CLINICAL OBSTETRICS AND GYNECOLOGY

Volume 50, Number 4, 850–867

r 2007, Lippincott Williams & Wilkins

Study Design to Evaluate the Safety

and Effectiveness of Hormonal
Contraception for Women
NATHALIE KAPP, MD, MPH,* KATHRYN M. CURTIS, PhD,w
and LYNN BORGATTA, MD, MPHz
*Department of Reproductive Health and Research, World Health
Organization, Geneva, Switzerland; w Division of Reproductive Health, Centers
for Disease Control and Prevention, Atlanta, Georgia; and z Department of
Obstetrics and Gynecology, Boston University School of Medicine, Boston,
Massachusetts

Abstract: A discussion of the issues that affect the
study design of hormonal contraception is presented,
including ethical issues, measurement of contraceptive
efficacy. Observational studies, experimental studies,
and meta-analysis in hormonal contraception are also
reviewed.
Key words: study design, hormonal contraception
Introduction
Rigorous, controlled experimentation is rela-
tively new to the field of medicine. The first
standardized use of research was in agriculture,
well before its adoption in medicine in the last
half of the 20th century.1 The US Department of
Health and Human Services defines research as a
systematic evaluation designed to produce gen-
eralizable results.2 According to this definition,
research must be systematic, or performed in a
rigorous manner, and the design must allow
interpretation that contributes to general scien-
tific knowledge. Therefore, the design of a study
is its most critical element. With thoughtful
application, a study’s results can be used to
answer a specific question.
In this report, we discuss issues that are
critical to the design of hormonal contraceptive
Correspondence: Dr Nathalie Kapp, MD, MPH,
Department of Reproductive Health and Research, World
Health Organization, 20 Avenue Appia, CH-1211 Geneva
27, Switzerland. E-mail: kappn@who.int
The findings and conclusions in the article are those of
the authors and do not necessarily represent the views of
the Centers for Disease Control and Prevention or the
World Health Organization.
studies. We begin with a discussion of ethical
considerations, some of which are unique to the
study of contraception. We then review exposure
and outcome measurements, followed by classi-
fications of study designs using published hor-
monal contraception studies as examples. We
conclude by considering the use of systematic
reviews to compile data from multiple studies.
Ethical Considerations in Studies
of Hormonal Contraception
Some unique ethical considerations pertaining
to randomization, efficacy assessment of, and
the risk of sexually transmitted infection (STI)
and HIV acquisition can affect the design of
hormonal contraception studies. We consider
these issues in the context of the general ethical
principle of beneficence, and the ethical obliga-
tion of researchers to maximize benefits and
minimize harms.3
The ethical acceptability of randomizing wo-
men to different contraceptive methods depends
on whether the method to which the woman is
assigned is safe for and acceptable to her and
provides her with adequate protection against
pregnancy. Women have demonstrated their
willingness to be randomized to dissimilar meth-
ods.4 Specifically, women generally accept ran-
domization to different methods with similar
routes of administration, efficacy, and side ef-
fects. For example, they are usually willing to
be randomized to different types of pills, the
patch and the vaginal ring, or different types of

CLINICAL OBSTETRICS AND GYNECOLOGY / VOLUME 50 / NUMBER 4 / DECEMBER 2007

850

injectable contraceptives. Considerations for
women who are deciding whether they are will-
ing to be randomized between methods include
whether the methods have similar high effective-
ness against pregnancy, similar return to fertility
after discontinuation, and similar degrees of
discrete use (use of a method without the knowl-
edge of her partner or family is important to
some women). However, randomization to dif-
ferent contraceptive methods must be consid-
ered carefully in study planning.
Studies of contraceptive efficacy also raise
important ethical considerations. A new method
that has biologic plausibility and safety data but
few or no data on efficacy should be tested in
people who are willing to accept the risk of an
unknown failure rate; specifically, they should
be prepared to rear a child or able to obtain a
safe abortion in the event of unintended preg-
nancy. These issues should be carefully ex-
plained during the informed consent process of
such studies.3
Efficacy studies should also include assess-
ment of actual risk for pregnancy by tracking
coital frequency. Confidentiality must be as-
sured for participants in trials to maximize the
accuracy of such data collected. The cultural
contexts of the study sites play a role in gathering
information such as types and frequency of
sexual acts and number of sexual partners.
Efficacy studies are reliant on the accuracy
of these data to some extent, as actual expo-
sure to the risk of pregnancy is inherent in
measuring the occurrence of pregnancy as a
study outcome.
Hormonal contraceptive studies must con-
sider the issue of acquisition of and protection
against STI/HIV in its participants. Sexually
active women using hormonal contraceptives
may be at risk of STI and HIV acquisition,
depending on their sexual behavior. In studies
measuring contraceptive efficacy, as a study
outcome use of dual protection with condoms
may reduce the power of the study to determine
the effectiveness of the contraceptive, as consis-
tent use of condoms markedly decreases the risk
of pregnancy.5,6 To minimize STI exposure
while gathering data on the method’s pregnancy
rates, efficacy studies are best conducted in
couples in monogamous, low-risk relationships.
In contraceptive studies, STI screening and
treatment by referral or by study centers should
be available to all participants. Standard coun-
seling for participants in hormonal contracep-
tive study groups should include that hormonal
contraceptives do not protect against STI.3
Hormonal Contraception 851
Measurement Issues
The study of hormonal contraception, regard-
less of its design, involves measuring exposures,
outcomes, and potential confounders, such as
adherence to the contraceptive method, bleeding
patterns, and sexual behaviors, which can be
difficult to ascertain accurately.
MEASURING EXPOSURE
Exposure is often poorly assessed in investiga-
tions of hormonal contraception.7 For example,
although the duration of exposure to hormonal
contraceptives has an impact on changes in bone
mineral density (BMD), many studies of the
impact of contraceptives on BMD compare
current users to nonusers without assessing past
history of use.8 Information on the length of use
is important; taking depot-medroxyprogester-
one acetate (DMPA) for 6 months produces a
different effect on bone than using DMPA for 4
years because DMPA’s effects are ongoing and
cumulative.9
In some cases, contraceptive use has a rela-
tionship with a medical condition that is main-
tained over time, such as total length of
combined oral contraceptive (COC) use and
ovarian cancer risk.10 Because women often
need to control their fertility over the 30 or more
years of their reproductive lives, investigators
should collect information on their cumulative
exposure in addition to present contraceptive
use to clarify the true differences in exposures
between groups.
MEASURING ADHERENCE
To assess exposure to a hormonal contraception
regimen accurately, investigators must take ad-
herence into account.11 Women who fail to take
some of their pills every month or who do not
receive their contraceptive injections on time
could have lower effectiveness and acceptance
rates and different side-effect profiles than wo-
men who follow their contraception regimens as
directed. Investigators should attempt to mea-
sure adherence because it influences such out-
comes as pregnancy and side effects, and
indicates whether use in women with a medical
problem is truly safe when used as prescribed.
Such methods as daily diaries, electronic diaries,
and pill counting have been employed to mea-
sure exposure and adherence to contraceptive
regimens while minimizing the potential for re-
call bias that can arise when women are asked at
a later time to estimate the number of missed
pills or late injections.11
852 Kapp et al
Pregnancy rates are the most obvious out-
come that can vary as a factor of nonadherence
with a contraceptive regimen, as nonadherence
contributes to the ‘‘typical use’’ estimate of
effectiveness. Typical use refers to the method’s
failure rate for the use of a method during actual
use (which includes incorrect and inconsistent
use); in contrast, a method’s perfect use failure
rate is based on correct and consistent use.5
Whether a pregnancy or any other outcome in
a study participant that can vary by degree of
adherence is due to the difficulty of adherence or
the contraceptive’s intrinsic properties should be
assessed by carefully measuring adherence.
Different adherence rates for different con-
traceptive methods can make it challenging to
compare efficacy rates, side effects, ease of use,
and safety. For example, a 2002 study that
randomized 610 healthy adult women to use
the contraceptive patch or an oral contraceptive
found that, based on diary cards kept over 4
cycles, women who used the patch had an ad-
herence rate of more than 90% compared with
77% in women taking the oral contraceptive
pill.12 This is unsurprising given the opportu-
nities to err in a 28-day cycle; once a week for
patch-users, and daily for pill-users. The differ-
ential adherence demonstrated did not translate
into outcome differences (rates of ovulation); it
merely suggested that a method that requires
conscious action once a week as opposed to once
a day is associated with higher adherence rates.
MEASURING METHOD SWITCHING AND
DISCONTINUATION
A constant challenge posed by contraceptive
trials, particularly those with several years of
follow-up, is contraceptive method discontinua-
tion and switching. Discontinuation rates of hor-
monal contraceptive methods vary by method.
For example, in the United States, discontinua-
tion rates at 1 year range from 16% for the
contraceptive implant to 44% for oral contra-
ceptives.13 Moreover, discontinuation and preg-
nancy rates during research studies tend to be
lower than in when the method is general use.7,14
If discontinuation rates are not measured, they
can threaten a study’s internal validity because
the outcomes measured will not reflect the contra-
ceptive agent’s effect. To control for the effects of
discontinuation and switching, researchers
should assess women’s contraceptive use fre-
quently. Investigators should consider using
time-varying variables to interpret such outcomes
as side effects, bleeding profiles, or adverse events
by varying contraceptive method use over time.
MEASURING VAGINAL BLEEDING
The disturbance in vaginal bleeding induced by
many methods of contraception affects their
acceptability and is a frequently cited reason
for discontinuation by users.15 The lack of stan-
dardized techniques for data collection, report-
ing, or analysis related to spotting and bleeding
has impeded researchers’ ability to combine and
compare data on bleeding and spotting from
different trials. The World Health Organization
instituted in 1987 the criteria of Table 1 as a
reference for bleeding and spotting terminology
and standardization.16 Although these criteria
created standard definitions of types of bleeding
and spotting, standardization of data collection,
reporting or analysis were not concurrently de-
veloped, and has impeded combining and com-
paring data between trials. However, new
guidance has recently been published.17 As this
guidance becomes integrated into studies asses-
sing bleeding patterns, it will be easier to com-
pare data from different studies thereby
improving interpretation of their results. Even-
tually, better interpretation of results will lead to
more accurate information about a contracep-
tive method, enhancing the counseling women
receive regarding its use.
MEASURING OUTCOMES
Outcomes in all clinical research should be clear,
measurable in all participants, and, preferably,
standardized.14 An example in contraceptive
studies of such an outcome is pregnancy; it has
TABLE 1. WHO Vaginal Bleeding Pattern
Definitions
Terminology Definition
Bleeding Vaginal blood loss requiring the
use of sanitary protection
Spotting Vaginal blood loss not requiring the
use of sanitary protection
Bleeding day Day on which bleeding is recorded
Spotting day Day on which spotting is recorded
Bleeding or One or more consecutive days of
spotting bleeding, spotting, or both
episode bounded on either end by days
without bleeding. Episodes can
be consecutive or separated by
only one bleeding-free day
Bleeding-free or Day without bleeding or spotting
spotting-free
day
Bleeding or A bleeding or spotting episode and
spotting the bleeding-free or spotting-free
segment interval that immediately follows
Adapted from Belsey et al.16

a standardized diagnosis and it is measurable.
Other examples may include cardiovascular out-
comes such as deep venous thrombosis, myocar-
dial infarction, or stroke, which have
standardized and validated measurements pub-
lished in the literature.18 The criteria for diag-
nosis of these diseases have been used in
evaluating cardiovascular outcomes in investi-
gations such as the Transnational European
study of ischemic stroke in COC users.19 As an
outcome, with the criteria referenced from its
validated source, it is a direct one: it measures the
event of interest, and it is clear and measurable.
Outcome issues explored thoroughly outside
the contraceptive literature include that of sur-
rogate end points.20 Rare, or time-intensive dis-
eases may be difficult to measure, making their
use attractive. Examples of the use of surrogates
are the measurement of BMD or lipid profiles in
the place of fracture or cardiovascular dis-
ease.9,20–24 The effects of hormonal contracep-
tion on the development of osteoporotic fracture
or cardiovascular disease, are difficult to assess
in a study, especially as these outcomes occur
decades after use of the contraceptive; for this
reason, investigators have assessed BMD or
lipid profile to evaluate future disease risk. How-
ever, a true surrogate marker is one that not only
correlates with the clinical event to be measured,
but can accurately predict its occurrence.25 Both
BMD and lipid profiles are not generally accu-
rate predictors of either fracture or cardiovas-
cular disease, particularly in young women. The
nonvalidated use of surrogate end points can
lead to inaccurate and irrelevant data. Their use
in research on hormonal contraceptive use
should therefore be minimized. Instead, research
in this field should focus on measuring clinically
relevant events whenever possible.
Efficacy
LIFE TABLES VERSUS PEARL INDEX
Measuring the outcome of contraceptive effi-
cacy trials varies greatly by the method of calcu-
lation chosen. The Pearl index is the ratio of
pregnancy occurrences divided by the total num-
ber of years in which all study participants were
studied. Pearl rates decline over time because
participants who are most likely to become
pregnant usually do so early on, leaving more
consistent, or less fecund, users in the calcula-
tion.5 Pearl indices were used in early trials of
contraceptive efficacy but their use is no longer
Hormonal Contraception 853
recommended because of this serious error: the
longer the observation time, the lower the esti-
mated pregnancy rate. Because having a bias in
favor of less fertile women and successful contra-
ceptive users, the Pearl rate is time dependent,
so it is impossible to compare the Pearl rate
results from studies with different durations of
follow-up.5,26
A more appropriate measure of contraceptive
efficacy is the life table, a type of survival ana-
lysis in which each month of the study is treated
as a separate data point. The monthly propor-
tions are multiplied together to estimate the
cumulative proportion of women who conceive
over a period of contraceptive use. Only partici-
pants at risk of pregnancy who remain in the
study are considered for that time period, while
all participants are accounted for (regardless of
the reason for leaving the study, because the
results from all months are included).27 Life
table pregnancy rates vary from 0 to 100, in-
crease over the length of the study, include
women who leave the study for reasons other
than pregnancy, and are comparable between
studies.26 For these reasons, statisticians recom-
mend the use of life tables in contraceptive
efficacy trials.5,26,27
EFFICACY AND EFFECTIVENESS
In the public health literature, efficacy studies
are those that enroll a homogeneous group and
administer a standardized, rigorous interven-
tion.28 Studies of efficacy are intense, complex,
and highly standardized. In contrast, effective-
ness studies enroll a heterogeneous group, such
as in postmarketing surveillance, and use a more
real-world approach.28 Perfect use failure rates
are based on data from efficacy studies of wo-
men who use the method as directed. In contrast,
typical use failure rates, which are derived from
effectiveness studies, indicate a method’s effec-
tiveness during actual use and are calculated
from data which include results from both
inconsistent and incorrect use.5,28
Study Design
In any research context, the appropriate study
design should be selected on the basis of the
study question the exposure and outcome to be
measured, and the appropriate study popula-
tion.14 In hormonal contraception studies, the
study population is generally women of repro-
ductive age although a narrower age range might
be of interest for some studies. In experimental
contraceptive studies, the intervention is usually
854 Kapp et al
the contraceptive method and the comparison
group could be one that is using a similar or
dissimilar type of contraceptive, depending on
the study objective. The usual outcomes of inter-
est are safety, efficacy, acceptability, side effects,
and discontinuation or continuation rates.
Two broad classifications of study design
exist: experimental and observational.29 We dis-
cuss both of these study design types in detail
below and provide examples of each type. Pre-
paring for a study of any design may involve
conducting a pilot study, which is a small ob-
servational study or randomized controlled clin-
ical trial (RCT) to establish feasibility, efficacy,
safety, and optimal recruitment techniques.14
EXPERIMENTAL STUDIES: ISSUES AND
EXAMPLES
The distinguishing feature of experimental stu-
dies is that the investigator assigns study parti-
cipants to an intervention or treatment. These
studies are, by definition, prospective and, when
properly executed, provide the strongest empiri-
cal evidence available to answer a single study
question. When experimental studies involve
randomization, they have a greatly reduced like-
lihood of selection bias (a bias introduced by
including people with different characteristics in
the study and control groups) because partici-
pants do not choose their exposure and, because
both known and unknown confounders are
theoretically balanced between the groups stu-
died, the only difference between the groups is
the type of treatment received. In addition to
RCTs, experimental study designs which are not
discussed in this paper include cross-over stu-
dies, in which subjects serve as their own con-
trols, and nonrandomized trials, in which
subjects are assigned to the study treatment or
intervention by a nonrandom method (such as
by birth dates or medical record numbers).30
RANDOMIZED CONTROLLED TRIALS
IN HORMONAL CONTRACEPTION
Unlike observational trials in which confound-
ing variables must be anticipated and controlled,
RCTs minimize the effect of both known and
unknown confounders, when randomization is
done correctly.30 This increases the likelihood
that the 2 comparison groups will differ only in
their intervention status. Disadvantages of
RCTs are their expense, the ethical issues asso-
ciated with randomization, the difficulty of
implementing and maintaining blinding proce-
dures in RCTs that utilize double-blinding (to
ensure that neither participants nor study per-
sonnel are aware of group assignments), and the
prohibitively large sample sizes that can be
required to determine differences between un-
common or rare events.31,32
Recent initiatives to improve the quality of
RCTs by a community of experts in the field
resulted in the creation of the Consolidated
Standards of Reporting Trials (CONSORT)
statement (Fig. 1).33 The CONSORT guidelines
have been adopted as necessary standards for the
publication of RCTs in most journals. Evidence
indicates that the CONSORT guidelines im-
prove the quality of published trials in journals
requiring their use.34
The advantages and disadvantages of RCTs
are demonstrated in a study of a progesterone
receptor modulator for emergency contracep-
tion.35 The study question was whether the
progesterone antagonist CDB-2914 was as ef-
fective as levonorgestrel for preventing preg-
nancy after unprotected intercourse. Secondary
outcomes included nausea and vomiting and
menstrual length changes. The study recruited
1672 women not taking hormonal contraception
who were seeking emergency contraception
within 72 hours of unprotected intercourse.
The women were randomized to receive either
2 doses of 0.75 mg levonorgestrel or 1 dose of
50 mg CDB-2914. Both investigators and parti-
cipants were blinded to regimen and, after treat-
ment, all women planned to abstain from
intercourse or use barrier protection for future
pregnancy prevention.
After a negative urine pregnancy test, the
women swallowed their medication in front of
the investigator. Participants took the second
capsule (either the second dose of levonorgestrel
or a placebo) 12 hours later at home and re-
corded the time in their diary. The women were
asked to return 5 to 7 days after their next
expected menses. Urine pregnancy tests were
repeated at follow-up and if the women’s menses
had occurred, their participation was complete.
If not, they were seen weekly until their menses
returned, they were declared amenorrheic, or
their pregnancy test was positive.
The question to be answered by this RCT was
whether the progesterone antagonist CDB-2914
prevents pregnancy among women who have
had unprotected intercourse and a levonorges-
trel regimen. This question defines the group to
be studied (women with recent unprotected in-
tercourse), the intervention (CDB-2914), the
outcome (pregnancy), and the comparison treat-
ment (levonorgestrel regimen). The control
group (women taking levonorgestrel) should
 Hormonal Contraception 855

Assessed for eligibility (n =…)

Excluded (n =….)
Enrolment

Did not meet inclusion

criteria (n = ….)
Refused to participate
(n =….)
Other reasons (n =….)

Randomized (n =….)

Allocated to intervention (n = ….) Allocated to intervention (n = ….)

Allocation

Received allocated Received allocated

intervention (n =….) intervention (n =….)
Did not receive allocated Did not receive allocated
intervention (give reasons) intervention (give reasons)
(n =….) (n =….)
Follow-up

Lost to follow-up (give reasons) Lost to follow-up (give reasons)

(n =….) (n =….)
Discontinued intervention Discontinued intervention
(give reasons) (n =….) (give reasons) (n =….)
Analysis

Analyzed (n =….) Analyzed (n =….)

Excluded from analysis Excluded from analysis
(give reasons) (n =….) (give reasons) (n =….)

FIGURE 1. The CONSORT flow diagram for randomized trials.21

have the same characteristics as the intervention
group with the exception of their treatment
allocation—women who were neither pregnant
nor using hormonal contraception and were
seeking emergency contraception immediately
after unprotected intercourse.
The outcome in this study was clear, measur-
able, and assessable equally in both groups.
These necessary elements allow for a power
calculation (Table 3). This trial used a noninfer-
iority design: the intervention could perform no
worse than the control, so it was calculated
somewhat differently. Still, the error rates apply;
an a rate of 5% and power of 80% were chosen,
and the maximal amount of acceptable differ-
ence in rates between treatment groups was
determined by the authors was chosen to be
2%. On the basis of this calculation, a sample
size of 1440 women could be used to detect an
effectiveness rate for the intervention that was
more than 2% lower than in the control group.
The recruited sample size was 10% higher than
the number needed according to this estimate to
account for drop-outs and losses to follow-up.
The 2 medications had similar results, with
pregnancies occurring in 0.9% of women in the
intervention group and 1.7% of women in the
control group. Nausea occurred more frequently
in the intervention group and all women, regard-
less of treatment, experienced variation in their
cycle length after treatment.35
When randomization, or random allocation,
is used successfully, outside factors cannot ac-
count for the group allocation and, given enough
information, a person cannot predict a particu-
lar allocation.30 Therefore, use of birthdays,
856 Kapp et al
alternating enrollment or use of the date of
enrollment or treatment are not random. Ran-
dom numbers may be generated in several ways,
such as flipping a coin. However, over time,
randomization may result in uneven numbers
between groups, which can make statistical ana-
lysis more difficult. Researchers prefer random
allocation methods that assign equal numbers,
or almost equal numbers, of participants to each
group.30 Some recommended randomization
methods include the use of random number
tables and computer-generated randomization,
which can generate permuted block schemes.
The CDB-2914 study used a block scheme of 8;
within each block, 4 indicated levonorgestrel,
whereas 4 indicated CDB-2914, randomly with-
in each segment. In this scheme, no more than 8
subjects in a row would ever receive the same
treatment.
Inadequate concealment of participant allo-
cation can introduce bias into a study7,36; trials
in which assignment is poorly concealed have the
most heterogeneous results. Unless the authors
stipulate the methods used to ensure that their
randomization scheme was free of bias by in-
dicating that participants and investigators did
not know the groups to which participants were
assigned, it is difficult to determine whether the
randomization scheme was biased.7,36 The most
well-accepted method is to use sequentially num-
bered, opaque, sealed envelopes that contain the
group allocation or drug. Each time a new
participant enrolls in the trial, the next envelope
is opened and the participant is assigned to the
group indicated in the envelope. In the CDB-
2914 study, the capsules taken by participants in
both the control and investigative groups looked
identical and were packaged in identical bottles.
This ensured that both participants and study
personnel were unaware of the group to which
each participant had been assigned.
Using an unpredictable, random sequence
should prevent preferential enrolment by mak-
ing it impossible for investigators to have
prior knowledge of the groups to which partici-
pants will be assigned.36 Allocation should be
concealed rigorously and cleverly to make it
impossible for investigators to determine assign-
ment. The security of each assignment can be
increased by using audit trails, such as docu-
menting the name of each participant on the
envelope before opening it, using carbon-copy
papers placed inside, central randomization that
delivers the allocation only after the subject’s
enrollment is assured, or pharmacy-controlled
allocation.36
Demographic profiling can be used to assess
externally whether the randomization procedure
was effective; an effective randomization scheme
should result in identical treatment groups (with
means and SDs), which implies that unmeasured
differences in each group are similar.30
The outcome of interest in contraceptive
trials can vary widely. Efficacy trials measure
pregnancy as an outcome. Traditionally, in
emergency contraceptive trials, the absence of
pregnancy is calculated by determining the prob-
ability of conception on the basis of the date of
unprotected intercourse37 and the efficacy is
calculated by subtracting the number of ob-
served pregnancies from the expected number
to yield the number of pregnancies averted by
emergency contraceptive use. Trials that do not
measure efficacy might use bleeding changes,
side effects, or measures of safety as outcomes.
Measuring nonstandardized outcomes makes
interpretation within the context of existing
research difficult.
Data should be analyzed by randomization
group, regardless of whether participants actu-
ally adhered to the protocol; this is known as an
intent-to-treat analysis.29 If participants are ex-
cluded from analysis after randomization, the
value of the randomization is threatened. Intent-
to-treat analysis is preferable for the final ana-
lysis because it guards against nonrandom losses
of participants from treatment groups.38 Post-
randomization protocol deviations should be
analyzed within the relevant treatment group
to honor the value of the a priori randomization
scheme, and authors should state this expli-
citly.38
A particularly difficult issue in RCTs of con-
traceptive methods is contraceptive method
switching. Women who have been randomized
to a particular contraceptive method sometimes
stop using the assigned method and start using a
new one or stop using contraception altogether.
Because method switching is so common, some
researchers argue that over time, women essen-
tially reassign themselves to their preferred
group. Although this is a common problem,
evidence indicates that women can be success-
fully randomized between different contracep-
tive methods, at least for 1 year of follow-up.4
However, if participants switch or discontinue
contraception frequently, both an intent-to-
treat analysis and an actual-use analysis (on
the basis of the actual reported exposure to the
contraceptive method) can be informative.
During analysis, steps should be taken to
control for confounding, which occurs when a

variable related to both the exposure and out-
come distorts the relationship between the two.
By gathering information about parameters that
may be confounders, it is possible to take this
into account in the analysis phase.29 We discuss
confounders in more detail in the context of
observational studies because they are more
common in such studies than in RCTs.
OBSERVATIONAL STUDIES: ISSUES
AND EXAMPLES
Descriptive observational studies include case
reports and series, descriptive incidence and
prevalence studies, and ecologic studies. These
studies are used to assess patterns of disease
occurrence and are useful in generating hypoth-
eses about associations between exposures and
health outcomes. Analytic observational studies
(eg, cross-sectional, case-control, and cohort
studies) are used to test hypotheses by evaluating
and quantifying the joint distribution between
an exposure and specific health outcomes. These
studies can be useful when introducing a parti-
cular exposure is unethical or for studying dis-
eases or health outcomes that are rare or take
several years to develop (case-control design).
Observational studies are also useful to examine
rare exposures (cohort design) or to elucidate
significant associations in preparation for devel-
oping the hypothesis for an experimental study.
In general, observational studies provide weaker
empirical evidence for a study question than
experimental studies because of the greater like-
lihood that differences observed between groups
are due to bias, including selection bias, mis-
classification, and confounding.
Most investigational studies, including hor-
monal contraception studies, are observational
studies.31 Guidance for reporting observational
studies is currently under development by an
international working group (Table 2).40 As
these guidelines are finalized and disseminated,
it is likely that journals will require authors to
adhere to them, as was done with the CON-
SORT guidelines for RCTs.33
Cross-sectional, case-control, and cohort stu-
dies are all types of analytic observational stu-
dies that will be discussed here in some detail. In
observational studies, the assignment of a parti-
cipant to a given treatment or exposure is not
under the control of the investigator. Outcomes
are usually presented as proportions or numer-
ical values with confidence intervals (CIs), which
are ranges of values that are likely to contain the
true measure of association. The confidence
level is determined arbitrarily but is usually
Hormonal Contraception 857
chosen to be 95%, which means that if the study
were conducted repeatedly, 95% of the CIs
would contain the true point estimate for the
study population.29
CROSS-SECTIONAL STUDIES IN
HORMONAL CONTRACEPTION
Cross-sectional studies provide a simultaneous
assessment of both the exposure and the out-
come.41 These studies therefore produce a
‘‘snapshot’’ of the frequency and characteristics
of exposures and health outcomes in a popula-
tion at a particular point in time. Cross-sectional
studies are generally inexpensive and are con-
ducted over a short time frame. However, be-
cause these studies measure the exposure and the
outcome at the same time, they cannot be used to
establish temporality (whether the exposure ex-
isted before the outcome). For this reason, cross-
sectional studies are most useful for generating
hypotheses that can be tested in subsequent
experimental studies.41
An example of a cross-sectional hormonal
contraception investigation is a study examining
whether hormonal contraceptive use among
HIV-1-infected women could affect cervical
shedding of HIV-1 DNA.42 In this study, the
investigators analyzed endocervical and vaginal
samples from women with previously diagnosed
HIV-1 who were seeking care at an STI clinic in
Kenya. The goal was to establish patterns of
cervical shedding of HIV-1 by hormonal contra-
ceptive use DMPA, COC, or none.42 The inves-
tigators found that women who used combined
oral contraception and, to a lesser degree,
DMPA had higher levels of HIV-1 infected cells
in their endocervical or vaginal samples.
This study on cervical shedding of HIV-1
DNA provided a hypothesis that hormonal con-
traception may lead to increased transmission of
HIV-1 to uninfected partners through the in-
creased presence of HIV-1 DNA in the genital
tract. However, this cross-sectional analysis
could not determine whether the hormonal con-
traceptive use occurred before the occurrence of
higher levels of HIV-1 DNA in the genital tract
or vice versa. To address this issue, a prospective
study was conducted to determine whether the
increased shedding was caused by the initiation
of hormonal contraception.43 This study found
that women initiating hormonal contraception
(DMPA or combined oral contraception) in
Kenya demonstrated an increase of HIV-1
DNA infected endocervical cells from 42% to
52% (P = 0.03); however, a corresponding
increase of HIV-1 RNA was not detected.
858 Kapp et al

TABLE 2. Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement:

Checklist of Essential Items (Version 3)39
Cohort Study Case-control Study Cross-sectional Study

Title and abstract Identify the report as a cohort Identify the report as a case-control Identify the report as a cross-
study. study. sectional study.
Introduction
Background/ Explain scientific background and rationale for the investigation being reported.
rationale
Objectives State specific objectives including prespecified hypothesis.
Methods
Study design Present key elements of study design. State purpose of original study, if article is from an ongoing
study.
Setting Describe setting, locations, and dates defining periods of data collection
Participants (a) Give inclusion and (a) For cases and controls, give (a) Give inclusion and
exclusion criteria, inclusion and exclusion criteria, exclusion criteria, sources
sources and methods of sources and methods of selection. and methods of selection.
selection.
(b) Give period and methods (b) Give precise diagnostics criteria for
of follow-up. cases, and rationale for choices of
controls.
Funding Give source of funding and role of funder(s) for the study and, if applicable, the original study on
which the present article is based.
Results
Participants (a) Report the number of individuals at each stage of the study, eg, numbers potentially eligible,
examined for eligibility, confirmed eligible, included in the study, completing follow-up, and
analyzed.
(b) Give reasons for nonparticipation at each stage.
(c) A flow diagram is recommended.
(d) Report dates defining period of recruitment.
(e) Give distribution of number of controls per case for matched studies.
Descriptive (a) Give characteristics of study participants and information on exposures and potential
data confounders.
(b) Indicate for each variable of interest the completeness of the data.
(c) Summarize average and total amount of follow-up and dates defining follow-up.
Outcome data Report number of outcome Report numbers in each exposure Report number of outcome
events or summary category. event or summary
measures over time. measures.
Main results (a) Give unadjusted and confounder adjusted measures of association and precision (eg, 95%
confidence intervals). Make clear which confounders were adjusted for and on what grounds they
were included.
(b) For comparisons using categories derived from quantitative variables, report the range of values
or median value in each group.
(c) Translate relative measures into absolute differences, for a meaningful risk period that does not
extend beyond the range of the data.
(d) Report results standardized to confounder and modifier distributions for realistic target
populations.
Other analyses Report any other analyses performed, including subgroup and sensitivity analyses.
Discussion
Key findings Summarize key results with reference to study hypotheses.
Limitations (a) Discuss limitations of the study, including sources of potential bias or imprecision, and problems
that could arise from multiplicity of analyses, exposures, and outcomes. Discuss both direction
and magnitude of any potential bias.
(b) Discussion of limitations should not be used as a substitute for quantitative sensitivity analyses.
Generalizability Discuss the external validity of the study findings.
Interpretation Give a cautious overall interpretation of the results in the context of current evidence and study
limitations, paying attention to alternative interpretations.

As HIV-1 RNA levels in the serum correlates to HIV-1 DNA. Further investigations of the role
infectivity, they may also play a role in the of hormonal contraception in women infected
genital tract, in combination or separately from with HIV are ongoing.

CASE-CONTROL STUDIES IN HORMONAL
CONTRACEPTION
In case-control studies, investigators compare a
group of people with a specific outcome to a
group of people without that outcome. De-
scribed as ‘‘research in reverse,’’ case-control
studies identify an outcome before determining
whether those who had the outcome (cases) and
those who did not (controls) had different
exposures.44 Case-control designs are used
to explore associations between outcomes and
exposures.41 They are relatively inexpensive and
can usually be completed in a short timeframe.
These studies are particularly suitable for study-
ing rare diseases and diseases with a long latency
period.29
In case-control studies, differences observed
between groups are expressed as odds ratios
(ORs), or ratios between the odds of exposure
among the cases and among the controls. If the
outcome is rare (<5%), ORs can approximate
relative risks (RRs) (the risk of disease in the
exposed group divided by risk of disease in
the unexposed group) because the events in the
control group are similarly infrequent.29
Despite the design’s advantages, it cannot be
used to measure disease incidence directly.44 In
addition, these studies are more likely to intro-
duce selection and recall bias than prospective
designs because of the manner in which the study
groups are chosen and how past exposures are
assessed.44,45
An example of a well-designed case-control
study is the Risk of Arterial Thrombosis in
Relation to Oral Contraceptives (RATIO)
study,35,46 which investigated whether women
using oral contraceptives who also had a pro-
thrombotic condition (such as factor V Leiden
mutation) had an increased risk of ischemic
stroke compared with women without these risk
factors. Using data from 9 hospitals in the
Netherlands, the investigators identified 179
cases of ischemic stroke over a 5-year period in
women aged 18 to 53 and 767 controls in the
same age group using random-digit dialing. The
exposure assessed in each group was the pre-
sence of a thrombogenetic mutation (by DNA
analysis) and the women’s use of oral contra-
ceptives in the month preceding the stroke for
cases or in the past month for controls. For the
analysis, cases were matched to controls by age,
geographic area, and year of stroke.
The resulting OR was 2.6 (95% CI 1.7-4) for
women who used COCs and had no factor V
Leiden mutation, 0.4 (95% CI 0.1-1.9) for those
who did not use COCs but had factor V Leiden
Hormonal Contraception 859
mutation, and 11.2 (95% CI 4.0-29.0) for women
who used COCs and had the factor V Leiden
mutation. Therefore, the risk of ischemic stroke
in women who used COCs and carried the factor
V Leiden mutation was 11.2 times as higher than
as for women without either risk factor.
Aspects of this study that made the case-
control design particularly suitable are the un-
common outcome, ie stroke in women during the
reproductive years. Identifying a control group
is often the most difficult aspect of designing a
case-control trial as the control group should be
representative of all those at risk for the outcome
described (in this case, all women of reproduc-
tive age); essentially, the group from which the
cases could have originated. The investigators in
the RATIO study approached this by randomly
calling households all over the Netherlands until
767 women of reproductive age were identified.
Random-digit dialing is an excellent method for
selecting a random sample of the population in a
study. Although this method can be biased in
favor of people who own a telephone and, there-
fore, could have higher socioeconomic status, its
use in the RATIO study probably did not lead to
a large selection bias. Population-based controls
are generally considered superior to other con-
trols, such as those formed from the same hos-
pital or clinic doctor as the cases.20 In the
RATIO study, use of hospital-based controls
would have included sicker women who would
be less likely to use oral contraception because of
their high blood pressure, or smoking habits.
Oversampling of the control group, or re-
cruiting a larger number of controls than cases,
is commonly done to increase a study’s statistical
power.20 This was particularly important in the
RATIO study because of the small number of
cases and the rarity of the disease. However,
oversampling has limitations because statistical
power increases, nonlinearly, only up to a ratio
of about 1:4 (cases to controls); higher ratios add
only time and expense to a study without addi-
tional power.20
Assessment of exposure in the RATIO study
was limited to COC use in the month preceding
the stroke. For the outcome of interest, a throm-
botic event culminating in a stroke, a month of
use is probably an adequate exposure duration
because COCs create prothrombotic effects as
soon as 4 days after initiation; this effect dis-
appears quickly after discontinuation.47
Case-control studies are particularly vulner-
able to recall bias, which occurs when partici-
pants are asked to remember events and one
group recalls past events differently than the
860 Kapp et al
comparative group.44 This can happen, for ex-
ample, when cases, who have more motivation
than healthy controls to do so, reflect on expo-
sures that could have led to their illness.44 Cases
might also be more likely to overreport possible
exposures in trying to find a reason for their
health event. Using hospital-based controls can
limit recall bias because all study participants
will have experienced a recent health event
although, as noted above, using hospital-based
controls has certain disadvantages that generally
make them less suitable for case-control studies
than population-based controls.20 Investigators
should include steps in their designs to minimize
recall bias as it is a pervasive threat to the validity
of the study. In the RATIO study, 2 important
steps were taken to minimize recall bias: the
standardized questionnaire used to collect infor-
mation from cases and controls included color
photographs of packages of different types of
contraceptives to help women recall which con-
traceptive(s) they had used during the designated
month, and the data were collected by well-
trained, blinded interviewers and data collec-
tors; even those performing the DNA analyses
for factor V Leiden mutation were blinded to
case or control status.
Confounding is of particular importance in
the evaluation of data from observational stu-
dies. Confounders are variables that are related
to both the exposure and the outcome, but not in
the causal chain.29 Their presence may distort
the relationship between the exposure and the
outcome if the researcher does not take them
into account in the design and/or in analysis. For
example, sexual behavior could be a confound-
ing variable in a study of the association between
oral contraceptive use and the acquisition of
STIs because any observed association between
oral contraceptive use and STIs could be pri-
marily due to oral contraceptive users engaging
in riskier sexual behaviors than nonusers. If so,
the observed increase in STI rates in contra-
ceptive users could be due to risky sexual beha-
viors rather than the use of contraceptives.48
Known confounders, if anticipated and the
relevant information collected, can be controlled
for in the design or the analysis phase.45 One
method to reduce confounding is by matching
cases and controls on particular variables; in the
RATIO study, cases and controls were matched
by age, geographic area, and year of stroke.46 In
addition, investigators can control confounding
by restricting the study population to, for exam-
ple, a specific age group or geographic area of
residence.45 However, after such restrictions,
baseline characteristics cannot be compared be-
tween groups, given their initial use in guiding
group enrollment. In the analysis phase, con-
founding can be controlled for using stratifica-
tion and multivariate analysis techniques.
The measure of association in case-control
studies is the OR.29 The RATIO study found a
moderately elevated OR (2.6, 95% CI: 1.7-4.0)
of having used COCs in the last month in women
who had had a stroke and no factor V Leiden
mutation and an OR of 11.2 (95% CI: 4.3-29.0)
among women who had had a stroke and were
carriers of factor V Leiden compared with wo-
men who had not had a stroke and did not have
factor V Leiden mutation. This suggests that
using COCs is significantly associated with the
occurrence of stroke in women of reproductive
age, particularly those carrying the factor V
Leiden mutation.
COHORT STUDIES IN HORMONAL
CONTRACEPTION
In cohort studies, investigators follow the course
of 2 or more groups that differ according to
exposure to some factor for the occurrence of an
outcome.49 Cohort studies can be conducted
concurrently (going forward in time), ambi-
directionally (looking backward in time to the
exposure, then continuing forward in time to the
outcome), or retrospectively (looking backward
in time from present).49 Participants are chosen
based on their exposure status and are followed
over time to observe the development of the
outcome of interest. The measure of association
for cohort studies is the RR, which can be either
a risk ratio (risk of disease or cumulative inci-
dence of cases in the exposed group divided by
risk of disease or cumulative incidence of cases in
the unexposed group) or, in cases with differen-
tial follow-up, the rate ratio (the rate of disease
or incidence density of cases in the exposed
group divided by the rate of disease or incidence
density of cases in the nonexposed group).29,49
RRs are used to measure the relative increase or
decrease in the number of events in the exposed
group compared with the unexposed popula-
tion. Both ORs (used in case-control designs)
and RRs are valid measurements of effect, but
RRs are easier to interpret and therefore
preferred.
Cohort studies are valuable for establishing
or exploring an outcome’s magnitude of risk and
modifying factors. Multiple outcomes, both ad-
verse and beneficial, from one exposure can be
investigated concurrently.29 However, following
participants over time can be particularly

challenging in cohort studies. In addition, these
studies can involve large sample sizes, complex
logistics, and great expense.29
In hormonal contraceptive research, cohort
studies are an important mechanism for post-
marketing surveillance of contraceptive meth-
ods. The postmarketing surveillance of the
Norplant levonorgestrel implant, for example,
was accomplished with a well-designed cohort
study.50 This study assessed Norplant’s long-
term safety and efficacy compared with 2 other
methods of long-term contraception, steriliza-
tion, and intrauterine device (IUD) use (includ-
ing both copper and noncopper IUDs but
excluding hormone-releasing IUDs) in 8 devel-
oping countries. The cohort included women
who were choosing long-term contraception;
7977 were using Norplant, 6625 an IUD, and
1419 sterilization. Participants were interviewed
and examined at the initial visit, 6 weeks later,
and then again semiannually for 5 years, regard-
less of whether they switched methods during
that time period. The investigators found that
pregnancy rates for each method were less than
1/100 woman-years and they documented no
significant excess risk of morbidity for Norplant
users except for small increases in gallbladder
disease (RR 1.52, 95% CI: 1.02-2.27) and hy-
pertension (RR 1.81, 95% CI: 1.12-2.92).
This study’s merits include the choice of the
design; preclinical studies of drugs are seldom
large enough with long-enough follow-up to
document rare adverse events. Moreover, the
setting of a clinical study is one with strict
follow-up and adherence, unlike the real-life
situation that women face outside a clinical
study. Therefore, observational studies after
marketing are crucial to identify rare events, or
those that occur only after years of use.49
Exposure must be clearly defined at the in-
itiation of a study, as in this case, the exposure
was the use of the contraceptive method Nor-
plant. Investigators should identify the compar-
ison group, or control cohort, by identifying a
group that has not been exposed to the defined
exposure but that resembles the exposed group
in every other relevant way.49 In the Norplant
study, women who sought long-term contracep-
tive regimens at the family planning clinics used
by the Norplant cohort were chosen as controls.
These women were age-matched to the women in
the exposure group. For other contraceptive
cohort studies, investigators sometimes choose
women for the control group who have never
used hormonal contraception, have never used
the hormonal contraceptive method of interest,
Hormonal Contraception 861
or are not currently using the hormonal contra-
ceptive of interest. However, in all other ways,
the comparison group should have the same
background as the exposed group and be at risk
of the outcome being studied.49
Losses to follow-up are inevitable over a
time-intensive study but the study’s validity is
endangered when unequal number of partici-
pants are lost in different groups and the reasons
are due to either the exposure or the outcome.38
Accordingly, prospective cohort studies must
minimize losses and incorporate into power
calculations the small losses that will inevitably
occur.49 Many large studies employ a research
coordinator whose full-time job is to track par-
ticipants (Table 3). Investigators should report
the final total participation rate at the comple-
tion of the study. They should also summarize in
the analysis the characteristics of participants
who were lost to follow-up and reasons why
participants dropped out of the study. The Nor-
plant study retained 94.6% of participants by
actively tracing women if they failed to attend a
semiannual clinic visit. Study personnel also
followed all clinic and hospital visits when events
were reported.
The cohort design allows investigators to
examine several predefined outcomes after an
exposure.29 The outcomes studied should be
expressed as incidences (occurrences per time
period) and dichotomous outcomes should be
expressed as RRs or rate ratios with CIs. The
inherent biases in observational studies should
TABLE 3. Required Elements for Power Calcula-
tions for all Observational and Experi-
mental Studies51
Required Definition
Type 1 error (a) The error of rejecting a null
hypothesis when it is true
(observing a difference when none
exists)
Power (1-b) The error of failing to accept an
alternative hypothesis when it is
true (failing to observe a difference
when one exists)
Outcome rate in A known rate from the literature
control group
Expected An estimation of treatment effect.
outcome rate Also may be determined by the
in treatment difference from the control rate
group that would be clinically relevant
Secondary outcomes that occur less frequently than the
primary outcome will be underpowered to detect a difference,
whereas those that occur as or more frequently will be
powered to do so.
862 Kapp et al

be addressed in the analysis to control for con-
founding. In addition to adjusted rate ratios,
crude rates should be presented.49 Finally, in-
vestigators should describe any biases and, if
possible, quantify them in the discussion section
of their manuscripts. They should also describe
the steps taken to minimize bias and the possible
direction such unaccounted biases would have
on the results.
Additional Tools to Assess
Hormonal Contraception
SYSTEMATIC REVIEWS AND META-
ANALYSES: ISSUES AND EXAMPLES
Systematic reviews and meta-analyses can pro-
vide useful summaries of large bodies of scien-
tific evidence regarding a specific question.
Meta-analyses, in particular, can provide more
robust and precise estimates of an effect than
individual studies alone.
The quality of a systematic review or meta-
analysis depends on both its rigor and the quality
of the studies summarized. If several of the
studies have the same systematic error, a sys-
tematic review or meta-analysis cannot correct
for that error. Therefore, the results of systema-
tic reviews and meta-analyses must be scruti-
nized in the same manner as other studies.
A major question that has arisen in the field
of meta-analysis is whether the primary goal is
synthesis (providing a single answer or effect
measure) or the identification and exploration
of heterogeneity (exploring why different studies
produce different findings).52,53 Although a syn-
thetic approach is simpler and therefore poten-
tially appealing, it might produce overly
simplistic results that could mask important
differences among subpopulations. In contrast,
exploring heterogeneity can lead to new hypoth-
eses regarding exposure and disease.52,53
The conduct and reporting of both meta-
analysis for observational studies and rando-
mized trials have been addressed in the
Meta-analysis of Observational Studies in Epi-
demiology (MOOSE)54 and Quality of Report-
ing of Meta-analyses (QUOROM)55 statements
(Fig. 2, Table 4). Many prominent journals have

Potentially relevant studies identified

and screened for retrieval (n=...)

