of
venous
thromboembolism
during the postpartum
period: a systematic
review. Obstet Gynecol
2011; 117: 691-703.

6. Update to CDC's U.S.

Medical
Eligibility
Criteria
for
Contraceptive
Use,
2010:
revised
recommendations for
the
use
of
contraceptive methods
during the post-partum
period. Morb Mortal
Wkly Rep 2011; 60:
878-883.

7. Heit

JA, Kobbervig

8.

CE, James AH, et al.

Haemost 2011; 106:
Trends in the incidence
304-309.
of
venous 11.
Lidegaard O,
thromboembolism during
Lokkegaard
E,
pregnancy or postpartum:
Svendsen AL, et al.
a 30-year populationHormonal
based study. Ann Intern
contraception and risk
Med 2005; 143: 697-706.
of
venous
McColl MD, Ramsay JE,
thromboembolism:
national
follow-up
Tait RC, et al. Risk factors
study. Br Med J 2009;
for pregnancy associated
339: b2890.
ve-nous thromboembolism.
Thromb Haemost
78: 1183-1188.

1997;

9. Salonen Ros H, Lichtenstein

P, Bellocco R, et al.
Increased risks of circulatory
diseases in late pregnancy
and
puerperium.
Epidemiology 2001; 12: 456460.

10.

12.

Lidegaard
O,
Nielsen LH, Skovlund
CW, et al. Risk of
venous
thromboembolism from
use
of
oral
contraceptives
containing
different
progestogens
and
oestrogen doses: Danish
cohort study, 2001-9. Br
Med J 2011; 343: d6423.

Virkus
RA,
Lokkegaard EC, Bergholt
T,
et
al.
Venous
Danish
thromboembolism
in 13.
guidelines
of
pregnant and puerperal
postpartum
care.
women in Denmark
Sundhedsstyrelsen;
1995-2005. A national
2001 [cited 2012
cohort study. Thromb
08.03];
Available

from:
http://www.sst.dk/pub
l/Publ2001/Svangreo
msorg/html/
index17.html.

14.

Sultan
AA,
Tata LJ, West J, et al.
Risk factors for first
venous
thromboembolism
around pregnancy: a
population-based
cohort study from the
United
Kingdom.
Blood 2013; 121:
3953-3961.

15.

Larsen
TB,
Johnsen SP, Moller
CI, et al. A review of
medical records and
dis-charge summary
data found moderate
to high predictive
values of discharge
diagnoses of venous
thromboembolism
during pregnancy and
postpartum. J Clin
Epidemiol 2005; 58:
316-319.

Thrombosis and
Haemostasis 112.1/2014
Schattauer 2014