Customize your JAMA Network experience by selecting one or more topics from the list below.
Audet M, Moreau M, Koltun WD, et al. Evaluation of Contraceptive Efficacy and Cycle Control of a Transdermal Contraceptive Patch vs an Oral Contraceptive: A Randomized Controlled Trial. JAMA. 2001;285(18):2347–2354. doi:10.1001/jama.285.18.2347
Context Oral contraceptive (OC) pills are effective, but poor compliance increases
rates of pregnancy during treatment.
Objective To compare the contraceptive efficacy, cycle control, compliance, and
safety of a transdermal contraceptive patch and an OC.
Design Randomized, open-label, parallel-group trial conducted October 1997
to June 1999.
Setting Forty-five clinics in the United States and Canada.
Participants A total of 1417 healthy adult women of child-bearing potential.
Interventions Participants were randomly assigned to receive a transdermal contraceptive
patch (n = 812) vs an OC (n = 605) for 6 or 13 cycles. Patch treatment consisted
of application of 3 consecutive 7-day patches followed by 1 patch-free week.
Main Outcome Measures Overall and method-failure Pearl Indexes (number of pregnancies/100
person-years of use) and life-table estimates of the probability of pregnancy
were calculated. Cycle control, compliance, patch adhesion, and adverse events
were also assessed.
Results Overall and method-failure Pearl Indexes were numerically lower with
the patch (1.24 and 0.99, respectively) vs the OC (2.18 and 1.25, respectively);
this difference was not statistically significant (P
= .57 and .80, respectively). The incidence of breakthrough bleeding and/or
spotting was significantly higher only in the first 2 cycles in the patch
group, but the incidence of breakthrough bleeding alone was comparable between
treatments in all cycles. The mean proportion of participants' cycles with
perfect compliance was 88.2% (811 total participants, 5141 total cycles) with
the patch and 77.7% (605 total participants, 4134 total cycles) with the OC
(P < .001). Only 1.8% (300/16 673) of patches
completely detached. Both treatments were similarly well tolerated; however,
application site reactions, breast discomfort, and dysmenorrhea were significantly
more common in the patch group.
Conclusion The contraceptive patch is comparable to a combination OC in contraceptive
efficacy and cycle control. Compliance was better with the weekly contraceptive
patch than with the OC.
Worldwide more than 100 million women choose hormonal contraception
for family planning,1 with more than 12 million
users in the United States alone.2,3
Combined oral contraceptives (OCs) are widely used because of the efficacy
demonstrated in clinical trials and the established safety from postmarketing
surveillance.4 However, while clinical trials
have shown that correct and consistent use of OCs results in a first-year
failure rate of 0.1%,5 the 1995 National Survey
of Family Growth (US data) estimated that actual first-year failure rates
during typical use of OCs range from 7.3% to as high as 8.5%.6
Noncompliance is the primary reason cited to explain this difference.7-9 There is clearly a need
for reversible contraceptives with a more convenient dosing schedule that
would enhance patient compliance and achieve high contraceptive efficacy in
The transdermal contraceptive patch has been evaluated as a new method
of contraception in several trials. Pharmacokinetic data from studies involving
the contraceptive patch, which delivers 150 µg of norelgestromin and
20 µg of ethinyl estradiol daily to the systemic circulation,10 have been published.11-14
This study is the first published clinical trial of transdermal contraception.
The objective of this study was to compare the transdermal contraceptive patch
and an OC in terms of contraceptive efficacy, cycle control, compliance, and
safety. The a priori hypotheses were that the contraceptive patch would be
comparable to the OC with respect to contraceptive efficacy, cycle control,
and safety but better than the OC with respect to compliance.
This study was a randomized, active control, multicenter clinical trial
comparing the contraceptive patch with an OC. The study was conducted at 39
centers in the United States and 6 centers in Canada from October 1997 to
June 1999. The protocol was approved by the institutional review board or
ethics committee at each center, and the study was conducted in compliance
with the regulations governing good clinical practice. All sites were visited
regularly during the trial by study monitors to ensure adherence to study
procedures. After giving informed consent, 1400 participants were to be randomized
in a ratio of 4:3 to receive either the contraceptive patch (ORTHO EVRA/EVRA,
a 20-cm2 patch designed to deliver 150 µg of norelgestromin
and 20 µg of ethinyl estradiol daily to the systemic circulation10; The R.W. Johnson Pharmaceutical Research Institute,
Raritan, NJ) or the OC (Triphasil, levonorgestrel 50 µg and ethinyl
estradiol 30 µg days 1 to 6, levonorgestrel 75 µg and ethinyl
estradiol 40 µg days 7 to 11, levonorgestrel 125 µg and ethinyl
estradiol 30 µg days 12 to 21, and placebo for days 22 to 28; Wyeth-Ayerst
Laboratories, Radnor, Pa). To facilitate study completion, the first third
of participants were enrolled for 13 cycles, and the remaining participants
for 6 cycles. Patch treatment was 3 consecutive 7-day patches followed by
1 patch-free week. Oral contraceptive treatment was administered according
to the US physician package insert.