Studies excluded, with reasons (n=...)

Studies retrieved for more detailed

evaluation (n=...)

Studies excluded, with reasons (n=...)

Potentially appropriate studies to be

included in the meta-analysis (n=...)

Studies excluded from meta-analysis,

with reasons (n=...)

Studies included in meta-analysis

(n=...)

Studies withdrawn, by outcome, with

reasons (n=...)

Studies with usable information, by

outcome (n=...)

FIGURE 2. QUOROM statement flow diagram.55

 Hormonal Contraception 863

TABLE 4. Guidelines for Meta-analyses and Systematic Reviews of Observational Studies (MOOSE)54
Title: Identify the study as a meta-analysis (or systematic review)
Abstract: Use the journal’s structured format
Introduction: present
 The clinical problem
 The hypothesis
 A statement of objectives that includes the study population, the condition of interest, the exposure or intervention,
and the outcome(s) considered
Sources: Describe
 Qualifications of searchers (eg, librarians and investigators)
 Search strategy, including time period included in the synthesis and keywords
 Effort to include all available studies, including contact with authors
 Databases and registries searched
 Search software used, name and version, including special features used (eg, explosion)
 Use of hand searching (eg, reference lists of obtained articles)
 List of citations located and those excluded, including justification
 Method of addressing articles published in languages other than English
 Method of handling abstracts and unpublished studies
 Description of any contact with authors
Study selection: Describe
 Types of study designs considered
 Relevance or appropriateness of studies gathered for assessing the hypothesis to be tested
 Rationale for the selection and coding of data (eg, sound clinical principles or convenience)
 Documentation of how data were classified and coded (eg, multiple raters, blinding, and interrater reliability)
 Assessment of confounding (eg, comparability of cases and controls in studies where appropriate)
 Assessment of study quality, including blinding of quality assessors; stratification or regression on possible predictors
of study results
 Assessment of heterogeneity
 Statistical methods (eg, complete description of fixed or random effects models, justification of whether the chosen
models account for predictors of study results, dose-response models, or cumulative meta-analysis) in sufficient detail
to be replicated
Results: Present
 A graph summarizing individual study estimates and the overall estimate
 A table giving descriptive information for each included study
 Results of sensitivity testing (eg, subgroup analysis)
 Indication of statistical uncertainty of findings
Discussion: Discuss
 Strengths and weaknesses
 Potential biases in the review process (eg, publication bias)
 Justification for exclusion (eg, exclusion of non-English-language citations)
 Assessment of quality of included studies
 Consideration of alternative explanations for observed results
 Generalization of the conclusions (ie, appropriate for the data presented and within the domain of the literature
review)
 Guidelines for future research
 Disclosure of funding source

adopted these guidelines as essential require-
ments for accepting systematic reviews for pub-
lication.
The Cochrane Collaborative’s Fertility Reg-
ulation Group has had a major impact on the
field of hormonal contraceptive systematic re-
views and meta-analyses. This group reviews the
effectiveness and safety of contraceptive agents,
contraceptive service delivery, contraceptive in-
formation acquisition and use, and reproductive
health and policy development.7 The group’s
reviews use a transparent search strategy to
assess both published and unpublished trials.
Data are aggregated by at least 2 independent
researchers, methodological quality is assessed,
and results are compiled in a meta-analysis when
possible. The reviews are peer reviewed and
published electronically, along with regular
updates.
A Cochrane Collaboration meta-analysis on
whether combined hormonal contraceptives af-
fect body weight provides a useful example of a
meta-analysis in the study of hormonal contra-
ception.56 The study question was whether wo-
men using combined hormonal contraceptives
had different weight change patterns than
864 Kapp et al
women not using combined hormonal contra-
ceptives. The meta-analysis included RCTs that
had been reported in English and were con-
ducted among women of reproductive age with
no contraindications for combined hormonal
contraceptive use. These RCTs compared a
combined hormonal contraceptive to a placebo
or another combined hormonal contraceptive
for at least 3 treatment cycles and measured
body weight changes. On the basis of a systema-
tic search strategy and further review by the
Cochrane Collaboration authors, 44 trials met
the criteria and were included in the review. The
authors also contacted contraceptive manufac-
turers and authors of published trials to identify
unpublished trials in an effort to avoid publica-
tion bias.
The Cochrane Collaboration authors used
statistical methods to combine the results from
several studies. When results were reported as
the mean change in weight between baseline and
follow-up, the authors estimated weighted mean
difference and 95% CIs using a fixed-effects
model. When results were reported as the pro-
portion of women who gained or lost more than
a certain amount of weight, the authors calcu-
lated the Peto OR as the summary measure.
Three placebo-controlled trials did not show
any difference in weight change between the
contraceptive and placebo groups. In addition,
most of the other studies that compared different
types of hormonal contraception found no sub-
stantial differences in weight. The authors con-
cluded that the existing evidence is insufficient to
rule out any effect of hormonal contraceptive
use on weight change, although no large effects
were found.
Interpreting Results
Determining causality borders on the realm of
philosophy. Epidemiologists generally define
causality as an event, condition, or characteristic
without which the disease would not have oc-
curred or would have occurred some time later.29
This is often the concern of the researcher; what
is the cause for this disease, or specifically, is this
hormonal contraceptive the cause of this dis-
ease? In answering this question, the issues of
chance, bias, truth, and, finally, causality must
be addressed.
First, investigators must determine whether
the findings are due to chance. The P value and
CI are both measures of chance, or the likelihood
of the study findings given the null hypothesis
(that no relationship exists between the interven-
tion and the outcome).29
If the likelihood that the findings are due to
chance is low, investigators must consider
whether the findings could be due to bias. Selec-
tion, information, confounding, and other types
of bias can systematically threaten a study’s
internal validity.57
If both chance and bias are unlikely to explain
a study’s findings, the results could indicate a
true association between the exposure and the
outcome, although a true association does not
necessarily indicate causation. A causal associa-
tion is more likely when the following criteria are
met: a strong association (greater measures of
association), consistent results (multiple studies
with similar results), specific results (a single
cause leads to a single response—this is rare),
temporality (the exposure occurs before the dis-
ease), biologic gradient (a dose-response rela-
tionship), biologic plausibility, coherence (the
result does not conflict with prior knowledge),
experimental evidence (results from experimen-
tal studies), and analogy.29
The RATIO study of stroke risk in women
taking hormonal contraceptives (described ear-
lier) meets many of these criteria. The women
who used COCs had slightly greater ORs of
stroke than those who did not, but women with
factor V Leiden mutation who used COCs had a
much higher OR (11.2) than women who did not
have the mutation and did not use COCs. An OR
of more than 4, in a well-conducted study, is felt
by some to have an implication of causality.58
Additionally, the association of oral contracep-
tives and factor V Leiden with the occurrence of
stroke implies a biologic gradient and clearly
demonstrates temporality (factor V Leiden mu-
tation and COC use were present before the
stroke), biologic plausibility, and consistency
(the mutation and COC use increase risk of
thrombosis).
Conclusions
Study design is critical to the accurate interpre-
tation of hormonal contraceptive study results.
Of the varied design types, all have utility for
certain questions, and there is no one design
suitable for all situations. In general, the good
conduct of hormonal contraceptive studies
should follow these guidelines:
1. Establish a clearly defined primary ques-
tion which is carefully selected and stated in
advance.

2. Carefully consider ethical issues and exist-
ing ethical guidelines in the study design
and the informed consent process.
3. Power calculations should be performed
for all studies, as they are critical for both
interpretation of results and plans for study
conduct.
4. Standard measurement of both exposure
and outcome should be used. This includes
the use of standardized, validated measure-
ment tools for clinical outcomes, avoidance
of surrogate markers, and accurate assess-
ment of exposures, including adherence.
5. In experimental studies, carefully address
random allocation, allocation conceal-
ment, and blinding to minimize bias in
randomization procedures. Describe the
steps taken in study publications to help
readers interpret the findings.
6. In observational studies, take steps to mini-
mize selection bias, information bias, and
confounding. Describe these steps in study
publications to help readers interpret the
findings.
7. The interpretation of findings for all design
types should carefully address the issues of
chance, bias, and causality.
8. Publish equivocal results as these data
could contribute to future meta-analyses
and withholding data is both unethical and
creates a publication bias.
9. Follow the standardized procedures for
reporting study results using the appropri-
ate guidelines, including: CONSORT,
STROBE, QUOROM, or MOOSE.
References
1. Rothman KJ. Show of confidence. N Engl J
Med. 1978;299:1362–1363.
2. Department of Health and Human Services.
Protection of Human Subjects. 45 CFR
46-102. 6-23-2005.
3. Council for International Organizations of
Medical Sciences. International ethical guide-
lines for biomedical research involving human
subjects. Bull Med Ethics. 2002;182:17–23.
4. Feldblum PJ, Caraway J, Bahamondes L, et al.
Randomized assignment to copper IUD or
depot-medroxyprogesterone acetate: feasi-
bility of enrollment, continuation and
disease ascertainment. Contraception. 2005;72:
187–191.
Hormonal Contraception 865
5. Trussell J. Contraceptive failure in the United
States. Contraception. 2004;70:89–96.
6. Steiner MJ, Dominik R, Rountree RW, et al.
Contraceptive effectiveness of a polyurethane
condom and a latex condom: a randomized
controlled trial. Obstet Gynecol. 2003;101:
539–547.
7. Helmerhorst FM, Belfield T, Kulier R, et al.
The Cochrane Fertility Regulation Group:
synthesizing the best evidence about family
planning. Contraception. 2006;74:280–286.
8. Martins SL, Curtis KM, Glasier AF.
Combined hormonal contraception and bone
health: a systematic review. Contraception.
2006;73:445–469.
9. Clark MK, Sowers M, Levy B, et al. Bone
mineral density loss and recovery during 48
months in first-time users of depot medroxy-
progesterone acetate. Fertil Steril. 2006;
86:1466–1474.
10. Grimes DA, Economy KE. Primary preven-
tion of gynecologic cancers. Am J Obstet
Gynecol. 1995;172:227–235.
11. Potter L, Oakley D, deLeonWong E, et al.
Measuring compliance among oral contra-
ceptive users. Fam Plann Perspect. 1996;28:
154–158.
12. Dittrich R, Parker L, Rosen JB, et al. Trans-
dermal contraception: evaluation of three
transdermal norelgestromin/ethinyl estradiol
doses in a randomized, multicenter, dose-
response study. Am J Obstet Gynecol. 2002;
186:15–20.
13. Trussell J, Vaughan B. Contraceptive failure,
method-related discontinuation and resump-
tion of use: results from the 1995 National
Survey of Family Growth. Fam Plann
Perspect. 1999;31:64–72, 93.
14. Friedman L, Furberg C, DeMaets D. Funda-
mentals of Clinical Trials. 3rd ed. New York:
Springer-Verlag New York Inc; 1998.
15. Anderson FD, Hait H. A multicenter, rando-
mized study of an extended cycle oral contra-
ceptive. Contraception. 2003;68:89–96.
16. Belsey EM, Machin D, Darcangues C. The
analysis of vaginal bleeding patterns induced
by fertility regulating methods. Contraception.
1986;34:253–260.
17. Mishell DR, Guillebaud J, Westhoff C, et al.
Recommendations for standardization of data
collection and analysis of bleeding in combined
hormone contraceptive trials. Contraception.
2007;75:11–15.
18. World Health Organization Collaborative
Study of Cardiovascular Disease and Steroid
Hormone Contraception. A multinational case-
control study of cardiovascular disease and
steroid hormone contraceptives. Description
866 Kapp et al
and validation of methods. J Clin Epidemiol.
1995;48:1513–1547.
19. Heinemann LAJ, Lewis MA, Spitzer WO, et al.
Thromboembolic stroke in young women. A
European case-control study on oral contra-
ceptives. Contraception. 1998;57:29–37.
20. Grimes DA, Schulz KF. Compared to what?
Finding controls for case-control studies.
Lancet. 2005;365:1429–1433.
21. Scholes D, LaCroix AZ, Ott SM, et al. Bone
mineral density in women using depot medrox-
yprogesterone acetate for contraception.
Obstet Gynecol. 1999;93:233–238.
22. Berenson AB, Radecki CM, Grady JJ, et al.
A prospective, controlled study of the effects
of hormonal contraception on bone mineral
density. Obstet Gynecol. 2001;98:576–582.
23. Cromer BA, Blair JM, Mahan JD, et al. A
prospective comparison of bone density in
adolescent girls receiving depot medroxypro-
gesterone acetate (Depo-Provera), levonorges-
trel (Norplant), or oral contraceptives. J
Pediatr. 1996;129:671–676.
24. Araujo FF, de Lima GR, Guazzelli CAF, et al.
Long-term evaluation of lipid profile and oral
glucose tolerance test in Norplant (R) users.
Contraception. 2006;73:361–363.
25. Grimes DA, Schulz KF. Surrogate end points
in clinical research: hazardous to your health.
Obstet Gynecol. 2005;105:1114–1118.
26. Farley TMM. Life-table methods for
contraceptive research. Stat Med. 1986;5:
475–489.
27. Potter RG. Application of life table techniques
to measurement of contraceptive effectiveness.
Demography. 1966;3:297–304.
28. Glasgow RE, Lichtenstein E, Marcus AC.
Why don’t we see more translation of health
promotion research to practice? Rethinking
the efficacy-to-effectiveness transition. Am J
Public Health. 2003;93:1261–1267.
29. Rothman K, Greenland S. Modern Epidemio-
logy. 2nd ed. Philadelphia: Lippincott-Raven;
1998.
30. Schulz KF, Grimes DA. Generation of alloca-
tion sequences in randomised trials: chance,
not choice. Lancet. 2002;359:515–519.
31. Brownson RC, Petitti DB. Applied Epidemiol-
ogy. New York, Oxford: Oxford University
Press; 1998.
32. Schulz KF, Grimes DA. Blinding in rando-
mised trials: hiding who got what. Lancet.
2002;359:696–700.
33. Moher D, Schulz KF, Altman DG. The CON-
SORT statement: revised recommendations
for improving the quality of reports of paral-
lel-group randomized trials. Ann Intern Med.
2001;134:657–662.
34. Plint AC, Moher D, Morrison A, et al. Does
the CONSORT checklist improve the quality
of reports of randomised controlled trials?
A systematic review. Med J Aust. 2006;185:
263–267.
35. Creinin MD, Schlaff W, Archer DF, et al.
Progesterone receptor modulator for emer-
gency contraception: a randomized con-
trolled trial. Obstet Gynecol. 2006;108:
1089–1097.
36. Schulz KF, Grimes DA. Allocation conceal-
ment in randomised trials: defending against
deciphering. Lancet. 2002;359:614–618.
37. von Hertzen H, Piaggio G, Peregoudov A,
et al. Low dose mifepristone and two regimens
of levonorgestrel for emergency contraception:
a WHO multicentre randomised trial. Lancet.
2002;360:1803–1810.
38. Schulz KF, Grimes DA. Sample size
slippages in randomised trials: exclusions and
the lost and wayward. Lancet. 2002;359:
781–785.
39. Strengthening the Reporting of Observational
Studies in Epidemiology. STROBE Statement:
checklist of essential items version 3 (Septem-
ber 2005). Available at: http://www.strobe-
statement.org/PDF/STROBE-Checklist-
Version3.pdf. Accessed August 17, 2007.
40. von Elm E. Failure to publish study outcomes:
can the STROBE statement help avoid non-
publication of outcomes? Eur J Epidemiol.
2006;21:24.
41. Grimes DA, Schulz KF. Descriptive studies:
what they can and cannot do. Lancet.
2002;359:145–149.
42. Mostad SB, Overbaugh J, DeVange DM, et al.
Hormonal contraception, vitamin A defi-
ciency, and other risk factors for shedding of
HIV-1 infected cells from the cervix and
vagina. Lancet. 1997;350:922–927.
43. Wang CC, McClelland RS, Overbaugh J, et al.
The effect of hormonal contraception on geni-
tal tract shedding of HIV-1. AIDS. 2004;
18:205–209.
44. Schulz KF, Grimes DA. Case-control studies:
research in reverse. Lancet. 2002;359:431–434.
45. Grimes DA, Schulz KF. Bias and causal asso-
ciations in observational research. Lancet.
2002;359:248–252.
46. Slooter AJC, Rosendaal FR, Tanis BC, et al.
Prothrombotic conditions, oral contracep-
tives, and the risk of ischemic stroke. J Thromb
Haemost. 2005;3:1213–1217.
47. Winkler UH, Holscher T, Schulte H, et al.
Ethinylestradiol 20 versus 30 micrograms
combined with 150 micrograms desogestrel:
a large comparative study of the effects
of two low-dose oral contraceptives on the

hemostatic system. Gynecol Endocrinol. 1996;
10:265–271.
48. Mohllajee AP, Curtis KM, Martins SL,
et al. Hormonal contraceptive use and
risk of sexually transmitted infections: a
systematic review. Contraception. 2006;73:
154–165.
49. Grimes DA, Schulz KF. Cohort studies:
marching towards outcomes. Lancet. 2002;
359:341–345.
50. Meirik O, Farley TMM, Sivin I. Safety and
efficacy of levonorgestrel implant, intrauterine
device, and sterilization. Obstet Gynecol.
2001;97:539–947.
51. Schulz KF, Grimes DA. Sample size calcula-
tions in randomised trials: mandatory and
mystical. Lancet. 2005;365:1348–1353.
52. Greenland S. Meta-analysis. In: Rothman KJ,
Greenland S, eds. Modern Epidemiology. 2nd
ed. Philadelphia: Lippincott-Raven Publish-
ers; 1998:643–673.
Hormonal Contraception 867
53. Lau J, Ioannidis JPA, Schmid CH. Summing
up evidence: one answer is not always enough.
Lancet. 1998;351:123–127.
54. Stroup DF, Berlin JA, Morton SC, et al. Meta-
analysis of observational studies in epidemio-
logy. A proposal for reporting. JAMA.
2000;283:2008–2012.
55. Moher D, Cook DJ, Eastwood S, et al. Im-
proving the quality of reports of meta-analyses
of randomised controlled trials: the QUOR-
OM statement. Lancet. 1999;354:1896–1900.
56. Gallo MF, Lopez LM, Grimes DA, et al.
Combination contraceptives: effects on weight.
Cochrane Database Syst Rev. 2006;1:CD003987.
57. Hennekens CH, Buring JE. Epidemiology
in Medicine. Boston: Little, Brown and
Company; 1987.
58. Sackett DL, Haynes RB, Guyatt GH, et al.
Clinical Epidemiology: A Basic Science for
Clinical Medicine. 2nd ed. Boston: Little,
Brown and Company; 1991.

Combined Oral Contraceptives:
A Comprehensive Review
JESSICA KILEY, MD and CASSING HAMMOND, MD
Department of Obstetrics and Gynecology, Northwestern University, Chicago,
Illinois
Abstract: Millions of women use birth control pills
for contraceptive and noncontraceptive reasons.
Although there have been reports of rare adverse
events, birth control pills do offer well-documented
health benefits, including a decrease in the risk of
ovarian and endometrial carcinoma. In addition,
manufacturers continue to modify birth control pills
to reduce side effects and medical risks.
Key words: oral contraceptives, hormonal contracep-
tion, birth control pill, noncontraceptive benefits,
progestin
Mechanisms of Action
Combined oral contraceptives (COCs) prevent
pregnancy in several ways, but primarily by
suppressing ovulation. The progestin compo-
nent prevents the luteinizing hormone surge
required for release of the ovum. It also thickens
cervical mucus and decreases tubal motility,
creating a difficult passage for sperm, and it
thins the endometrium, resulting in tissue less
receptive to implantation.1 The estrogen com-
ponent of the pill augments contraceptive effi-
cacy and improves cycle control. It increases
efficacy by inhibiting the release of follicle-sti-
mulating hormone from the pituitary, prevent-
ing the development of the dominant follicle,
and it simultaneously potentiates the progestin’s
inhibition of the luteinizing hormone surge.
Estrogen improves cycle control by stabilizing
the endometrium, minimizing irregular shedding
interpreted by patients as ‘‘breakthrough bleed-
ing.’’2
Correspondence: Jessica Kiley, MD, Department of
Obstetrics and Gynecology, Northwestern University,
680 N. Lake Shore Drive, Suite 1015, Chicago, IL
60611. E-mail: j-kiley@md.northwestern.edu
CLINICAL OBSTETRICS AND GYNECOLOGY / VOLUME 50
868
CLINICAL OBSTETRICS AND GYNECOLOGY
Volume 50, Number 4, 868–877
r 2007, Lippincott Williams & Wilkins
Efficacy and Compliance
Numerous clinical trials have documented the
efficacy of COCs to prevent pregnancy. Theore-
tical and ‘‘typical use’’ failure rates differ be-
cause COC effectiveness relies on correct patient
use. Although failure rates with perfect use are as
low as 0.3%, the National Survey of Family
Growth demonstrates typical use failure rates
of 8% in the first year.3
Given the discrepancy between theoretical
and actual failure rates, investigators seek meth-
ods that will improve patient compliance. Curtis
et al3 recently performed a meta-analysis evalu-
ating factors related to pill failure. Extension of
the pill-free interval beyond 7 days may increase
ovulation, whereas missing pills other than those
adjacent to the pill-free interval is less risky.3
Unfortunately, studies included in the meta-
analysis used different end points to define ovu-
lation, rarely actual rates of unintended preg-
nancy. When managing patients, clinicians must
urge adherence to the pill-taking regimen and
suggest cues that help individual patients re-
member to restart active pills on time.
A recent study by Huber et al4 examined
factors related to discontinuation of COCs. Of
128 subjects who discontinued COCs during the
course of the study, a majority listed ‘‘medical
side effects’’ such as nausea, headache, and
breakthrough bleeding as the reason for discon-
tinuation. Many patients discontinuing COCs
switched to less effective regimens without first
consulting their provider. Indeed, 10% used
the withdrawal method.4 Clinicians should
advise patients beginning COCs of common
side effects and suggest consultation before
discontinuation. Counseling allows proper con-
version to either a more suitable COC regimen
or to another reasonably effective method. New
formulations alleviate side effects, thereby
/ NUMBER 4 / DECEMBER 2007

improving compliance, by modifying hormone
dosage and progestin type.
Pharmacology
ESTROGEN
Oral contraceptives contain 2 active compo-
nents: estrogen and progestin. Orally active
forms of estrogen contain a modified structure
of estradiol. Ethinyl estradiol (EE) adds an
ethinyl group at position 17. Mestranol is the
3-methyl ether of EE. Nearly all modern COC
pills contain EE.2
Estrogen dose correlates with efficacy, ad-
verse events, and numerous side effects. Risk of
thrombosis, for example, increases with higher
doses of estrogen, whereas efficacy diminishes in
concert with reductions in estrogen dose. The
original contraceptive pills contained 50 mcg or
more of EE. Manufacturers decreased EE do-
sages to minimize many potential adverse
events, particularly thrombosis, without sacrifi-
cing contraceptive efficacy. Most products now
contain 20 to 35 mcg of EE.
In 2006, Gallo et al5 reviewed 18 randomized
controlled trials comparing the efficacy, rate of
discontinuation, bleeding patterns, and side ef-
fects of 20 mcg versus standard low-dose formu-
lations. Although the authors reported no
difference in contraceptive efficacy, the studies
were underpowered to examine pregnancy rates
as an end point. Formulations studied often
contained differing progestins making it difficult
to reliably compare side effects and bleeding
patterns. The authors concluded that potential
benefits of ‘‘very low-dose’’ 20 and 25 mcg EE
pills versus standard ‘‘low-dose’’ 30 to 35 mcg
pills remain theoretical.5
PROGESTIN
Most modern COCs contain 1 of 9 androgen-
derived progestins. The 19-nortestosterones dif-
fer in potency and androgenicity, which, in turn,
results in various clinical effects.3 See Table 1 for
a listing of progestins by class.
Newer progestins such as norgestimate,
desogestrel, and gestodene offer reduced
androgenic activity. The clinical activity of nor-
gestimate is primarily a function of its conver-
sion to levonorgestrel metabolites. Desogestrel is
converted in vivo to etonogestrel, its active form,
and gestodene into a number of progestational
derivatives. Only desogestrel and gestodene are
properly termed ‘‘third-generation progestins,’’
Combined Oral Contraceptives 869
TABLE 1. Progestins in Oral Contraceptives
19-nortestosterone derivatives
Estranes (norethindrone family)
Norethindrone
Norethindrone acetate
Ethynodiol
Lynestrenol
Gonanes (levonorgestrel family)
Levonorgestrel
Norgestrel
Desogestrel
Gestodene
Norgestimate
Spironolactone derivative
Drospirenone
because the biologic activity of norgestimate
derives from its conversion to levonorgestrel.
Drospirenone is a progestin derived from
spironolactone. It possesses antiandrogenic
and antimineralocorticoid activity. Efficacy of
pills containing EE/drospirenone resembles that
of other formulations, but the combination of
lower androgenicity and diuretic effect might
render such pills particularly effective for
patients suffering premenstrual dysphoric dis-
order, nausea, fluid retention, acne, and
even polycystic ovarian syndrome. Given the
antimineralocorticoid effect and potential for
hyperkalemia, clinicians should prescribe dros-
pirenone-containing COC cautiously in patients
with impaired renal or adrenal function.2
A Cochrane review of randomized trials com-
pared contraceptive efficacy, cycle control, side
effects, and discontinuation rates among formu-
lations containing different progestins. Second-
generation pills offered better cycle control and
continuation rates compared with first-genera-
tion formulations. Second and third-generation
preparations offered similar acceptability. Ges-
todene and desogestrel were comparable and
offered efficacy and cycle control that resembled
those of second-generation pills. Pills containing
gestodene might offer better cycle control than
those containing levonorgestrel.6
Regimens
Multiphasic COC regimens vary estrogen or
progestin dose during the cycle in an attempt
to decrease metabolic effects, minimize side
effects, and improve cycle control.
VARIATIONS IN PROGESTERONE DOSE
DURING CYCLE
A number of regimens with varying progestin
dosage during the cycle are available. Triphasic
870 Kiley and Hammond
formulations provide escalating progestin do-
sage weekly during weeks 1 to 3. Efficacy is
maintained with the goal of decreasing break-
through bleeding and amenorrhea.
In 2006, Van Vliet et al7 reviewed 21 rando-
mized controlled trials that compared monopha-
sic and triphasic regimens. Eighteen trials
demonstrated similar efficacy between groups.
Although many studies reported favorable bleed-
ing patterns with triphasic regimens, the studies
were heterogeneous in measurement and report-
ing of data and in progestin type and medication
dosages; meta-analysis was not possible. The
authors concluded that monophasic and triphasic
regimens offered similar rates of abnormal bleed-
ing, discontinuation, and adverse events.7
VARIATIONS IN ESTROGEN DOSE
DURING CYCLE
Estrostep (Pfizer, New York, NY) is an ‘‘estro-
phasic’’ COC that alters EE dose during the cycle,
whereas the progestin dose remains constant. It
delivers 20-mcg EE on cycle days 1 to 5, 30 mcg on
days 6 to 12, and 35 mcg on days 13 to 21, in an
effort to decrease estrogenic side effects while
maintaining good cycle control. The escalating
estrogen dose markedly increases sex hormone-
binding globulin levels, decreasing the amount of
circulating free androgens. The antiandrogenic
activity may benefit patients with acne vulgaris.2,8
Several new formulations add EE during the
traditional pill-free interval to decrease the like-
lihood of breakthrough ovulation. However,
failure rates and cycle control resembles those
of traditional regimens, leading some to ques-
tion whether such formulations offer a distinct
advantage.2
EXTENDED-CYCLE REGIMENS
Although the traditional 7-day pill-free or pla-
cebo interval imitates the menstrual cycle, the
physiologic benefit of intermittent ‘‘pill breaks’’
remains unproven. Continuous or ‘‘extended-
cycle’’ regimens offer particular advantage to
women with gynecologic conditions that re-
spond to menstrual suppression, including
chronic pelvic pain, dysmenorrhea, or endome-
triosis. They also provide a lifestyle advantage to
women who choose to have fewer menses. Most
COCs may be used continuously by omitting the
pill-free interval. Seasonale (Duramed, Pomona,
NY) is a 91-day extended-use regimen contain-
ing EE 30 mcg and levonorgestrel 150 mcg over
84 days followed by a 7-day pill-free interval.
Multicenter, open-label trials have confirmed its
safety and efficacy.9
A recent Cochrane review compared cyclic
and extended-cycle COCs. Compliance and dis-
continuation rates were similar overall between
the 2 groups. Studies reporting on patient satis-
faction demonstrated adequate satisfaction in
both groups. Menstrual symptoms were re-
ported in a few of the studies, and extended-
cycle users had better control of headaches,
fatigue, bloating, menstrual pain, and genital
irritation than did cyclic users. Bleeding patterns
were similar between groups or improved in
extended-cycle patients, with some studies show-
ing fewer bleeding and spotting days in contin-
uous users. The authors concluded that there
was good evidence regarding safety and accept-
ability of extended-cycle COCs, but further stu-
dies on specific formulations and studies of
patient satisfaction are necessary.10
INITIATION OF PILL USE
Traditionally, patients commencing COC use
are instructed to wait until their next menses to
start the pill. Most patients take the first pill on
the first day of menses or the first Sunday after
menstrual bleeding begins. Relating COC initia-
tion to menses is to avoid administering hormo-
nal contraceptives to a patient who has already
conceived; however, inadvertent COC use in
early pregnancy is of negligible risk.
In 2002, Westhoff et al11 introduced a novel
reconsideration of the traditional modes of pill
initiation with their ‘‘Quick Start’’ regimen.
With this method, patients begin taking the first
pill during their clinic visit. Subjects using Quick
Start are more likely to continue COC use to at
least the second pill pack than subjects awaiting
the next spontaneous menses, demonstrating
improved compliance.
Major Adverse Events
Concern that COCs predispose to major cardi-
ovascular and carcinogenic morbidities
prompted extensive investigation and dramatic
changes in formulation throughout the 1970s
and 1980s. As the lay press sometimes overstates
the risk associated with COC use—and grossly
understates the competing risk of unintended
pregnancy—clinicians must be able to help pa-
tients at risk for thromboembolic, cardiovascu-
lar, or cancer complications determine the
validity of new information.
BREAST CANCER
Whether COCs meaningfully alter short-term or
long-term risk of developing breast carcinoma

remains a source of controversy, one in which
the best evidence fails to support any of the
alarmist reports that frequently surface in the
media.
The Oxford Family Planning Study, a large
cohort study of 17,032 women followed for up to
16 years, found no difference in breast cancer
risk—and no increase in incidence of other
malignancies—in COC users versus never-users,
with relative risk (RR) 1.0 [95% confidence
interval (CI), 0.8-1.1].12
In 1996, the Collaborative Group on Hormo-
nal Factors in Breast Cancer released results of a
meta-analysis of 54 worldwide studies, including
53,297 subjects with breast cancer and 100,239
controls. The report generated 2 important con-
clusions. First, it supported a modest increase in
breast cancer risk during COC use (RR = 1.24;
CI 1.15-1.33), 1 to 4 years after cessation of use
(RR = 1.16; CI 1.08-1.23), and 5 to 9 years after
cessation of COC use (RR = 1.07; CI 1.02-1.13).
Second, the authors found no increase in risk of
breast carcinoma 10 or more years after cessation of
use. Cancers found in COC users were less ad-
vanced than those in women never using COCs.13
A subsequent review in 2004 reported similar
findings. The authors evaluated 51 epidemiolo-
gic studies and concluded that transient in-
creases in rates of breast carcinoma during or
near the time of COC use do not contribute
significantly to increased rates of breast cancer.
These findings reflect the very low risk for breast
carcinoma in the age groups most likely to use
COCs.14
VENOUS THROMBOEMBOLISM
Researchers have studied the relationship be-
tween COCs and venous thromboembolism
(VTE) extensively. Both estrogen and progestin
increase the risk of VTE among COC users, as
do patient characteristics such as smoking, age,
hereditary thrombophilias, and other medical
comorbidities.
Estrogen increases the risk of VTE in a dose-
dependent fashion, particularly among patients
ingesting pills containing 50 mcg or more of
EE.15 The incidence of VTE in women aged 15
to 44 not using COCs is 5 to 10 per 100,000
woman-years. Women ingesting low-dose COCs
experience VTE at a rate of 12 to 20 cases per
100,000 woman-years. The rate increases to 24
to 50 cases per 100,000 woman-years with higher
dose preparations. Certainly, none of these risks
approach the typical baseline risk of 60 cases per
100,000 woman-years of use associated with
normal pregnancy.16
Combined Oral Contraceptives 871
Progestin type also influences risk of VTE,
although the clinical significance of this associa-
tion remains a matter of debate. In the mid-
1990s, new data associated the third-generation
progestins with increased occurrence of throm-
boembolic events. In nonusers aged 20 to 24, the
incidence of VTE was estimated at 3 events per
100,000 women per year. The risk rose to ap-
proximately 9 per 100,000 woman-years for
users of pills with second-generation progestins,
and it increased to as high as 21 per 100,000
woman-years in users of pills containing deso-
gestrel or gestodene. Studies comparing rates of
VTE among users of pills containing levonor-
gestrel versus those containing either gestodene
or desogestrel demonstrated a RR for VTE of
1.3 to 2.2.16 These data (summarized in Table 2)
are derived from large case-control studies and
resulted in the so-called Pill Scare of 1995, when
many patients discontinued their COCs after
hearing media reports of higher rates of poten-
tially fatal thromboembolic complications. As a
result, many parts of Europe documented in-
creased rates of unplanned pregnancy and abor-
tion during the same time period.17
Studies implicating desogestrel and gesto-
dene as ‘‘higher risk’’ progestins suffered from
sources of bias and inconsistent methods used to
diagnose VTE among subjects. New case-con-
trol studies, which attempted to adjust for bias
and confounding variables, showed no increase
in VTE risk with third-generation progestins.
Conversely, 2 separate meta-analyses confirmed
a higher risk of VTE with a RR of 1.5 to 2. Some
newer data suggest that COC-related VTE rates
increase within the first year of use and decrease
thereafter. Overall, the absolute risk of VTE
remains low given the very low incidence of
VTE in patients without other significant risk
factors.17 The risk is acceptable to most patients
and clinicians. Table 2 provides a summary of
the incidence of VTE in patients using oral
contraception.
Hereditary thrombophilias clearly influence
the risk of VTE, but the clinical relevance of
TABLE 2. Incidence of VTE Among Women
Aged 15 to 44, as Related to Oral
Contraceptive Use*
No contraception 5-10
High-dose oral contraceptive 24-50
Low-dose oral contraceptive 12-20
Third-generation oral contraceptive 9-21
Pregnancy 60-70
* Figures represent incidence per 100,000 women per year.
872 Kiley and Hammond
these mutations among users of COCs remains
unclear. A 2006 meta-analysis reviewed 16 case-
control studies of COC use in patients with
inherited hypercoagulable states. Eleven studies
evaluated Factor V Leiden mutation, 6 studied
the prothrombin gene mutation, 3 assessed both
Factor V Leiden and prothrombin gene muta-
tion, and 3 studied other mutations.17
The presence of Factor V Leiden mutation
clearly increases the risk of VTE among COC
users with odds ratios (OR) ranging from 6.4 to
99 in the reported literature. It remains unclear
how heterozygosity versus homozygosity influ-
ences risk remains. The Expert Working Group
of the World Health Organization (WHO) re-
commends avoiding COCs among women
known to possess the mutation. However, it
simultaneously cautions against universal
screening for thrombogenic mutations before
initiating COCs. Such a practice would withhold
oral contraception from 3% to 6% of women,
99.9% of whom would never develop VTE.18
Routine screening is neither cost-effective nor
clinically warranted.
MYOCARDIAL INFARCTION
Myocardial infarction (MI) is a rare event in
young women, and the incidence increases with
age and the presence of other well-documented
risk factors. Past COC users experience no in-
creased risk of MI, but evidence suggests a high-
er risk of MI in current users. In their 2003 meta-
analysis, Khader et al18 evaluated 19 case-con-
trol and 4 cohort studies assessing MI risk
among current COC users. In this review, COC
users had an increased risk of MI compared with
nonusers, with an OR of 2.48 (95% CI, 1.91-
3.22). Past users had a nonsignificant increase in
risk with OR 1.15 (0.98-1.35). The risk of MI
among COC users increased when accompanied
by other risk factors for cardiovascular disease,
especially smoking, hypertension, and hyperch-
olesterolemia.18
The WHO holds that COC use alone is not
associated with increased risk of MI in healthy,
nonsmoking patients.19 The overall risk of MI
among young women is low whether they smoke
or do not smoke. Recommendations for COC
use among smokers are reviewed below.
STROKE
Stroke is also a rare event among users of COCs.
Data demonstrate a relationship between in-
creasing estrogen dose and risk of ischemic
stroke. One meta-analysis demonstrated a high-
er risk among those formulations containing
50 mcg or more EE with a RR 3.95 (95% CI,
2.4-6.5) than with dosages less than 50 mcg, RR
2.19 (95% CI; 1.1-4.2). Unfortunately, a subse-
quent large cohort study failed to corroborate
this data, suggesting that the risk of stroke
doubled among current COC users but did not
correlate with increasing dose of estrogen or type
of progestin. Subjects most at risk included those
with hypertension (7.6, 3.5 to 16.3), diabetes
(5.3, 0.7 to 42.6), hypercholesterolemia (10.8,
2.3 to 49.9), and tobacco use (4.4, 2.7-7.3).20
Other studies in US and Europe suggest no
increased risk of stroke among women on COCs
who monitored their blood pressure.20
As with MI and VTE, the overall incidence of
stroke among women using COCs remains so
low that significant changes in RR often trans-
late into clinically insignificant changes in abso-
lute risk. In all cases, risks of using COCs must
be weighed against the risks posed by other
contraceptive methods, and the risk of unin-
tended pregnancy.
METABOLIC EFFECTS
Estrogens and progestins influence carbohy-
drate and lipid metabolism. Estrogens usually
increase high-density lipoprotein (HDL) choles-
terol and decrease low-density lipoprotein
(LDL) cholesterol. In contrast, progestins tend
to decrease HDL cholesterol and increase LDL
and total cholesterol.20 The clinical impact of
these effects among COC users remains unclear.
Although COC preparations containing 100 mg
of levonorgestrel exert undesirable effects on
blood lipids, current triphasic levonorgestrel
formulations have no similar effects. Moreover,
pills containing desogestrel, norgestimate, and
gestodene improve HDL/LDL ratios.2,20
Oral contraceptives increase peripheral resis-
tance to insulin, primarily through action of the
progestin component. The relative glucose intol-
erance produced by COCs is dose-dependent
and clinically insignificant with most currently
used COCs. COC use does not influence subse-
quent development of diabetes, although it
might affect current diabetic control. COC use
is safe among diabetics without other significant
cardiovascular risk factors.2
Noncontraceptive Benefits
Many women knowingly or unknowingly rely
on COCs for their noncontraceptive benefits.
Indeed, patients who suddenly stop the pill ow-
ing to lack of contraceptive need often realize in
hindsight the dramatic impact COCs exert on

quantity of menstrual flow, dysmenorrhea, acne,
benign breast disease (BBD), and several other
physical parameters. Patients who contemplate
cessation of COCs in favor of other regimens—
particularly surgical sterilization—should first
undergo counseling that includes a description
of the noncontraceptive benefits forfeited should
they stop taking COCs.
CANCER REDUCTION
Although some media reports still attempt to
link cancer to oral contraceptives, a signi-
ficant body of evidence suggests that COC
use decreases the risk of some gynecologic
malignancies.
COC users are 40% less likely to develop
ovarian cancer than women never using COCs,
and women using COCs for over 10 years reduce
their overall risk of ovarian carcinoma by as
much as 80% for nearly 2 decades post-COC
ingestion.21 COC use might confer even greater
protection against ovarian cancer in the very
women at highest risk for the disease, particu-
larly women with a family history of the illness.
In 1 study, women whose first degree relatives
had ovarian cancer had a 90% reduction in risk
after using COCs for 4 years.22 It is unclear
whether COC use confers similar benefit in
patients with the BRCA1 or BRCA2 mutations.
The mechanism responsible for lowering the
risk of ovarian cancer among COC users re-
mains unclear. Some authors suggest that
chronic suppression of ovulation decreases re-
current ovarian injury. Other authors, noting an
association between high gonadotropin level
and ovarian cancer in other species, believe
COCs reduce ovarian cancer through chronic
suppression of gonadotropin secretion. Regard-
less of mechanism, the overall protective effect
appears to apply to all current doses of low-dose
COCs regardless of progestin. Indeed, COCs
with higher doses of progestin may confer even
greater protection against ovarian cancer than
other preparations.23
COC use also reduces the risk of endometrial
carcinoma, the third most common gynecologic
malignancy among American women. The long-
er women use COCs, the greater the reduction in
risk of endometrial cancer. Women taking
COCs for 1 year have a 20% reduction in risk
of endometrial cancer, whereas women taking
COCs for 10 years or more experience a reduc-
tion of approximately 80% in risk. Risk reduc-
tion may last as long as 20 years after
discontinuation of COCs.24 This reduction
holds for most common histologic types of
Combined Oral Contraceptives 873
endometrial carcinoma, including adenocar-
cinoma and adenosquamous carcinoma. The
protective effect occurs in low and high-dose
progestin preparations, but is most prominent
among women of low parity, particularly nulli-
parous patients.
MENSTRUAL BENEFITS
Millions of women suffer from menstrual dis-
orders. Primary dysmenorrhea, to cite only one
of many menstrual disturbances, afflicts 40% to
50% of young women. Menorrhagia, premenstr-
ual syndrome and dysphoria, periovulatory
pain, and catamenial migraine together consti-
tute one of the most prevalent groups of disor-
ders potentially relieved through the use of oral
contraceptive therapy.
COCs significantly decrease menstrual and
periovulatory pain, most likely by decreasing
prostaglandin release. Cross-sectional surveys
and clinical trials suggest improvement in pain
among women using a variety of preparations,
regardless of progestational component, estro-
gen dose, and whether the agent was monopha-
sic or multiphasic. COC use also decreases the
rate of absence from school and work among
women with dysmenorrhea and results in use of
fewer analgesic agents, particularly nonsteroidal
anti-inflammatory drugs. Use of extended-cycle
regimens might further ameliorate dysmenor-
rhea, as they decrease the frequency of timed
withdrawal bleeding.
Although studies supporting the use of COCs
to control menorrhagia often used high-dose
preparations, noncomparative and randomized
controlled trials similarly substantiate the effi-
cacy of low-dose preparations. Larsson et al25
documented significant decreases in average
blood loss among women using 30 mcg of EE/
150 mcg of desogestrel. Fraser and McCarron,26
in their randomized controlled cross-over trial,
demonstrated that patients using 30 mcg of EE/
150 mcg of levonorgestrel exhibited a 43% re-
duction in menstrual flow. Decreases in men-
strual flow and duration reduce the risk of iron
deficiency anemia. Perimenopausal patients not
only reduce the likelihood of abnormal uterine
bleeding, but avoid the pain and inconvenience
of tests to evaluate abnormal uterine bleeding,
such as endometrial biopsy.
COCs reduce the frequency and severity of
many other moliminal complaints, including
menstrual-associated mastalgia, mood dys-
phoria and, especially in the case of drospire-
none-containing formulations, fluid retention.
Among women with catamenial migraine,
874 Kiley and Hammond
continuous or extended-cycle use often de-
creases the frequency of symptoms, because such
migraines are related to estrogen withdrawal.
BBD
Although physicians have long advised users of
COCs to expect improvement in symptoms re-
lated to BBD, benefits vary unpredictably de-
pending on the estrogen dose, progestin type,
and histology of the underlying breast pathol-
ogy. Studies cited to support the use of COCs for
relief of breast symptoms often fail to address
these differences.27 Furthermore, although some
epidemiologic data suggest an association be-
tween BBD and breast carcinoma, COC use does
not seem to demonstrably reduce risk of breast
cancer but might even slightly augment the risk
of breast carcinoma in certain subgroups of
patients with BBD.13
Charreau and colleagues’28 case-control
study was an early effort to evaluate the relation-
ship between COC use and BBD controlling for
cumulative hormonal dose and breast histology.
The authors compared 286 patients with biopsy-
proven BBD and 382 age-matched controls and
concluded that the risk for nonproliferative
BBD decreased with long-term COC use. There
was no detectable effect on proliferative BBD.28
In 1999, Rohan and Miller29 investigated the
association between COC use and the risk of
BBD overall and by histologic subtype using
cohorts from the Canadian National Breast
Screening Study (NBSS), the latter a rando-
mized controlled trial of breast cancer screening
among Canadian women. The authors com-
pared 2166 women with biopsy-proven benign
proliferative disease with more than 5000 NBSS
patients lacking breast disease. The risk of BBD
was decreased in COC users, proliferative dis-
ease without atypia being most notably de-
creased. Moreover, the degree of risk reduction
positively correlated with duration of use. Wo-
men using COCs more than 7 years reduced their
risk by as much as 40% (RR 0.64, CI 0.47-
0.87).29
BONE MINERAL DENSITY
Abnormalities of bone mineral density afflict
roughly 44 million Americans, or 55% of people
50 years of age and older. Ten million indivi-
duals have frank osteoporosis, while 34 million
exhibit osteopenia, placing them at increased
risk for myriad health consequences that inflict
an escalating public health toll as the American
population gentrifies. Women comprise ap-
proximately 80% of all osteoporosis patients.30
Estrogen helps prevent osteoporosis through
many mechanisms. It increases calcium absorp-
tion from the gut, decreases urinary excretion of
calcium, and directly inhibits osteoclast-induced
bone resorption. Because COCs contain estro-
gen, physicians have long questioned whether
their use, particularly in patients without hy-
poestrogenism, might help prevent or delay the
development of osteoporosis.
Studies to date yield conflicting information
regarding the ability of COCs to prevent osteo-
porosis.31 Some studies document increased cor-
tical and trabecular bone mass in both
premenopausal and postmenopausal women
who used COCs. The beneficial impact of COCs
on bone mass correlates with increasing dura-
tion of use, particularly among women over the
age of 40 using COCs for more than 5 years.
Other studies have found no positive effect of
COCs on bone mass. To date, no study has
suggested an adverse effect.32
PELVIC INFLAMMATORY DISEASE
Whether COCs reduce the risk of acute pelvic
inflammatory disease remains unclear. Some
series report a 50% to 80% reduction in risk of
acute salpingitis among pill users, noting parti-
cularly a reduction in risk for hospitalization
related to pelvic inflammatory disease.33,34 Pro-
posed mechanisms for such a reduction include
thickening of cervical mucus and altered tubal
motility, impeding ascent of pathogens and de-
creased menstruum to serve as a nidus for infec-
tion. Whether such a risk applies equally to all
pathogens or only to specific pathogens, such as
gonorrhea, remains unclear. Indeed, more recent
data questions the protective effect of COCs in
its entirety, even going so far as to implicate
COCs in a higher rate of unrecognized endome-
tritis.35
Providers should advise COC users at risk for
sexually transmitted infection of the need for
concomitant barrier methods to prevent infec-
tion, such as condoms and diaphragms.
DERMATOLOGIC BENEFITS
Oral contraceptive preparations reduce the in-
cidence and severity of acne, most likely by
reducing levels of circulating androgen. COCs
suppress pituitary gonadotropin secretion and
induce the production of sex hormone-binding
globulin, binding circulating testosterone.36
Placebo-controlled trials support the use of
several different contraceptive preparations to
ameliorate mild to moderate acne.37,38 Although
triphasic preparations containing norgestimate

and 35-mg EE (Ortho Tri-Cyclen, Ortho-
McNeil, Raritan, NJ) were the first preparations
to receive FDA approval for this indication,
more recent placebo-controlled, randomized
trials have confirmed the efficacy of other pre-
parations. On this basis, FDA has now approved
the estrophasic/norethindrone (Estrostep, Pfi-
zer, New York, NY) agent for acne relief. Data
also support the use of 20-mg EE/LNG (Alesse,
Loette, Wyeth, Madison, NJ) agents, particu-
larly in patients desiring lower estrogen formu-
lations to avoid other troublesome side
effects.38,39
PREVENTION OF ECTOPIC PREGNANCY
Like all other forms of contraception, oral con-
traceptives reduce the risk of ectopic pregnancy,
probably by inhibiting ovulation.39 Although
controlled trials have not compared rates of
ectopic pregnancy among users of different for-
mulations, the effect is most likely similar when
using lower dose pills and pills containing differ-
ing progestins.
UTERINE LEIOMYOMATA
Uterine leiomyomata are the most common
benign tumors of the uterus. Because fibroids
contain both estrogen and progestin receptors,
clinicians have traditionally worried that use of
oral contraceptives might influence their devel-
opment or growth. Clinical studies suggest that
this is not the case.40 Indeed, many noncontra-
cepting women with small to moderate sized
fibroids use low-dose oral contraceptives to
control menstrual flow and menstrual pain,
avoiding more invasive forms of therapy.
Special Groups
WOMEN WITH DISABILITIES
Approximately 28.6 million women and girls
with disabilities live in the United States, com-
prising roughly 20% of the total female popula-
tion. Women with cognitive and physical
disabilities are often viewed as asexual, resulting
in deprivation of adequate reproductive health
care, including access to contraception. Physi-
cians erroneously assume that women with var-
ious disabling conditions cannot take oral
contraceptives, or they fail to weigh theoretic
risks, particularly the risk of thromboembolism,
against the concomitant risk of unintended preg-
nancy. Women with disabilities often benefit
from both contraceptive and noncontraceptive
applications of oral contraceptives, especially
Combined Oral Contraceptives 875
hygienic improvement associated with control
of menstrual flow. Given the lack of published
clinical trials and inherent difficulty studying
women with pure, representative disabilities,
physicians desiring information about providing
contraception to this group of patients should
consider contacting one of several multidisci-
plinary clinics specializing in reproductive health
care for disabled women.
SMOKERS
Smokers using any estrogen-containing form of
hormonal contraception have an increased risk
of coronary artery disease. Older women, wo-
men with preexisting cardiac disease, and heavy
smokers suffer the greatest risk. However, even
light smoking on the order of 1 to 4 cigarettes/d
increases the risk of coronary events substan-
tially among women 40 and older.41
Although most clinicians avoid prescribing
COCs to women over 35, overall morbidity
resulting from smoking and COC use among
women younger than 35 remains less than the
risk of undesired pregnancy. Many pharmaceu-
tical companies have even marketed lower estro-
gen and less androgenic preparations specifically
for smokers, although smoking could theoreti-
cally reduce circulating estrogen levels owing to
increased metabolism by the cytochrome P-450
system rendering those preparations less effec-
tive. Unfortunately, no controlled trials exist
to guide clinicians when prescribing COCs to
younger smokers.
PERIMENOPAUSAL WOMEN
Clinicians have often hesitated to prescribe oral
contraceptives to older, perimenopausal pa-
tients, despite the numerous noncontraceptive
benefits specifically sought by this age group.
Women throughout their 40s commonly experi-
ence abnormal uterine bleeding, resulting in
numerous patient visits, expensive tests, and
time lost from work. Low-dose oral contracep-
tives offer one of the most effective means to
control abnormal uterine bleeding in this age
group. Simultaneously, combination pills ame-
liorate vasomotor symptoms, improve bone
density, and provide protection against both
endometrial and ovarian carcinoma——all sig-
nificant concerns of the reproductively mature
patient.
Healthy, nonsmoking women with normal
blood pressure require no special screening
before initiating low-dose contraceptives. Cer-
tainly, all patients should receive standard
annual examination including cervical cytology,
876 Kiley and Hammond
mammography, and other age-appropriate
screens, but none of these tests are necessary
prerequisites for prescribing oral contraceptives.
POSTPARTUM, POSTABORTION, AND
LACTATING WOMEN
More than half of all American women resume
sexual activity within 1 month of delivery, in-
dicating a significant need for safe, effective
means of postpartum contraception that does
not hinder breastfeeding. Because sex steroids
might inhibit breast milk production and
are excreted in breast milk, the WHO and
Planned Parenthood promote hormonal meth-
ods as second-line agents among breast-feeding
patients. Unfortunately, many other methods—
particularly barrier methods—fail to offer the
convenience and efficacy many women require
during this stressful time period. In postpartum
patients who are not breast-feeding, the WHO
recommends starting COCs 3 weeks after deliv-
ery, when the risk of venous thrombosis
decreases.42
References
1. Speroff L, Darney P. A Clinical Guide for
Contraception. 4th ed. Philadelphia PA: Lip-
pincott Williams & Wilkins; 2005:21–138.
2. Hatcher RA, Trussell J, et al. Contraceptive
Technology. 18th ed. New York, NY: Ardent
Media; 2004:391–460.
3. Curtis KM, Chrisman CE, Mohllajee AP, et al.
Effective use of hormonal contraceptives: part
I: combined oral contraceptive pills. Contra-
ception. 2006;73:115–124.
4. Huber LR, Hogue CJ, Stein AD, et al. Contra-
ceptive use and discontinuation: findings form
the contraceptive history, initiation, and
choice study. Am J Obs Gynecol. 2006;194:
1290–1295.
5. Gallo MF, Nanda K, Grimes DA, et al.
Twenty micrograms vs. >20 mcg estrogen oral
contraceptives for contraception: systematic
review of randomized controlled trials. Contra-
ception. 2005;71:162–169.
6. Maitra N, Kulier R, Bloemenkamp KWM,
et al. Progestogens in combined oral contra-
ceptives for contraception. Cochrane Database
Syst Rev. 2004:CD004861.
7. Van Vliet HA, Grimes DA, Lopez LM, et al.
Triphasic versus monophasic oral contracep-
tives for contraception. Cochrane Database
Syst Rev. 2006;3:CD003553.
8. Boyd RA, Zegarac EA, Posvar EL, et al.
Minimal androgenic activity of a new oral
contraceptive containing norethindrone acet-
ate and graduated doses of ethinyl estradiol.
Contraception. 2001;63:71–76.
9. Anderson FD, Gibbons W, Portman D. Long-
term safety of an extended-cycle oral contra-
ceptive (Seasonale): a 2-year multicenter open-
label extension trial. Am J Obstet Gynecol.
2006;195:92–96. [Epub ahead of print April
21, 2006.]
10. Edelman AB, Gallo MF, Jensen JT, et al.
Continuous or extended cycle vs. cyclic use of
combined oral contraceptives for contracep-
tion. Cochrane Database Syst Rev. 2005:
CD004695.
11. Westhoff C, Kerns J, Morroni C, et al. Quick
start: a novel oral contraceptive initiation
method. Contraception. 2002;66:141–145.
12. Vessey M, Painter R. Oral contraceptive use
and cancer. Findings from a large cohort
study, 1968-2004. Br J Cancer. 2006;95:
385–389.
13. Collaborative Group on Hormonal Factors in
Breast Cancer. Breast cancer and hormonal
contraceptives: collaborative reanalysis of in-
dividual data on 53,297 women with breast
cancer and 100,237 women without breast
cancer from 54 epidemiological studies. Lan-
cet. 1996;347:1713–1727.
14. ESHRE Capri Workshop Group. Hormones
and breast cancer. Hum Reprod Update.
2004;10:281–293.
15. Burkman RT. Venous thromboembolism and
oral contraceptives. Current status and clinical
implications. Treat Endocrinol. 2002;1:
143–147.
16. Shulman LP, Goldheizer JW. The truth about
oral contraceptives and venous thromboembo-
lism. J Reprod Med. 2003;48:930–938.
17. Mohllajee AP, Curtis KM, Martins SL, et al.
Does use of hormonal contraceptives among
women with thrombogenic mutations increase
their risk of venous thromboembolism?
A systematic review. Contraception. 2006;73:
166–178.
18. Khader YS, Rice J, John L, et al. Oral contra-
ceptives use and the risk of myocardial infarc-
tion: a meta-analysis. Contraception. 2003;68:
11–17.
19. Burkman RT, Schlesselman J, Zieman M.
Safety concerns and health benefits associated
with oral contraception. Am J Obs Gyn.
2004;190:S5–S22.
20. Bushnell CD. Oestrogen and stroke in women:
assessment of risk. Lancet Neurol. 2005;4:
743–751.
21. Vessey MP, Painter R. Endometrial and ovar-
ian cancer and oral contraceptives—findings in
a large cohort study. Br J Cancer. 1995;71:
1340–1342.

22. Walker GR, Schlesselman JJ, Ness RB. Family
history of cancer, oral contraceptive use, and
ovarian cancer risk. Am J Obstet Gynecol.
2002;186:8–14.
23. Schildkraut JM, Calingaert B, Marchbanks
PA, et al. Impact of progestin and estrogen
potency in oral contraceptives on ovarian can-
cer risk. J Natnl Cancer Inst. 2002;94:32–38.
24. Schlesselman JJ. Risk of endometrial cancer in
relation to use of combined oral contracep-
tives. A practitioner’s guide to meta-analysis.
Hum Reprod. 1997;12:1851–1863.
25. Larsson G, Milsom I, Lindstedt G, et al. The
influence of a low-dose combined oral contra-
ceptive on menstrual blood loss and iron sta-
tus. Contraception. 2002;46:327–334.
26. Fraser I, McCarron G. Randomized trial of 2
hormonal and 2 prostaglandin-inhibition agents
in women with a complaint of menorrhagia. Aust
NZ J Obstet Gynecol. 1991;31:66–70.
27. Goehring C, Morabia A. Epidemiology of
benign breast disease with special attention to
histologic types. Epidemiol Rev. 1997;19:
310–327.
28. Charreau I, Plu-Bureau G, Bachelot A, et al.
Oral contraceptive use and risk of benign
breast disease in a French case-control study
of young women. Eur J Cancer Prev. 1993;
2:147–154.
29. Rohan TE, Miller AB. A cohort study of oral
contraceptive use and risk of benign breast
disease. Int J Cancer. 1999;82:191–196.
30. National Osteoporosis Foundation Website.
Available at http://www.nof.org. Accessed
January 2, 2007.
31. Kuohung W, Borgatta L, Stubblefield P. Low-
dose oral contraceptives and bone mineral
density: an evidence-based analysis. Contra-
ception. 2000;61:77–82.
32. Vessey M, Mant J, Painter R. Oral contra-
ception and other factors in relation to hospital
Combined Oral Contraceptives 877
referral for fracture. Contraception. 1998;57:
231–235.
33. Rubin GL, Ory HW, Layde PM. Oral contra-
ceptives and pelvic inflammatory disease. Am J
Obstet Gynecol. 1982;144:630–635.
34. Wolner-Hanssen P, Svensson L, Mardh P,
et al. Laparoscopic findings and contraceptive
use in women with signs and symptoms sug-
gestive of acute salpingitis. Obstet Gynecol.
1985;66:233–238.
35. Ness RB, Keder LM, Soper DE, et al. Oral
contraception and the recognition of endome-
tritis. Am J Obstet Gynecol. 1997;176:580–585.
36. Huber J, Walch K. Treating acne with oral
contraceptives: use of lower doses. Contracep-
tion. 2006;73:23–29.
37. Thiboutot D, Archer DF, Lemay A, et al.
A randomized, controlled trial of a low-dose
contraceptive containing 20 mg of ethinyl estra-
diol and 100 mg of levonorgestrel for acne
treatment. Fertil Steril. 2001;76:461–468.
38. Leyden J, Shalita A, Hordinsky M, et al.
Efficacy of a low-dose oral contraceptive con-
taining 20 mg of ethinyl estradiol and 100 mg of
levonorgestrel for the treatment of moderate
acne: a randomized, placebo-controlled trial.
J Am Acad Dermatol. 2002;47:399–409.
39. Mol BW, Ankum WM, Bossuyt PM, et al.
Contraception and the risk of ectopic preg-
nancy: a meta-analysis. Contraception. 1995;
52:337–341.
40. Parazzini F, Negri E, La Vecchia C, et al. Oral
contraceptive use and risk of uterine fibroids.
Obstet Gynecol. 1992;79:430–433.
41. Willett WC, Green A, Stampfer MJ, et al.
Relative and absolute excess risks of coronary
heart disease among women who smoke cigar-
ettes. N Engl J Med. 1987;317:1303–1309.
42. Lesnewski R, Prine L. Initiating hormonal
contraception. Am Fam Physician. 2006;74:
105–112.

Contraceptive Vaginal Ring
TESSA MADDEN, MD, MPH* and PAUL BLUMENTHAL, MD, MPHw
*Department of Obstetrics and Gynecology, Washington University School of
Medicine, St. Louis, Missouri; and w Department of Obstetrics and Gynecology,
Stanford University, Stanford, California
Abstract: The vaginal contraceptive ring is comprised
of ethinyl vinyl acetate that releases ethinyl estradiol
and the third-generation progestin, etonorgestrel. It
was approved by the Food and Drug Administration
in 2001. The vaginal contraceptive ring is highly
efficacious with a Pearl Index of 1.18 and an efficacy
of 99.1%. Effectiveness is similar with actual use.
Key words: contraception, vaginal contraceptive ring,
etonogestrel, combined hormonal contraception,
extended contraceptive regimen
Introduction
Combined oral contraceptives are an effective
and reversible method of contraception. They
are a popular choice for many women, but some
women have difficulty with a contraceptive
method that requires daily administration, and
compliance is an essential component contribut-
ing to the effectiveness of the method.1 Other
disadvantages of combined oral contraceptives
include fluctuating hormonal levels and unac-
ceptable cycle control at doses less than 20 mcg.
Development of alternative hormonal method
delivery systems, such as the contraceptive
patch, the depo-medroxyprogesterone injection,
the etonogestrel implant, and the levonorgestrel
intrauterine system (IUS), has aimed to over-
come some of the disadvantages of combined
oral contraceptives. The higher bioavailability
of nonoral routes may allow lower dose of
hormones to be equally effective while offering
reasonable cycle control. Recently, concern has
been raised about higher than expected estrogen
Correspondence: Tessa Madden, MD, MPH, Depart-
ment of Obstetrics and Gynecology, Washington
University School of Medicine, 4533 Clayton Ave,
Box 8219, St. Louis, MO 63110. E-mail: maddent@
wustl.edu
CLINICAL OBSTETRICS AND GYNECOLOGY
878
CLINICAL OBSTETRICS AND GYNECOLOGY
Volume 50, Number 4, 878–885
r 2007, Lippincott Williams & Wilkins
levels with the contraceptive patch and a possi-
ble increase in the risk of venous thrombotic
events. Disadvantages of the available long-act-
ing progestin-only methods require administra-
tion/insertion by a trained health-care
professional and irregular bleeding patterns.
The contraceptive vaginal ring (NuvaRing,
Organon USA, Roseland, NJ) is a flexible ring
composed of ethinyl vinyl acetate measuring
54 mm in diameter and 4 mm in cross-section.
Ethinyl vinyl acetate has been approved by the
US Food and Drug Administration (FDA) for
use in other biomedical products, such as Im-
planon, the etonogestrel-containing contracep-
tive implant. The ring releases 15 mcg of
ethinylestradiol (EE) and 120 mcg of etonoges-
trel per day.2 Etonogestrel is 3-ketodesogestrel,
the biologically active metabolite of desogestrel,
a third-generation progestin (Fig. 1).
The delivery system provides distinct advan-
tages over oral contraceptives, including avoid-
ance of first-pass metabolism through the liver,
constant serum steroid levels, and once-a-month
insertion. NuvaRing is the only FDA-approved
contraceptive vaginal ring available in the United
States.
DEVELOPMENT
Since the 1960s attempts have been made to
develop an effective, safe, and convenient vagi-
nal contraceptive ring. However, difficulties
with steroid formulations and concerns about
possible ring side effects such as vaginal lesions
and lipoproteins effects slowed product devel-
opment.3 NuvaRing was eventually developed
by NV Organon in The Netherlands. Initially,
the ring was comprised of Silastic tubing; how-
ever, development was suspended due to a man-
ufacturing problem. Subsequently, a ring was
developed comprised of ethinyl vinyl acetate
releasing EE and etonogestrel, which had accep-
table cycle control and excellent contraceptive
/ VOLUME 50 / NUMBER 4 / DECEMBER 2007

FIGURE 1. Contraceptive vaginal ring (Photo
supplied courtesy of NuvaRing, Organon USA,
Roseland, NJ).
efficacy in clinical trials.4 The FDA approved
the NuvaRing for contraceptive use in October
2001. Additionally, a progesterone-only releas-
ing ring is currently available in Peru and Chile
for breast-feeding women, and a 1-year ring
releasing EE and Nestorone (a 19-norprogester-
one derivative) is now undergoing clinical trials
in the United States.
Efficacy
In clinical trials the vaginal contraceptive
ring had high contraceptive efficacy, and is equal
to that of oral contraceptives. The European
1-year phase 3 study demonstrated that the
vaginal ring is highly efficacious at preventing
pregnancy. There were 6 pregnancies out of 1145
women in the intent-to-treat group, giving an
efficacy rate of 99.5% and a Pearl Index of 0.65
(the number of pregnancies expected per 100
woman-years of use). Three of these women
were believed to have violated the research pro-
tocol, and thus may represent user-related as
opposed to method-related failure.5 A subse-
quent study was also published which combined
data from the above study and a multicenter US
trial. There were a total of 21 pregnancies in 2392
women, giving an efficacy rate for the vaginal
ring of 99.1% and a Pearl Index of 1.18. There
were 10 pregnancies in women with perfect use
for an efficacy rate of 99.6% and a Pearl Index of
0.77.2
Contraceptive Vaginal Ring 879
A multicenter, randomized trial of 1030 wo-
men was conducted in Europe and South Amer-
ica to compare the efficacy and safety of the ring
to an oral contraceptive pill containing 150 mcg
of levonorgestrel and 30 mcg of EE. There were a
total of 10 pregnancies, 5 in each arm. The Pearl
Index was 1.23 for the ring and 1.19 for the
combined oral contraceptives. This difference
was not statistically significant. The estimated
cumulative probability of pregnancy over 1 year
was 1.2% and 1.07%, respectively,6 also not a
statistically significant difference. This observed
Pearl Index is similar to that observed in the
large clinical trials mentioned above.
Only one study has looked at the effectiveness
of the ring in actual clinical use. This was an
observational study of 1130 women in The
Netherlands conducted over the first 3 months
of ring use. Only 1 pregnancy was observed in
282 woman-years of use suggesting that, in
actual use, the ring can be as effective as the
prior efficacy trials demonstrated.7
MECHANISM OF ACTION
The contraceptive vaginal ring has been shown
to completely inhibit ovulation.8 Similar to com-
bined oral contraceptives, this most likely repre-
sents the primary mechanism of action. The
administration of an estrogen and a progestin
inhibits follicular development and ovulation. A
reduction of ovarian estradiol secretion is ob-
served and the lack of corpus luteum formation
results in the absence of endogenous progester-
one production. The ovarian effects are due to
the inhibitory action of combined hormonal
contraceptives on the pituitary production and
secretion of both follicle-stimulating hormone
and luteinizing hormone.9 In addition, the
atrophic endometrial changes and thickened
cervical mucus likely contribute to reduced
fertility.
PHARMACOKINETICS
Vaginal administration of contraceptive hor-
mones allows low, steady, and continuous dosing
and results in stable serum concentrations. The
ring has been shown to produce mean serum
ethinyl estradiol concentrations of 19 pg/mL and
maximum serum concentrations (Cmax) of 35 pg/
mL.10 The ring produces peak serum concentra-
tions that are approximately 20% of those seen
with the 30 mcg EE oral contraceptives and 60%
of those seen with the contraceptive patch (Fig. 2).
In a nonblinded, randomized trial of the
vaginal ring compared with combined oral con-
traceptives containing 30 mcg of EE and
880 Madden and Blumenthal

FIGURE 2. Mean EE C-t curves for subjects treated with Nu-

vaRing (n = 8), the transdermal contraceptive patch (n = 6), and
the combined oral contraceptive (n = 8). From Contraception.
2005;72:168–174. Used with permission.

150 mcg of desogestrel, the maximum serum
concentrations of etonogestrel and EE were
achieved in the ring group in approximately 1
week. These serum concentrations subsequently
showed a gradual linear decrease in time. The
maximum serum concentrations of etonogestrel
and EE were approximately 40% and 30%,
respectively, of those for the EE/desogestrel
combined oral contraceptives. The absolute
bioavailability of etonogestrel was higher in ring
users (102.9% vs. 79.2%) and similar for EE
(55.6% vs. 53.8%), however, the overall sys-
temic exposure to EE was approximately 50%
lower with the ring. Levels of etonogestrel and
ethinyl estradiol likely remain therapeutic for up
to 25 days of use in ring users.11
Acceptability
Overall, the contraceptive vaginal ring has been
found to be highly acceptable to women with
96% being satisfied and 97% reporting they
would recommend the ring to a friend. The level
of acceptability increases with duration of use; at
baseline 66% of women found the ring to be
acceptable, this increased to 81% after 3 months
of use. Additionally, the ring was found to be
acceptable to partners with 71% reporting never
or rarely feeling the ring during intercourse and
94% reporting never or rarely minding that the
woman was using the ring.12 Compliance with
use is also high; one study found patients were
compliant with the ring in 90% of cycles over 1
year.5
BLEEDING AND CYCLE CONTROL
The combined hormonal ring has been shown to
have excellent cycle control with very few wo-
men experiencing unscheduled bleeding. The
incidence of irregular bleeding is 5.5% per cycle
with the majority of the bleeding being defined
as spotting (Fig. 3).
Irregular bleeding was also noted to decrease
with duration of use.1 A large European efficacy
trial showed that the incidence of unscheduled
bleeding was low, occurring in 2.6% to 6.4% of
cycles. The unscheduled bleeding reported was
predominantly spotting (r1 pad/d). Scheduled
bleeding occurred in 97.9% to 99.4% of cycles,
and the average duration of the withdrawal
bleed was 4.7 to 5.3 days.5
An Italian study of 280 women randomized
to a low-dose combined oral contracep-
tives (20 mcg EE/100 mcg levonorgestrel), a

FIGURE 3. Incidence of breakthrough bleeding and spotting
(% of women). From Obstet Gynecol. 2002;100:585–593. Used with
permission.
very-low-dose combined oral contraceptives
(15 mcg EE/60 mcg gestodene), or the vaginal
ring, found that women using the vaginal ring
had significantly lower rates of irregular bleed-
ing than the women taking either of the com-
bined oral contraceptives. Additionally, women
using the vaginal ring had lower rates of negative
mood, vaginal dryness, and higher rates of sex-
ual desire, leading the authors to conclude that
the vaginal ring has fewer negative effects on
sexual desire and satisfaction compared with
combined oral contraceptives.13
EFFECTS ON THE VAGINA, CERVIX, AND
ENDOMETRIUM
Several studies have shown that the vaginal ring
has no detrimental effects on the vaginal and
cervical mucosa or the endometrial lining. An
open-label study of 58 women over 13 cycles
found no unfavorable effects of the ring on the
cervical or vaginal mucosa. The majority of
women showed no adverse cervical changes
when viewed colposcopically. Over time, the
same number of women was found to have a
normal to abnormal change in colposcopy
(11%), as those with an abnormal to normal
change (11%). Cytology was normal in 74% of
women at both visits. Fifteen percent of woman
had a worsening of cervical cytology and 9%
showed an improvement. Vaginal microbiology
assessed by Nugent score showed improvement
in 40% of women.14
Contraceptive Vaginal Ring 881
A prospective study of 103 women using the
ring examined the endometrial histology after 13
and 26 cycles. The investigators found that the
majority of women had atrophy of the endome-
trial lining with either inactive or atrophic en-
dometrium, with secretory changes. No
hyperplastic or neoplastic changes were seen in
the endometrial histology.15 These changes are
similar to those seen with oral contraceptives
and likely to provide a protective effect against
endometrial cancer.
Use
The ring is easily inserted and removed by the
woman at home. In typical use, the ring is
inserted continuously for 21 days and then re-
moved for 7 days during which the woman has a
scheduled bleed. Each ring is used for a single 3-
week period. The ring may be removed from the
vagina for up to a 3-hour period without de-
crease in efficacy. Off label, the ring can be worn
continuously for a 28-day cycle or a calendar
month cycle. This may produce similar patterns
of amenorrhea as continuous use combined oral
contraceptives, but may also produce intermit-
tent, unscheduled spotting and bleeding (see
below).
STARTING THE RING
Women who are not using another contraceptive
method should insert the ring between days 1
and 5 of the menstrual cycle. If they start the
882 Madden and Blumenthal
method after day 5 of their cycle, they should use
a back-up method for the first 7 days of ring use.
Conventionally, patients are instructed to begin
the ring on the first Sunday of their menstrual
period. However, day 1 start is also both safe and
effective. The ring may be inserted within 5 days
of a pregnancy termination or miscarriage with-
out increased risk of infection.
SWITCHING FROM ANOTHER METHOD
Women who were previously using combined
oral contraceptive pills should start the ring
within 7 days after the last active pill. Women
who were previously using a progestin-only pill
should begin the ring the day following the last
pill. Women using depo-medroxyprogesterone
should start the ring 10 to 13 weeks after the last
injection. Women previously using the levonor-
gestrel IUS should start the ring as soon as they
discontinue the IUS. Women previously using
the copper intrauterine device should start the
ring as soon as they discontinue the intrauterine
device and use a back-up method for 7 days.
SAME DAY START
An alternative to conventional day 1, day 5, or
‘‘Sunday’’ start is for the woman to start the ring
at the time she is seen by a health-care provider;
this has previously been described as ‘‘Quick
Start.’’ The woman inserts the ring immediately
at the time of clinic visit after a negative preg-
nancy test [or using World Health Organization
(WHO) criteria to be ‘‘reasonably sure’’ that the
woman is not pregnant] and uses a back-up
method for a week. An open-label study com-
paring 201 women randomized to immediate
start of low-dose oral contraceptives compared
with the vaginal ring. The investigators found
that women using the ring had fewer total num-
ber of bleeding days and fewer days of prolonged
bleeding compared with the conventional-start
combined oral contraceptive users.16
EXTENDED USE
There is no physiologic requirement for a
monthly withdrawal bleed while using combined
hormonal contraceptives and extended regimens
of oral contraceptives have been shown to be
safe, effective, and well tolerated. Many women
when offered a choice would prefer to eliminate
their monthly period and multiple clinical trials
have shown that continuous combined oral con-
traceptives regimens induce amenorrhea in 80%
to 100% of women by 10 to 12 months of use.
Extended use has also been associated with
improved quality of life and decreased frequency
of dysmenorrheal, menorrhagia, and migraine
headaches.17 Shortening the pill-free interval
decreases ovarian function, whereas extending
the pill-free interval increases the risk of ovula-
tion and may lead to method failure, therefore
extended use regimens may be more effective.18
An open-label, randomized trial of 429 wo-
men conducted in the United States and Europe
compared monthly ring use with 3 different
extended regimens; every other month (49-d
cycle), every third month (91-d cycle), or con-
tinuous (364-d cycle) over 1 year of use. Each
ring was only used for 21 days. Women in the 2
longer extended cycle regimens had decreased
number of bleeding days compared with the 28-
day regimen, however, they had an increased
number of spotting days. The 49-day group had
very similar bleeding profile to the 28 day, with
only an occasional unscheduled bleeding or
spotting day reported. Overall, 32% of women
discontinued the study early. The discontinua-
tion rates were higher in the 91-day and 364-day
groups and the most common reason given for
discontinuation was irregular uterine bleeding.
Overall, satisfaction with the ring was lower in
the 91-day and 364-day groups compared with
the 28-day and 49-day groups, 76.6% and 77.5%
versus 91.8% and 89.8%, respectively.17
A recent open-label study conducted in Brazil
of 75 women using a 91-day cycle of vaginal ring
over 1 year of use found that the mean number of
unscheduled bleeding days was 7.0, 4.0, 3.5, and
5.5 days per 90-day reference period. The dis-
continuation rate was 17.3% with only 6.6%
reporting a change in bleeding patterns as the
reason for discontinuation.19
In our clinical practice, we offer women the
option of leaving the ring in for 4 weeks, removing
it, and then immediately inserting a new ring. She
may then remove the second or third ring after 3
weeks and leave it out for a week if she desires a
withdrawal bleed or to avoid break through
bleeding. Patients should be counseled about
possible irregular bleeding and that while this
may be a nuisance, it is normal and not indicative
of a problem. This may increase the woman’s
willingness to tolerate unscheduled bleeding.
CALENDAR SCHEDULE
Another option for use is for women to insert the
ring on the first day of a calendar month and
remove it for the last 4 days of the month when
she will have the scheduled bleed. She should use
a back-up method for 7 days unless she is start-
ing the ring within 5 days of the first day of her
last menstrual period.

Adverse Effects
In a large phase 3 trial, the most frequently
reported events were vaginal complaints (vagi-
nitis 13.7% and leukorrhea 5.9%) and head-
aches (11.8%). There were no serious adverse
events. In total, 29.6% of women stopped the
ring early, 15% of these discontinued use of the
ring due to adverse events, the most frequently
reported events leading to discontinuation were
device-related events (2.6%), headache (2.1%),
vaginal discomfort (1.0%), and nausea (1.0%)5
(Fig. 4).
In the combined United States and European
trial 15.9% of women discontinued due to ad-
verse events, most commonly reported events
were headache (5.8%), vaginitis (5.6%), leukor-
rhea (4.8%), and device related events (4.4%).2
Figure 4 shows the rates of discontinuation by
reason.
Women should be counseled regarding the
possibility of expulsion. If the ring is expelled it
should be rinsed off in tepid water and reinserted
within 3 hours. Anecdotally, some women re-
port noting that the ring is absent from the
vagina after intercourse. If the ring is left in place
during intercourse, women are advised to check
for its presence in the vagina. If it is not felt,
expulsion may have occurred during inter-
course, particularly with withdrawal of the pe-
nis. Clients are advised to check their linens in
this instance. The ring may be out of the vagina
for up to a 3-hour period without any decrease in
efficacy. If the ring is out of the vagina for
greater than 3 hours a back-up method is advised
for the following 7 days.
FIGURE 4. Cumulative discontinuation rate during the study.
From Obstet Gynecol. 2002;100:585–593. Used with permission.
Contraceptive Vaginal Ring 883
The WHO has created evidence-based criter-
ia for medical eligibility for contraceptive meth-
ods. The contraceptive vaginal ring is classified
with the contraceptive patch and combined in-
jectable contraceptives (not currently available
in the United States). These contraceptive meth-
ods carry similar medical eligibility criteria as
combined oral contraceptives, however, the
WHO recognizes that there is little long-term
data about safety for these methods.20 Table 1
shows guidelines for use of the ring for many
common medical conditions.
Conclusions
The vaginal contraceptive ring is a safe and
effective contraceptive method that provides
good cycle control with a low incidence of un-
scheduled bleeding when used either for a 21-day
or 49-day cycle. Both women and their partners
report high levels of satisfaction with the ring,
which no doubt contributes to compliance with
use. Irregular bleeding is often cited as reason for
discontinuation of hormonal methods of contra-
ception. However, extended regimens may offer
women the opportunity for amenorrhea and
control the timing of their menses. It is possible
that appropriate counseling of irregular bleeding
may increase continuation rates and further
investigation of extended regimens will be useful
in further characterizing an appropriate ex-
tended regimen approach. Multiple studies show
that the discontinuation rate of the vaginal ring
is anyway from 17% to 32% over the course of a
year or use, which is better than has often been
884 Madden and Blumenthal

TABLE 1. WHO Eligibility Criteria for use of the reported for the pill. Still, further investigation
Vaginal Contraceptive Ring needs to be carried out to determine if there are
ways that the discontinuation rates can be de-
Condition Category
creased.
Breastfeeding
<6 wk postpartum 4
6 wk to 6 mo postpartum 3 References
>6 mo postpartum 2
Smoking 1. Rosenberg MJ, Waugh MS, Meehan TE. Use
Age <35 y 2 and misuse of oral contraceptives: risk indica-
Age >35 y tors for po or pill taking and discontinuation.
<15 cigarettes/d 3 Contraception. 1995;51:283–288.
>15 cigarettes/d 4 2. Dieben TOM, Roumen FJ, Apter D. Efficacy,
BMI >30 m2/kg 2 cycle control, and user acceptability of a novel
Multiple risk factors for CVD 3/4 combined contraceptive vaginal ring. Obstet
Hypertension 3 Gynecol. 2002;100:585–593.
SBP >160, DBP >100 4
3. Johansson EDB, Sitruk-Ware R. New delivery
History of DVT/PE 4
Current DVT/PE 4 systems in contraception: vaginal rings. Am J
Family history of DVT/PE 2 Obstet Gynecol. 2004;190:S54e9.
Major surgery 4. Sarkar NN. The combined contraceptive va-
With prolonged immobilization 4 ginal device (NuvaRing): a comprehensive re-
Without prolonged immobilization 2 view. Eur J Contracept Reprod Health Care.
Minor surgery without immobilization 1 2005;10:73–78.
Known thrombogenic mutations 4 5. Roumen FJME, Apter D, Mulders TMT, et al.
Current or history of ischemic heart disease 4 Efficacy, tolerability and acceptability of a
Current or history of CVA 4
novel contraceptive vaginal ring releasing eto-
Migraine
Without aura nogestrel and ethinyl oestradiol. Hum Reprod.
Age <35 2 2001;16:469–475.
Age >35 3 6. Oddsson K, Leifels-Fischer B, Roberto de Melo
With aura, at any age 4 N, et al. Efficacy and safety of a contraceptive
Breast cancer vaginal ring (NuvaRing) compared with a com-
Current 4 bined oral contraceptive: a 1-year randomized
Past and no evidence of disease for 5 y 3 trial. Contraception. 2005;71:176–182.
Diabetes 7. Roumen FJME, op ten Berg MMT, Hoomans
No vascular disease
EHM. The combined contraceptive vaginal
Noninsulin dependent 2
Insulin dependent 2 ring (NuvaRing): first experience in daily clin-
Nephropathy/retinopathy/neuropathy 3/4 ical practice in The Netherlands. Eur J Contra-
Viral hepatitis cept Reprod Health Care. 2006;11:14–22.
Active 4 8. Mulders TMT, Dieben TOM. Use of the novel
Carrier 1 combined contraceptive vaginal ring NuvaR-
Cirrhosis ing for ovulation inhibition. Fertil Steril.
Mild 3 2001;75:865–870.
Severe (decompensated) 4 9. Rivera R, Yacobson I, Grimes D. The mechan-
Liver tumors
ism of action of hormonal contraceptives and
Benign 4
Malignant 4 intrauterine contraceptive devices. Am J Ob-
stet Gynecol. 1999;181:1263–1269.
The WHO categories for contraceptive use are as follows: 10. van den Heuvel MW, van Bragta AJM,
1. A condition for which there is no restriction for the use of Alnabawyb AKM, et al. Comparison of
the contraceptive method. ethinylestradiol pharmacokinetics in three
2. A condition where the advantages of using the method
generally outweigh the theoretical or proven risks. hormonal contraceptive formulations: the
3. A condition where the theoretical or proven risks usually vaginal ring, the transdermal patch and an
outweigh the advantages of using the method. oral contraceptive. Contraception. 2005;72:
4. A condition which represents an unacceptable health risk if 168–174.
the contraceptive method is used.
BMI indicates body mass index; CVA, cerebrovascular acci- 11. Timmer CJ, Mulders TM. Pharmacokinetics
dent; CVD, cardiovascular disease; DBP, diastolic blood of etonogestrel and ethinylestradiol released
pressure; DVT, deep vein thrombosis; PE, pulmonary embo- from a combined contraceptive vaginal ring.
lism; SBP, systolic blood pressure. Clin Pharmokinet. 2000;39:233–242.
Table is derived from Medical Eligibility for Contraceptive
Use: Third Edition. Reproductive Health and Research, 12. Novak A, de la Loge C, Abetz L, et al.
World Health Organization, Geneva, 2004. The combined contraceptive vaginal ring,

NuvaRings: an international study of
user acceptability. Contraception. 2003;67:
187–194.
13. Sabatinia R, Cagianob R. Comparison profiles
of cycle control, side effects and sexual satis-
faction of three hormonal contraceptives. Con-
traception. 2006;74:220–223.
14. Archer D, Raine T, Darney P, et al. An open-
label noncomparative study to evaluate the
vagina and cervix of NuvaRings users. Fertil
Steril. 2002;78:S25.
15. Bulten J, Grefte J, Siebers B, et al. The com-
bined contraceptive vaginal ring (NuvaRing)
and endometrial histology. Contraception.
2005;72:362–365.
16. Westhoff C, Osborne LM, Schafer JE, et al.
Bleeding patterns after immediate initiation of
an oral compared with a vaginal hormonal
Contraceptive Vaginal Ring 885
contraceptive. Obstet Gynecol. 2005;106:
89–96.
17. Archer DF. Menstrual-cycle-related symp-
toms: a review of the rationale for continuous
use of oral contraceptives. Contraception.
2006;74:359–366.
18. Miller L, Verhoeven CHJ, in’t Hout J. Ex-
tended regimens of the contraceptive vaginal
ring: a randomized trial. Obstet Gynecol.
2005;106:473–482.
19. Barreiros FA, Guazzelli CAF, de Araú jo FF,
et al. Bleeding patterns of women using ex-
tended regimens of the contraceptive vaginal
ring. Contraception. 2007;75:204–208.
20. World Health Organization. Medical Eligibil-
ity for Contraceptive Use: Third Edition. Re-
productive Health and Research. Geneva:
World Health Organization; 2004.

The Levonorgestrel-releasing
Intrauterine System: An Updated
Review of the Contraceptive and
Noncontraceptive Uses
CAMARYN CHRISMAN, MD, MPH, PRICILLA RIBEIRO, BA, and
VANESSA K. DALTON, MD, MPH
Department of Obstetrics and Gynecology, University of Michigan Medical
School, Ann Arbor, Michigan
Abstract: The levonorgestrel containing intrauterine
system is an effective and safe form of long-term yet
reversible birth control. Intrauterine contraception
use in the United States fell dramatically after early
studies reported an association between intrauterine
contraception use and later tubal infertility. Subse-
quent evaluation suggests that these early studies were
biased. Users often experience menstrual distur-
bances. Informing patients of these common side
effects is important to improve compliance. In addi-
tion to its contraceptive effect, the levonorgestrel
intrauterine system offers potential therapeutic bene-
fits in other clinical contexts, including menorrhagia,
symptomatic fibroids, endometriosis, and endome-
trial suppression.
Key words: levonorgestrel intrauterine system, intra-
uterine device, contraception
Background
Intrauterine devices (IUDs) or intrauterine con-
traception (IUC) accounted for almost 10% of
birth control methods used by US women in the
mid-1970s. By comparison, 2 decades later IUC
is used by less than 2% of women in the United
States.1 This decline followed concern generated
by studies linking salpingitis and tubal infertility
with IUC use, especially the Dalkon Shield. In
subsequent years, IUC safety has been exten-
sively studied and the validity of these early
studies has been questioned. As such, the elig-
Correspondence: Vanessa K. Dalton, MD, MPH, De-
partment of Obstetrics and Gynecology, University of
Michigan, 1500 E. Medical Center Dr, L4000 Women’s
Hospital, Ann Arbor, MI 48109. E-mail: daltonvk@
med.umich.edu
CLINICAL OBSTETRICS AND GYNECOLOGY / VOLUME 50
886
CLINICAL OBSTETRICS AND GYNECOLOGY
Volume 50, Number 4, 886–897
r 2007, Lippincott Williams & Wilkins
ibility criteria in 2006 include nulliparous
women, reflecting some liberalization in its use.2
However, both provider and patient factors
continue to contribute to the low use rates.3–8
IUC is highly effective, requires little patient
effort and is cost-effective.9,10 Furthermore, hor-
mone-containing devices have many potential
noncontraceptive benefits and could address
both medical and contraceptive needs of an
individual patient. This paper summarizes the
contraceptive uses of the levonorgestrel contain-
ing intrauterine system and reviews current
evidence of its noncontraceptive benefits.
Contraceptive Uses
The hormonally impregnated IUD was first
introduced in the 1970s. Adding a progestin
was predicted to increase its contraceptive effec-
tiveness and to reduce expulsion rates by de-
creasing uterine contractility.11,12 Furthermore,
hormone-containing systems were hypothesized
to increase acceptability and compliance by
improving bleeding patterns and lessening dys-
menorrhea over copper-containing IUDs.13
Mirena (Berlex, Montville, NJ) is the only pro-
gestin-containing device currently approved in
the United States. It is a T-shaped device with a
reservoir containing 52 mg of levonorgestrel. It
delivers 20 mg of levonorgestrel per day and
maintains its contraceptive effectiveness for at
least 5 years (LNG-20 IUS).14 The maximum
plasma levels are reached within a few hours and
plateau between 100 and 200 pg/mL, which is
lower than those with implants or oral contra-
ceptives. Three large, multicenter trials reported
/ NUMBER 4 / DECEMBER 2007
 The Levonorgestrel Intrauterine System
pregnancy rates between 0.1/100 and 0.3/
100 women/y.11,15,16 A recent Cochrane review
concluded that there is insufficient evidence to
conclude that the levonorgestrel intrauterine
system (LNG IUS) is more effective than copper
IUDs, however, the power of the available stu-
dies might be insufficient to find small differ-
ences between the 2 highly effective methods.12
Despite endometrial suppression, there does not
seem to be a delay in the return of fertility with
conception rates 79.1/100 women at 12 months
after removal.17
The exact mechanism of action of the contra-
ceptive effects of IUC is not well defined. The
best evidence suggests that the LNG IUS pre-
vents pregnancy by both prefertilization and
postfertilization actions. On the basis of studies
of hormone levels and follicular progression,
women with hormonally active IUC continue
to ovulate.18 Specifically among LNG-20 users,
it seems that ovulatory cycles occur in most, even
amenorrheic users.19,20 Therefore, ovulation
suppression is not the primary mode of action.
IUC seems to prevent fertilization by reducing
the ability of sperm to fertilize an ovum, possibly
by creating a foreign body reaction in the uterine
cavity. On the basis of 2 small studies, fertilized
embryos are less frequently recovered from IUC
users as compared with those not using contra-
ception.18 In contrast to copper IUDs, altered
cervical mucus does not seem to be a dominant
mechanism in LNG IUS users.18 Endometrial
suppression is profound with the LNG-20 IUS
and likely contributes to its contraceptive effec-
tiveness. Like copper-containing IUDs, a strong
inflammatory reaction is seen in the endome-
trium of LNG-20 users. These endometrial ef-
fects are thought to underlie the primary
mechanism of action and lead to a reduction in
menstrual bleeding seen among LNG IUS
users.19,20 Some have suggested that the endo-
metrial effects may be spermicidal and would
likely prevent implantation if fertilization were
to occur.18
SIDE EFFECTS/RISKS
Side effects, such as menstrual abnormalities,
are important reasons for early discontinuation
of a contraceptive method. In part, hormone-
containing devices were hoped to increase con-
tinuation rates by decreasing menorrhagia and
dysmenorrhea. However, early studies reported
discontinuation rates up to 20% and menstrual
disturbances were a common reason.11,15,16,21
Several recent studies confirm that dissatisfac-
tion with bleeding patterns contribute to early
887
discontinuation rates.22–24 Bleeding abnormal-
ities seem to be particularly common during the
first 3 months of LNG IUS use. Compared with
the Nova T, LNG IUS users reported more total
days of bleeding, including spotting, but had
fewer days of menstrual-like flow. Over time,
bleeding frequency decreases and the incidence
of amenorrhea increases.21,25,26 After the first
year of use, up to one half of the users report
amenorrhea or infrequent bleeding.21,25,26 Ad-
dressing patient preferences and assessing accep-
tance of menstrual disturbances are integral to
efforts aimed at reducing early discontinuation
rates. Adequate and specific counseling before
placement is paramount.27
Expulsion is the most common cause of IUD
failure and occurs in about 5% of users. There is
insufficient evidence to indicate that expulsion
rates are lower with LNG IUS. Those at in-
creased risk of expulsion include nulliparous
women, women with severe dysmenorrhea, and
those with insertions immediately postpartum or
postabortion.28–30 Still, the convenience and
immediacy of postpartum and postabortal pla-
cement may outweigh the risk of expulsion in
some circumstances. The World Health Organi-
zation (WHO) recommends immediate insertion
after first (category 1) and second (category 2)
trimester abortions (Table 1). Postpartum inser-
tion is recommended after a 4-week interval,
with more immediate insertion only if the clin-
ician decides that the risks outweigh the bene-
fits.31 As expulsion generally occurs within first
few months, women are encouraged to follow-
up with their care provider within 12 weeks of
insertion (Mirena Product Monograph 2006).
Uterine perforation is an uncommon but
potentially serious complication of IUC place-
ment. Perforation is estimated to occur in 0 to
1.3 per 1000 insertions. Initial clinical trials
reported only 1 perforation after over 3000
insertions.11,15,16 A recent Dutch study esti-
mated that the rate was closer to 2.6 per 1000
insertions; however, this study relied on physi-
cian reported perforations and regional sales
data.32 Other studies have suggested that per-
foration rates may be increased in postpartum
and lactating women.33,34 As such, follow-up is
TABLE 1. WHO Medical Eligibility Criteria31
Category 1: no restriction
Category 2: advantages generally outweigh the risks
Category 3: risks generally outweigh the advantages
Category 4: unacceptable level of risk
888 Chrisman et al
recommended to assess for signs or symptoms of
perforation and to visualize the strings.
UPPER GENITAL-TRACT INFECTION
The concern for upper genital-tract infection
continues to limit IUC use worldwide. After
case-control studies in the 1970s and 1980s
suggested an association between IUC and in-
fection, use in the United States decreased tre-
mendously.1 More recently, evaluation of these
early studies identified numerous biases, which
could explain the apparent associations. For
instance, IUC users were compared with groups
of women using contraceptives known to lower
the risk of pelvic inflammatory disease (PID).
There also seemed to be systematic overdiagno-
sis of PID among IUC users and a failure to
correct for important confounding factors, such
as number of sexual partner.35 Furthermore,
most subsequent studies have not confirmed
these findings.35 However, although pelvic in-
fections are rare with LNG IUD use, there seems
to be a slightly increased risk of infection within
the first 20 days of insertion.36
Several strategies have been used in an effort
to reduce the risk of infection, including anti-
biotic prophylaxis at the time of placement,
avoiding placement during active cervicitis,
and device removal if PID is diagnosed with an
IUD in situ. The evidence supporting these
practices is limited. Recently, a Cochrane review
of available studies concluded that antibiotic
prophylaxis is not beneficial in low-risk wo-
men.37 Another systematic review concluded
that while women with chlamydia or gonorrhea
at time of IUD insertion were at increased risk
for PID relative to women without infection,
both had low absolute risks. Further, no studies
compare PID rates between women with active
cervicitis having an IUC device placed with a
similar group who did not undergo IUC place-
ment. Therefore, it is unknown whether women
with active cervicitis increase their risk of PID by
having IUC placed.38 Current clinical guidelines
do not recommend prophylactic antibiotics,
although they may be considered in a high-risk
population.2,31 Appropriate treatment before
insertion is recommended if an infection is found
at time of evaluation. But if a woman was to
develop PID with IUC in place, some guidelines
support initial treatment without immediate re-
moval.31
IUC use in the presence of other infections
such as Actinomyces israelii or HIV carries par-
ticular concerns. A. israelii is a gram-positive
anaerobic bacterium normally found in the gas-
trointestinal tract and may be a normal compo-
nent of the vaginal flora. It’s association with a
rare but significant pelvic infection has led to
recommendations, suggesting IUC removal in
patients with A. israelii identified on Pap smear.
A single randomized trial concluded that device
removal in conjunction with antibiotic treat-
ment was more successful at clearing the coloni-
zation than antibiotics alone.39 However, the
importance of clearing the colonization is still
not established. With regard to IUC use in HIV-
positive women, available evidence suggests that
use does not increase a women’s risk of overall
complications or infection-related complica-
tions.40,41 However, although IUC seem safe
for women who are HIV positive on antiviral
therapy (category 2), alternative options should
be considered in patients with AIDS not taking
antiretroviral therapy and those without regular
access to medical care (category 3).31,40 Accord-
ing to the WHO recommendations, other infec-
tious contraindications to IUC initiation include
puerperal sepsis, recent septic abortion, current
or recent PID, current or recent sexually trans-
mitted diseases, or known pelvic tuberculosis.31
INFERTILITY
Despite growing consensus that early studies
linking IUC with infertility were flawed, contro-
versy continues. As recently as 2001, some in-
vestigators have advised against IUC in
nulliparous women because of the risk of tubal
infertility.42 However, the evidence behind these
recommendations is limited and studies on the
subject have been inconsistent. Two large case-
control studies conducted in the 1980s reported a
2.0 to 2.6-fold increase in the risk of tubal
infertility among users of all types of devices,
including the Dalkon Shield.43–45 Other case-
control studies found no such association.35 A
more recent large case-control study of women
with tubal infertility found a significant associa-
tion between antibodies to Chlamydia trachoma-
tis and infertility but not between infertility and
IUC use.46 Cohort studies, in general, have not
demonstrated an increased risk of tubal inferti-
lity among IUC users, including among nulli-
parous women although several were
insufficiently powered.47–51 Evaluation of these
studies suggests that bias can explain the appar-
ent associations found in the case-control stu-
dies.52 Multiple other studies have shown no
such association, including among nulli-
paras.35,46,53,54 The WHO concludes that
although ‘‘there are conflicting data regarding
whether IUD use is associated with infertility
 The Levonorgestrel Intrauterine System
among nulliparous womenymore recent well-
conducted studies suggest no increased risk.’’31
In light of the available data, both the American
College of Obstetrician Gynecologists and the
WHO include nulliparous women as IUC can-
didates (category 2).2,31
METABOLIC/SYSTEMIC EFFECTS
Because there are low systemic levels of levonor-
gestrel in LNG IUS users, there are theoretical
concerns that glucose control, lipid profile, and
blood pressure could be negatively affected.
However, there are few studies examining these
questions. One of the early randomized trials
found no difference in blood pressure between
LNG IUS and Nova T users at 12 months.11 A
single randomized controlled trial (RCT) demon-
strated that the LNG IUS has no effect on glucose
control among well-controlled type I diabetes.55
A large population-based study found no change
in total/high density lipoprotein (HDL) choles-
terol ratio as compared with a group not using
hormones, after controlled for identified risk
factors56 Additionally, a small study of postme-
nopausal women reported a slight increase in
mean HDL concentration in the LNG IUS group
as compared with a group on oral progestins57
and another reported a decrease in HDL choles-
terol from baseline among women started on
transdermal estrogen along with LNG IUS.58
More studies are needed to improve our under-
standing of the effect exogenous hormones on
lipid profiles, particularly considering differential
effects of routes of admission. Currently, there
are no data suggesting that LNG IUS should not
be used in patients with diabetes, hypertension, or
hyperlipidemia. The WHO places these condi-
tions in at most category 2.31
Thromboembolism is a known risk of estro-
gen-containing hormone therapy, both contra-
ceptives and hormone replacement therapies.
What risk is posed by progestin only is uncertain.
There are no studies that have specifically exam-
ined the comparative risk of thromboembolism
among LNG IUS users and others. Even so, the
WHO guidelines consider the use of hormone-
containing IUDs/IUS to be risk category 2 in
women with prior thromboembolic events and
category 3 in those with active disease.31
Noncontraceptive Uses
MENORRHAGIA
The best-studied noncontraceptive use of LNG
IUS is in patients with menorrhagia. There are
889
currently 14 published RCT comparing LNG
IUS with other treatment modalities, including
hysterectomy, balloon ablation, transcervical
endometrial resection, and medical management
(Table 2). In addition, 2 cohort and 5 prospective
observational studies and also numerous case
reports have been published.14
Ten RCT met criteria to be included in a 2006
Cochrane review, examining the effectiveness
and acceptability progesterone-releasing intrau-
terine systems in menorrhagia treatment.71 This
review concluded that LNG IUS was more
effective at reducing menstrual blood loss than
oral progestins and mefenamic acid but less
effective than endometrial ablation. With regard
to acceptability, LNG IUD users are more sa-
tisfied and willing to continue treatment as
compared with oral progestins but there were
no differences in satisfaction rates as compared
with endometrial ablation. However, more pa-
tients reported systemic side effects, irregular
bleeding, and breast tenderness in the LNG
IUD group as compared with either the oral
progestins or endometrial ablation groups.71
Included in this Cochrane review was a large
study comparing LNG IUS with abdominal
hysterectomy to assess effects on quality of life
and cost.59,60 One hundred and nineteen women
with menorrhagia were randomized to receive
LNG IUS and 117 were randomized to hyster-
ectomy. Health-related quality of life scores
improved in both groups, and there was no
detectable difference in psychosocial well-being
between women who had received an LNG IUS
and those who had undergone a hysterectomy at
12 months. However, at 12 months only 68% of
women in the IUS group still had the device in
situ, and 20% had undergone a hysterectomy.
After 5 years, their findings continued to
demonstrate no difference between the hyster-
ectomy and LNG IUS groups in health-related
quality of life measures, with both groups show-
ing improvements in anxiety and depression.
Even though 42% of women originally assigned
to LNG IUS group eventually underwent hys-
terectomy, indirect and direct health care
costs were still significantly lower at 5 years in
the LNG IUS group ($2817 per participant)
than the hysterectomy group ($4660 per partici-
pant).59,60
Since the Cochrane review was conducted, an
additional study comparing the cost-effective-
ness of LNG IUS with thermal balloon ablation
using data generated in the same groups’ clinical
trial.63,64 Using direct and indirect costs, includ-
ing those incurred owing to treatment failure,
890 Chrisman et al

TABLE 2. RCT Evaluating Therapeutic Effects of the LNG IUS and Menorrhagia
Study Sample
Authors Design Size Comparison Groups Outcome
59
Hurskainen et al RCT 236 LNG IUS vs. total 68% of women in the IUS group still had the
abdominal hysterectomy, device in situ, 20% had undergone a
followed for 12 mo hysterectomy. Both groups had
comparable improvements in HRQL
Hurskainen and RCT 232 LNG IUS vs. total 42% of women originally assigned to LNG
Paavonen60 abdominal hysterectomy IUS group eventually underwent
followed for 5 y hysterectomy; however, no difference was
detected between groups in HRQL scores,
and indirect and direct health care costs
were substantially lower at 5 y in the IUS
group than the hysterectomy group
Lahteenmaki et al61 RCT 56 LNG IUS vs. medical 64% of participants randomized to receive
management while LNG IUS canceled their hysterectomy at
awaiting scheduled 6 mo while 14% canceled surgery in
hysterectomy medical group. At 12 mo, 12 of 28 (57%)
in the IUS group elected to proceed with
hysterectomy
Soysal et al62 RCT 72 LNG IUS vs. thermal Larger reduction in bleeding scores in
balloon ablation ablation group than LNG IUS group. No
differences in mean hemoglobin. Greater
improvements in HRQL in ablation
group
Busfield et al63 RCT 79 LNG IUS vs. thermal At 12 and 24 mo, bleeding scores were
balloon ablation significantly lower in LNG IUS group.
Quality of life scores increased in both
groups, and remained comparable
through 24 mo
Brown et al64 RCT 79 LNG IUS vs. thermal Quality of life scores improved in both
balloon ablation groups at 24 mo, while LNG IUS was less
costly (US $869) than ablation (US
$1608). Analysis included treatment,
posttreatment medical and failed
treatment costs
Irvine et al65 RCT 44 LNG IUS vs. cyclical oral Significantly higher reduction of blood loss
progestin in LNG IUS group vs. oral progestin
(94% vs. 87%). After 3 mo, 76% of LNG
IUS group wanted to continue vs. 22% of
oral progestin
Kittelsen and Istre66 RCT 60 LNG IUS vs. transcervical Equivalent reductions in bleeding scores.
resection of endometrium Satisfaction scores not different
Crosignani et al67 RCT 70 LNG IUS vs. transcervical Larger reduction in bleeding scores with
resection of endometrium ablation. Satisfaction improved in both
groups
Istre and Trolle68 RCT 59 LNG IUS vs. transcervical At 12 mo, treatment success achieved in
resection of endometrium 90% ablation group and 67% of the LNG
IUS group
Rauramo et al69 RCT 57 LNG IUS vs. transcervical Comparable reduction in bleeding scores at
resection of endometrium 36 mo
Reid and RCT 52 LNG IUS vs. mefenamic Both LNG IUS and mefenamic acid
Virtanen-Kari70 acid reduced bleeding scores at 6 mo. LNG
IUS led to a greater improvement in all
3 parameters

and patient reported quality of life scores, they
concluded that the LNG IUS was most cost-
effective. At 24-month follow-up, the total cost
of LNG IUS use was less than balloon ablation
with comparable increases in quality of life.64
However, they based their analysis on an esti-
mated failure rate of 25% in both treatment
groups. Failure rates among women initially
treated with LNG IUS have been reported closer
to 50%.60,61 Had they based their analysis on
 The Levonorgestrel Intrauterine System
these rates, it would likely diminish the apparent
advantage of LNG IUS.
In summary, LNG IUS seems to be a reason-
able option for women with menorrhagia. Over-
all, reductions in blood loss seem to be slightly
less than ablation methods but most studies
indicate a comparable increase in health-related
quality of life measures. However, several stu-
dies indicated a significant number of women
will ‘‘fail’’ LNG IUS treatment and move on to
definitive surgery. Still, costs appear lower in a
treatment protocol that begins with LNG IUS
versus hysterectomy alone.
SYMPTOMATIC FIBROIDS
There are few studies examining the use of LNG
IUD specifically for fibroid-related symptoms.
One cohort and 5 observational studies were
identified that examined fibroid-related menor-
rhagia and LNG IUS use. Additionally, 3 ran-
domized controlled clinical trials examined
LNG IUS use in women with fibroid-related
menorrhagia as a secondary analysis.
In the only comparative study designed to
examine fibroid-related symptoms, 32 women
with fibroid-related menorrhagia who received
LNG IUS were compared with 32 matched
historical controls consisting of women who
had undergone thermal balloon ablation.72
There was less bleeding improvement from 0 to
6 months in the LNG IUS group as compared
with the ablation group, but at 6 and 12 months
there were comparable reductions in menorrha-
gia and increases in hemoglobin.72 Several non-
comparative observational studies have also
reported a decrease in menorrhagia.14 Similarly,
indirect evidence is available from a RCT de-
signed to examine LNG IUS as a means of
endometrial protection among tamoxifen
users.73 Six of 8 patients noted to have fibroids
at entry in the LNG IUS group were observed to
have a decrease in fibroid size at 12 months.73
Size reduction, however, has not been consistent
across studies.14
Overall, studies designed to examine the use
of LNG IUS in the treatment of fibroids re-
ported decreased blood loss but were inconsis-
tent on whether LNG IUS was associated with a
decrease in fibroid size. Additional trials are
needed to define its role in the treatment of
symptomatic fibroids.
ENDOMETRIOSIS
Two RCT and 3 prospective observational stu-
dies have examined LNG IUS as a treatment
modality for women with endometriosis and
891
pelvic pain or dysmenorrhea. All studies were
limited by small sample sizes and varied signifi-
cantly with regard to enrollment criteria as well
as stage and location of disease.
In 1 RCT, 82 Brazilian women with surgically
and histologically diagnosed endometriosis were
randomized to receive either a LNG IUS or a
monthly GnRH analog.74 Both groups experi-
ence similar improvement in pain scores after 6
month of therapy but, as expected, the LNG IUS
group was less likely to be amenorrheic at 6
months as compared with the GnRH group.
The second published RCT enrolled 40 women
undergoing operative laparoscopy for endome-
triosis and randomized 20 to receive a LNG IUS
postoperatively and 20 to surgery alone. Post-
operative recurrence of moderate or severe dys-
menorrhea was reduced in the LNG IUS group
(P<0.03) and there was a 35% absolute risk
reduction of recurrence of dysmenorrhea in
women allocated to LNG IUS.75 A reduction
in the severity of endometriosis by standardized
staging criteria and improvement in pelvic pain
was found in 3 observational studies as well.76–78
Because of the limited number of studies
available, the small number of patients enrolled
and the heterogeneity of the study populations,
the efficacy of LNG IUS in the management of
endometriosis is uncertain. But, studies consis-
tently find a benefit of LNG IUS use for the
treatment of endometriosis-associated pain. Ad-
ditional studies are needed to confirm these
studies findings and to determine how to best
integrate its use into current clinical protocols.
Additionally, as expected, all studies also re-
ported increased menstrual disturbances in the
LNG IUS groups. Patient preferences regarding
side effect profiles could impact compliance and
treatment acceptance significantly, and should
be better understood.
ADENOMYOSIS
Only 1 RCT, 1 prospective observational study,
and 1 case report have been published that
examine LNG IUS use in patients with adeno-
myosis. Maia et al79 randomized 95 women with
adenomyosis-associated menorrhagia and pain
to either endometrial resection followed by
LNG IUS placement or surgery alone. They
found significant reductions in pain (10% vs.
80%) and bleeding. The observational study of
25 women also reported improvements in both
bleeding and pain after treatment with LNG
IUS.80 Because of the few number of women
studied thus far, it is premature to draw conclu-
sions about the role of LNG IUS in adenomyosis.
892 Chrisman et al

Future studies should take into consideration
the difficulties attributing pain and bleeding to
adenomyosis and attempt to eliminate other
etiologies of pain and bleeding.
UTERINE PROTECTION WITH ESTROGEN
REPLACEMENT THERAPY
There is good evidence supporting the use of
LNG IUS for endometrial suppression during
estrogen replacement therapy. Seven rando-
mized controlled clinical trials have examined
the efficacy and acceptability of LNG IUS as a
method of endometrial protection in peri and
postmenopausal period among women taking
estrogen replacement therapy (Table 3). In addi-
tion, 4 cohort and 7 prospective observational
studies have been published on the subject.
Various doses of LNG IUS have been studied,
including 20, 10, and 5 mcg of levonorgestrel per
24-hour period.
The largest trial randomized 200 perimeno-
pausal women to either oral estradiol and a
LNG IUS or a cyclic oral regimen of estradiol
and norethisterone-acetate. Bleeding patterns
were more regular in the oral progestin group.
After 2 years, 62% of participants with the LNG
IUD were amenorrheic. No cases of hyperplasia
in either group were identified.81 Raudaskoski
et al83 randomized 163 postmenopausal women
on oral estrogen to receive either 10 mcg/24 h
LNG IUD (n = 54), 20 mcg/24 h LNG IUD
(n = 56), or oral progestogen (n = 53). At 12
months, 91% of 20-mcg group and 76% of the
10-mcg group were amenorrheic. Endometrial
suppression was documented in over 98% of the
LNG IUS group, which was significantly greater
than the suppression rate observed in the oral
progestin group.83 Another trial compared en-
dometrial suppression rates between groups ran-
domized with different LNG doses. This study
randomized 51 women to receive a LNG IUS
releasing 10 mcg of levonorgestrel/d plus oral/
transdermal estrogen, and 45 women to receive
LNG IUS 5-mcg levonorgestrel/d levonorgestrel
plus oral/transdermal estrogen. Ninety-five of
96 women had histologic endometrial suppres-
sion and amenorrhea rates were comparable
between the groups (61% and 62%, respec-
tively).82
The 4 remaining RTCs included 135 perime-
nopausal and postmenopausal women and com-
pared a variety of hormone replacement
regimens. No cases of endometrial hyperplasia
were identified after 1 year of use in any study.84–
87
Similarly, the 11 observational studies exam-
ined a range of regimens in both perimenopausal
and postmenopausal women and no cases of
endometrial hyperplasia were reported.14,32,88–91

TABLE 3. Randomized Trials Evaluating the LNG IUS and Endometrial Protection in Women Taking
Estrogen Replacement Therapy
Study Sample
Authors Design Size Comparison Groups Outcome
81
Boon et al RCT 200 LNG IUS and oral estrogen vs. No hyperplasia in either group. More spotting
combined HRT in first 3 cycles in IUS group as compared
with oral HRT, by 12 months 68% of IUS
group was amenorrheic
Wollter- RCT 96 10 mcg LNG IUS plus oral/transdermal 95/96 participants had histologic endometrial
Svensson estrogen vs. 5 mcg LNG IUS plus oral/ suppression. Amenorrhea rates comparable
et al82 transdermal estrogen between groups (61% and 62%)
Raudaskoski RCT 163 Patients on oral estrogen plus either 109/110 participants randomized to receive an
et al83 20 mcg LNG IUS, 10 mcg LNG IUS IUS had documented histologic endometrial
or oral medroxyprogesterone suppression. 91% amenorrhea rate in the
20 mcg IUS group, 76% in the 10 mcg IUS
group
Raudaskoski RCT 40 LNG-20 IUS+ estrogen patch, vs. oral No difference in endometrial suppression
et al84 estrogen/progesterone
Andersson RCT 40 LNG-20 IUS+ oral estrogen vs. oral No difference in endometrial suppression
et al85 estrogen/progesterone
Suhonen RCT 36 LNG-16 IUS+ 1 subdermal estrogen No difference in endometrial suppression
et al86 implant vs. LNG-20 IUS+ 3
subdermal estrogen implants
Suhonen RCT 19 LNG IUS+ oral estrogen vs. LNG No difference in endometrial suppression
et al87 subdermal implant+ oral estrogen

HRT indicates hormone replacement therapy.

 The Levonorgestrel Intrauterine System
Studies on the effectiveness of LNG IUS in
endometrial suppression among women taking
hormone replacement therapy differ greatly with
regard to methodology, inclusion criteria, meth-
ods of hormone administration, dosing, and
methods of assessing endometrial suppression.
However, these studies consistently indicate that
LNG IUS adequately suppresses the endome-
trium during hormone replacement therapy.
Notably, several studies reported high drop out
rates among LNG IUS recipients, which could
indicate the side effects were intolerable to some
women. Further, acceptability studies are
needed. Additionally, more studies are needed
to elicit the lowest effective dose.
UTERINE PROTECTION WITH
TAMOXIFEN USE
One RCT examined LNG IUS use during
tamoxifen treatment for breast cancer.73 This
study randomized 122 postmenopausal women
being treated for breast cancer with adjuvant
tamoxifen for at least 12 months to receive LNG
IUS (n = 64), or to a control group (n = 58). At
12-month follow-up, 91% women in the LNG
IUS group had endometrial suppression on his-
tology as compared with 75% in the control
group. In addition, there were fewer endometrial
polyps in the LNG IUS treatment group. The
LNG IUS group reported more bleeding
throughout the study period but there was no
difference in visual analog scale ratings of pain,
embarrassment, acceptability, or recommenda-
tion between groups.73
ENDOMETRIAL HYPERPLASIA/
CARCINOMA
The best treatment for early endometrial cancer
is hysterectomy. Progestins are most commonly
used as palliative or adjuvant therapy. However,
systemic progestins are sometimes preferred,
specifically in women who are poor surgical
candidates. There are no RCT examining the
role of LNG IUS in the treatment of either
endometrial hyperplasia or carcinoma. There
are, however, 2 cohort studies and 1 prospective
observational study that have examined the
effectiveness of LNG IUS in endometrial hyper-
plasia and carcinoma.
Montz et al92 enrolled 12 women with FIGO
grade I endometrioid carcinoma without evi-
dence of extracorporeal disease spread or myo-
metrial invasion, and were classified as high-risk
surgical candidates. All were treated with the
Progestasert IUS (65 mcg of levonorgestrel re-
lease per 24 h). This group was compared with a
893
historical control group of 15 women who ful-
filled the study criteria but who had undergone
total abdominal hysterectomy, bilateral salpin-
go-oophorectomy. At 6 months, 7 of 11 had
negative biopsies. Of the 8 who were followed
to 12 months, 6 had no malignant tissue on
biopsy. Sixteen complications occurred, includ-
ing 1 death in matched historical controls who
underwent surgical management.92
One observational study compared LNG
IUS with oral progestin treatment for endome-
trial hyperplasia in 57 women found higher
regression rates in the LNG IUS versus the oral
progestin group at 3 months.93 Several small
observational studies without comparison
groups have also documented high regression
rates in women treated with progestin-contain-
ing IUS for hyperplasia.94–96
There are insufficient data to determine what
role LNG IUS has in the treatment of endome-
trial hyperplasia or carcinoma, if any, when
surgery needs to be avoided. In fact, several
cases of endometrial cancer after LNG IUS
placement have been reported.97,98 Larger stu-
dies are needed to better elicit the value of LNG
IUS in this context.
Conclusions
There is growing consensus that its time to
broaden our use of IUDs, including LNG IUS.
One of the greatest challenges will be to effec-
tively counsel patients seeking contraceptives to
best meet their needs, and to prepare them for
possible side effects. There are a number of
demonstrated and potential noncontraceptive
uses of LNG IUS. Research supports its use to
reduce menorrhagia and protect the endome-
trium in women on hormone replacement thera-
pies. Other uses, such as for endometriosis,
adenomyosis, or endometrial hyperplasia/carci-
noma are less clear. Resultant bleeding patterns
emerges again as an important consideration as
the LNG IUS often results in more bleeding
abnormalities than other treatment options. Pa-
tient acceptance of irregular bleeding patterns
might affect its long-term use in these contexts as
well. More research is needed to help providers
and patients make acceptable treatment choices.
References
1. Mosher WD, Martinez GM, Chandra A, et al.
Use of contraception and use of family plan-
ning services in the United States: 1982-2002.
Adv Data. 2004:1–36.
894 Chrisman et al
2. ACOG. (American College of Obstetrician
Gynecologist). ACOG Practice Bulletin: In-
trauterine device. 2005.
3. Stanwood NL, Garrett JM, Konrad TR. Ob-
stetrician-gynecologists and the intrauterine
device: a survey of attitudes and practice.
Obstet Gynecol. 2002;99:275–280.
4. Ogburn JA, Espey E, Stonehocker J. Barriers
to intrauterine device insertion in postpartum
women. Contraception. 2005;72:426–429.
5. Espey E, Ogburn T, Espey D, et al. IUD-
related knowledge, attitudes and practices
among Navajo Area Indian Health Service
providers. Perspect Sex Reprod Health. 2003;
35:169–173.
6. Stanback J, Grimes D. Can intrauterine device
removals for bleeding or pain be predicted at a
one-month follow-up visit? A multivariate
analysis. Contraception. 1998;58:357–360.
7. Asker C, Stokes-Lampard H, Beavan J, et al.
What is it about intrauterine devices that wo-
men find unacceptable? Factors that make
women non-users: a qualitative study. J Fam
Plan Reprod Health Care. 2006;32:89–94.
8. Postlethwaite D, Shaber R, Mancuso V, et al.
Intrauterine contraception: evaluation of clin-
ician practice patterns in Kaiser Permanente
Northern California. Contraception. 2007;75:
177–184.
9. Chiou CF, Trussell J, Reyes E, et al. Economic
analysis of contraceptives for women. Contra-
ception. 2003;68:3–10.
10. Varney SJ, Guest JF. Relative cost effective-
ness of Depo-Provera, Implanon, and Mirena
in reversible long-term hormonal contracep-
tion in the UK. Pharmacoeconomics. 2004;22:
1141–1151.
11. Luukkainen T, Allonen H, Haukkamaa M, et
al. Effective contraception with the levonor-
gestrel-releasing intrauterine device: 12-month
report of a European multicenter study. Con-
traception. 1987;36:169–179.
12. French R, Van Vliet H, Cowan F, et al. Hor-
monally impregnated intrauterine systems
(IUSs) versus other forms of reversible contra-
ceptives as effective methods of preventing
pregnancy. Cochrane Database Syst Rev.
2004:CD001776.
13. Nilsson CG, Luukkainen T, Diaz J, et al.
Intrauterine contraception with levonorges-
trel: a comparative randomised clinical perfor-
mance study. Lancet. 1981;1(8220 Pt 1):
577–580.
14. Varma R, Sinha D, Gupta JK. Non-contra-
ceptive uses of levonorgestrel-releasing hor-
mone system (LNG-IUS)—a systematic
enquiry and overview. Eur J Obstet Gynecol
Reprod Biol. 2006;125:9–28.
15. Sivin I, Stern J, Diaz J, et al. Two years of
intrauterine contraception with levonorgestrel
and with copper: a randomized comparison of
the TCu 380Ag and levonorgestrel 20 mcg/day
devices. Contraception. 1987;35:245–255.
16. Baveja R, Bichille LK, Coyaji KJ, et al. Ran-
domized clinical trial with intrauterine devices
(levonorgestrel intrauterine device (LNG),
CuT 380Ag, CuT 220C and CuT 200B). A
36-month study. Indian Council of Medical
Research Task Force on IUD. Contraception.
1989;39:37–52.
17. Andersson K, Batar I, Rybo G. Return to
fertility after removal of a levonorgestrel-re-
leasing intrauterine device and Nova-T.
Contraception. 1992;46:575–584.
18. Stanford JB, Mikolajczyk RT. Mechanisms of
action of intrauterine devices: update and esti-
mation of postfertilization effects. Am J Obstet
Gynecol. 2002;187:1699–1708.
19. Xiao B, Zeng T, Wu S, et al. Effect of levo-
norgestrel-releasing intrauterine device on
hormonal profile and menstrual pattern
after long-term use. Contraception. 1995;51:
359–365.
20. Lahteenmaki P, Rauramo I, Backman T. The
levonorgestrel intrauterine system in contra-
ception. Steroids. 2000;65:693–697.
21. Andersson K, Odlind V, Rybo G. Levonorges-
trel-releasing and copper-releasing (Nova T)
IUDs during five years of use: a randomized
comparative trial. Contraception. 1994;49:
6–72.
22. Tewari R, Kay VJ. Compliance and user sa-
tisfaction with the intra-uterine contraceptive
device in Family Planning Service: the results
of a survey in Fife, Scotland, August 2004. Eur
J Contracept Reprod Health Care. 2006;11:
28–37.
23. Dubuisson JB, Mugnier E. Acceptability of the
levonorgestrel-releasing intrauterine system
after discontinuation of previous contracep-
tion: results of a French clinical study in wo-
men aged 35 to 45 years. Contraception. 2002;
66:121–128.
24. Diaz J, Bahamondes L, Monteiro I, et al.
Acceptability and performance of the levonor-
gestrel-releasing intrauterine system (Mirena)
in Campinas, Brazil. Contraception. 2000;62:
59–61.
25. Hidalgo M, Bahamondes L, Perrotti M, et al.
Bleeding patterns and clinical performance of
the levonorgestrel-releasing intrauterine sys-
tem (Mirena) up to two years. Contraception.
2002;65:129–132.
26. Datey S, Gaur LN, Saxena BN. Vaginal bleed-
ing patterns of women using different contra-
ceptive methods (implants, injectables, IUDs,
 The Levonorgestrel Intrauterine System
oral pills)—an Indian experience. An ICMR
Task Force Study. Indian Council of
Medical Research. Contraception. 1995;51:
155–165.
27. Backman T, Huhtala S, Luoto R, et al. Ad-
vance information improves user satisfaction
with the levonorgestrel intrauterine system.
Obstet Gynecol. 2002;99:608–613.
28. Grimes D, Schulz K, Stanwood N. Immediate
postabortal insertion of intrauterine devices.
Cochrane Database Syst Rev. 2004:CD001777.
29. Grimes D, Schulz K, Van Vliet H, et al. Im-
mediate post-partum insertion of intrauterine
devices. Cochrane Database Syst Rev.
2003:CD003036.
30. Eroglu K, Akkuzu G, Vural G, et al. Compar-
ison of efficacy and complications of IUD
insertion in immediate postplacental/early
postpartum period with interval period: 1 year
follow-up. Contraception. 2006;74:376–381.
31. WHO. Medical eligibility criteria for contra-
ceptive use. World Health Organization; 2004.
Available at: http://www.who.int/reproductive-
health/publications/mec/7_iud.pdf. Accessed
January 10, 2007.
32. Van Houdenhoven K, van Kaam KJ, van
Grootheest AC, et al. Uterine perforation in
women using a levonorgestrel-releasing in-
trauterine system. Contraception. 2006;73:
257–260.
33. Andersson K, Ryde-Blomqvist E, Lindell K,
et al. Perforations with intrauterine devices.
Report from a Swedish survey. Contraception.
1998;57:251–255.
34. Caliskan E, Ozturk N, Dilbaz BO, et al.
Analysis of risk factors associated with uterine
perforation by intrauterine devices. Eur
J Contracept Reprod Health Care. 2003;8:
150–155.
35. Grimes DA. Intrauterine device and upper-
genital-tract infection. Lancet. 2000;356:
1013–1019.
36. Farley TM, Rosenberg MJ, Rowe PJ, et al.
Intrauterine devices and pelvic inflammatory
disease: an international perspective. Lancet.
1992;339:785–788.
37. Grimes DA, Schulz KF. Antibiotic prophy-
laxis for intrauterine contraceptive device
insertion. Cochrane Database Syst Rev. 2001:
CD001327.
38. Mohllajee AP, Curtis KM, Peterson HB. Does
insertion and use of an intrauterine device
increase the risk of pelvic inflammatory disease
among women with sexually transmitted infec-
tion? A systematic review. Contraception.
2006;73:145–153.
39. Bonacho I, Pita S, Gomez-Besteiro MI. The
importance of the removal of the intrauterine
895
device in genital colonization by actinomyces.
Gynecol Obstet Invest. 2001;52:119–123.
40. Morrison CS, Sekadde-Kigondu C, Sinei SK,
et al. Is the intrauterine device appropriate
contraception for HIV-1-infected women?
BJOG. 2001;108:784–790.
41. Sinei SK, Morrison CS, Sekadde-Kigondu C,
et al. Complications of use of intrauterine
devices among HIV-1-infected women. Lancet.
1998;351:1238–1241.
42. Doll H, Vessey M, Painter R. Return of ferti-
lity in nulliparous women after discontinua-
tion of the intrauterine device: comparison
with women discontinuing other methods of
contraception. BJOG. 2001;108:304–314.
43. Daling JR, Weiss NS, Metch BJ, et al. Primary
tubal infertility in relation to the use of an
intrauterine device. N Engl J Med. 1985;312:
937–941.
44. Daling JR, Weiss NS, Voigt LF, et al. The
intrauterine device and primary tubal inferti-
lity. N Engl J Med. 1992;326:203–204.
45. Cramer DW, Schiff I, Schoenbaum SC, et al.
Tubal infertility and the intrauterine device.
N Engl J Med. 1985;312:941–947.
46. Hubacher D, Lara-Ricalde R, Taylor DJ, et al.
Use of copper intrauterine devices and the risk
of tubal infertility among nulligravid women.
N Engl J Med. 2001;345:561–567.
47. Vessey MP, Lawless M, McPherson K, et al.
Fertility after stopping use of intrauterine con-
traceptive device. Br Med J (Clin Res Ed).
1983;286:106.
48. Wilson JC. A prospective New Zealand study
of fertility after removal of copper intrauterine
contraceptive devices for conception and be-
cause of complications: a four-year study. Am
J Obstet Gynecol. 1989;160:391–396.
49. Skjeldestad F, Bratt H. Fertility after compli-
cated and non-complicated use of IUDs. A
controlled prospective study. Adv Contracept.
1988;4:179–184.
50. Belhadj H, Sivin I, Diaz S, et al. Recovery of
fertility after use of the levonorgestrel 20 mcg/d
or Copper T 380 Ag intrauterine device. Con-
traception. 1986;34:261–267.
51. Diaz S, Pavez M, Cardenas H, et al. Recovery
of fertility and outcome of planned pregnan-
cies after the removal of Norplant subdermal
implants or Copper-T IUDs. Contraception.
1987;35:569–579.
52. Grimes DA. Intrauterine devices and inferti-
lity: sifting through the evidence. Lancet.
2001;358:6–7.
53. Hov GG, Skjeldestad FE, Hilstad T. Use of
IUD and subsequent fertility-follow-up after
participation in a randomized clinical trial.
Contraception. 2007;75:88–92.
896 Chrisman et al
54. Duenas JL, Albert A, Carrasco F. Intrauterine
contraception in nulligravid vs parous women.
Contraception. 1996;53:23–24.
55. Rogovskaya S, Rivera R, Grimes DA, et al.
Effect of a levonorgestrel intrauterine system
on women with type 1 diabetes: a randomized
trial. Obstet Gynecol. 2005;105:811–815.
56. Graff-Iversen S, Tonstad S. Use of progesto-
gen-only contraceptives/medications and lipid
parameters in women age 40 to 42 years: results
of a population-based cross-sectional Norwe-
gian Survey. Contraception. 2002;66:7.
57. Kayikcioglu F, Gunes M, Ozdegirmenci O, et
al. Effects of levonorgestrel-releasing intrau-
terine system on glucose and lipid metabolism:
a 1-year follow-up study. Contraception.
2006;73:528.
58. Wollter-Svensson L-O, Stadberg E, Andersson
K, et al. Intrauterine administration of levo-
norgestrel in two low doses in HRT. A rando-
mized clinical trial during one year: effects on
lipid and lipoprotein metabolism. Maturitas.
1995;22:199.
59. Hurskainen R, Teperi J, Rissanen P, et al.
Quality of life and cost-effectiveness of levo-
norgestrel-releasing intrauterine system versus
hysterectomy for treatment of menorrhagia: a
randomised trial. Lancet. 2001;357:273–277.
60. Hurskainen R, Paavonen J. Levonorgestrel-
releasing intrauterine system in the treatment
of heavy menstrual bleeding. Curr Opin Obstet
Gynecol. 2004;16:487–490.
61. Lahteenmaki P, Haukkamaa M, Puolakka J,
et al. Open randomised study of use of
levonorgestrel releasing intrauterine system
as alternative to hysterectomy. BMJ. 1998;
316:1122–1126.
62. Soysal M, Soysal S, Ozer S. A randomized
controlled trial of levonorgestrel releasing
IUD and thermal balloon ablation in the treat-
ment of menorrhagia. Zentralbl Gynakol.
2002;124:213–219.
63. Busfield RA, Farquhar CM, Sowter MC, et al.
A randomised trial comparing the levonorges-
trel intrauterine system and thermal balloon
ablation for heavy menstrual bleeding. BJOG.
2006;113:257–263.
64. Brown PM, Farquhar CM, Lethaby A, et al.
Cost-effectiveness analysis of levonorgestrel
intrauterine system and thermal balloon abla-
tion for heavy menstrual bleeding. BJOG.
2006;113:797–803.
65. Irvine GA, Campbell-Brown MB, Lumsden
MA, et al. Randomised comparative trial of
the levonorgestrel intrauterine system and nor-
ethisterone for treatment of idiopathic menor-
rhagia. Br J Obstet Gynaecol. 1998;105:
592–598.
66. Kittelsen N, Istre O. A randomized study
comparing levonorgestrel intrauterine system
(LNG IUS) and transcervical resection of the
endometrium (TCRE) in the treatment of me-
norrhagia: preliminary results. Gynaecol En-
dosc. 1998;7:61–65.
67. Crosignani PG, Vercellini P, Mosconi P, et al.
Levonorgestrel-releasing intrauterine device
versus hysteroscopic endometrial resection in
the treatment of dysfunctional uterine bleed-
ing. Obstet Gynecol. 1997;90:257–263.
68. Istre O, Trolle B. Treatment of menorrhagia
with the levonorgestrel intrauterine system
versus endometrial resection. Fertil Steril.
2001;76:304–309.
69. Rauramo I, Elo I, Istre O. Long-term treat-
ment of menorrhagia with levonorgestrel in-
trauterine system versus endometrial resection.
Obstet Gynecol. 2004;104:1314–1321.
70. Reid PC, Virtanen-Kari S. Randomised
comparative trial of the levonorgestrel intrau-
terine system and mefenamic acid for the
treatment of idiopathic menorrhagia: a multi-
ple analysis using total menstrual fluid loss,
menstrual blood loss and pictorial blood
loss assessment charts. BJOG. 2005;112:
1121–1125.
71. Lethaby AE, Cooke I, Rees M. Progesterone
or progestogen-releasing intrauterine systems
for heavy menstrual bleeding. Cochrane Data-
base Syst Rev. 2005:CD002126.
72. Soysal S, Soysal ME. The efficacy of levonor-
gestrel-releasing intrauterine device in selected
cases of myoma-related menorrhagia: a pro-
spective controlled trial. Gynecol Obstet Invest.
2005;59:29–35.
73. Gardner FJ, Konje JC, Abrams KR, et al.
Endometrial protection from tamoxifen-sti-
mulated changes by a levonorgestrel-releasing
intrauterine system: a randomised controlled
trial. Lancet. 2000;356:1711–1717.
74. Petta CA, Ferriani RA, Abrao MS, et al.
Randomized clinical trial of a levonorgestrel-
releasing intrauterine system and a depot
GnRH analogue for the treatment of chronic
pelvic pain in women with endometriosis. Hum
Reprod. 2005;20:1993–1998.
75. Vercellini P, Frontino G, De Giorgi O, et al.
Comparison of a levonorgestrel-releasing in-
trauterine device versus expectant manage-
ment after conservative surgery for sympto-
matic endometriosis: a pilot study. Fertil Steril.
2003;80:305–309.
76. Lockhat FB, Emembolu JO, Konje JC. The
efficacy, side-effects and continuation rates in
women with symptomatic endometriosis un-
dergoing treatment with an intra-uterine ad-
ministered progestogen (levonorgestrel): a 3
 The Levonorgestrel Intrauterine System
year follow-up. Hum Reprod. 2005;20:
789–793.
77. Vercellini P, Aimi G, Panazza S, et al. A
levonorgestrel-releasing intrauterine system
for the treatment of dysmenorrhea associated
with endometriosis: a pilot study. Fertil Steril.
1999;72:505–508.
78. Fedele L, Bianchi S, Zanconato G, et al. Use of
a levonorgestrel-releasing intrauterine device
in the treatment of rectovaginal endometriosis.
Fertil Steril. 2001;75:485–488.
79. Maia H Jr, Maltez A, Coelho G, et al. Insertion
of mirena after endometrial resection in pa-
tients with adenomyosis. J Am Assoc Gynecol
Laparosc. 2003;10:512–516.
80. Fedele L, Bianchi S, Raffaelli R, et al. Treat-
ment of adenomyosis-associated menorrhagia
with a levonorgestrel-releasing intrauterine de-
vice. Fertil Steril. 1997;68:426–429.
81. Boon J, Scholten PC, Oldenhave A, et al.
Continuous intrauterine compared with cyclic
oral progestin administration in perimenopau-
sal HRT. Maturitas. 2003;46:69–77.
82. Wollter-Svensson LO, Stadberg E, Andersson
K, et al. Intrauterine administration of levo-
norgestrel 5 and 10 microg/24 hours in perime-
nopausal hormone replacement therapy. A
randomized clinical study during one year.
Acta Obstet Gynecol Scand. 1997;76:449–454.
83. Raudaskoski T, Tapanainen J, Tomas E, et al.
Intrauterine 10 microg and 20 microg levonor-
gestrel systems in postmenopausal women re-
ceiving oral oestrogen replacement therapy:
clinical, endometrial and metabolic response.
BJOG. 2002;109:136–144.
84. Raudaskoski TH, Lahti EI, Kauppila AJ, et al.
Transdermal estrogen with a levonorgestrel-
releasing intrauterine device for climacteric
complaints: clinical and endometrial re-
sponses. Am J Obstet Gynecol. 1995;172
(1 Pt 1):114–119.
85. Andersson K, Mattsson LA, Rybo G, et al.
Intrauterine release of levonorgestrel—a new
way of adding progestogen in hormone repla-
cement therapy. Obstet Gynecol. 1992;79:
963–967.
86. Suhonen SP, Allonen HO, Lahteenmaki P.
Sustained-release estradiol implants and a le-
vonorgestrel-releasing intrauterine device in
hormone replacement therapy. Am J Obstet
Gynecol. 1995;172(2 Pt 1):562–567.
87. Suhonen SP, Holmstrom T, Allonen HO, et al.
Intrauterine and subdermal progestin admin-
istration in postmenopausal hormone replace-
ment therapy. Fertil Steril. 1995;63:336–342.
897
88. Wildemeersch D, Janssens D, Schacht E, et al.
Intrauterine levonorgestrel delivered by a fra-
meless system, combined with systemic estro-
gen: acceptability and endometrial safety
after 3 years of use in peri- and postmeno-
pausal women. Gynecol Endocrinol. 2005;20:
336–342.
89. Sturdee DW, Rantala ML, Colau JC, et al. The
acceptability of a small intrauterine progesto-
gen-releasing system for continuous combined
hormone therapy in early postmenopausal wo-
men. Climacteric. 2004;7:404–411.
90. Varila E, Wahlstrom T, Rauramo I. A 5-year
follow-up study on the use of a levonorgestrel
intrauterine system in women receiving hor-
mone replacement therapy. Fertil Steril.
2001;76:969–973.
91. Suhonen S, Holmstrom T, Lahteenmaki P.
Three-year follow-up of the use of a levonor-
gestrel-releasing intrauterine system in hor-
mone replacement therapy. Acta Obstet
Gynecol Scand. 1997;76:145–150.
92. Montz FJ, Bristow RE, Bovicelli A, et al.
Intrauterine progesterone treatment of early
endometrial cancer. Am J Obstet Gynecol.
2002;186:651–657.
93. Vereide AB, Arnes M, Straume B, et al. Nucle-
ar morphometric changes and therapy moni-
toring in patients with endometrial
hyperplasia: a study comparing effects of in-
trauterine levonorgestrel and systemic me-
droxyprogesterone. Gynecol Oncol. 2003;91:
526–533.
94. Wildemeersch D, Dhont M. Treatment of non-
atypical and atypical endometrial hyperplasia
with a levonorgestrel-releasing intrauterine
system. Am J Obstet Gynecol. 2003;188:
1297–1298.
95. Scarselli G, Tantini C, Colafranceschi M, et al.
Levo-norgestrel-nova-T and precancerous le-
sions of the endometrium. Eur J Gynaecol
Oncol. 1988;9:284–286.
96. Perino A, Quartararo P, Catinella E, et al.
Treatment of endometrial hyperplasia with
levonorgestrel releasing intrauterine devices.
Acta Eur Fertil. 1987;18:137–140.
97. Ndumbe FM, Husemeyer RP. Endometrial
adenocarcinoma in association with a levonor-
gestrel-releasing intrauterine system (Mirena).
J Fam Plann Reprod Health Care. 2006;32:
113–114.
98. Jones K, Georgiou M, Hyatt D, et al. Endo-
metrial adenocarcinoma following the inser-
tion of a Mirena IUCD. Gynecol Oncol.
2002;87:216–218.
 CLINICAL OBSTETRICS AND GYNECOLOGY
Volume 50, Number 4, 898–906
r 2007, Lippincott Williams & Wilkins

Injectable Contraception
SADIA HAIDER, MD, MPH* and PHILIP D. DARNEY, MD, MScw
*Department of Obstetrics and Gynecology, Beth Israel Deaconess Hospital,
Harvard Medical Center, Boston, Massachusetts; and w Obstetrics,
Gynecology and Reproductive Sciences, San Francisco General Hospital,
University of California, San Francisco, California

Abstract: Injectable contraceptive methods are safe,
highly efficacious, and commonly used worldwide.
Depo-Provera (depot-medroxyprogesterone acetate)
is the most commonly used injectable in the United
States. It is a convenient, discrete, and low-mainte-
nance method, and is ideal for patients with contra-
indications to estrogen use and certain medical
conditions. In addition, there are many noncontra-
ceptive benefits to Depo-Provera use. Side effects with
this method including irregular bleeding, breast ten-
derness, weight gain, and the impact on bone mineral
density should be taken into consideration when
prescribing the method. Other injectables such as
Mesigyna and norethindrone enanthate are widely
used outside the United States.
Key words: injectable contraceptive, Depo-Provera
Depot-Medroxyprogesterone
Acetate
According to the 2002 National Survey of Fa-
mily Growth, 5.3% of all US women between
the ages of 15 and 44 currently use depot-
medroxyprogesterone acetate (DMPA or
Depo-Provera) as their chosen method of con-
traception.1 A higher proportion of women in
younger age groups use DMPA with 13.9% of
women aged 15 to 19 using DMPA compared
with 10.1% between the age of 20 and 24, and
1.6% between the ages of 40 and 44. A decline in
adolescent pregnancy rates between 1995 and
2002 has been attributed to increased use of
contraception in general, including DMPA use
by younger women.
Correspondence: Philip D. Darney, MD, MSc, Profes-
sor and Chief, Obstetrics, Gynecology and Reproduc-
tive Sciences, San Francisco General Hospital,
University of California, San Francisco, 1001 Potrero
Avenue 6D, San Francisco, CA 94110. E-mail: darneyp
@obgyn.ucsf.edu
Depo-Provera is one the most widely used
and thoroughly studied contraceptive methods.
It has been available in many countries other
than the United States since the 1960s. Evidence
about the safety, efficacy, and acceptability of
DMPA comes from other countries including Sri
Lanka, Indonesia, Thailand, and Mexico where
DMPA has been used and studied for decades. It
was approved for use as contraception in the
United States in 1992. It is composed of micro-
crystals suspended in an aqueous solution. Stu-
dies comparing 100-mg intramuscularly (IM)
doses compared with 150 mg IM doses found
that 150 mg IM dose (gluteal or deltoid) is
required to effectively prevent pregnancy with
a first-year pregnancy rate of 0.3%.2
Other injectables including Mesigyna (50 mg
norethindrone enanthate and 5 mg estradiol
valerate), and norethindrone enanthate (200 mg
every 2 mo) are widely used outside the United
States. A new formulation of DMPA, 104 mg,
has recently been approved for subcutaneous
administration in the United States.
Mechanism of Action
Depo-Provera, like other progestin-only meth-
ods, alters the endometrial lining, thickens the
cervical mucus, and blocks the luteinizing hor-
mone surge to prevent ovulation. After method
discontinuation, ovulation resumes at 14 weeks,
but can take up to 18 months. In comparison to
combined hormonal methods DMPA does not
impact follicle-stimulating hormone as consis-
tently or as intensely. Thus, for one-third of
DMPA users, estradiol levels remain unchanged
from early-to-mid-follicular phase levels—ap-
proximately 40 to 50 pg/mL. For most DMPA
users, estradiol levels vary throughout the injec-
tion period and may reach postmenopausal

CLINICAL OBSTETRICS AND GYNECOLOGY / VOLUME 50 / NUMBER 4 / DECEMBER 2007

898

levels of <20 pg/mL, but vasomotor symptoms
such as hot flashes and vaginal atrophy are
uncommon in DMPA users. Absence of these
symptoms is likely due to the fact that proges-
tins, like estrogens, also suppress them. Onset of
contraceptive efficacy of Depo-Provera is within
24 hours after injection and is maintained for at
least 14 weeks providing a margin of protection
if reinjection typically scheduled at 12 weeks is
delayed. If reinjection is delayed beyond 14
weeks, women should be questioned regarding
unprotected intercourse since they were due for
their injection and should be evaluated for emer-
gency contraception.3 Because there is an asso-
ciation between higher neonatal and infant
mortality rates associated with initial DMPA
injection at the time of early pregnancy, the
timing of the first injection is important. The
first injection should be administered within the
first 5 days of the menstrual cycle. DMPA
(150 mg) can be given IM into the deltoid or
gluteal muscle using a z-track technique; the area
should not be massaged postinjection. A new
subcutaneous formulation is now available in
the United States. DMPA (104 mg/0.65 mL) is
administered subcutaneously every 3 months.
The 30% lower dose of progestin in the subcu-
taneous formulation is because it provides
slower and more sustained absorption of the
progestin; the duration of the effect is un-
changed.4 The risks of using the IM injection
and the subcutaneous injection are similar. The
subcutaneous formulation is less painful for
patients and has the potential for allowing pa-
tient self-administration whereas the IM formu-
lation is available in the generic form, and thus is
less costly.
DMPA takes 6 to 8 months after the last
injection to disappear and clearance is slower in
heavier women. Approximately half of women
who discontinue Depo-Provera can expect nor-
TABLE 1. Failure Rates During the First Year
of Use, United States6 (Percent of
Women With Pregnancy)
Lowest
Method Expected Typical
Combination pill 0.1% 7.6%
Progestin only 0.5% 3.0%
IUDs
Progesterone IUD 1.5% 2.0%
Levonorgestrel IUD 0.1% 0.1%
Copper T 380A 0.6% 0.8%
Implants 0.05 0.2
Injectable 0.3% 3.1%
Injectable Contraception 899
mal menses to return in 6 months after the last
injection, but 25% will wait a year before re-
sumption of a normal pattern. This delay should
be taken into consideration when counseling
patients about method choice5 (Table 1).
Efficacy
The efficacy of DMPA is equal to that of ster-
ilization and better than that of oral hormonal
contraceptive methods. The theoretical effec-
tiveness is 99.7% with a lowest expected rate of
pregnancy of 0.3%.6 Serum concentrations of
DMPA are high so efficacy of the method is not
compromised by increasing body weight or med-
ication-induced increased hepatic enzyme activ-
ity patients with contraindications to estrogen
use, DMPA, a progestin-only method, is a safe
and effective alternative.
Indications for DMPA use:
1. At least 1 year of birth spacing desired.
2. Highly effective long-acting contraception
not linked to coitus desired.
3. Private, coitally independent method desired.
4. Estrogen-free contraception needed.
5. Breastfeeding.
6. Sickle cell disease.
7. Seizure disorder.
Absolute Contraindications:
1. Pregnancy.
2. Unexplained genital bleeding.
3. Severe coagulation disorders.
4. Previous sex steroid-induced liver adenoma.
Relative Contraindications:
1. Liver disease.
2. Severe cardiovascular disease.
3. Rapid return to fertility desired.
4. Difficulty with injections.
5. Severe depression.
Advantages
Depo-Provera is an especially good choice for
contraception in women who find it challenging
to use contraception regularly. It is a convenient,
discrete, and low-maintenance method. These
characteristics make it ideal for adolescents
seeking contraception. Furthermore, the effi-
cacy of DMPA does not rely on daily compli-
ance. The intramuscular and subcutaneous
injections of DMPA are required only once
every 3 months and it does not depend on
900 Haider and Darney
partner cooperation or user action at the time of
intercourse. The availability of the newer sub-
cutaneous injection of DMPA provides an even
more convenient option for women who cannot
make a clinic visit for reinjection every 3 months.
Compared with the oral contraceptive pill
DMPA continuation rates are increased and
repeat pregnancy rates are reduced in teenagers.
DMPA as a progestin-only method can be
used in patients with contraindications to estro-
gen such as women over age 35 who are at risk
because of smoking, hypertension, or diabetes.
DMPA can also be used safely in women with
seizure disorder, sickle cell anemia, congenital
heart disease, migraines with aura, and a pre-
vious history of venous thromboembolism. Ob-
servational data from epidemiologic studies and
a World Health Organization (WHO) case-
control study have not shown increased risks
of stroke, myocardial infarction, or venous
thromboembolism.7
Depo-Provera is a good contraceptive option
for women with a history of seizure disorder and
sickle cell anemia. In patients with seizure dis-
order antiepileptic drugs increase hepatic enzyme
activity and the metabolism of contraceptive
steroids. Oral contraceptives are not recom-
mended in patients on antiepileptic drugs per
WHO guidelines. However, DMPA hormone
levels are high enough that contraceptive efficacy
is not compromised by increased metabolic ac-
tivity of hepatic enzymes. Furthermore, Depo-
Provera can have a beneficial effect on seizure
control as it is thought that progestins increase
the seizure threshold in patients with seizure
disorders. In patients with sickle cell anemia,
there is evidence that DMPA results in decreased
sickling of red blood cells, and a reduction in the
frequency and intensity of sickle cell crisis.
Noncontraceptive benefits of DMPA include
a decreased risk of endometrial cancer.8 Reduc-
tion in risk of endometrial cancer is similar to
that observed with oral contraceptive pills. In
addition, like oral contraceptive pills, DMPA
decreases the risk of pelvic inflammatory dis-
ease, endometriosis, and uterine fibroids.
Lactation
The physiologic changes required for breastfeed-
ing are dependent on the drop in serum concentra-
tions of progesterone after the delivery of the
placenta. This mechanism causes some concern
regarding the use of DMPA during breastfeeding.
A prospective cohort study evaluated the effect of
early administration of DMPA on initiation of
breastfeeding. This study found no adverse effects
on breastfeeding. However, the mean time of
initiation of DMPA was 50 hours postpartum.9
This delay should be kept in mind when initiating
DMPA postpartum as there is a theoretical risk of
decreasing breastfeeding if progesterone is intro-
duced too soon after delivery not allowing for the
natural physiologic drop in progesterone to occur
for initiation of milk production. Studies have
shown no adverse effect on breastfeeding in wo-
men who receive DMPA at postpartum follow-up
visits. In particular, there was no effect on the
quantity and quality of milk production while on
Depo-Provera. Thus, DMPA can safely be admi-
nistered at the 2-week or 6-week postpartum visit.
This schedule will provide full contraceptive pro-
tection for lactating mothers in whom ovulation
does not occur until at least 12 weeks after delivery.
Use of DMPA does not affect the duration of
breastfeeding. The concentration of DMPA in the
breast milk is negligible, and no effects of the drug
on infant growth and development have been
observed. One group of women, obese Latina
women with prior gestational diabetes who used
DMPA postpartum require extracaution when
prescribing DMPA. In this group of breastfeeding
women who were taking the progestin-only oral
contraceptive pill, a 3-fold increased risk of non-
insulin-dependent diabetes was observed.10 It is
unknown whether other progestin-only methods
might pose a risk in obese Latinas who have a
history of gestational diabetes.
Summary of Advantages
1. Long-acting contraceptive benefit is not
dependent on daily user action or partner
cooperation at time of intercourse.
2. Discrete, use not detectable.
3. No serious health side effects.
4. Highly effective, efficacy is comparable to
sterilization, intrauterine contraception, and
implant contraception.
5. Multiple noncontraceptive benefits.
6. Good alternative for women with contraindi-
cations to estrogen.
7. Safe for breastfeeding mothers.
8. Positive impact on sickle cell anemia and
seizure disorders.
Disadvantages
There are a few major side effects associated with
Depo-Provera use. These include irregular men-
strual bleeding, breast tenderness, weight gain,

depression, acne, and headache. These side ef-
fects combined resulted in over 70% disconti-
nuation at 1 year in 1. One-quarter of women
initiating DMPA use discontinue it due to irre-
gular bleeding in the first year.11 If bleeding
occurs shortly after beginning DMPA, patients
should be counseled and advised that irregular
bleeding decreases with use of DMPA for a few
more cycles. Overall incidence of irregular bleed-
ing is 70% in the first year with a decrease in
irregular bleeding with each reinjection, and
rates as low as 10% after the first year. Fifty
percent of the patients report amenorrhea by 1
year with 10% becoming amenorrheic by the
first injection interval, 40% reporting amenor-
rhea by the fourth injection cycle, and 80%
reporting amenorrhea at 5 years.5
Irregular bleeding which begins several
months after initiation is due to decidualization
of the endometrium and can be treated with
exogenous estrogen such as 1.25 mg conjugated
estrogens, or 2 mg estradiol, given daily for 7 to
14 days. In addition, nonsteroidal anti-inflam-
matory drugs given for a week or oral contra-
ception pill supplementation during the first 1 to
3 months of DMPA injection, which is when
irregular bleeding tends to be worst, can be used
for treating it.
Studies of exogenous estrogens including
supplementation with 50 mcg ethinyl estradiol,
2.5 mg estrone sulfate, or transdermal estrogen
have shown limited effect on bleeding with no
change in continuation rates of DMPA.12 Over-
all continuation of DMPA at year 1 is about
33%, 50% at year 2, and as low as 20% at the
end of 3 years.13
Weight gain is a common complaint among
DMPA users with approximately 2.1% of users
discontinuing the method due to weight in-
crease. As is the case with other hormonal con-
traceptive methods, weight gain while using
DMPA may be a result of dietary changes and
aging and not hormonally induced. Studies of
weight gain and DMPA use have shown mixed
results. Prospective controlled studies evaluat-
ing weight gain in normal weight women showed
no relationship between DMPA use and food
intake, energy expenditure, or body weight.
However, in certain subgroups DMPA use has
been correlated with weight gain. Obese adoles-
cent DMPA users gained significantly more
weight when compared with obese adolescents
using oral contraceptive pills or nonhormonal
contraception. Similarly, it has been show that
Navajo women using DMPA have significant
weight gain.14 Weight change should be taken
Injectable Contraception 901
TABLE 2. Most Common Reasons for Disconti-
nuing Depo-Provera in the First 2
Years of Use15
Headaches 2.3%
Weight gain 2.1%
Dizziness 1.2%
Abdominal pain 1.1%
Anxiety 0.7%
into consideration when counseling patients
about using DMPA (Table 2).
Cancer
Evidence from animal data indicates that
estrogens and progestins could be carcinogenic,
and that progestins could serve as tumor pro-
moters.
Concerns about Depo-Provera use and in-
creased risk of liver, breast, ovarian, and cervical
cancer have not been confirmed in humans.
Earlier population-based case-control studies
indicate a possible association between breast
cancer and DMPA. These studies had small
numbers with only 19 cases in a study conducted
in Costa Rica. Another earlier study in New
Zealand did not find an increased relative risk
in those who had ever used DMPA, but did find
an increased risk among those who recently
initiated the method and were under age 25.
Given the fact that these studies demonstrated
increased risk on the basis of a small number of
cases it is not clear whether the risk, which seems
to be small, is actually due to confounding
variables.16 Other data suggesting a possible link
between DMPA use and breast cancer come
from a large, hospital-based case-control study
conducted by the WHO over 9 years in 3 devel-
oping countries. This study found that exposure
to DMPA is associated with a very slight in-
crease risk in breast cancer in the first 4 years of
use, but there was no evidence for an increase in
risk with duration of use.17 The limitation of this
study was that the number of cases in recent
users was small and limited the interpretation of
conclusions. As is the case with oral contracep-
tive pills and hormone replacement therapy, it is
possible that DMPA accelerated the growth of
an already present tumor or that detection of the
tumor occurred earlier because of study-related
surveillance. More recent data from a popula-
tion-based multinational case-control study,
qthe Women’s Contraceptive and Reproductive
902 Haider and Darney
Experiences Study of the National Institute
of Child Health and Human Development,
assessed the association between exposure to
injectable contraceptives and risk of breast
cancer. The use of injectable contraceptives
was not associated with an increased risk
of breast cancer [odds ratio (OR) = 0.9, 95%
confidence interval (CI) = 0.7, 1.2], and risk was
not increased in current users (method used
within 1 y of the reference date) (OR = 0.7,
95% CI = 0.4, 1.3), or in those initiating the
method within 5 years of the reference date
(OR = 0.9, 95% CI = 0.5, 1.4).18
These studies did not find an increased risk
for breast cancer or increase in risk with dura-
tion of use. However, there is some evidence
suggesting that recent use maybe related to
breast cancer development, thus clinicians
should consider counseling patients that DMPA
might accelerate the growth of an already pre-
sent cancer rather than cause a new disease.
If such cases occur, disease would likely be
detected earlier and result in better outcomes.
Earlier detection is the probable explanation for
lower breast cancer mortality rates in women
who use oral contraceptive pills.
Other Cancers
Increased risk of cervical cancer while on Depo-
Provera has not been confirmed. The WHO
Collaborative Study of Neoplasia and Steroid
Contraceptives conducted a hospital-based case-
control study in Thailand, Mexico, and Kenya.
This study did not find a statistically significant
increase in invasive cervical cancer with DMPA
use. Furthermore, there was no association with
increasing risk with duration of use.19 In one
study conducted in New Zealand, a slight in-
crease in the risk of cervical dysplasia among
users of Depo-Provera was noted, but could be
attributed to an increased prevalence of known
risk factors (ie, multiple sexual partners, history
of sexually transmitted infections) for dyspla-
sia.20 Although there is no increase in overall risk
of invasive cervical cancer with DMPA use,
clinicians should provide recommended cervical
cytologic surveillance in all users of contracep-
tion including DMPA users, but Pap smears
should not be required before initiation of
DMPA or other methods.
As with oral contraceptive use, there is no
evidence for increased risk of liver cancer with
use of Depo-Provera. Strong evidence supports
a reduction in risk of endometrial cancer in
DMPA users at least as great as that seen with
oral contraceptive users relative risk = 0.21
(95% CI = 0.06, 0.79) in women who had ever
used DMPA in a Thai study.21 There is no
evidence for increased risk of ovarian cancer
with DMPA use, and potentially a modest de-
crease in overall risk.
Depression
Although DMPA labeling suggests that depres-
sion may worsen with use of the method, studies
regarding DMPA use and depression are incon-
clusive. No significant increase in depressive
symptoms were observed among young, poor,
Hispanic inner city women who are known to
have a high prevalence of depression. There is
little evidence for increasing depression with
short-term or long-term DMPA use and it is
not a contraindication to DMPA use.22 A pro-
spective study set in an urban clinic compared
adolescents on DMPA to controls and used 2
standardized questionnaires, the Beck Depres-
sion Inventory (BDI) and the Multiple Affect
Adjective Checklist-Revised (MAACL-R), and
found no statistically significant differences
in depressive symptoms when using DMPA
as a contraceptive agent over a period of
12 months and no significant changes in negative
or positive affect.23 Depression should not be a
reason to withhold DMPA administration in
patients for whom DMPA is an ideal contra-
ceptive method.
Metabolic Effects
The overall impact of Depo-Provera on lipopro-
teins remains uncertain. There seems to be no
clinically significant alterations in lipoprotein
profiles while on Depo-Provera. It is thought
that there may be no impact on lipoproteins due
to the avoidance of a first-pass metabolism
through the liver. However, there does seem to
be a theoretical risk of increased total cholesterol
and low-density lipoprotein-cholesterol and a
decrease in high-density lipoprotein-cholesterol
during the initial period after initiation of
DMPA. A multicenter clinical trial by the
WHO found a transient adverse impact in the
first few weeks after injection when hormonal
levels in the blood were high.24 The clinical
significance of these findings is unclear. It seems
that following lipid profiles of patients with
hyperlipidemia or with a family history of
it would be prudent. Depo-Provera does not
affect carbohydrate metabolism or coagulation
factors.

Effect on Bone Density
Owing to the decrease in endogenous serum
estradiol levels secondary to DMPA, a relative
hypoestrogenic state exists among users that is,
associated with increased bone resorption. Gi-
ven this physiologic change in DMPA users,
there has been increasing concern about the
impact of DMPA on bone health. In 2004, the
US Food and Drug Administration (FDA)
added a black box warning to DMPA labeling
which stated that women who use DMPA might
lose significant bone mineral density (BMD).
The warning also cautioned providers that bone
loss increases with greater duration of DMPA
use and may not be completely reversible.
Furthermore, the warning stated that it is un-
known whether use of DMPA during adoles-
cence or early adulthood will reduce peak bone
mass or increase risk for osteoporotic fracture
later in life. The FDA then concluded that
DMPA should be used as birth control for long-
er than 2 years only if other contraceptive meth-
ods are not advisable.
The BMD effects of DMPA are equivalent to
those of pregnancy and lactation. A hypoestro-
genic state occurs in women after their deliveries
and is known to result in a decline in BMD.
During lactation breastfeeding women experi-
ence progressive declines in BMD compared
with non-breastfeeding women and are known
to regain BMD after cessation of breastfeeding.
Similarly, in the case of Depo-Provera use, most
studies have found a decrease in BMD in adults
compared with their own baseline BMD or
compared with nonusers. There were similar
findings among adolescent DMPA users com-
pared with nonusers. Only some of these studies
found a statistically significant difference in
bone density changes between DMPA users
and nonusers whereas others did not. Available
data do not suggest that DMPA will reduce peak
bone mass and increase the risk of osteoporotic
fracture rate. A systematic review of DMPA
studies revealed that longitudinal data from
adult DMPA users had results that differed
about the rates of BMD reduction and seemed
to change over time. Most studies showed re-
duction in BMD of less than 1% per year of
use.25 In addition, most of the decrease in BMD
occurs in the first 2 years of DMPA use, with
little incremental BMD loss during additional
years of use.26 There are no data on postmeno-
pausal fracture rates in women who used DMPA
as adolescents or adults thus, the clinical signifi-
cance of DMPA associated decline in BMD is
unknown. Furthermore, most of the evidence
Injectable Contraception 903
currently available indicates that women regain
BMD after discontinuation of DMPA use. A 7-
year prospective study found at 96 weeks (1.8 y)
after DMPA was discontinuous, total hip BMD
had returned almost to baseline levels, only
– 0.20% lower than baseline, and BMD mea-
surements at the lumbar spine showed partial
recovery ( – 1.19%).27 After discontinuation of
DMPA in adult women users, BMD recovery
differed by anatomic site and length of
DMPA use. All women had return to levels
comparable to nonuser controls by 76 months
and return to baseline levels by 92 months.
Two years after discontinuation of long-term
(median 10 y; range: 3 to 17 y) DMPA use,
BMD in the lumbar spine increased significantly
(P<0.002), with mean increases overall of 3%
(0.4% to 5.6%) at 1 year and 6.4% (3.4% to
9.4%) at 2 years.28
Because adolescence is a period of critical
bone formation, there has been even more con-
cern about the use of DMPA in adolescent users.
Evidence regarding DMPA use in adolescents
and BMD recovery is similar to that in adults,
and is overall reassuring. The only study to look
at BMD changes in adolescents after DMPA
discontinuation is a population-based prospec-
tive cohort study of users age 14 to 18, the mean
BMD at all anatomic sites reached or exceeded
the levels of never-users by 12 months after
DMPA discontinuation, a faster rate of recovery
than that seen in the adult studies.29
After the FDA black box warning resulted in
changes in Depo-Provera labeling, many profes-
sional organizations including the American
College of Obstetrician Gynecologists (ACOG),
the Society for Adolescent Medicine (SAM), and
the WHO reviewed the existing evidence and
endorsed guidelines differing from DMPA pro-
duct labeling. All 3 of these organizations stated
that there should be no restriction on initiation
or duration of Depo-Provera use in adult women
who are eligible to use this method as the risk of
osteoporotic fractures is only theoretical, and
does not warrant alteration in current practice
patterns. Furthermore, adolescents should not
be denied DMPA as the risk of unintended
pregnancy far outweighs the theoretical con-
cerns about fracture risk. Overall risks and
benefits for continuing DMPA use should be
reconsidered over time with individual adoles-
cent users.
The decline in BMD in Depo-Provera users is
similar to that associated with lactation. There is
no evidence of clinically significant adverse out-
comes (ie, increased fracture rates) associated
904 Haider and Darney
with current, past, or prolonged DMPA use.
Bone densitometry testing or antiresorptive
agents should not be used in otherwise healthy
adolescents or adults using Depo-Provera. All
users of contraception, including DMPA users
should be encouraged to perform weight-bear-
ing exercise, not to smoke, and have a sufficient
amount of dietary calcium (1300 mg) and vita-
min D (400 IU) daily in order for adolescents to
attain peak bone mass, and adults to decrease
osteoporotic risk. Concerns about skeletal
health in adults and adolescents should not
restrict initiation, continuation, or duration of
DMPA use.
Effect on Future Fertility
Despite concern about the delay in becoming
pregnant after cessation of DMPA use, Depo-
Provera does not permanently suppress ovarian
function, and the pregnancy rate in women with
a desire to become pregnant after discontinuing
DMPA is normal. By 18 months after disconti-
nuation of the method 90% of Depo-Provera
users have become pregnant which is the same
proportion as other methods.30 There is a delay
in conception after DMPA. This delay is ap-
proximately 9 months after discontinuation of
the method and there is no increase in the delay
with duration of use. Need for pregnancy plan-
ning should be taken into consideration when
counseling patients as those patients wanting to
conceive immediately after stopping the method
should not use DMPA. If menstrual function
remains suppressed beyond 18 months after
discontinuation further evaluation of other
etiologies unrelated to DMPA administration
should be pursued.
Short-term Injectable
Contraceptives
Short-term injectable contraception of estrogen
and progestin administered on a monthly or
alternate month basis has been available around
the world for some time. These injections are
popular methods in China, Latin America, and
Eastern Asia.
Lunelle (or Cyclofem)
Lunelle, a monthly injectable, consists of 25 mg
of DMPA and 5 mg estradiol cypionate, and is
administered every 28 to 30 days (not to exceed
33 d) as a deep intramuscular injection. Lunelle
was taken off the market in the United States in
2002 by the manufacturer because of concerns
that there may not have been enough medication
in the syringe containing the medication, but is
still available in other countries. Lunelle is as
efficacious as DMPA, but has a better bleeding
profile with less irregular bleeding and amenor-
rhea.31 In addition, fertility rates after disconti-
nuation of the method are similar to oral
contraceptives due to rapid reversibility of the
method. Owing to the estrogen and progestin
components of the method, the same risks,
benefits, and contraindications as oral contra-
ceptive pills apply to Lunelle.
Norethindrone Ethanate
Another progestin-only injectable, norethin-
drone ethanate, is given in a dose of 200 mg IM
every 2 months. This progestin has the same
mechanism of action as DMPA as well as the
same advantages and disadvantages.
Mesigyna
Mesigyna is a combination of norethindrone
ethanate (50 mg) and estradiol valerate (5 mg)
given monthly. It has a better bleeding
profile than even Lunelle, provides effective
contraception and cycle control, and rapid re-
turn to fertility within 1 month after disconti-
nuation.
Dihydroxyprogesterone
Acetophenide and Estradiol
Enanthate
A combination injectable, 150 mg dihydroxy-
progesterone acetophenide with 10 mg estradiol
enanthate is the most commonly used injectable
in Latin America. It is also given monthly like
Mesigyna and Lunelle and has similar bleeding
profiles.
References
1. Mosher WD, Martinez GM, Chandra A, et al.
Use of contraception and use of family plan-
ning services in the United States: 1982-2002.
Adv Data. 2004;350:1–35.
2. World Health Organization. A multicentered
phase III comparative clinical trial of depot-
medroxyprogesterone acetate given three-
monthly at doses of 100 mg or 150 mg: I.
Contraceptive efficacy and side effects.
Contraception. 1986;34:223.

3. Mishell DR Jr. Pharmacokinetics of depot
medroxyprogesterone acetate contraception.
J Reprod Med. 1996;41:381–390.
4. Jain J, Dutton C, Nicosia A, et al. Pharmaco-
kinetics, ovulation suppression and return to
ovulation following a lower dose subcutaneous
formulation of Depo-Provera. Contraception.
2004;70:11–18.
5. Gardner JM, Mishell DR Jr. Analysis of bleed-
ing patterns and resumption of fertility follow-
ing discontinuation of a long-acting injectable
contraceptive. Fertil Steril. 1970;21:286.
6. Fu H, Darroch JE, Haas T, et al. Contracep-
tive failure rates: new estimates from the 1995
National Survey of Family Growth. Fam Plann
Perspect. 1999;31:58.
7. WHO. Cardiovascular disease and use of oral
and injectable progestogen-only contracep-
tives and combined injectable contraceptives.
Results of a an international, multicenter,
case-control study. Contraception. 1998;57:
315–324.
8. WHO. Depot-medroxyprogesterone acetate
(DMPA) and risk of endometrial cancer. The
WHO Collaborative Study of Neoplasia and
Steroid Contraceptives. Int J Cancer.
1991;49:186–190.
9. Truitt ST, Fraser AB, Grimes DA, et al. Com-
bined hormonal versus nonhormonal versus
progestin-only contraception in lactation. Co-
chrane Database Syst Rev. 2003:CD003988.
10. Kjos SL, Peters RK, Xiang A, et al. Contra-
ception and the risk of type 2 diabetes in Latino
women with prior gestational diabetes. JAMA.
1998;280:533.
11. Cromer BA, Smith RD, Blair JM, et al.
A prospective study of adolescents who choose
among levonorgestrel implant (Norplant),
medroxyprogesterone acetate (Depo-Provera),
or the combined oral contraceptive pill as
contraception. Pediatrics. 1994;94:687.
12. Goldberg AB, Cardenas LH, Hubbard AD,
et al. Post-abortion depomedroxyprogesterone
acetate continuation rates: a randomized trial
of cyclic estradiol. Contraception. 2002;66:
215.
13. Said S, Sadek W, Rocca M. Clinical evaluation
of the therapeutic effectiveness of ethinyl oes-
tradiol and oestrone sulphate on prolonged
bleeding in women using depot medroxypro-
gesterone acetate for contraception. Hum
Reprod. 1996;11(suppl 2):1–13.
14. Smith RD, Cromer BA, Hayes JR, et al. Me-
droxyprogesterone acetate (Depo-Provera) use
in adolescents: uterine bleeding and blood
pressure patterns, patient satisfaction, and
continuation rates. Adolesc Pediatr Gyencol.
1995;8:24.
Injectable Contraception 905
15. Espey E, Steinhart J, Ogburn T, et al. Depo-
Provera associated with weight gain in Navajo
women. Contraception. 2000;62:55.
16. WHO. Multinational comparative clinical eva-
luation of two long-acting injectable contra-
ceptive steroids: norethisterone enanthate and
medroxyprogesterone acetate. Contraception.
1983;28:1.
17. Paul C, Skegg DC, Spears GF. Depot-medrox-
yprogesterone (Depo-Provera) and risk of
breast cancer. Br Med J. 1989;299:7591.
18. WHO. Collaborative study of neoplasia and
steroid contraceptives, breast cancer and de-
pot-medroxyprogesterone acetate: a multina-
tional study. Lancet. 1991;338:833.
19. Strom BL, Berlin JA, Weber AL. Absence
of an effect of injectable and implantable pro-
gestin-only contraceptives on subsequent risk of
breast cancer. Contraception. 2004;69:353–360.
20. WHO. Depot-medroxyprogesterone acetate
(DMPA) and risk of invasive squamous cell
cervical cancer. The WHO Collaborative
Study of Neoplasia and Steroid Contracep-
tives. Contraception. 1992;45:299–312.
21. The New Zealand Contraception and Health
Study Group. History of long-term use of de-
pot-medroxyprogesterone acetate (DMPA) in
patients with cervical dysplasia; case-control
analysis nested in a cohort-study. Contracep-
tion. 1994;50:443.
22. WHO. Depot-medroxyprogesterone acetate
(DMPA) and risk of endometrial cancer. Int
J Cancer. 1991;49:186–190.
23. Westoff C, Truman C, Kalmuss D, et al.
Depressive symptoms and Depo-Provera.
Contraception. 1995;51:351.
24. Gupta N, O’Brien R, Jacobson LJ, et al. Mood
changes in adolescents using depot-medroxy-
progesterone acetate for contraception: a
prospective study. J Pediatr Adolesc Gynecol.
2001;14:71.
25. WHO. A multicentre comparative study of
serum lipids and apolipoproteins in long-term
users of DMPA and a control group of IUD
users. Contraception. 1993;47:177.
26. Curtis KM, Martins SL. Progestogen-only
contraception and bone mineral density: a
systematic review. Contraception. 2006;73:
470–487.
27. Scholes D, Lacroix AZ, Ott SM, et al. Bone
mineral density in women using depot medro-
xyprogesterone acetate for contraception.
Obstet Gynecol. 1999;93:233–238.
28. Kaunitz AM, Miller PD, Rice VM, et al. Bone
mineral density in women aged 25-35 years
receiving depot medroxyprogesterone acetate:
recovery following discontinuation. Contra-
ception. 2006;74:90–99.
906 Haider and Darney
29. Cundy T, Cornish J, Roberts H, et al.
Menopausal bone loss in long-term users of
depot medroxyprogesterone acetate contra-
ception. Am J Obstet Gynecol. 2002;186:
978–983.
30. Scholes D, Lacroix AZ, Ichikawa LE, et al.
The association between depot medroxypro-
gesterone acetate contraception and bone
mineral density in adolescent women. Contra-
ception. 2004;69:99–104.
31. Schwallie P, Assenze J. The effect of depo-
medroxyprogesterone acetate on pituitary
and ovarian function, and the return to fertility
following its discontinuation. Contraception.
1974;10:181.
32. Kaunitz AM, Garceau RJ, Cromie MA, et al.
Comparative safety, efficacy, and cycle control
of Lunelle monthly contraceptive injection
and Ortho-Novum 7/7/7 oral contraceptive.
Contraception. 2000;62:289.

The Contraceptive Implant
HEATHER HOHMANN, MD* and
MITCHELL D. CREININ, MD* w
*Department of Obstetrics, Gynecology, and Reproductive Sciences, University
of Pittsburgh School of Medicine; and w Department of Epidemiology,
University of Pittsburgh Graduate School of Public Health, Pittsburgh,
Pennsylvania
Abstract: Contraceptive implants provide long-acting,
highly effective reversible contraception. Currently,
the only subdermal implant available to women in the
United States is the single rod etonogestrel implant,
Implanon (N.V. Organon, Oss, the Netherlands) ap-
proved by the Food and Drug Administration in July
2006. Implanon is currently approved for 3 years of
use, provides excellent efficacy throughout its use, and
is easy to insert and remove. Similar to other proges-
tin-only contraceptives, Implanon can cause irregular
vaginal bleeding. Implanon has been shown to be safe
to use during lactation, may improve dysmenorrhea,
and does not significantly affect bone mineral density,
lipid profile, or liver enzymes.
Key words: contraception, Implanon, progestin-only,
ENG implant
The Contraceptive Implant
Contraceptive implants provide long-acting,
highly effective reversible contraception. All
subdermal implants for clinical use in humans
use progestins. These methods offer an excellent
contraceptive option for women who have con-
traindications to combined hormonal methods
and an option for any woman who desires long-
term protection against pregnancy that is rapidly
reversible.
Background
Clinical research on subdermal implant contra-
ception has been carried out since the late 1960s.
Norplant System (Schering, Berlin) is a levonor-
Correspondence: Heather Hohmann, MD, Department
of Obstetrics, Gynecology, and Reproductive Sciences,
University of Pittsburgh School of Medicine, Magee-
Womens Hospital, 300 Halket Street, Pittsburgh, PA
15213. E-mail: hhohmann@mail.magee.edu
Sources of support: none.
CLINICAL OBSTETRICS AND GYNECOLOGY
CLINICAL OBSTETRICS AND GYNECOLOGY
Volume 50, Number 4, 907–917
r 2007, Lippincott Williams & Wilkins
gestrel (LNG) implant, which first received ap-
proval in Finland in 1983. Although it had been
extensively used in Finland and in parts of
Southeast Asia, it did not receive approval from
the United States Food and Drug Administra-
tion (FDA) until 1990. Norplant System consists
of 6 silicone polymer capsules containing a total
of 216 mg of LNG and is FDA approved for 5
years of use, although evidence exists for effec-
tiveness through 7 years.1 One long-term study,
the Norplant Postmarketing Surveillance Study
(NPMS), enrolled 16,000 women in 8 different
countries for a 5-year controlled cohort study to
compare safety and contraception efficacy of
Norplant System, intrauterine device (IUD),
and sterilization. Follow-up was completed by
95% of women. Over 39,000 women-years of
observation were completed for Norplant Sys-
tem users. Efficacy rates reported using a Pearl
index were 0.27 per 100 women-years for Nor-
plant System, compared with 0.88 per 100 wo-
men-years for copper IUD users and 0.17 per
100 women-years for sterilization.2 In a multi-
national trial of 1210 women who were offered
Norplant System for 7 years, the mean duration
of use was 4.16 years but 22.6% of women
continued use for the entire 7 years.3 At 5 years,
the cumulative pregnancy rate was 1.1/100 wo-
men-years. The cumulative rate increased to 1.9/
100 women-years at the completion of 7 years,
giving women aged 18 to 33 a similar median
pregnancy rate as expected with tubal steriliza-
tion3 and providing evidence for the effective-
ness of the Norplant System for up to 7 years.1
Despite this excellent efficacy profile, experi-
ence with the millions of worldwide Norplant
System users prompted manufacturers to devel-
op advances in implant technology, namely
reducing the number of implants to 1 or 2 instead
/ VOLUME 50 / NUMBER 4 / DECEMBER 2007
907
908 Hohmann and Creinin
of 6 to facilitate easier insertion and removal. A
2-rod LNG implant was developed in the 1980s
and is currently available in European countries
marketed under the name Norplant II or Jadelle
(Schering, Berlin). This 2-rod system was ap-
proved by the FDA in 1996 but has never been
marketed in the United States. Jadelle contains a
total of 150 mg of LNG in two 4-cm rods.
Similar efficacy rates are seen with this implant
as with the original 6-rod system.1,4 Other
implants are available and undergoing research
in countries outside of the United States.
Nestorone5,6 containing implants are marketed
in Brazil on a limited basis and Uniplant7,8
implants are still under investigation. None of
these methods are expected to be available in
the United States in the near future.
Although Norplant System entered the US
market with initial enthusiasm, by the mid-1990s
the implant was surrounded by controversy
secondary to allegations of misuse including
coercion directed against low-income women.
Wyeth-Ayerst, the manufacturer of Norplant
System was also dealing with lawsuits alleging
significant side effects and claims of complicated
and painful removals.9 Ironically, none of these
claims were substantiated and all class action
lawsuits were decertified. In 2000, Wyeth-Ayerst
advised physicians to stop insertions secondary
to lower than expected hormone release rates
noted from a single manufacturing lot.1
Although further testing revealed that this lower
rate was not problematic for efficacy, US dis-
tribution was officially halted in 2002.
In July, 2006 the FDA-approved Implanon
(N.V. Organon, Oss, The Netherlands), also re-
ferred to as the etonogestrel (ENG) implant. This
single rod subdermal implant is currently the only
implant available to women in the United States,
and will be the further focus of this review. The
ENG implant has been extensively used in
Australia, Indonesia, the Netherlands, and more
than 30 other countries. Implanon is currently
approved for 3 years of use, provides excellent
efficacy throughout its use, and is easy to insert
and remove. Additionally, Implanon can be used
during lactation10,11 and may improve dysmenor-
rhea.12 However, similar to other progestin-only
contraceptives, Implanon can cause side effects
such as irregular vaginal bleeding.13
Description and Pharmacology
The ENG implant is a single rod implant that
releases the gonane progestin ENG, also known
as 3-ketodeosogestrel, the biologically active
FIGURE 1. Implanon in palm.16
metabolite of desogestrel.14 ENG is the same
progestin used in the contraceptive vaginal ring.
This new implant measures 4-cm long and 2 mm
in diameter and has a core made from a non-
biodegradable solid composed of ethylene vinyl
acetate impregnated with 68 mg of ENG15
(Fig. 1). The ethylene vinyl acetate copolymer
of Implanon allows controlled release of hor-
mone over 3 years of use.17 Each implant is
provided in a disposable sterile inserter for sub-
dermal application (Fig. 2).
After ENG implant insertion, serum levels of
ENG rapidly rise to a mean serum concentration
of 265.9 ± 80.9 pg/mL by 8 hours,18 a level that
exceeds the 90 pg/mL needed to prevent ovula-
tion.19 Maximum serum concentrations are
usually seen by day 4 after implant insertion,
with a variation from day 1 to day 13. ENG
levels decrease slightly to a mean serum
concentration of 196 pg/mL at 1 year of
use15,18,20 and 156 pg/mL by 3 years.14,15 After
removal, serum levels are undetectable (less
than 20 pg/mL) by 1 week in the majority of
users, with most women demonstrating ovula-
tion within 6 weeks of implant removal.18
Serum ENG levels showed less individual varia-
tion over time compared with LNG levels de-
tected in Norplant System users.18 Potential
explanations include differences in the release
mechanism or that ENG is chiefly bound to
albumin which is not affected by varying phy-
siologic endogenous estradiol concentration,
whereas LNG is mainly bound to sex hormone
binding globulin.17
Estradiol (E2) levels initially decrease to early
follicular-phase range after insertion of the ENG
implant.21 This initial decrease is followed by a
gradual rise in E2 levels. Given the ENG im-
plants ability to suppress ovulation, that is, the
LH surge, a situation of anovulation with nor-
mal endogenous E2 synthesis occurs with use.

FIGURE 2. Implanon with applicator and parts.15
Wenzl et al17 examined the bioavailability
and accumulation of drug in 8 healthy women
aged 18 to 40, weighing between 80% and 130%
of ideal body weight [body mass index (BMI)
ranging from 19.6 to 27.5]. The study demon-
strated that the ENG implant has an absorption
rate of approximately 60 mg/d after 3 months,
which decreases to 30 mg/d at 2 years. The elim-
ination half-life was approximately 25 hours
compared with over 41 hours for Norplant
System. One subject with a BMI of 27.5 demon-
strated an increased elimination half-life of 40
hours. This result is most likely secondary to
ENG’s lipophilic nature.
Mechanism of Action
The ENG implant was specifically designed to
provide contraceptive efficacy by inhibiting ovu-
lation.14 Makarainen et al18 reported data from
32 women randomized to either receive Nor-
plant System or Implanon. All women had con-
firmed ovulation before the start of the study
and weighed between 80% and 120% of ideal
body weight, with a mean body weight of
60 ± 6.7 kg. Subjects received the ENG implant
between day 1 and 5 of menses or Norplant
System between day 1 and day 7 of menses.
Women were scheduled to complete 36 months
of observation; 7 women in the ENG implant
group and 3 women in the Norplant System
group completed the full time course. Research-
ers evaluated evidence for ovulation using
progesterone concentrations Z16 nmol/L with
ultrasound confirmation of ovulation if possi-
ble. Progesterone concentrations were elevated
above the 16 nmol/L threshold for the first time
after 30 months in 2 ENG implant users and
Contraceptive Implant 909
again after 33 months. In 1 subject, confirmation
of ovulation was demonstrated with ultrasound
at both time periods, but for the other subject
ovulation was only demonstrated with ultra-
sound at month 30. The first progesterone con-
centration above 16 was seen at 12 months in 1
Norplant user, although ovulation was not con-
firmed via ultrasound. The first confirmed ovu-
lation was seen in a Norplant System user at 18
months. Further, in both groups estradiol levels
decreased to early follicular range after implant
insertion. During treatment, no subjects had
continuously low or high estradiol levels.
Davies et al22 demonstrated a similar effect
on ovulation suppression in an assessment of 15
women during a 1-year period. Women with a
mean age of 32 years had leached ENG implants
inserted on day 1 to day 5 of their menstrual
cycle. The implants were partially leached of
hormone to assure a release rate of 40 mg/d. No
luteal activity was demonstrated with either
ultrasound evaluation or serum hormone levels.
Additionally, 6 weeks after implant insertion,
cervical mucus assessment demonstrated a sig-
nificant thickening based on a mean Insler value
that had fallen significantly from 13, preinser-
tion, to 3.5 (P = 0.0001). Although these studies
used small numbers of ENG implant users, the
results combined with the knowledge that ENG
levels remain above the ovulation-inhibiting
levels for the full 3 years of use, suggests that
ovulation inhibition is the ENG implant’s pri-
mary mechanism of action.
Efficacy
Several large trials have demonstrated the
efficacy of the ENG implant. A review of ENG
910 Hohmann and Creinin
implant studies published in 1998 found no
pregnancies among 1716 women who used the
implant between 2 and 5 years [Pearl index 0.0
(95% confidence interval (CI) 0.00, 0.09)].21
Subsequent studies have confirmed this high
efficacy level for the implant. An American
open-label single treatment study of 330 women
followed ENG implant users for 2 years. All
women were sexually active, between the ages of
18 and 40, and weighed between 80% and 130%
of their ideal body weight.23 About one fourth of
the originally enrolled women had a BMI greater
than 26. Four hundred and seventy-four
women-years of ENG implant use were ob-
served, with 68% of subjects continuing the
implant for 1 year and 51% completing 2 years.
No pregnancies occurred while the ENG im-
plant was in place. Importantly, of 46 subjects
who chose not to use any contraception after
having the implant removed, almost 24% be-
came pregnant between 1 and 18.5 weeks after
removal; thus, supporting the pharmacologic
data that demonstrate the lack of ENG accumu-
lation in the body.17
A multicenter efficacy trial performed in
Europe and South America24 examined ENG
implant use in 635 women with a mean age of 29
years. The 436 women (68.6%) who completed 2
years were asked if they were willing to extend
ENG implant use for an additional year. Ap-
proximately one third (n = 147) agreed and
completed the third year of use. Although this
study population included women who were
between 80% and 130% of ideal body weight,
only 9.0% of women followed for 2 years
were over 75 kg at the start of the study and
only 3.4% of women followed for 3 years
weighed more than 75 kg at the start of the
study. Overall, total exposure to the ENG
implant was for 1200 women-years, which is
equivalent to 15,653 28-day cycles of use. Similar
to other ENG implant studies, no pregnancies
were recorded in this trial yielding a Pearl
index of 0.0 (95% CI, 0.0-0.2). Thus, this study
provides clinical data to support excellent con-
traceptive efficacy of the product through 3
years of use.
A multicenter Mexican study followed a total
of 417 women during 3 years of ENG implant
use with 256 women (61.4%) completing the full
3-year use.25 Overall, this study enrolled women
with a mean weight of 59.4 ± 9.3 kg and a mean
BMI of 24.9 ± 3.9, with 19.4% of subjects cate-
gorized as overweight (BMI>25) and 8.9%
categorized as obese (BMI>30). Once again,
no pregnancies were recorded in this study,
which corresponded to 958.5 women-years of
observation.
Of course data from controlled clinical trials
can differ from what is seen after a new contra-
ceptive is brought onto the market. One large
body of postmarketing data comes from Aus-
tralia during a 3-year time period after the ENG
implant was first introduced to that country. A
total of 218 confirmed pregnancies during ENG
implant use were reported.26 Of the cases re-
ported, 21% of patients were found to have been
pregnant before the time of implant insertion
and 39% of pregnancies were due to ‘‘noninser-
tion.’’ Although the study reports that some
physicians recognized their noninsertions, the
series did not include the number of providers
who recognized or failed to recognize these
events. For the total number of confirmed preg-
nancies, 21% of cases had insufficient data to
detect the reason for failure and the remaining
19% of pregnancies were due to method failure.
Thus, this data set gives a failure rate of 1.07 per
1000 insertions. It is important to understand the
reasons for method failure. Of the 43 women
who experienced a contraceptive failure, 8 were
determined to be secondary to interactions with
other medications, carbamazepine being the
most notable. Interestingly, one of the failures
was in a woman who had a reported weight gain
of over 10 kg between the time of insertion and
pregnancy, thus highlighting the fact that there is
limited data about ENG implant efficacy in
overweight and obese women. Furthermore, this
information emphasizes the importance of hav-
ing proper training in inserting the implant so
the clinician can better recognize instances of
noninsertion. Lastly, although no specific stu-
dies have been carried out to examine the ENG
implant’s interactions with hepatic enzyme in-
ducing medication, the package insert instructs
women to use an additional contraceptive
method during and for at least 7 days after
stopping such drugs. Until further research is
performed in this particular area, the ENG
implant should not be considered first-line con-
traception for women chronically on these types
of medications.
The results of the above studies demonstrate
the ENG implant’s excellent efficacy. Even when
accounting for the failures noted in the Austra-
lian postmarketing surveillance, the implant
continues to have one of the highest efficacies
of any method available. Of course, the trials
discussed above were all open-label studies and
did not directly compare the ENG implants
efficacy with any other methods. A Chinese

study randomized 200 women to directly com-
pare the ENG implant with the 6 capsule LNG
implant over a period of 4 years.27 Of 153 women
who completed the full trial, no pregnancies were
reported in either group. These data further
support the ENG implant’s effectiveness and
suggests that further research may demonstrate
that the device has acceptable efficacy for greater
than 3 years of use.
Safety and Side Effect Profile
BLEEDING PATTERNS
Similar to other continuous progestin-only con-
traceptives, irregular and unpredictable bleeding
patterns are reported with ENG implant use.
A review of noncomparative trials and open
randomized trials of the ENG implant demon-
strates that bleeding patterns were similar
through out the various trials. The following
bleeding patterns were reported for 1716 ENG
implant users. Amenorrhea increased from low
levels early in the studies to between 30% and
40% at 12 months. Infrequent bleeding occurred
in about 50% of subjects at 3 months and
decreased to 30% after 6 months. Prolonged
bleeding, although high during the first 3 months
of use, decreased to 10% to 20% after 3 months
of use.13 Unfortunately, although bleeding pat-
terns were similar between the studies no con-
sistent bleeding pattern can be demonstrated for
any individual woman.
A retrospective Swiss study was performed at
12 centers to assess acceptability and side effect
profile of the ENG implant.28 A total of 1183
women had the ENG implant inserted, of which
991 (84%) completed 1 follow-up visit. Mean
time from insertion to first follow-up was 224
days. Normal bleeding patterns were reported
by only 11% of women. Infrequent bleeding
was seen in 28% of women, where as pro-
longed bleeding was reported in 15% of
women and metromenorrhagia was reported in
16% of women. Of women with 1 follow-up
visit, 23.7% had the implant removed prema-
turely. The most frequently reported adverse
event leading to removal was prolonged and
frequent bleeding, comprising 45% of removals
for side effects.
In a prospective US study of 330 women, 43
discontinued the ENG implant secondary to
bleeding irregularities as their primary adverse
event.23 Episodes of prolonged and frequent
bleeding was highest during the first 3 months
of use (36% and 14%, respectively) and decreased
Contraceptive Implant 911
during the remainder of the study (14% and 7%,
respectively). The number of subjects discontinu-
ing implant use was highest during the first 8
months of use. This number is similar to results
found for the above Swiss study in which the
mean time from insertion to removal for those
who chose to discontinue the product was 9.2
months.28
Given the potential of prolonged and fre-
quent bleeding to lead to discontinuation of
the ENG implant, researchers have examined
different regimens to improve bleeding pro-
files.29–31 Earlier studies of Norplant System
users demonstrated that providing combined
oral contraceptive pills (COC) with 50 mg of
ethinyl estradiol (EE)/250 mg LNG for 20 days
shortened bleeding episodes to 2.6 days com-
pared to 12.3 days with placebo. Oral doses of
EE 50 mg also shortened bleeding episodes, but
not to the same extent as seen with the COC.31
Despite reduction in bleeding duration, these
high doses of estrogen are associated with side
effects such as nausea.30 Other data on Norplant
System users demonstrated decreased duration
of bleeding episodes without change in number
of episodes with the administration of EE 30 mg/
150 mg LNG for 21 days.29 Currently, Weisberg
et al30 has the only study published that exam-
ined treatments for prolonged and frequent
bleeding specifically in ENG implant users. This
study randomized 179 women into 1 of 4 treat-
ments. Subjects were women who had used the
implant for greater than 3 months and experi-
enced prolonged or frequent bleeding. An EE
arm alone was excluded in this study because of
the increased number of subjects needed for
adequate power. Treatments were mifepristone
25 mg taken twice on day 1 followed by 4 days of
placebo, mifepristone 25 mg twice on day 1
followed by 4 days EE 20 mg in the AM and
placebo at night, doxycycline 100 mg twice a day
for 5 days, or placebo twice a day for 5 days.
Mifepristone was used because of prior data
showing effect in Norplant System users32 and
doxycycline was used secondary to its known
anti-inflammatory properties. Mifepristone
combined with EE and doxycycline both signifi-
cantly reduced bleeding episodes [mean, 4.3 d
(95% CI 3.5-5.3) and 4.8 d (95% CI 3.9-5.8),
respectively].30 Despite the results of this study,
the limited availability of mifepristone in the
United States decreases the utility of this study’s
results. Doxycycline may be considered as a
method to decrease bleeding episodes in ENG
implant users, keeping in mind the risk of side
effects. For women without contraindications to
912 Hohmann and Creinin
estrogen, data from Norplant System users sug-
gest COC use as a method to decrease duration
of bleeding episodes. Given the possibility of side
effects with COCs containing EE 50 mg, a COC
containing a lower EE doses is recommended.
ACNE
Clinical results regarding acne are mixed. In a 3-
year study of 635 women, acne was the second
most common (12.6%) nonbleeding adverse
event associated with ENG implant use.12 This
number is consistent with the results of a Swiss
retrospective study of ENG implant users in
which 12% of side-effect removals were second-
ary to reports of acne.28 Croxatto12 also
reported an opposing trend in women who re-
ported having acne at baseline, with 78 out of
133 reporting improvements in this skin condi-
tion during ENG implant use.
Funk et al23 had 315 subjects provide baseline
and posttreatment acne information. About
26% of women had acne at baseline and 24%
had acne after treatment. From the total popu-
lation, 16% reported a decrease from baseline,
70% reported no change, and 14% reported
increased acne. Of subjects with baseline acne,
61% reported decreased acne posttreatment and
only 7% reported an increase in acne posttreat-
ment. For those women without acne at base-
line, 84% reported no change and 16% reported
increase in this skin condition.
The opposing nature of these data, and the
lack of control group, makes it difficult to
provide patients with a clear expectation regard-
ing the incidence or severity of acne while using
the ENG implant. Patients should be counseled
that there is no apparent trend with regard to
acne incidence or improvement while using the
ENG implant.
BONE MINERAL DENSITY
The role of hormonal contraception’s influence
on bone mineral density (BMD) has become an
area of controversy since the FDA required an
inclusion of a black box warning on the package
insert for depot medroxyprogesterone acetate.
Beerthuizen et al33 reported on a comparative
study of BMD in users of the ENG implant verse
users of nonhormonal IUDs. Forty-four ENG
users and 29 IUD users, aged 18 to 40 years, were
followed for 2 years. BMD was measured
through the use of dual energy x-ray absorptio-
metry at the lumbar spine, proximal femur, and
distal radius. Estradiol levels were comparable
between the groups at baseline and showed
no correlation to baseline BMD. No clinically
significant difference was seen in BMD between
the ENG implant users and the IUD users. No
relationship was noted between estradiol levels
and changes in BMD in this study.
Bahamondes et al34 performed a trial of ENG
implant and Norplant System users comparing
BMD at the ulna and distal radius. At 18
months, both groups demonstrated a decrease
in BMD at the midshaft of the ulna, but no
difference at the distal radius. It should be noted
that although the BMD was significantly de-
creased, the decrease never went outside of the
limit of 1 standard deviation. Also, the study
only looked at the BMD of the forearm, which is
not the best site to use to predict fracture risk.
Further, there are no long-term data, which
show that this result has any clinical significance.
DYSMENORRHEA
Data regarding incidence of dysmenorrhea in
ENG implant users suggests that use of the
implant may improve this condition. Funk et
al23 showed that the percentage of subjects with
dysmenorrhea decreased from a baseline of 59%
to a posttreatment level of 21%. Of the total
population, 48% reported decreased dysmenor-
rhea with ENG implant use, whereas 8%
showed an increase in the condition. Of those
women who reported a history of dysmenorrhea
at baseline, 81% showed improvement with im-
plant use. In a similar study, Croxatto12 reported
a 35% incidence of dysmenorrhea among sub-
jects at baseline with 82% of these women
reporting improvement in symptoms at the end
of the study. When complying data from multi-
ple ENG implant studies, differences in bleeding
pattern during ENG implant use was not corre-
lated to reported incidence or severity of dysme-
norrhea.13
WEIGHT CHANGES
Weight changes attributed to the use of the ENG
implant have been described in a number of
clinical trials, although the percentage of women
who ultimately have the implant removed for
this reason is low. On the basis of results of the
large American trial of the implant over a 2-year
period, weight increase was reported in about
12% of subjects,23 but only 3.3% of women
withdrew because of this weight increase. The
mean increase in BMI from baseline to last
measurement was 0.7 kg/m2 in this trial. Crox-
atto et al24 reported that approximately 20% of
women reported a greater than 10% increase in
BMI over baseline at one or more measures. The
mean increase in BMI over the study’s 3-year

time period was 3.5%, but the mean change in
BMI was only 0.8 kg/m2, similar to the results
from the American trial. Zheng et al27 reported a
change in body weight in 100 Chinese women
using the ENG implant with a breakdown of
0.82, 1.15, 2.5, and 3.1 kg for years 1, 2, 3, and 4
of use. Importantly, no women withdrew from
the study secondary to weight gain. In a retro-
spective Swiss study of implant users, 9% of 991
women at first visit (mean time of 272 d since
implant insertion; range, 1 to 677 d) and 9% of
306 women at second visit (mean time of 347 d
since implant insertion; range, 15 to 709 d)
reported weight gain during implant use, but
only 7% of women requesting implant removal
reported the primary reason as weight gain.28
Lastly, in a retrospective British study following
ENG implant use in 324 women, of the 277 for
whom information was available, 14 (5%)
women who discontinued implant use with-in 1
year cited weight change as their primary rea-
son.35 Thus, the overall removal rates because of
weight change seem to be in the 3% to 7% range
in non-Asian populations.
OVARIAN CYSTS
Progestin-only contraceptive methods have been
associated with ovarian cyst formation.36 Hidal-
go et al37 performed a prospective study com-
paring ovarian cyst formation in 344 women
using the ENG implant, Jadelle, or the copper
IUD. One-year follow-up was available for 90%
of ENG implant users, 84% of Jadelle users, and
75% of IUD users. Throughout the study, cop-
per IUD users had an approximate 2% or less
rate of ovarian cyst formation compared with
5% and 13% of ENG implant and Jadelle users,
respectively, at 3 months and 27% and 15% of
ENG implant and Jadelle users at 12 months,
respectively. All ENG implant users and all but 1
Jadelle user were anovulatory based on proges-
terone levels. Notably the authors reported that
the time to ovarian cyst resolution in ENG
implant users ranged from 7 to 72 days. In
conclusion, although ovarian cysts may be asso-
ciated with ENG implant use, the majority of
cysts will regress spontaneously and do not need
additional treatment.
BREAST FEEDING
ENG implant use seems to be safe for breast-
feeding women. Reinprayoon et al10 performed
an open-label nonrandomized group compari-
son study of breast-feeding women using either
the ENG implant or a copper IUD for contra-
ception. A total of 80 women and infants parti-
Contraceptive Implant 913
cipated in this study and were initially followed
for 4 months. The study found that there was no
significant difference in total fat, protein, or
lactose content in the breast milk of the 2 groups.
Further, 24-hour milk production was not dif-
ferent between the 2 groups. Although infants’
ENG dose was highest during the first month of
implant use (19.86 ng/kg/d, equivalent to 1.7%
of the maternal dose), the dose significantly
decreased by months 2 and 4. Infant growth rate
did not differ significantly during the first 4
months of use between the ENG implant group
and the copper IUD group. As a secondary
outcome of the study, the researcher continued
to follow the breast-fed infants over a 3-year
period to evaluate any differences in long-term
outcomes.11 A total of 81% of the ENG implant
exposed infants and 86.8% of the copper IUD
exposed infants completed the study. Over the 3-
year period, there was no difference between the
2 group’s growth rates or biparietal head cir-
cumferences. Even though the original study was
not powered for this outcome, the results further
support the initial conclusion of ENG implant
safety for breast-feeding women. Clinicians
should feel comfortable recommending this
form of contraception to breast-feeding patients.
EFFECT ON LIPID PROFILE
Biswas et al38 performed a randomized trial with
80 subjects receiving either the Norplant System
or the ENG implant and followed subjects for 2
years. Overall, only 3 ENG implant users dis-
continued before the study’s completion. For the
ENG implant users, there was a significant
decrease in serum total cholesterol (TC), high-
density lipoprotein (HDL) cholesterol, and low-
density lipoprotein (LDL) cholesterol, which
was similar for women receiving the Norplant
System. A similar decrease in TC in 21% of
subjects was noted in the American trial of the
ENG implant,23 but unlike the results from
Biswas et al,38 this trial also reported a decrease
in triglycerides in 33% of women using the ENG
implant. No significant change in the HDL/TC
ratio was reported by Biswas et al,38 but a
significant decrease in the HDL/LDL ratio for
ENG implant users was seen at 1 year with a
return to preinsertion level at 2 years. Despite
these results, the change in HDL was only 5.8%
lower at 2 years compared with baseline and the
HDL/LDL ratio was never within a range asso-
ciated with increase risk of cardiovascular dis-
ease. Accordingly, the authors concluded
that ENG implant use should not significantly
increase the risk of cardiovascular disease.
914 Hohmann and Creinin
Manufactures of the implant recommend fol-
lowing cholesterol values in women with know
elevated lipid profiles throughout the use of the
implant.15
EFFECT ON LIVER FUNCTION
Biswas et al39 reported the effects of the ENG
implant on liver function tests using the same
study design as described above. The authors
reported a significant increase in mean total and
unconjugated bilirubin in both Norplant System
and ENG implant users, although levels never
exceeded the normal range. The researchers did
notice an initial significant increase in aspartate
transaminase levels in ENG implant users at 6
months, but after a year of use levels decline
toward baseline. An increase in total bilirubin
was seen in a similar prospective randomized
trial of 86 subjects randomized to either Nor-
plant System or the ENG implant for 6 months
of use.40 Contrary to the above trials, Funk et
al23 reported no significant change in liver func-
tion parameters over a 2-year use of the ENG
implant. Thus, even though results are mixed,
clinicians should be aware that there may be mild
changes in liver function tests during ENG im-
plant use. Although these changes may not be
clinically significant in health women, these
changes may have serious consequences in
women with preexisting liver disease.
Counseling
Given the number of contraceptive options
available to women, it is essential that providers
concentrate their efforts not only on helping
women chose the best contraceptive method,
but also focus on counseling that helps improve
continuation. The best birth control method is
one that is safest and most effective for that
TABLE 1. Comparison of Long-acting, Reversible Contraceptives42
Perfect Use Typical Use
Failure Rate Failure Rate
(%) (%)
DMPA 0.3 3 Approximate
(DepoProvera)
Copper-T IUD 0.6 0.8
(ParaGard T 380A)
LNG-IUD (Mirena) 0.1 0.1
Single-Rod ENG 0.1 0.1
Implant
(Implanon)
DMPA indicates medroxyprogesterone acetate.
individual woman.41 This approach places a
strong value on medical considerations, but also
includes consideration of a woman’s lifestyle,
preferences, and level of prevention desired.
When discussing long-term, reversible contra-
ception, physicians are obligated to present all
suitable options to their patients (Table 1).
Before providing the ENG implant, provi-
ders should review the indications and contra-
indications for its use. Contraindications to
ENG implant use listed on the package insert
include known or suspected pregnancy, active
venous thromboembolic disease, active liver dis-
ease, undiagnosed genital tract bleeding, known
or suspected breast cancer, progesterone depen-
dent tumors, or allergy to any of the implants
components.15 Although the package insert lists
venous thromboembolic disease as a contraindi-
cation to ENG implant use, there is no evidence
to support this restriction. Also, as mentioned
above, women chronically using hepatic enzyme
inducing medications are not proper candidates
for this type of contraception.
When explaining the ENG implant, the phy-
sician needs to address any concern and fears a
woman may have about this method of contra-
ception. In particular, women may have con-
cerns about implant removal based on media
coverage regarding the Norplant System. Also,
side effects such as irregular bleeding should be
discussed in advance with your patients, as an
unexpected side effect may cause women to
request early removal of the implant. The bot-
tom line is that bleeding patterns are irregularly
irregular. Clinicians should consider discussing
plans for treatment of unsatisfactory bleeding
patterns in advance of insertion in the hope
that this may minimize discontinuation. Lastly,
all women need to be reminded about safe
sexual practices, as the implant does not provide
Return to Hormonal Long-term
Fertility Method Effectiveness
Yes, progestin At least 3 mo
6-mo delay only
Immediate No Up to 10 y
Immediate Yes, progestin Up to 5 y
only
1-mo delay Yes, progestin Up to 3 y
only

protection against sexually transmitted dis-
eases.15 Patients who are good candidates for
this form of contraception are those that desire
long-term reversible birth control, have no con-
traindications to ENG implant use, accept im-
plant insertion and removal, and are ready to
accept a change in menstrual bleeding patterns.
Insertion and Removal
Proper insertion and removal techniques are
essential for clinical efficacy and for the preven-
tion of complications. Timing of insertion is
dependent on the patient’s prior use of contra-
ception and the clinician’s evaluation of the
appropriateness for each individual. For women
without preceding hormone use, the ENG im-
plant should be inserted within 5 days from the
start of menses. When switching from a COC,
insertion should occur within 7 days of the last
active pill. Patients switching from another pro-
gestin-only method can have the implant placed
at anytime while on the progestin-only pill, at the
time of IUD or implant removal, or on the due
date of the next contraception injection. Im-
plants may be inserted within 5 days of a first
trimester abortion, within 6 weeks of a second
trimester abortion, or within 6 weeks of child-
birth.15 Additional, a clinician may prescribe
and insert the ENG implant at any time during
a woman’s menstrual cycle with recommenda-
tions adopted from the ‘‘quick start’’ guidelines
for oral contraceptives.43 Before insertion the
patient needs a negative, high-sensitive urine
pregnancy test. Additionally, emergency contra-
ception should be provided if she has had un-
protected intercourse within the last 120 hours.
Patients should also be instructed to be abstinent
or use a backup method of contraception or
abstain for 1 week after insertion and perform
a urine pregnancy test 3 to 4 weeks after ENG
implant insertion.
The ENG implant is inserted with a single
use, sterile applicator. Each implant is preloaded
in the needle of the applicator, minimizing hand-
ling of the implant before insertion. The implant
is typically inserted in the nondominate arm 6 to
8 cm above the elbow. It is essential that the
clinician have proper training to decrease com-
plications. It is critical that the physician verify
proper placement of the implant after insertion
through palpation of the patient’s arm.44 Results
from a large American trial of 330 women
demonstrated the mean time to ENG implant
insertion was 0.5 minutes (range, 0.05 to
15 min).23 Most other studies report insertion
Contraceptive Implant 915
times averaging 2 minutes or less.44 However,
none of these studies clearly stated who was
doing the insertions or what level of training
the inserter had. Also, the authors of the above
studies did not account for a learning curve as
the inserters gained more experience.
Low numbers of ENG implant site complica-
tions are reported in the literature. In 1 study of
330 American women, only 2.5% reported inter-
mittent pain at the insertion site over a 2-year
period.23 In a large multicenter trial, Croxatto et
al24 reported a 1.3% complication rate with
insertion with the author citing examples such
as a visible implant tip and blood loss from
injection site.
Follow-up after ENG implant insertion can
be based on a physician’s individual practice. A
study of early Norplant System users found
routine follow-up to be of no clinical benefit.45
The package labeling indicates that the ENG
implant needs to be removed at the end of 3 years
of use.15 However, there are no known risks for
leaving the implant in longer unless the patient
desires pregnancy. Unless future data demon-
strate otherwise, the patient can only rely on the
ENG implant for contraception for 3 years.
Before removal, the clinician needs to palpate
the implant. Under sterile conditions, a 2 to 3-
mm incision is made vertically over the implant.
The rod is than removed using the ‘‘pop-out’’
technique previously described for Norplant
System removal.46 If inserted correctly, removal
has been shown to be simple. The American trial
published by Funk et al23 had an average re-
moval time of 3.5 minutes (range, 0.2 to 60 min)
and reported difficulties in 2 of the 330 removals,
including a implant that broke during removal
necessitating a second attempt for complete
removal. In a large multicenter study, removal
difficulties were reported in 3% of case.24 The
most common reason for difficulty was second-
ary to implants being placed to deep. If the
implant is unable to be palpated by the clinician
before removal, imaging techniques may be
necessary before proceeding. Case reports have
used high frequency (10 MHz) ultrasound to
detect the acoustic shadow associated with the
implant47 and magnetic resonance imaging as a
second line modality if needed.48
Conclusions
The ENG implant provides women with an addi-
tional highly effective non–user-dependent re-
versible contraceptive option. With greater con-
traceptive options available, we as providers may
916 Hohmann and Creinin
be better able to match women’s contraceptive
needs and desires with the appropriate method.
The primary advantage of the ENG implant
over other types of contraception is the lack of
contraceptive failure in women who have the
implant inserted. The trade-off for women
is irregularly irregular bleeding that occurs
throughout the lifespan of use. Because of the
high efficacy, we need to find ways to minimize
bleeding issues to improve continuation rates.
References
1. Meckstroth KR, Darney PD. Implant contra-
ception. Semin Reprod Med. 2001;19:339–354.
2. Meirik O, Farley TM, Sivin I. Safety and
efficacy of levonorgestrel implant, intrauterine
device, and sterilization. Obstet Gynecol.
2001;97:539–547.
3. Sivin I, Mishell DR Jr, Diaz S, et al. Prolonged
effectiveness of Norplant(R) capsule implants:
a 7-year study. Contraception. 2000;61:
187–194.
4. Sivin I. Risks and benefits, advantages and
disadvantages of levonorgestrel-releasing
contraceptive implants. Drug Saf. 2003;26:
303–335.
5. Haukkamaa M, Laurikka-Routti M, Heikin-
heimo O, et al. Contraception with subdermal
implants releasing the progestin ST-1435: a
dose-finding study. Contraception. 1992;45:
49–55.
6. Sivin I, Croxatto H, Bahamondes L, et al.
Two-year performance of a Nestorone(R)-
releasing contraceptive implant: a three-center
study of 300 women. Contraception. 2004;69:
137–144.
7. Coutinho EM, de Souza JC, Athayde C, et al.
Multicenter clinical trial on the efficacy and
acceptability of a single contraceptive implant
of nomegestrol acetate, Uniplant. Contracep-
tion. 1996;53:121–125.
8. Barbosa IC, Maia H Jr, Coutinho E, et al.
Effects of a single Silastic contraceptive
implant containing nomegestrol acetate
(Uniplant) on endometrial morphology and
ovarian function for 1 year. Contraception.
2006;74:492–497.
9. Boonstra H, Duran V, Northington Gamble V,
et al. The ‘‘boom and bust phenomenon’’: the
hopes, dreams, and broken promises of the
contraceptive revolution. Contraception. 2000;
61:9–25.
10. Reinprayoon D, Taneepanichskul S, Bunya-
vejchevin S, et al. Effects of the etonogestrel-
releasing contraceptive implant (Implanon on
parameters of breastfeeding compared to those
of an intrauterine device). Contraception. 2000;
62:239–246.
11. Taneepanichskul S, Reinprayoon D, Thaithu-
myanon P, et al. Effects of the etonogestrel-
releasing implant Implanon and a nonmedi-
cated intrauterine device on the growth of
breast-fed infants. Contraception. 2006;73:
368–371.
12. Croxatto HB. Clinical profile of Implanon: a
single-rod etonogestrel contraceptive implant.
Eur J Contracept Reprod Health Care. 2000;
5(suppl 2):21–28.
13. Edwards JE, Moore A. Implanon. A review of
clinical studies. Br J Fam Plann. 1999;24
(4 suppl):3–16.
14. Croxatto HB. Mechanisms that explain the
contraceptive action of progestin implants for
women. Contraception. 2002;65:21–27.
15. Implanon (package insert). 2006, Organon
USA Inc: Roseland, NJ.
16. [cited; 8/15/07]. Available from: http://www.
implanon-usa.com/media/imagebank/index.asp?
C = 54991393098966782407&svarqvp2 = 0.
17. Wenzl R, van Beek A, Schnabel P, et al.
Pharmacokinetics of etonogestrel released
from the contraceptive implant Implanon.
Contraception. 1998;58:283–288.
18. Makarainen L, van Beek A, Tuomivaara L,
et al. Ovarian function during the use of a
single contraceptive implant: Implanon com-
pared with Norplant. Fertil Steril. 1998;69:
714–721.
19. Diaz S, Pavez M, Moo-Young AJ, et al. Clin-
ical trial with 3-keto-desogestrel subdermal
implants. Contraception. 1991;44:393–408.
20. IMPLANON approval letter, U.F.a.D. Ad-
ministration, Editor. 2006, US Food and Drug
Administration.
21. Croxatto HB, Makarainen L. The pharma-
codynamics and efficacy of Implanon. An
overview of the data. Contraception. 1998;
58(6 suppl):91S–97S.
22. Davies GC, Feng LX, Newton JR, et al.
Release characteristics, ovarian activity and
menstrual bleeding pattern with a single con-
traceptive implant releasing 3-ketodesogestrel.
Contraception. 1993;47:251–261.
23. Funk S, Miller MM, Mishell DR Jr, et al.
Safety and efficacy of Implanon, a single-rod
implantable contraceptive containing etono-
gestrel. Contraception. 2005;71:319–326.
24. Croxatto HB, Urbancsek J, Massai R, et al.
A multicentre efficacy and safety study of
the single contraceptive implant Implanon.
Implanon Study Group. Hum Reprod. 1999;
14:976–981.
25. Flores JB, Balderas ML, Bonilla MC, et al.
Clinical experience and acceptability of the

etonogestrel subdermal contraceptive implant.
Int J Gynaecol Obstet. 2005;90:228–233.
26. Harrison-Woolrych M, Hill R. Unintended
pregnancies with the etonogestrel implant
(Implanon): a case series from postmarketing
experience in Australia. Contraception. 2005;
71:306–308.
27. Zheng SR, Zheng HM, Qian SZ, et al. A
randomized multicenter study comparing the
efficacy and bleeding pattern of a single-rod
(Implanon) and a six-capsule (Norplant) hor-
monal contraceptive implant. Contraception.
1999;60:1–8.
28. Bitzer J, Tschudin S, Alder J. Acceptability and
side-effects of Implanon in Switzerland: a retro-
spective study by the Implanon Swiss Study
Group. Eur J Contracept Reprod Health Care.
2004;9:278–284.
29. Witjaksono J, Lau TM, Affandi B, et al.
Oestrogen treatment for increased bleeding
in Norplant users: preliminary results. Hum
Reprod. 1996;11(suppl 2):109–114.
30. Weisberg E, Hickey M, Palmer D, et al. A pilot
study to assess the effect of three short-term
treatments on frequent and/or prolonged
bleeding compared to placebo in women using
Implanon. Hum Reprod. 2006;21:295–302.
31. Alvarez-Sanchez F, Brache V, Thevenin F,
et al. Hormonal treatment for bleeding irregu-
larities in Norplant implant users. Am J Obstet
Gynecol. 1996;174:919–922.
32. Cheng L, Zhu H, Wang A, et al. Once a month
administration of mifepristone improves
bleeding patterns in women using subdermal
contraceptive implants releasing levonorges-
trel. Hum Reprod. 2000;15:1969–1972.
33. Beerthuizen R, van Beek A, Massai R, et al.
Bone mineral density during long-term use of
the progestogen contraceptive implant Impla-
non compared to a non-hormonal method of
contraception. Hum Reprod. 2000;15:118–122.
34. Bahamondes L, Monteiro-Dantas C, Espejo-
Arce X, et al. A prospective study of the
forearm bone density of users of etonogestrel-
and levonorgestrel-releasing contraceptive
implants. Hum Reprod. 2006;21:466–470.
35. Lakha F, Glasier AF. Continuation rates of
Implanon in the UK: data from an observa-
tional study in a clinical setting. Contraception.
2006;74:287–289.
Contraceptive Implant 917
36. Croxatto HB. Progestin implants. Steroids.
2000;65:681–685.
37. Hidalgo MM, Lisondo C, Juliato CT, et al.
Ovarian cysts in users of Implanon and Jadelle
subdermal contraceptive implants. Contracep-
tion. 2006;73:532–536.
38. Biswas A, Viegas OA, Roy AC. Effect of
Implanon and Norplant subdermal contracep-
tive implants on serum lipids—a randomized
comparative study. Contraception. 2003;68:
189–193.
39. Biswas A, Biswas S, Viegas OA. Effect
of etonogestrel subdermal contraceptive
implant (Implanon) on liver function tests—
a randomized comparative study with
Norplant implants. Contraception. 2004;70:
379–382.
40. Egberg N, van Beek A, Gunnervik C, et al.
Effects on the hemostatic system and liver
function in relation to Implanon and Nor-
plant. A prospective randomized clinical trial.
Contraception. 1998;58:93–98.
41. Farrington A. Today’s contraceptive options:
finding the right fit. Health Sexuality. 2001;6:
4–6.
42. Hatcher R, Trussell J, Stewart F, et al. Contra-
ceptive Technology. 18th Revised ed. New
York: Ardent Media, Inc.; 2004.
43. Westhoff C, Kerns J, Morroni C, et al. Quick
start: novel oral contraceptive initiation
method. Contraception. 2002;66:141–145.
44. Shulman LP, Gabriel H. Management and
localization strategies for the nonpalpable
Implanon rod. Contraception. 2006;73:
325–330.
45. Cullins VE, Blumenthal PD, Remsburg RE,
et al. Preliminary experience with Norplant
in an inner city population. Contraception.
1993;47:193–203.
46. Pymar HC, Creinin MD, Schwartz JL. ‘‘Pop-
out’’ method of levonorgestrel implant
removal. Contraception. 1999;59:383–387.
47. Lantz A, Nosher JL, Pasquale S, et al.
Ultrasound characteristics of subdermally
implanted Implanon contraceptive rods.
Contraception. 1997;56:323–327.
48. Westerway SC, Picker R, Christie J. Implanon
implant detection with ultrasound and mag-
netic resonance imaging. Aust N Z J Obstet
Gynaecol. 2003;43:346–350.

The Transdermal Contraceptive
Patch: An Updated Review of the
Literature
KATHARINE O’CONNELL, MD, MPH* and
RONALD T. BURKMAN, MDw
*Columbia University College of Physicians and Surgeons,
New York, New York; and w Division of General Obstetrics and Gynecology,
Department of Obstetrics and Gynecology, Tufts University School of
Medicine, Baystate Medical Center, Springfield, Massachusetts
Abstract: The transdermal contraceptive patch which
contains ethinyl estradiol and norelgestromin has an
efficacy similar to current oral contraceptives (OCs).
The major advantages include transdermal applica-
tion and maintenance of adequate hormonal levels for
at least 7 days. Side effects are similar to OC except for
breast tenderness in the first 2 months of use and skin
irritation at the application site. Although concern has
been raised about a possible increased risk of venous
thromboembolism, current available data comparing
the patch to a norgestimate-containing OC ranges
from no increase in risk to a 2.4-fold increase.
Key words: transdermal, estradiol, norelgestromin,
venous thromboembolism
Despite the explosive growth of new birth con-
trol methods in the United States, the unplanned
pregnancy rate continues to be high—49% of
the 6 million pregnancies each year in the United
States are unplanned.1 Many women with an
unplanned pregnancy were using a method of
contraception during the month they got preg-
nant.2 This chapter is a review of some of the
literature of the contraceptive patch (Ortho-
Evra, Ortho-McNeil Pharmaceutical Inc), one
of the new methods of contraception that has
entered the marketplace in the past decade.
Correspondence: Katharine O’Connell, MD, MPH,
Obstetrics and Gynecology, Columbia University Col-
lege of Physicians and Surgeons, 630 West 168th Street,
PH16-80, New York, NY 10032. E-mail: ko2032@
columbia.edu
CLINICAL OBSTETRICS AND GYNECOLOGY / VOLUME 50
918
CLINICAL OBSTETRICS AND GYNECOLOGY
Volume 50, Number 4, 918–926
r 2007, Lippincott Williams & Wilkins
Description
Transdermal delivery of estradiol was initially
developed for use in estrogen replacement ther-
apy.3 It was found to have a similar adverse effect
profile to oral estradiol, with skin irritation the
most commonly reported side effect. Yet, the
estradiol patch seemed to lead to greater compli-
ance than oral regimens. More recently, reservoir
patches have been replaced by matrix systems
with a significant reduction in skin irritability and
improved adhesion of the patch to the skin.
The contraceptive patch contains 2 hor-
mones, 0.75 mg of ethinyl estradiol (EE) and
6 mg of norelgestromin (NGMN), with daily
release rates of about 20 mg and 150 mg, respec-
tively. This dosing provides serum levels of both
hormones at a range likely to be efficacious for
contraception (EE, 25 to 75 pg/mL; NGMN, 0.6
to 1.2 ng/mL).4 NGMN was previously known
as 17-deacetylnorgestimate, the primary active
metabolite of norgestimate (NGM). The patch
itself is a thin square, 20 cm2 on each side,
composed of 3 layers. The outer protective layer
is water-resistant, overlying a medicated and
adhesive middle layer. There is a clear release
liner that is removed before patch application.
Each contraceptive patch is worn for 1 week,
and then replaced with a new patch. Three
consecutive patches are supplied in each box
and are worn, for a total of 21 days of use. There
is then a patch-free week during which with-
drawal bleeding occurs. Each patch is removed
and replaced on the same day of the week.
The patch is placed on an area of clean,
abrasion-free skin. Four placement sites were
/ NUMBER 4 / DECEMBER 2007
 The Transdermal Contraceptive Patch
studied for absorption of hormones: lower abdo-
men, upper arm, buttock, upper torso (excluding
breast tissue). The absorption of EE/NGMN on
the abdomen site was approximately 20% less
than that observed for the other 3 sites, although
the mean serum concentrations were still within
reference ranges (thus therapeutically equiva-
lent).5
Multiple studies of the contraceptive patch’s
adhesion properties have been performed. The
rates of complete detachment, requiring patch
replacement, range from 0.5% to 2.4%.5–9 The
rates of partial detachment, where the patch
could be reapplied, range from 1.4% to
2.9%.9,10 Most detachments occurred in the first
cycle of use, and the percentage of patches per
cycle that detached tended to decrease with
experience.10 In an exercise study, only one
patch (1.1%) completely detached, and none
partially detached. Mean peel force measure-
ments did not differ across normal activity; use
of a sauna, whirlpool, or treadmill; cool-water
immersion; or combination of aerobic activ-
ities.11 It should be noted that because this is a
matrix patch, reattachment with some other type
of adhesive will adversely affect the patch prop-
erties because the contraceptive hormones are an
integral part of the adhesive material.
The product labeling calls for starting the
patch within the first 7 days of the menstrual
cycle. In a randomized study comparing women
starting the patch on the day of their clinic visit
(Quick Start) compared with starting it on the
day of their first menses, the continuation rates
for both groups through three cycles was 90% or
higher. The continuation rates were slightly
higher with the Quick Start group but not at a
statistically significant level.12
Pharmacokinetics
Delivery of contraceptive hormones through the
skin avoids first-pass metabolism via the gastro-
intestinal tract or the liver that is experienced
with oral administration. In addition, the peaks
and troughs typically seen with oral dosing are
not seen with transdermal use. Both EE and
NGMN reach steady-state conditions over 3
cycles of treatment, and mean serum concentra-
tions for both hormones remained within refer-
ence ranges throughout the 3-month study.4
Both hormones remained within reference
ranges for up to 9 days of use providing users
with a 2-day ‘‘grace period’’ should they fail to
change the patch on time.13
919
Other differences in the biologic effects of the
patch when compared with oral contraceptive
(OC) users include less follicular development
and a lower incidence of presumptive ovulation
(on the basis of the serum progesterone Z3 ng/
mL).8 The patch’s effects on androgenic markers
are comparable to that of OCs.14
Efficacy
Three trials were initially conducted to deter-
mine the efficacy of the contraceptive patch. The
first, by Hedon et al,15 was a controlled com-
parative trial of the patch versus an OC (EE
20 mg, desogestrel 150 mg). The authors rando-
mized 1517 subjects in a 4:3 ratio; 861 subjects
received the patch and the remaining women the
OC. One-third of subjects used the method for
13 cycles and the remainder for 6 cycles. Overall
and method failure Pearl indices for the patch
were 0.88 and 0.66, respectively, and were not
different from the comparator OC (0.56 and
0.28). There were 3 method failure pregnancies
on the patch, with a cumulative probability of
pregnancy of 0.5% at 13 cycles.
The second trial by Audet et al16 was an active
control, multicenter clinical trial comparing the
contraceptive patch with an OC (EE 30/40/
30 mg, levonorgestrel 50/75/125 mg). The authors
randomized 1417 subjects in a 4:3 ratio; 811
subjects received the patch with the remainder
receiving the OC. Again, one-third of subjects
used the method for 13 cycles and the remainder
for 6 cycles. Overall and method-failure Pearl
indices in the patch group were 1.24 and 0.99,
and were 2.18 and 1.25 in the pill group, respec-
tively. These differences were not statistically
significant. There were 4 method-failure preg-
nancies and one user-failure pregnancy during
5240 cycles of patch use. The cumulative prob-
ability of pregnancy was 0.6% at 6 cycles, and
1.3% at 13 cycles.
The final trial by Smallwood et al17 was an
open-label, single-arm, multicenter, clinical trial
that included 1672 subjects in its final analysis.
There were 501 subjects who received 13 cycles
of treatment, and 1171 subjects who received 6
cycles of treatment. Overall and method-failure
probabilities of pregnancy were 0.7% and 0.4%
through 13 cycles, and overall and method-fail-
ure probabilities of pregnancy were 0.4% and
0.4% through 6 cycles. There were 5 method-
failure pregnancies and 1 user-failure pregnancy
during 10,994 cycles of patch use. The overall
Pearl Index was 0.71, and the method-failure
Pearl Index was 0.59.
920 O’Connell and Burkman
The most recent efficacy trial was a rando-
mized, open-label, parallel-group trial at 65
centers in Europe and South Africa.9 The patch
was compared with multiple OC formulations;
1489 women were enrolled in a 4:3 randomiza-
tion pattern. Subjects used the patch (n = 846)
or an OC (n = 643) for 6 or 13 cycles. The overall
and method-failure probabilities of pregnancy
were 0.5% and 0.4%, respectively, through 13
cycles. The overall and method-failure probabil-
ities of pregnancy were 0.5% and 0.4%, respec-
tively, through 6 cycles. The overall Pearl Index
for the patch was 0.88, and the method-failure
Pearl Index was 0.66. For the OC group, the
respective Pearl indices were 0.56 and 0.28,
which were not significantly different from those
experienced by the patch users.
Efficacy and Body Weight
Recent studies have raised the issue of the impact
of body weight on the efficacy of contraception.
In the original efficacy trials of the patch, sub-
jects needed to be within 35% of ideal body
weight for trial entry. In the pooled analysis of
these trials,10 a significant association between
greater baseline body weight and pregnancy was
found (P<0.001). Five of fifteen on-treatment
pregnancies occurred in the subgroup of women
with a baseline body weight Z90 kg (Z198 lb).
In women below 90 kg, no association between
body weight and pregnancy was found. These
results, however, come from secondary analyses
of studies that were designed to examine method
efficacy. The observed associations are based on
small numbers and are unadjusted for potential
confounding factors. Trials are clearly needed
that directly address the question of the impact
of body weight on the patch’s effectiveness.
Cycle Control and Tolerability
Breakthrough bleeding (BTB) and spotting with
the patch has been comparable to or less than
that seen with OCs. In one of the original studies,
subjects in the patch group had less BTB than the
subjects in the comparator OC group (EE 35 mg,
NGM 250 mg).6 The patch efficacy studies had
similar findings. In the 2001 study, there were no
statistically significant differences between the
patch and the OC (EE 30/40/30 mg, levonorges-
trel 50/75/125 mg) groups with respect to BTB
during any cycle.16 BTB and spotting during
cycle 3 was 14% in the patch group and 15%
in the OC group.15 In the single-arm study
(without an OC comparator group), the rate of
BTB among patch users at 13 cycles was 1.7%,
and the rate of all unscheduled bleeding (BTB or
spotting) at 13 cycles was 9.2%.17 In the Eur-
opean and South African study, rates of BTB
during cycle 3 were 14% in the patch group and
15% in OC group; at cycle 13, the BTB rates
were 8.2% and 12.0%, respectively.9
Breast tenderness, conversely, is more com-
mon among patch users in the initial months of
use. In Audet’s 2001 study, the overall occur-
rence was higher for the patch than for the OC,
though the difference was significant only in
cycles 1 and 2 [15.4% vs. 3.5% in cycle 1 (P,
0.001) and 6.6% vs. 1.5% in cycle 2 (P, 0.001)].16
For cycles 3 to 13, the incidence of breast dis-
comfort was not significantly different between
treatments. Eighty-five percent of subjects rated
the discomfort as mild-to-moderate in severity.
Hedon et al15 and Urdl et al9 both noted the
overall occurrence of breast tenderness was
higher for the patch than for OC users (19%
for patch, 6% for OCs and 25.1% for patch,
8.9% for OCs, respectively). The Urdl study also
found that breast tenderness was treatment lim-
iting in 3.0% of patch users and 0.2% of OC
users.
Mild-to-moderate and transient application
site reactions have also been reported. The in-
cidence ranges from 6.7% to 13.8% of subjects,
with the reaction being treatment limiting for
1.2% of subjects.6,9 Another study found higher
rates of skin reaction—33.7% at cycle 1, declin-
ing to 14.7% by cycle 9; this side effect was
treatment limiting for 5.6% of subjects.18 There
is 1 case report of a 15-year-old girl who devel-
oped allergic contact dermatitis after 5 months
of regular patch use.19
The adverse effect profile among patch users
is otherwise similar to OC users.6 Table 1 pro-
vides an overview of the rate of common side
effects for patch users compared with users of an
OC.
Major Risks
Within the 3 largest published clinical trials
involving the transdermal contraceptive patch,
the serious adverse events potentially attributa-
ble to use of this system include 2 cases of
nonfatal pulmonary embolism (1 case involved
a woman wearing the patch until the day of
abdominal surgery); 1 case of menorrhagia; 1
case of arm pain, paresthesia, and hypesthesia; 1
case of migraine headache; 1 case of cholecysi-
tits; and 1 case of carcinoma in situ of the
cervix.11 With more widespread use, however,

TABLE 1. Most Common Adverse Events in the Comparative Study of the Patch Versus an OC20
Adverse Event
Headache
Nausea
Application site reactions
Breast symptoms*
Upper respiratory tract infection
Dysmenorrhea
Abdominal pain
* Breast symptoms include breast discomfort, engorgement, and pain.
Adapted from Fertil Steril. 2002;77(2 suppl 2):S19–S26.20 With permission from American Society for Reproductive Medicine.
additional information on risk of venous throm-
boembolism and cardiovascular events has
emerged.
Venous Thromboembolic Disease
and Cardiovascular Events
Reports of patch users experiencing blood clots
have received recent media attention, and raised
concern whether patch users have a higher in-
cidence of venous thromboembolic disease
(VTE) compared with users of other combina-
tion hormonal contraceptive methods. A nested
case-control study was conducted by i3 Drug
Safety using an insurance claims database in
addition to verification of cases through medical
record review.21 The objective of this study was
to evaluate the combined risk of heart attack and
stroke and the occurrence of VTE in users of the
contraceptive patch compared with users of
NGM-containing OCs with 35 mcg of EE. For
VTE, there were 61 total cases, 22 among patch
users and 39 among the OC users; 57 controls
and 186 controls were matched in each treatment
group, respectively. The odds ratio (OR) for
VTE comparing current users of the patch and
OCs was 2.42 [95% confidence interval (CI):
1.07-5.46]. The estimated incidence of venous
thromboembolism per 100,000 women-years
was 40.8 for contraceptive patch users and 18.3
for users of the NGM-containing OC.
Another database study published in 2006
also used a nested case-control design and a
database of a large managed-care organiza-
tion.22 The authors identified women aged 15
to 44 years who were current users of the patch
or an OC containing 35 mg of EE and NGM and
who had a first-time recorded claim (index date)
for a clinically diagnosed deep vein thrombosis
The Transdermal Contraceptive Patch 921
Overall Incidence (%)
Patch (N = 812) OC (N = 605) P
21.9 22.1 0.95
20.4 18.3 0.34
20.2 NA NA
18.8 6.1 <0.001
13.3 17.9 0.02
13.3 9.6 0.04
8.1 8.4 0.85
or pulmonary embolus. Potential cases were
excluded if important clinical risk factors for
VTE were present in the 3 months before the
index date. Up to 4 women who did not have a
diagnosis of VTE were matched to each case by
year of birth and the index date of the case. The
final study population consisted of 68 cases of
VTE and 266 controls; cases were 31 patch users
and 37 OC users. The unadjusted matched OR
for VTE for the patch versus OC was 0.9 (95%
CI: 0.5-1.6). There was no change in OR after
adjusting for duration of exposure or switching
from another hormonal contraceptive. The
overall incidence rate for VTE was 52.8 per
100,000 women-years (95% CI: 35.8-74.9)
among patch users and 41.8 per 100,000 wo-
men-years (95% CI: 29.4-57.6) for users of
NGM-containing OCs. Adjusted for age, VTE
incidence rate ratio (IRR) for current use of the
patch versus OC was 1.1 (95% CI: 0.7-1.8), with
no effect modification by age. Thus, the risk of
nonfatal idiopathic VTE among new users of the
patch was similar to that of new users of the
NGM OC.
Jick and Jick23 also examined the same data-
base to evaluate the risk of cerebrovascular
accident and myocardial infarction. They iden-
tified 18 women from age 19 to 43 years who had
a first ischemic stroke (embolic in 4, thrombotic
in 14) while taking a study contraceptive. At the
index date, 8 women were exposed to the patch
and 10 were exposed to a NGM OC. The crude
incidence rate of ischemic stroke was 13.6 per
100,000 woman-years (95% CI: 5.9-26.8) among
patch users, and 11.3 per 100,000 woman-years
(95% CI: 5.4-20.8) for users of NGM-containing
OCs. The crude IRR in patch users compared
with users of a NGM OC was 1.2 (95% CI: 0.41-
3.4). There were not enough cases in either group
to adjust for age or calendar time. Although the
922 O’Connell and Burkman
women were current contraceptive users, several
had risk factors for stroke, such as diabetes
mellitus (4), hypertension (2), and atrial septal
defects (3). Nine women (5 exposed to a NGM
OC and 4 exposed to the patch) had no known
risk factors documented in their records. The
authors identified 8 women from age 20 to 43
years with acute myocardial infarction who sa-
tisfied the study criteria. At the index date, 1
woman was exposed to the patch and 7 women
were exposed to a NGM OC. The crude inci-
dence rate of acute myocardial infarction among
current patch users was 1.7 per 100,000 woman-
years (95% CI: 0.04-9.5) and among current
users of a NGM OC was 7.9 per 100,000 wo-
man-years (95% CI: 3.2-16.3). The IRR for the
patch users compared with users of a NGM OC
was 0.2 (95% CI: 0.004-1.7). Four women had
one or more risk factors for myocardial infarc-
tion such as diabetes (1), hypertension (2), hy-
perlipidemia (2), or obesity (2), whereas the
other 4 women had no risk factors documented.
All 4 of the women without risk factors were
exposed to a NGM OC. The authors concluded
that the data do not suggest an increased risk of
either ischemic stroke or acute myocardial in-
farction in users of the patch compared with
users of a NGM OC, and that these events are
rare among young women who use hormonal
contraceptives.
These 3 studies have limitations due to their
design. The nested case control design uses
restricted databases, limiting access to all poten-
tial confounders. There was no information on
body weight (obesity known to be a risk factor
for VTE) or on compliance with the method. The
databases include only short-term use of the
methods. Furthermore, the studies may suffer
from both reporting bias and nondifferential
misclassification. Although the i3 Drug Safety
data differs from the Jick study, it should be
noted that VTE is a relatively rare event and is a
potential risk with all combination hormonal
contraceptive therapy. Further, the level of dif-
ference in VTE risk reported in the i3 Drug
Safety study is similar to that reported in studies
comparing desogestrel-containing OCs to levo-
norgestrel-containing OCs.
Finally, given the higher total exposure of
estrogen with 3 months of continuous use, some
researchers have raised concerns about the pos-
sible higher cumulative levels of estrogen and
potential risk in particular of VTE and poten-
tially other cardiovascular events. Van den Heu-
vel et al24 conducted a randomized, open-label,
parallel-group single center trial of the pharma-
cokinetics of 3 methods of combined hormonal
contraception. They enrolled 24 total subjects, 8
in each of 3 groups: a single 21-day cycle of
treatment of the patch, the vaginal contraceptive
ring (NuvaRing, Organon), and an OC (EE
30 mg, levonorgestrel 150 mg). The mean estra-
diol (EE) area under the curve in the patch group
was 1.6 times (60%) higher than in the pill group,
and 3.4 times higher than in the NuvaRing group
(P<0.05 for both comparisons). There was a
trend of increasing EE concentration over time
for the patch (higher during the third patch than
the first patch). Peak concentrations for EE are
approximately 25% lower in women using the
patch; however, patch users had overall higher
steady state concentrations and lower peak con-
centrations than OC users. However, it should
also be recognized that there have been no
studies to date in the field of hormonal contra-
ception that have shown that any pharmacoki-
netic parameter can predict risk of an event such
as VTE. These results from a single small study
may have implications for women who have
other cardiovascular risk factors for thrombotic
disease; for most healthy women, the patch still
poses an acceptable level of risk. However, the
United States Food and Drug Administration
has required that the manufacturer of the patch
include such pharmacokinetic information in its
package label.
No study of the patch has yet been conducted
among women who have other cardiovascular
risk factors for thrombotic disease. The WHO
guide to medical eligibility for contraception25
rates the patch a category 4 (unacceptable health
risk) in the presence of history of deep vein
thrombosis/pulmonary embolus, known throm-
bogenic mutations, history of ischemic heart
disease or stroke, complicated valvular heart
disease—the same classifications as oral contra-
ception. As with OCs, the WHO rates the patch
at least a category 3 (risks usually outweigh
benefits) in the setting of multiple risk factors
for arterial cardiovascular disease (such as older
age, smoking, diabetes, and hypertension). Clin-
icians should proceed with caution in prescrib-
ing the patch to patients with multiple risk
factors until more is known about the VTE risks
of the patch as compared with the pill.
Extended Use
A recent innovation in oral contraception for-
mulations is extended use, where withdrawal
bleeding occurs every several months (Seaso-
nale,26,27 Seasonique28) or not at all (Lybrel29).
 The Transdermal Contraceptive Patch
Several studies of continuous use of the contra-
ceptive patch have also been published.
Stewart et al30 randomized 239 women in a
2:1 ratio, to a standard (cyclic) or extended
regimen of patch use. The extended regimen
was a weekly application of the patch for 12
consecutive weeks, followed by 1 patch-free
week, then finally 3 more consecutive weekly
applications (n = 158). Subjects in the extended
regimen group had fewer median bleeding days
(14 vs. 6, P<0.001), fewer bleeding episodes (1
vs. 3, P<0.001), and fewer bleeding or spotting
episodes (3 vs. 2, P<0.001). The median number
of bleeding or spotting days were similar be-
tween groups (16 vs. 14, P = 0.407). Addition-
ally, there was delayed median time to first
bleeding (25 vs. 54 days, P<0.001) and higher
amenorrhea rates (1% vs. 12%, P<0.003) in the
extended regimen group. The groups had similar
high satisfaction rates (88.6%, 86.3%); though
there were more reports of headache, nausea,
and breast discomfort in the extended regimen
group, they were not to a statistically significant
degree.
LaGuardia et al31 studied the same popula-
tion to look at the incidence of headache. Across
both regimens, the mean number of headache
days per week was 0.63 when patches were ‘‘on’’
and was 1.19 when patches were ‘‘off’’
(P<0.0001). The headache rate during ‘‘on-
patch’’ weeks decreased significantly in both
groups over the 16-week study (P = 0.0002);
mean headache days per week almost doubled
in ‘‘off-patch’’ weeks. Importantly, in the ex-
tended-cycle group, the headache frequency in
the ‘‘off-patch’’ weeks did not exceed that seen at
baseline. The authors concluded that withdra-
wal of contraception after prolonged exposure
did not result in an exaggerated increase in
headache frequency.
Compliance
Although all forms of combined hormonal con-
traception have high efficacy rates, there is a
difference between perfect use and typical use
failure rates. The daily adherence required for
optimal OC use may impede the method’s effi-
cacy. Lengthening the dosing interval may help
to close the gap between perfect and typical use.
Early studies showed better adherence among
patch users than OC users.6 Pooled data from
the original Food and Drug Administration
trials showed perfect patch use ranged from
88.1% to 91.0% and was not related to age, in
contrast to the OC groups.32 A summary of 3
923
international trials examined compliance by
users of the patch (either the patch alone, or
with an OC comparator group).33 There were
3329 total users of the patch in these trials; one-
third of subjects used the method for 13 cycles,
the remainder for 6 cycles. Perfect patch com-
pliance ranged between 92.7% and 94.4%, as
compared with 78.4% to 87.8% for OCs. The
overall patch compliance over 22,177 cycles was
93.3% (95% CI: 92.9, 93.6). A reanalysis of the
Audet data explored the impact of the increase in
the proportion of perfect dosing cycles on the
overall contraceptive efficacy of the patch.34
Efficacy was better in cycles with perfect dosing
compared with cycles with imperfect dosing
(P = 0.007 in the comparative study). The Pearl
Index was 0.73 for perfect dosing cycles, as
compared with 2.33 for imperfect dosing cycles.
Other studies confirmed these findings. A
randomized, open-label, parallel-group trial at
65 centers in Europe and South Africa enrolled
1489 women.9 Patch users had significantly
more cycles with perfect compliance than OC
users (P<0.001). There were no dosing errors in
96.5% of 5922 total cycles of patch use and
90.6% of 4671 cycles of OC use. For each age
group up to 40 years, the percentage of cycles
with perfect compliance was higher for the patch
group than the OC group. Similarly, a multi-
center, single-arm, open label descriptive cohort
study in Canada enrolled 392 women, age 18 to
45, who used the patch over 9 cycles.18 Eighty-
eight percent of subjects recorded perfect com-
pliance; overall compliance rates did not differ
among different age groups.
A more recent study, however, had a different
conclusion regarding compliance with the
patch.35 Subjects were 1230 women seen at 3
Planned Parenthood clinics; all were first time
users of OCs or the patch, and self-selected their
contraceptive method. Eighty-nine percent of
subjects were considered high risk for contra-
ceptive nonadherence, for example, under age
16, nonwhite and/or Hispanic, covered by gov-
ernment health insurance, or had a prior termi-
nation of pregnancy. The initial loss-to-follow-
up rate was higher among patch users compared
with OC users (45.2% vs. 29.5%, P<0.001).
Continued use beyond the first 70 days was
achieved in 67% of patch users and 89% of pill
users (P<0.001). The Pearl Index was 3.62 for
OC users, and 14.84 for the patch users. The
‘‘long-term’’ unintended pregnancy rate was
higher for patch users. For example, among
patch users the relative risk of pregnancy before
70 days of use was 0.73 (95% CI: 0.41-4.13),
924 O’Connell and Burkman
whereas the relative risk of pregnancy after 70
days of use was 3.23 (95% CI: 1.43-7.31,
P = 0.005). The absolute pregnancy risk after
70 days was 3.46%, which likely represents
‘‘typical use effectiveness’’ of the patch. At 1
year of follow-up, 76% of pill users and 57% of
patch users continued their method (P = 0.004);
there was no difference in continuation by high-
risk status.
How do we account for the difference in
findings? Studies that found higher compliance
with patch use may have ‘‘research bias’’; sub-
jects may have been more likely to use the
method due to the higher frequency of contact
with study staff. Subjects in the Planned Parent-
hood study who were considered ‘‘high risk’’ for
contraceptive noncompliance may or may not
reflect actual usage in the general population.
Another key factor in a possible decrease in
method compliance is counseling. Providers of-
ten spend more time discussing a method with
patients when the method is new. Over time, as
familiarity with the patch became more wide-
spread, providers may not spend sufficient time
explaining how to use the method. The patch is
not as intuitive as the pill, however, and may
actually be more difficult to use consistently
(anecdotally, women get confused if bleeding
begins early when patch is on, or runs late, and
they may alter their schedule of patch applica-
tion). Patients too may believe that the patch is
superior to the pill because of its less-frequent
application; patch users may therefore be less
likely to select strict contraceptive regimens and
may be more likely to fail, even with the less-
frequent dosing.
Adolescents
Studies of adolescents and the patch have
yielded mixed results. In addition to Bakhru
and Stanwood35 detailed above, 2 studies have
examined adolescents’ experiences with the
patch. Harel et al36 conducted interviews and
chart reviews of 28 adolescents (mean age 18 y)
who had used the patch for 7 months on average.
Ninety-three percent of subjects were sexually
active, and 50% had at least 1 child. All of the
subjects reported regular menses, and 14% re-
ported BTB. Many subjects reported a decrease
in dysmenorrhea (39%), headaches (29%), and
acne (33%). There were no significant body mass
index changes during patch use; 29% of subjects
perceived weight loss, and 14% had a perceived
weight gain. Correct patch use was self-reported
by 93% of subjects. Twenty-one percent of sub-
jects reported at least 1 episode of complete
detachment of a patch, with 32% reporting
partial peeling of the patch corners. Mild tem-
porary application site reactions were reported
by 64% of subjects, and breast tenderness in
18%; of subjects who experienced tenderness,
39% discontinued after 6 months of use.
Rubinstein et al37 conducted a longitudinal
trial in 50 adolescents (mean age 16.4 y). Four-
teen percent of subjects had been pregnant but
no information on parity was given. The authors
had 80% follow-up at 1 month, and 62% follow-
up at 3 months. Eighty-seven percent of subjects
self-reported perfect compliance at the 3-month
follow-up. Complete or partial detachment of a
patch was reported by 35.5% of subjects. This
study has a high incidence of skin complaints:
itching (29% at 3 months), skin irritation (13%),
and adhesive deposition after patch removal
(23%). Subjects with prior OC experience felt
that the patch was easier to use (P<0.001),
easier to remember (P<0.001), and they were
more likely to recommend this method to friends
(P = 0.012); 77% of subjects wanted to continue
the patch at study’s end.
Both studies are limited by small number of
subjects, short-term follow-up, and self-report-
ing of compliance data.
Specific Management Issues
Contraceptive counseling needs to be adapted
for this transdermal system. Women without
contraindications to OC use are appropriate
for the patch. Women with a history of signifi-
cant skin allergy or exfoliative dermatologic
disorders may not be ideal candidates. Obese
women also need to be counseled about the
potential for reduced efficacy. Counseling to
ensure appropriate weekly changing of the
patch, using a different site for the next applica-
tion, and avoiding use of lotions or occlusive
dressings is important.
When patches become detached, users should
attempt to reattach them if possible, without
using adhesives or tape. If detachment is 24 hours
or less in duration, the cycle continues as usual
with the patch being changed on the previously
determined change day. If detachment has been
for more than 24 hours, a new patch should be
applied, backup contraception should be used for
1 week, and the day that the new patch is applied
now becomes the patch change day.
When a new patch cycle is delayed beyond the
scheduled start day, users should be instructed
that they need to apply a patch as soon as they
 The Transdermal Contraceptive Patch
remember and use backup contraception for at
least 1 week. In addition, the day they apply the
patch becomes the new patch change day. If a
user forgets to change the first or second patch
on time, there are different strategies that are
dependant upon the circumstances. There is a 2-
day period of continued release of adequate
contraceptive steroid levels when the patch is
left on for 2 extra days. If the patch is changed
within this window, the patch change day re-
mains the same and there is no need for backup
contraception. Failure to replace a patch after
this 2-day time period increases the risk for
pregnancy. Therefore, users will need to use
backup contraception or in some instances
emergency contraception if this occurs. In addi-
tion, the day she remembers to apply the patch
becomes the new change day. Forgetting to
remove the third patch on time carries less risk.
One should instruct the user to remove the patch
when she remembers and tell her that the change
day is not altered. Finally, if one wishes to switch
to a new patch change day, this should be done
during the last week of a cycle when patches are
not usually used.
Conclusions
Since its introduction in 2002, the contraceptive
patch has had both champions and detractors.
Its weekly dosing has made it attractive for some
patients, though this regimen may not lead to the
increased method adherence some have hoped
for. The patch, like the ring and newest OCs, has
no data to indicate whether any or all of the
noncontraceptive benefits associated with OC
use will accrue. As the patch works through
similar mechanisms as the OC, long-term studies
will, nevertheless, likely show similar benefits. A
woman’s birth control method choices are influ-
enced by many factors, and frequently change
over the course of her reproductive years.
Although the patch may not be a contraceptive
panacea, increased birth control choices are
good for all women and their partners.
References
1. Finer LB, Henshaw SK. Disparities in unin-
tended pregnancy in the United States, 1994
and 2001. Perspect Sex Reprod Health.
2006;38:90–96. Referenced in www.guttma-
cher.org/presentations/ab_slides.html.
2. Mosher W, Martinez GM, Chandra A, et al.
Use of contraception and use of family planning
925
services in the United States, 1982-2002. Ad-
vance Data From Vital and Health Statistics;
no. 350. Hyattsville (MD): National Center for
Health Statistics; 2004.
3. Jewelewicz R. New developments in topical
estrogen therapy. Fertil Steril. 1997;67:1–12.
4. Abrams LS, Skee DM, Natarajan J, et al.
Pharmacokinetics of a contraceptive patch
(Evra/Ortho Evra) containing norelgestromin
and ethinyloestradiol at four application sites.
Br J Clin Pharmacol. 2002;53:141–146.
5. Dittrich R, Parker L, Rosen JB, et al, Ortho
Evra/Evra 001 Study Group. Transdermal
contraception: evaluation of three transdermal
norelgestromin/ethinyl estradiol doses in a
randomized, multicenter, dose-response study.
Am J Obstet Gynecol. 2002;186:15–20.
6. Abrams LS, Skee DM, Natarajan J, et al.
Multiple-dose pharmacokinetics of a contra-
ceptive patch in healthy women participants.
Contraception. 2001;64:287–294.
7. Creasy GW, Fisher AC, Hall N, et al. Trans-
dermal contraceptive patch delivering norel-
gestromin and ethinyl estradiol. Effects on the
lipid profile. J Reprod Med. 2003;48:179–186.
8. Pierson RA, Archer DF, Moreau M, et al.
Ortho Evra/Evra versus oral contraceptives:
follicular development and ovulation in nor-
mal cycles and after an intentional dosing
error. Fertil Steril. 2003;80:34–42.
9. Urdl W, Apter D, Alperstein A, et al. Ortho
Evra/Evra 003 Study Group. Contraceptive
efficacy, compliance and beyond: factors re-
lated to satisfaction with once-weekly trans-
dermal compared with oral contraception. Eur
J Obstet Gynecol Reprod Biol. 2005;121:
202–210.
10. Zieman M, Guillebaud J, Weisberg E, et al.
Contraceptive efficacy and cycle control with
the Ortho Evra/Evra transdermal system: the
analysis of pooled data. Fertil Steril. 2002;
77(2 suppl 2):S13–S18.
11. Zacur HA, Hedon B, Mansour D, et al. Inte-
grated summary of Ortho Evra/Evra contra-
ceptive patch adhesion in varied climates and
conditions. Fertil Steril. 2002;77(2 suppl 2):
S32–S35.
12. Murthy AS, Creinin MD, Harwood B, et al.
Same-day initiation of the transdermal hormo-
nal delivery system (contraceptive patch) ver-
sus traditional initiation methods. Contra-
ception. 2005;72:333–336.
13. Abrams LS, Skee DM, Wong FA, et al. Phar-
macokinetics of norelgestromin and ethinyl
estradiol from two consecutive contraceptive
patches. J Clin Pharmacol. 2001;41:1232–1237.
14. White T, Jain JK, Stanczyk FZ. Effect of oral
versus transdermal steroidal contraceptives on
926 O’Connell and Burkman
androgenic markers. Am J Obstet Gynecol.
2005;192:2055–2059.
15. Hedon B, Helmerhorst FM, Cronje HS, et al.
Comparison of efficacy, cycle control, compli-
ance, and safety in users of a contraceptive
patch vs. an oral contraceptive. Int J Gynecol
Obstet. 2000;70(suppl 1):78.
16. Audet MC, Moreau M, Koltun WD, et al.
Ortho Evra/Evra 004 Study Group. Evalua-
tion of contraceptive efficacy and cycle control
of a transdermal contraceptive patch vs an oral
contraceptive: a randomized controlled trial.
JAMA. 2001;285:2347–2354.
17. Smallwood GH, Meador ML, Lenihan JP, et
al. Ortho Evra/Evra 002 Study Group. Effi-
cacy and safety of a transdermal contraceptive
system. Obstet Gynecol. 2001;98(5 Pt 1):
799–805.
18. Weisberg F, Bouchard C, Moreau M, et al.
Preference for and satisfaction of Canadian
women with the transdermal contraceptive
patch versus previous contraceptive method:
an open-label, multicentre study. J Obstet Gy-
naecol Can. 2005;27:350–359.
19. Stricker T, Sennhauser FH. Allergic contact
dermatitis due to transdermal contraception
patch. J Pediatr. 2006;148:845.
20. Sibai BM, Odlind V, Meador ML, et al. A
comparative and pooled analysis of the safety
and tolerability of the contraceptive patch
(Ortho Evrat/Evrat). Fertil Steril. 2002;
77(2 suppl 2):S19–S26.
21. Cole JA, Norman H, Doherty M, et al. Venous
thromboembolism, myocardial infarction,
and stroke among transdermal contraceptive
system users. Obstet Gynecol. 2007;109:
339–346.
22. Jick SS, Kaye JA, Russmann S, et al. Risk of
nonfatal venous thromboembolism in women
using a contraceptive transdermal patch and
oral contraceptives containing norgestimate
and 35 microg of ethinyl estradiol. Contracep-
tion. 2006;73:223–228.
23. Jick SS, Jick H. The contraceptive patch in
relation to ischemic stroke and acute myocar-
dial infarction. Pharmacotherapy. 2007;27:
218–220.
24. van den Heuvel MW, van Bragt AJM,
Alnabawy AKM, et al. Comparison of
ethinylestradiol pharmacokinetics in three
hormonal contraceptive formulations: the
vaginal ring, the transdermal patch and an
oral contraceptive. Contraception. 2005;72:
168–174.
25. World Health Organization. Medical Eligibil-
ity Criteria for Contraceptive Use. Geneva:
World Health Organization; 2004.
26. Anderson FD, Hait H. A multicenter, rando-
mized study of an extended cycle oral contra-
ceptive. Contraception. 2003;68:89–96.
27. Anderson FD, Gibbons W, Portman D. Long-
term safety of an extended-cycle oral contra-
ceptive (Seasonale): a 2-year multicenter open-
label extension trial. Am J Obstet Gynecol.
2006;195:92–96.
28. Anderson FD, Gibbons W, Portman D. Safety
and efficacy of an extended-regimen oral con-
traceptive utilizing continuous low-dose ethi-
nyl estradiol. Contraception. 2006;73:229–234.
29. Archer DF, Jensen JT, Johnson JV, et al.
Evaluation of a continuous regimen of levo-
norgestrel/ethinyl estradiol: phase 3 study re-
sults. Contraception. 2006;74:439–445.
30. Stewart FH, Kaunitz AM, Laguardia KD, et
al. Extended use of transdermal norelgestro-
min/ethinyl estradiol: a randomized trial. Ob-
stet Gynecol. 2005;105:1389–1396.
31. LaGuardia KD, Fisher AC, Bainbridge JD, et
al. Suppression of estrogen-withdrawal head-
ache with extended transdermal contraception.
Fertil Steril. 2005;83:1875–1877.
32. Archer DF, Bigrigg A, Smallwood GH, et al.
Assessment of compliance with a contraceptive
patch (Ortho Evrat/Evrat) among North
American women. Fertil Steril. 2002;77(2
suppl):S27–S31.
33. Creasy G, Hall N, Shangold G. Patient adher-
ence with the contraceptive patch dosing sche-
dule versus oral contraceptives. Gynecol Obstet
Fertil. 2000;28:317–323.
34. Archer DF, Cullins V, Creasy GW, et al. The
impact of improved compliance with a weekly
contraceptive transdermal system (Ortho
Evra) on contraceptive efficacy. Contracep-
tion. 2004;69:189–195.
35. Bakhru A, Stanwood N. Performance of con-
traceptive patch compared with oral contra-
ceptive pill in a high-risk population. Obstet
Gynecol. 2006;108:378–386.
36. Harel Z, Riggs S, Vaz R, et al. Adolescents’
experience with the combined estrogen and
progestin transdermal contraceptive method
Ortho Evra. J Pediatr Adolesc Gynecol.
2005;18:85–90.
37. Rubinstein ML, Halpern-Felsher BL, Irwin
CE Jr. An evaluation of the use of the trans-
dermal contraceptive patch in adolescents. J
Adolesc Health. 2004;34:395–401.

Emergency Contraception:
A Clinical Review
REBECCA H. ALLEN, MD, MPH* and
ALISA B. GOLDBERG, MD, MPH* w
*Department of Obstetrics, Gynecology, and Reproductive Biology, Brigham
and Women’s Hospital, Harvard Medical School; and w Clinical Training and
Research, Planned Parenthood League of Massachusetts Inc, Boston,
Massachusetts
Abstract: Emergency contraception is defined as a
drug or device used to prevent pregnancy after un-
protected sexual intercourse (including sexual assault)
or after a recognized contraceptive failure. In the
United States, 1.5 mg of levonorgestrel, packaged as
Plan B, and the Copper T 380A intrauterine device are
the most common emergency contraceptives available
to women and are effective up to 5 days after unpro-
tected sexual intercourse. In August 2006, Plan B was
approved for over-the-counter sale to women aged 18
and older in the United States. It is not yet known
whether the increased availability of emergency con-
traception will decrease unintended pregnancy and
induced abortion rates.
Key words: emergency contraception, levonorgestrel,
intrauterine device, Plan B
Introduction
Unintended pregnancies, those that were either
mistimed or undesired at the time of conception,
are common in the United States. Of the 6.4
million pregnancies recorded in 2001, approxi-
mately half were unintended. Of these 3.1 million
unintended pregnancies, 44% resulted in births,
42% in induced abortions, and 14% in miscar-
riages. Among women with unintended preg-
nancies, 48% were using contraceptives during
the month of conception, although not always
correctly.1,2 Emergency contraception (EC) is
defined as a drug or device used to prevent
pregnancy after unprotected sexual intercourse
(including sexual assault) or after a recognized
Correspondence: Rebecca H. Allen, MD, MPH,
Department of Obstetrics, Gynecology, and Reproduc-
tive Biology, Brigham and Women’s Hospital, Harvard
Medical School, 1620 Tremont Street, 3rd Floor,
Boston, MA 02120. E-mail: rallen@partners.org
CLINICAL OBSTETRICS AND GYNECOLOGY / VOLUME 50
CLINICAL OBSTETRICS AND GYNECOLOGY
Volume 50, Number 4, 927–936
r 2007, Lippincott Williams & Wilkins
contraceptive failure.3 EC is intended as a back-
up method for occasional use, rather than a
regular method of contraception. Although EC
is sometimes called postcoital contraception or
the morning-after pill, these labels are confusing,
because they imply that EC can only be taken the
morning after unprotected sexual intercourse.4
On the contrary, women can use EC up to 5 days
afterwards.
EC has the potential to reduce unintended
pregnancy rates thereby reducing the number of
induced abortions. Providing education about
how to use and access EC, whether over-the-
counter, from a medical provider or pharmacist,
or by advance prescription, is an integral part of
women’s health care. Many medical organiza-
tions, such as the American College of Obste-
trician Gynecologists, the American Medical
Association, the Society for Adolescent Medi-
cine, the American Academy of Family Physi-
cians, and the American Academy of Pediatrics
have policies endorsing its use and advocating its
increased availability. Currently, women in the
United States can be offered oral hormonal
interventions, consisting of levonorgestrel
(LNG)-only or ethinyl estradiol (EE) plus
LNG, or a copper-bearing intrauterine device
(IUD) for EC. This review will focus on LNG-
only and the copper-bearing IUD as emergency
contraceptives.
Regimens
The oral EC regimen most commonly used in the
United States consists of 1.5 mg of the progestin,
LNG. This product, packaged and marketed as
Plan B (Barr Pharmaceuticals Inc, Pomona, NY)
/ NUMBER 4 / DECEMBER 2007
927
928 Allen and Goldberg

instructs women to take one 0.75-mg LNG tablet
as soon as possible after unprotected intercourse
and to take the second 0.75-mg tablet 12 hours
after the first dose.5 The Food and Drug Admin-
istration (FDA) approved Plan B for use as an
emergency contraceptive in 1999. Studies have
also shown that taking both tablets at the same
time or up to 24 hours apart is as effective as
taking them 12 hours apart and does not increase
side effects.6–8 An alternative regimen can be
created from oral contraceptives that contain
EE and the progestin LNG or norgestrel, which
contains LNG as an isomer. A total of 22 brands
of combined oral contraceptives in the United
States can be used to create this regimen, which
consists of 2 doses taken 12 hours apart, each
containing 100 mg of EE plus 0.5 mg of LNG
(Table 1). This treatment is often called the Yuzpe
regimen after the physician who first described it.9
LNG-only EC is considered the first-line therapy,
however, because it is more effective and causes
fewer side effects.10,11 Worldwide, about 50 emer-
gency contraceptive products are specifically
packaged, labeled, and marketed. For example,
the LNG-only products, Postinor-2 (Gedeon
Richter, Budapest, Hungary) and NorLevo
(HRA Pharma, Paris, France), each consisting
of a 2-pill strip with each pill containing 0.75 mg
LNG are marketed in many countries. LNG-only
EC is available either over-the-counter or from a
pharmacist without having to see a clinician in 42
countries.12
Initial studies concluded that oral EC should
be taken within 72 hours of unprotected inter-
course. However, recent trials have shown that it
is still effective, although less so, when the first
dose is taken up to 120 hours after intercourse
(Fig. 1).7,13 Given this reduction in effectiveness

TABLE 1. Oral Contraceptives That Can be Used for EC in the United States* w
EE per LNG per
Brand Company Pills per Dosez Dose (lg) Dose (mg)y

Progestin-only pills: take one dosez

Plan-B Barr/Duramed 2 white pills 0 1.5
Combined progestin and estrogen pills: take 2 doses 12 h apart
Alesse Wyeth-Ayerst 5 pink pills 100 0.50
Aviane Barr/Duramed 5 orange pills 100 0.50
Cryselle Barr/Duramed 4 white pills 120 0.60
Enpresse Barr/Duramed 4 orange pills 120 0.50
Jolessa Barr/Duramed 4 pink pills 120 0.60
Lessina Barr/Duramed 5 pink pills 100 0.50
Levlen Berlex 4 light-orange pills 120 0.60
Levlite Berlex 5 pink pills 100 0.50
Levora Watson 4 white pills 120 0.60
Lo/Ovral Wyeth-Ayerst 4 white pills 120 0.60
Low-Ogestrel Watson 4 white pills 120 0.60
Lutera Watson 5 white pills 100 0.50
Nordette Wyeth-Ayerst 4 light-orange pills 120 0.60
Ogestrel Watson 2 white pills 100 0.50
Ovral Wyeth-Ayerst 2 white pills 100 0.50
Portia Barr/Duramed 4 pink pills 120 0.60
Quasense Watson 4 white pills 120 0.60
Seasonale Barr/Duramed 4 pink pills 120 0.60
Seasonique Barr/Duramed 4 light-blue-green pills 120 0.60
Tri-Levlen Berlex 4 yellow pills 120 0.50
Triphasil Wyeth-Ayerst 4 yellow pills 120 0.50
Trivora Watson 4 pink pills 120 0.50

* NOT-2-LATE.com: the EC website. Princeton University Office of Population Research. Princeton (NJ): Office of Population
Research; 2005. Available at: http://ec.princeton.edu/questions/dose.html. Retrieved January 25, 2007.
w Plan-B is the only dedicated product specifically marketed for EC. Alesse, Aviane, Cryselle, Enpresse, Jolessa, Lessina, Levlen,
Levlite, Levora, Lo/Ovral, Low-Ogestrel, Lutera, Nordette, Ogestrel, Ovral, Portia, Quasense, Seasonale, Seasonique, Tri-
Levlen, Triphasil, and Trivora have been declared safe and effective for use as ECPs by the US FDA.
z The label for Plan B says to take 1 pill within 72 h after unprotected intercourse, and another pill 12 h later. However, recent
research has found that both Plan B pills can be taken at the same time. Research has also shown that that all of the brands listed
here are effective when used within 120 h after unprotected sex.
yThe progestin in Cryselle, Lo/Ovral, Low-Ogestrel, Ogestrel, and Ovral is norgestrel, which contains 2 isomers, only 1 of which
(LNG) is bioactive; the amount of norgestrel in each tablet is twice the amount of LNG.

4.5
4
Pregnancy Rate (%)
3.5
3 2.6
2.5
2 1.5
1.5
1 0.5
0.5
0
0-12
Time Interval Before Treatment (hours)
FIGURE 1. Impact of delay in treatment on oral EC effectiveness.14
with delayed use, women should also be informed
of the copper-bearing IUD as an alternative if
appropriate.15 An IUD containing at least
380 mm2 of copper, inserted within 5 days of
unprotected intercourse, is an effective emergency
contraceptive and provides long-term protection
against pregnancy.16 Unlike hormonal EC, there
does not appear to be a decrease in efficacy with
delayed insertion of the IUD, as long as use begins
within 5 days of unprotected intercourse.15 The
device available in the United States is the Copper
T 380A (ParaGard, Barr Pharmaceuticals Inc,
Pomona, NY).
International researchers have also studied a
range of doses of the antiprogestin, mifepris-
tone, for EC. Most commonly, a low dose of
mifepristone, 10, 25, or 50 mg, is taken within
120 hours of unprotected intercourse. These
regimens are as effective as 1.5 mg of LNG.2,7
In the United States, mifepristone is not used for
EC because of its high cost and because it is not
manufactured or marketed in an appropriate
dose. Other progesterone receptor modulators
for EC are currently undergoing clinical trials.
Mechanism of Action
The mode of action of hormonal EC is multi-
factorial and not completely understood. Inves-
tigators have theorized that LNG-only EC
interferes with ovulation, sperm migration, fal-
lopian tube motility, corpus luteum function,
and endometrial physiology.5 Because sperm
are viable in the female reproductive tract for
up to 5 days, while eggs can only be fertilized
within 1 day of ovulation, the mechanism of
action most likely differs depending on when
Emergency Contraception 929
4.1
3.1
1.8
13-24 25-36 37-48 49-60 61-72
LNG-only EC is given in relation to the time of
intercourse and time of ovulation.17 None of the
current evidence supports the theory that hor-
monal EC interferes with postfertilization
events. Because LNG-only EC does not inter-
rupt an established pregnancy, defined as begin-
ning with implantation,18 hormonal EC is not
considered an abortifacient.19
The primary mechanism of action of LNG-
only EC, if taken in the first half of the cycle, is by
preventing or postponing ovulation through
suppressing, blunting, or delaying the surge in
luteinizing hormone. A secondary mechanism is
interference with sperm transport by thickening
cervical mucus and alkalinizing the uterine cav-
ity to immobilize sperm.20 LNG-only EC does
not, however, directly act on sperm to prevent
them from fertilizing the egg.21 There is conflict-
ing evidence to support impairment of the
endometrium, and thus, interference with im-
plantation, as a mechanism of action.19 There
are no data to support hindrance of fallopian
tube motility, supported by the fact that the rate
of ectopic pregnancies in women who had used
LNG-only EC was found to be lower than the
baseline rate.22 In sum, although the mechanism
of action of LNG-only EC is not fully known,
the preponderance of evidence to date suggests
that it prevents pregnancy by preventing con-
ception. Although it is highly unlikely that
LNG-only EC interferes with postfertilization
events, it has not been possible to prove whether
EC also prevents pregnancy by affecting events
between fertilization and implantation.19
The same modes of action that make the
Copper T 380A IUD useful as a conven-
tional contraceptive also make it suitable as an
930 Allen and Goldberg
emergency contraceptive. The Copper T 380A
can be inserted up to the time of implantation,
5 to 7 days after ovulation, to prevent pregnancy.
However, most providers limit insertion to with-
in 5 days of intercourse rather than ovulation,
because it is frequently difficult to estimate the
day of ovulation. The IUD likely has both
prefertilization and postfertilization effects that
include toxicity to sperm, ovum, and embryo.23
The Copper T 380A releases copper ions that, in
turn, create a chronic inflammatory reaction in
the uterus and fallopian tubes. This may impair
gamete viability before fertilization or hinder the
embryo’s ability to implant. Although the pre-
vention of fertilization is the primary mechanism
of action of the Copper T 380A IUD, postferti-
lization effects do contribute. However, because
it does not disrupt an already implanted preg-
nancy, it is not considered an abortifacient.15
Efficacy
The probability of pregnancy after a single act of
unprotected intercourse varies according to the
day of the menstrual cycle and the couple’s
fertility status. In 1 study, the probability of
pregnancy after a single random act of inter-
course was 3%, increasing to 8% if intercourse
occurred 1 to 2 days before ovulation.24 The
efficacy of EC is calculated as the prevented
fraction: 1 – (number of pregnancies observed
after treatment/estimated number of pregnan-
cies that would occur without treatment).5 How-
ever, this calculation supposes that every woman
requesting EC and her partner are fertile, that
her cycles are ovulatory and regular, that inter-
course occurred only once during the treatment
cycle, and that any pregnancy is the result of the
act of intercourse for which EC was provided.25
Another limitation of this statistic is that the
denominator is based on published, estimated
probabilities of pregnancy per menstrual cycle
day (relative to ovulation) of women who were
trying to become pregnant, not using contra-
ception. That population is different from EC
users, who often request treatment because of a
condom break, missed pill, or unsuccessful with-
drawal before ejaculation.26 Moreover, the nu-
merator may be erroneous because of inclusion
of pregnancies established before EC was admi-
nistered or pregnancies, resulting from inter-
course after treatment.15 Because of this,
estimates of effectiveness vary from trial to trial
and are inexact. It is known, however, that the
use of LNG-only EC is more effective than no
treatment at all after a single episode of unpro-
tected intercourse.26
Studies of women using the LNG-only regi-
men have calculated preventive fractions ran-
ging from 60% to 94%.5 Plan B labeling cites a
preventive fraction of 89%. In contrast, the use
of emergency contraceptives containing a com-
bination of estrogen and progestin (Yuzpe) re-
duces the risk of pregnancy by approximately
75%. The preventive fraction is often interpreted
as follows: if 100 women had unprotected inter-
course once during the second or third week of
their cycle and were not treated with EC, about 8
would become pregnant; after treatment with
EC, only 1 or 2 women would become pregnant,
a 75% to 89% reduction.3 A recent meta-analy-
sis concluded that 1.5 mg LNG (Plan B) was
superior to the Yuzpe regimen with a combined
relative risk (RR) of pregnancy of 0.51 [95%
confidence interval (CI): 0.31, 0.83] compared
with Yuzpe.2 Taking a total of 1.5 mg of LNG in
a single-dose is as effective as the 0.75 mg split-
dose (Fig. 2) and more user-friendly.2
The timing of oral EC influences its effective-
ness significantly. Waiting 12 hours to initiate
treatment after unprotected intercourse in-
creases the odds of pregnancy by almost 50%
and its efficacy decreases linearly with time
(Fig. 1).14 Experts recommend, therefore, that
oral EC be taken as soon as possible after
unprotected intercourse or contraceptive failure.
There are currently no data regarding the effi-
cacy of EC when treatment is initiated beyond
120 hours after intercourse. EC can be used more
than once in the same menstrual cycle if needed,
although a regular form of contraception is
recommended. The minimum time interval be-
tween consecutive treatments is unknown, but
experts opine that unprotected intercourse with-
in 12 hours of EC dosing does not require repeat
treatment.15
Indications
EC is indicated for any woman at risk for unin-
tended pregnancy from an identified episode of
contraceptive failure or unprotected intercourse.
However, surveys have found that the level of
knowledge and use of EC in the United States by
women and their health care providers is rela-
tively low.10 In 2000, only 1.3% of women
presenting for an induced abortion in the United
States reported using EC to prevent the preg-
nancy.27 A more recent study from Scotland
reported that 11.8% of women presenting
for abortion and 1% presenting for antenatal

FIGURE 2. LNG-only EC: single (treatment) versus split-dose (control), observed number of pregnancies
(events). Reprinted from Cochrane Database Syst Rev. 2004:CD001324.2 Copyright r 2006, with permission
from The Cochrane Collaboration and John Wiley & Sons, Ltd.
care used EC in the cycle where conception
occurred.28 This lack of use is not surprising
given that one of the main reasons women cite
for not using regular and EC is their perception
that they are at low risk for pregnancy.29 There
are no data yet to evaluate the impact of over-
the-counter access to Plan B on EC knowledge
and use in the United States. There are several
situations in which EC can be used (Table 2).
Surveys have shown that most women request-
ing EC had either not used contraception or
experienced a failure with a barrier method,
and the remaining minority experienced a failure
of withdrawal, were sexually assaulted, or had
missed taking an oral contraceptive pill.3
No clinical examination or pregnancy testing
is required before using oral hormonal EC.31
There are no medical contraindications to the
use of oral EC regimens because of their short
duration of use. Plan B’s labeling lists 3 contra-
indications: pregnancy, hypersensitivity to any
component of the product, and undiagnosed
abnormal genital bleeding.3 Pregnancy is a re-
lative contraindication because EC is ineffective
if a pregnancy is already established. On the
basis of studies of the teratogenic risk of com-
bined oral contraceptives, however, there is no
increased risk to the fetus if oral EC is taken
while pregnant.15 Despite the labeling, there is
no medical rationale to justify denying women
with undiagnosed genital bleeding access to
EC.15
World Health Organization (WHO) recom-
mendations state that women with previous
ectopic pregnancy, cardiovascular disease, mi-
graines, or liver disease and women who are
breastfeeding can use EC.32 There is no evidence
that hormonal EC increases the risk of venous
Emergency Contraception 931
thromboembolic disease.33 There may be
reduced efficacy of oral EC in women with
severe malabsorption syndromes and those tak-
ing liver-enzyme–inducing drugs. There have
also been case reports of LNG-only EC interfer-
ing with warfarin anticoagulant action.15,34
Therefore, women taking liver-enzyme–indu-
cing medications (including but not limited to
rifampin, carbamazepine, phenytoin, phenobar-
bital, and St John’s Wort) or warfarin should
seek clinical advice before, or shortly after, using
EC. Emergency contraceptive use should not be
TABLE 2. Potential Indications for Use of EC*
Lack of contraceptive use during coitus
Mechanical failure of male condom (breakage, slippage,
or leakage)
Dislodgment, breakage, or incorrect use of diaphragm,
cervical cap, or female condom
Failure of spermicide tablet or film to melt before
intercourse
Error in practicing withdrawal (coitus interruptus)
Missed combined oral contraceptives (any 2 consecutive
pills)
Missed progestin-only oral contraceptive (one or more)
Expulsion or partial expulsion of an IUD
Exposure to potential teratogen (such as isotretinoin or
thalidomide while not using effective contraception)
Late injection of injectable contraceptive (>2 wk late for
a progestin only formulation such as depot
medroxyprogesterone acetate or norethindrone
enanthate, or >3 d late for a combined estrogen plus
progestin formulation)w
Rape
*Reprinted from Ann Intern Med. 2002;137:180–189.30 Copy-
right r 2002, with permission from the American College of
Physicians.
wThe usual interval for use of depot medroxyprogesterone
acetate as contraception is every 12 wk; the interval for the
combined monthly injectable formulation is every 28 to 30 d.
932 Allen and Goldberg
delayed in these women, for whom an unin-
tended pregnancy may carry additional health
risks. The use of the Copper T 380A IUD for EC
has the same contraindications, as does custom-
ary IUD insertion.32
Side Effects
The FDA has determined that Plan B is safe
enough to be available over-the-counter. In con-
trast to other over-the-counter medications,
such as acetaminophen or aspirin, there have
been no deaths or serious complications directly
linked to EC.5 The most common short-term
side effects of hormonal EC include nausea,
vomiting, and irregular bleeding. Other minor
adverse effects reported by women in clinical
trials included dizziness, fatigue, breast tender-
ness, headache, and abdominal pain.3 The
LNG-only regimen results in less nausea and
vomiting than the Yuzpe regimen. In a com-
bined analysis of 2 trials comparing the Yuzpe
regimen with 0.75 mg of LNG given twice 12
hours apart, there was a lower incidence of
nausea (RR: 0.43, 95% CI: 0.39 to 0.48) and
vomiting (RR: 0.24, 95% CI: 0.18 to 0.31).2,35,36
In absolute terms, nausea and vomiting occur in
approximately 18% and 4% of women taking
LNG-only EC, respectively.6,7,35 There is no
difference in nausea and vomiting between the
single-dose and split-dose LNG-only EC regi-
men.2 Routine use of antiemetics with LNG-
only EC is not recommended. If vomiting occurs
within 1 to 2 hours of taking Plan B, most experts
recommend that the dose be repeated. Vaginal
administration is another option if the woman is
experiencing severe vomiting.3,37
Irregular bleeding caused by hormonal EC
typically resolves by the next menstrual cycle.
There is a low incidence of intermenstrual spot-
ting after using hormonal EC, ranging from 3%
to 37% in trials.38 After hormonal EC is taken,
menses usually occurs within 1 week before or
after the expected time. If the delay in the onset
of menses is greater than 1 week or if the
expected menses is lighter than usual, a preg-
nancy test should be performed.3 Because hor-
monal EC can postpone ovulation, making a
woman vulnerable to pregnancy later in the
cycle, women should be counseled to begin a
regular method of contraception immediately
after using EC (Table 3).38 A woman should
also be advised to seek medical attention for
continued irregular bleeding or abdominal pain,
because these symptoms may be a sign of a
spontaneous abortion or ectopic pregnancy.5
The side effects of the Copper T 380A IUD
include those normally seen with insertion: pain,
spotting, and a small increase in risk of pelvic
infection in the first 20 days after placement. An
emergency IUD can be removed after the next
menstrual cycle without risk of pregnancy pro-
vided the woman has not had sexual intercourse
since menses or if in addition to the IUD,
hormonal contraception was started within the
first 5 days of that cycle.3,15
Availability
In the United States, access to EC has faced
multiple barriers. Studies have shown that many
women and health care providers have insuffi-
cient knowledge about EC and how it works.39,40
Despite the approval by the FDA of 2 dedicated
EC products, Preven (now off the market) and
Plan B, use remains limited. Access is also limited
by difficulties obtaining prescriptions for those
women who require or request one, unavailability
of EC pills at local pharmacies, and the refusal of
some pharmacists to dispense EC.41 Before the
change of Plan B to over-the-counter status,
organizations such as Planned Parenthood began
establishing programs using telephone and online
screening to streamline EC prescribing practices,
avoid an office visit, and to increase access to EC.
Some states (AK, CA, HI, ME, MA, MT, NH,
NM, VT, and WA) passed legislation to allow
pharmacists to dispense EC directly without a
prescription. Not all pharmacies stock Plan B or
participate in this program (see www.ec-help.org/
PharmacyLocations.asp or www.not-2-late.com
for a list of participating pharmacies).
One concern often voiced about increased
availability of EC is that it might encourage
imprudent sexual behaviors and reduce the use
of regular contraception.42 These concerns have
not been supported in multiple studies.5,43 Trials
comparing the advance provision of EC to the
regular practice of contacting a clinician if EC
was needed have shown no difference between
groups in self-reported incidence of unprotected
intercourse or use of regular contraception. All
trials showed that having an advance prescrip-
tion of EC resulted in increased and often earlier
use.43,44 Studies have also demonstrated that
advance provision of EC and its accompanying
contraceptive education results in better adher-
ence to regular contraception.44
Many countries allow the sale of LNG-only
EC without a prescription, and no adverse events
have been reported.45 In June 2003, the FDA
agreed to review Plan B for over-the-counter sale.
 Emergency Contraception 933

TABLE 3. Initiating Ongoing Contraception After EC Use*

Because oral EC pills can delay ovulation, a woman could be at risk of pregnancy in the first few days after treatment. Women
should use an effective method of contraception for the remainder of the treatment cycle and thereafter3

Method When to Initiate

Condom Can be used immediately

Diaphragm or cervical Can be used immediately
cap
Spermicide Can be used immediately
Combined oral Begin taking 1 tablet daily the day after EC treatment is completed, use condoms along with
contraceptive pill the pills for the first 7 d, or initiate a new pack within 5 d of beginning the next menstrual
period
Contraceptive patch Initiate the day after EC treatment is completed, use condoms along with the patch for the first
7 d, or initiate within 5 d of beginning the next menstrual period
Vaginal ring Initiate the day after EC treatment is completed, use condoms along with the ring for the first
7 d, or initiate within 5 d of beginning the next menstrual period
Progestin-only pill Begin taking 1 tablet daily the day after EC treatment is completed, use condoms along with
progestin-only pills for the first 7 d, or initiate a new pack within 5 d of beginning the next
menstrual period
Injectables The day after EC treatment is completed, check a pregnancy test and if negative, initiate
injection, use condoms along with the injectable for the first 7 d, and check a repeat
pregnancy test 2 to 3 wk after the injection,3 or initiate within 5 d of beginning the next
menstrual period
Implants Initiate within 5 d of beginning the next menstrual period
IUD Consider inserting a copper-releasing IUD for EC for any woman who intends to use a copper
IUD for ongoing contraception, or initiate use during the next menstrual period
Fertility awareness Initiate after onset of the next normal menstrual period and after the patient has been trained
in using the method
Female sterilization Perform the operation any time after beginning the next menstrual period

* Program for Appropriate Technology in Health (PATH). Resources for Emergency Contraceptive Pill Programming:
A Toolkit. Seattle, WA: PATH, 2004.

After an unusually long process, the FDA ap-
proved over-the-counter status for Plan B on
August 24, 2006, only for women age 18 and
older.46,47 Women age 17 and younger must
either obtain Plan B with a prescription or with-
out a prescription in states with direct pharmacy
access. Both men and women who can show
proof of age with a government-issued identifica-
tion can purchase Plan B over-the-counter. Be-
cause the same packaging contains both the
prescription and nonprescription form of Plan
B, it must be kept where customers do not have
direct access to the product (behind the counter),
although it can be fully visible to them.47 The
average wholesale price of Plan B is $31/unit,48
however, for public clinics, the price will be
substantially lower. Because insurance plans vary
in their coverage of over-the-counter medica-
tions, women should check with their insurance
provider about coverage.
The controlled availability of Plan B over-the-
counter in the United States is only a partial
solution to access. The fact that Plan B will be
stored behind the counter poses barriers to access
owing to lack of privacy, because women may be
embarrassed to ask a pharmacist for the drug.42
The requirement to show a government-issued ID
may also impose a burden on undocumented
women and women without proper identifica-
tion. Furthermore, there is no scientific or med-
ical reason to impose an age restriction or to
withhold EC from adolescents.49 The age cutoff
of 18 was chosen not for a medical reason but
simply because it was easier for pharmacists
to remember and enforce because it is similar to
restrictions on nicotine-containing products. As
shown in many other countries, EC is safe for
over-the-counter use by women of all ages.
Given the limitations of over-the-counter
access in the United States, prescriptions may
make it easier for women to obtain Plan B if they
do not have a government-issued ID, if they are
embarrassed to ask the pharmacist at the coun-
ter for Plan B, or if they want to avoid speaking
to the pharmacist about their intended use of
EC.50 Therefore, it is important to continue to
provide advance prescriptions of Plan B to all
women when they present for care. At the same
time, women should be encouraged to use a
regular method of contraception and not rely
934 Allen and Goldberg
on repeated dosing of EC. Providers should also
make women aware of the emergency contra-
ceptive properties of the Copper T 380A IUD.
To date, research has shown that although
increased access to EC improves use, it does not
seem to reduce the rates of unintended preg-
nancy or abortion.51 Although access is an im-
portant factor, how women use EC may be a
stronger determinant of its ultimate effect. Stu-
dies show that even when women have EC at
home, they often fail to use it after unprotected
sexual intercourse. The most common reason for
this is a lack of recognition of the risk of preg-
nancy or a neglect of the perceived risk.51 In sum,
there are a myriad of factors that influence EC
use in the real world. Given these realities, the
effect that widespread access to EC would have
on reducing unintended pregnancy and abortion
rates may have been overestimated.
Conclusions
In sum, EC reduces the chance of pregnancy
when used correctly, up to 5 days after an
identified episode of contraceptive failure or
unprotected sexual intercourse. The most com-
mon regimens used in the United States are
LNG-only EC (Plan B) and the Copper T
380A IUD (ParaGard). In the United States,
given limitations in access, advance prescrip-
tions for Plan B are still recommended for
women 18 and older despite its over-the-counter
status. The recent labeling changes of the Para-
Gard IUD, as suitable for use in nulliparous
women, might increase its use as an emergency
contraceptive. Although increasing knowledge
of and access to EC is a public health priority,
the use of regular contraception must also be
emphasized. The worldwide experience to date
has not shown that improved access to EC has a
population level effect of reduced unintended
pregnancies. This is mainly because of the fact
that a low proportion of women with access to
EC actually take the pills to prevent pregnancy.
Thus, along with increased access, more effort
toward educating women about how to use EC
and encouraging its use is required if there is to
be a demonstrable population level effect.
References
1. Finer LB, Henshaw SK. Disparities in rates of
unintended pregnancy in the United States,
1994 and 2001. Perspect Sex Reprod Health.
2006;38:90–96.
2. Cheng L, Gulmezoglu AM, Van Oel CJ, et al.
Interventions for emergency contraception.
Cochrane Database Syst Rev. 2004:CD001324.
3. Stewart F, Trussell J, Van Look PF. Emer-
gency contraception, 18th revised edtion. In:
Hatcher RA, Trussell J, Stewart F, et al, eds.
Contraceptive Technology. New York: Ardent
Media Inc; 2004.
4. Reader FC. Emergency contraception. BMJ.
1991;302:801.
5. ACOG Practice Bulletin. Clinical management
guidelines for obstetrician-gynecologists, num-
ber 69, December 2005 (replaces Practice Bul-
letin Number 25, March 2001). Emergency
contraception. Obstet Gynecol. 2005;106:
1443–1452.
6. Arowojolu AO, Okewole IA, Adekunle AO.
Comparative evaluation of the effectiveness
and safety of two regimens of levonorgestrel
for emergency contraception in Nigerians.
Contraception. 2002;66:269–273.
7. von Hertzen H, Piaggio G, Ding J, et al. Low
dose mifepristone and two regimens of levo-
norgestrel for emergency contraception: a
WHO multicentre randomised trial. Lancet.
2002;360:1803–1810.
8. Ngai SW, Fan S, Li S, et al. A randomized trial
to compare 24 h versus 12 h double dose regi-
men of levonorgestrel for emergency contra-
ception. Hum Reprod. 2005;20:307–311.
9. Yuzpe AA. Postcoital contraception. Int J
Gynaecol Obstet. 1978;16:497–501.
10. Trussell J, Ellertson C, Stewart F, et al. The
role of emergency contraception. Am J Obstet
Gynecol. 2004;190:S30–S38.
11. International Consortium for Emergency Con-
traception. Emergency Contraceptive Pills:
Medical and Service Delivery Guidelines. 2nd
ed. Washington, DC: International Consor-
tium for Emergency Contraception; 2004.
12. Not-2-Late.Com: The Emergency Contracep-
tion Website. Dedicated emergency contracep-
tion products worldwide. Available at: http://
ec.princeton.edu/questions/dedicated.html.
Accessed on: February 7, 2007.
13. Rodrigues I, Grou F, Joly J. Effectiveness
of emergency contraceptive pills between
72 and 120 hours after unprotected sexual
intercourse. Am J Obstet Gynecol. 2001;184:
531–537.
14. Piaggio G, von Hertzen H, Grimes DA, et al.
Timing of emergency contraception with levo-
norgestrel or the Yuzpe regimen. Task Force
on Postovulatory Methods of Fertility Regula-
tion. Lancet. 1999;353:721.
15. FFPRHC Guidance (April 2006). Emergency
contraception. J Fam Plann Reprod Health
Care. 2006;32:121–128.

16. Zhou L, Xiao B. Emergency contraception
with Multiload Cu-375 SL IUD: a multicenter
clinical trial. Contraception. 2001;64:107–112.
17. Croxatto HB, Devoto L, Durand M, et al.
Mechanism of action of hormonal prepara-
tions used for emergency contraception: a re-
view of the literature. Contraception. 2001;63:
111–121.
18. Office for Human Research Protections. Addi-
tional Protections for Pregnant Women, Human
Fetuses, and Neonates Involved in Research
Part 46 Protection of Human Subjects, Subpart
B, Section 46.202. Washington, DC: United
States Department of Health and Human
Services; 2005.
19. Davidoff F, Trussell J. Plan B and the politics
of doubt. JAMA. 2006;296:1775–1778.
20. Gemzell-Danielsson K, Marions L. Mechan-
isms of action of mifepristone and levonorges-
trel when used for emergency contraception.
Hum Reprod Update. 2004;10:341–348.
21. Brito KS, Bahamondes L, Nascimento JA,
et al. The in vitro effect of emergency contra-
ception doses of levonorgestrel on the acro-
some reaction of human spermatozoa.
Contraception. 2005;72:225–228.
22. Trussell J, Hedley A, Raymond E. Ectopic
pregnancy following use of progestin-only
ECPs. J Fam Plann Reprod Health Care.
2003;29:249.
23. Stanford JB, Mikolajczyk RT. Mechanisms of
action of intrauterine devices: update and esti-
mation of postfertilization effects. Am J Obstet
Gynecol. 2002;187:1699–1708.
24. Wilcox AJ, Dunson DB, Weinberg CR, et al.
Likelihood of conception with a single act of
intercourse: providing benchmark rates for
assessment of post-coital contraceptives. Con-
traception. 2001;63:211–215.
25. Stirling A, Glasier A. Estimating the efficacy of
emergency contraception—how reliable are
the data? Contraception. 2002;66:19–22.
26. Raymond E, Taylor D, Trussell J, et al.
Minimum effectiveness of the levonorgestrel
regimen of emergency contraception. Contra-
ception. 2004;69:79–81.
27. Jones RK, Darroch JE, Henshaw SK. Contra-
ceptive use among US women having abor-
tions in 2000–2001. Perspect Sex Reprod
Health. 2002;34:294–303.
28. Lakha F, Glasier A. Unintended pregnancy
and use of emergency contraception among a
large cohort of women attending for antenatal
care or abortion in Scotland. Lancet. 2006;
368:1782–1787.
29. Moreau C, Bouyer J, Goulard H, et al. The
remaining barriers to the use of emergency
contraception: perception of pregnancy risk
Emergency Contraception 935
by women undergoing induced abortions.
Contraception. 2005;71:202–207.
30. Grimes DA, Raymond EG. Emergency con-
traception. Ann Intern Med. 2002;137:180–189.
31. Program for Appropriate Technology in
Health. Resources for Emergency Contracep-
tive Pill Programming: A Toolkit. Seattle, WA:
PATH; 2004.
32. World Health Organization. Medical Eligibil-
ity Criteria for Contraceptive Use. 3rd ed.
Geneva: WHO; 2004.
33. Vasilakis C, Jick SS, Jick H. The risk of venous
thromboembolism in users of postcoital
contraceptive pills. Contraception. 1999;59:
79–83.
34. Ellison J, Thomson AJ, Greer IA, et al. Drug
points: apparent interaction between warfarin
and levonorgestrel used for emergency contra-
ception. BMJ. 2000;321:1382.
35. Task Force on Postovulatory Methods of Fer-
tility Regulation. Randomised controlled trial
of levonorgestrel versus the Yuzpe regimen of
combined oral contraceptives for emergency
contraception. Lancet. 1998;352:428–433.
36. Ho PC, Kwan MS. A prospective randomized
comparison of levonorgestrel with the Yuzpe
regimen in post-coital contraception. Hum
Reprod. 1993;8:389–392.
37. Alvarez F, Faundes A, Johansson E, et al.
Blood levels of levonorgestrel in women fol-
lowing vaginal placement of contraceptive
pills. Fertil Steril. 1983;40:120–123.
38. Raymond EG, Goldberg A, Trussell J, et al.
Bleeding patterns after use of levonorgestrel
emergency contraceptive pills. Contraception.
2006;73:376–381.
39. van Royen AR, Calvin CK, Lightner CR.
Knowledge and attitudes about emergency
contraception in a military population. Obstet
Gynecol. 2000;96:921–925.
40. Golden NH, Seigel WM, Fisher M, et al.
Emergency contraception: pediatricians’
knowledge, attitudes, and opinions. Pediatrics.
2001;107:287–292.
41. Trussell J, Duran V, Shochet T, et al. Access to
emergency contraception. Obstet Gynecol.
2000;95:267–270.
42. Wynn LL, Trussell J. Images of American
sexuality in debates over nonprescription ac-
cess to emergency contraceptive pills. Obstet
Gynecol. 2006;108:1272–1276.
43. Gold MA, Wolford JE, Smith KA, et al. The
effects of advance provision of emergency con-
traception on adolescent women’s sexual and
contraceptive behaviors. J Pediatr Adolesc
Gynecol. 2004;17:87–96.
44. Jackson RA, Schwarz EB, Freedman L, et al.
Advance supply of emergency contraception.
936 Allen and Goldberg
Effect on use and usual contraception—a ran-
domized trial. Obstet Gynecol. 2003;102:
8–16.
45. Grimes DA, Raymond EG, Scott Jones B.
Emergency contraception over-the-counter:
the medical and legal imperatives. Obstet
Gynecol. 2001;98:151–155.
46. United States Government Accountability Of-
fice. Food and Drug Administration: Decision
process to deny initial application for over-the-
counter marketing of emergency contraceptive
drug Plan B was unusual, 2005.
47. Food and Drug Administration. Plan B: Ques-
tions and Answers. Available at: http://
www.fda.gov/cder/drug/infopage/planB/plan
BQandA20060824.htm. Accessed on: Febru-
ary 7 2007.
48. Thomson Healthcare Inc. Red Book. Mon-
tvale, NJ: Thomson PDR; 2006.
49. American Academy of Pediatrics. Emergency
contraception. Pediatrics. 2005;116:1026–1035.
50. Physicians for Reproductive Choice and
Health. Emergency Contraception: A practi-
tioner’s guide. Available at: http://www.
prch.org/med_ed/ECguide/. Accessed on:
February 7 2007.
51. Raymond EG, Trussell J, Polis CB. Population
effect of increased access to emergency contra-
ceptive pills: a systematic review. Obstet
Gynecol. 2007;109:181–188.