An interactive voice-activated randomization system was used to randomize
participants to treatment within centers by permuted blocks. Randomization
was performed by The R.W. Johnson Pharmaceutical Research Institute.
A total of 1417 healthy women aged 18 to 45 years were randomized and
received the study drug. Participants had to be sexually active and at risk
of pregnancy; have regular menstrual cycles; have at least 1 normal menses
since removal of an intrauterine device or progestin implant (Norplant) if
applicable; have a negative serum β-human chorionic gonadotropin pregnancy
test at screening and a negative urine pregnancy test at admission; have seated
systolic/diastolic blood pressure less than 140/90 mm Hg; be within 35% of
ideal body weight; agree to use only the assigned study drug as contraception
and not to use any other steroid hormonal therapy other than topical corticosteroids;
and provide written informed consent. Exclusion criteria included lactation
or pregnancy within 42 days of study admission; any disorders that were contraindications
to steroid hormonal therapy; uncontrolled thyroid disorder; Papanicolaou test
evidence of squamous intraepithelial lesions or adenocarcinoma or other malignancy;
history or presence of dermal hypersensitivity in response to topical application;
smoking in women older than 35 years; alcohol or substance abuse within 12
months of screening; receipt of injectable progestin (Depo-Provera) within
6 months of screening; and receipt of any experimental drug, device, or hepatic
enzyme-inducing drug within 30 days of screening.
The matrix patch is thin and consists of 3 layers: an outer protective
layer of polyester; a medicated, adhesive middle layer; and a clear, polyester
release liner that is removed prior to patch application. Participants in
the patch group were instructed to apply the patch to the buttocks, upper
outer arm, lower abdomen, or upper torso (excluding the breast). New patches
could be applied to sites near the patch that was removed, but not to the
same site as the preceding patch. Participants could maintain their usual
activities, including bathing and swimming, while wearing the patch, but were
instructed not to apply oils, creams, or cosmetics on or around the area of
patch placement. Participants were instructed to apply the patch on the same
day of each week (1 patch per week) for 3 consecutive weeks followed by a
patch-free week. Patches were to adhere on their own, and no supplemental
tape or adhesive was permitted. In the event of detachment, patches could
be immediately reapplied. Patches that did not fully adhere on their own were
to be replaced. The replacement patch would then be worn for the remainder
of that week.
Sample Size Determination. The sample size was chosen to provide 400 person-years of observation
for the contraceptive patch. This was achieved among participants who were
enrolled for either 13 cycles or 6 cycles. Current regulatory guidance15 specifies that studies should be at least large enough
to provide an overall Pearl Index with a 95% confidence interval (CI) so that
the difference between the upper limit of the CI and the point estimate does
not exceed 1 pregnancy per 100 person-years. For a Pearl Index of 1, this
is achieved with 400 person-years.
The primary end point for cycle control was breakthrough bleeding and/or
spotting during cycle 3. It was anticipated that 11% of the participants in
the contraceptive patch group would have breakthrough bleeding and/or spotting
during cycle 3. The study was designed with a 4:3 ratio of participants using
the patch to the OC, so that a difference between treatment groups of 6% (17%
of participants in the OC group) could be detected at the .05 level of significance
with power of 0.82, assuming available data for 80% of the participants enrolled.
Contraceptive Efficacy. The primary outcome measure for this trial was contraceptive efficacy.
The criteria for evaluating contraceptive efficacy included calculation of
the Pearl Index (number of pregnancies per 100 person-years of use) and life-table
estimates of the probability of pregnancy. Each participant was to have a
urine pregnancy test performed 10 days after the scheduled termination cycle
or on early withdrawal. All pregnant participants were to have an ultrasound.
Pregnancies were classified into the following 4 categories based on available
data from the investigators (ultrasound, diary cards, and narratives): (1)
pretherapy, in which the estimated date of conception preceded start of the
study drug; (2) posttherapy, in which the estimated date of conception was
after the last cycle of therapy; (3) method failure during therapy, in which
the estimated date of conception was during the cycles of therapy and there
was information that the participant complied with dosing; and (4) user failure
during therapy, defined as in number 3 except that available information indicated
that the participant failed to comply with dosing on, or immediately contiguous
with, the date of conception. Follow-up information regarding the pregnancy
outcome and any postnatal sequelae in the infant were obtained in all cases.
All participants who did not return for a scheduled visit were to be contacted
first by telephone and then by certified letter requesting study data. Only
participants who failed to respond to telephone calls and the certified letter
were considered lost to follow-up.
Cycle Control. Cycle control was evaluated from information recorded on diary cards,
which were completed daily by participants. Breakthrough bleeding and spotting
was any bleeding and spotting occurring on days 1 through 21, excluding bleeding
contiguous with menses. Breakthrough bleeding was defined as requiring sanitary
protection of more than 1 pad or tampon on any of those days. Amenorrhea was
defined as 2 continuous cycles without any bleeding or spotting.
Compliance. Compliance was determined by daily dosing (and patch replacement) noted
on diary cards and included all cycles in which adequate dosing information
was available. Perfect compliance was defined as 21 consecutive days of drug
taking, which could have included the use of replacement patches, followed
by a 7-day drug-free interval. For patch users, no patch could be worn for
more than 7 days.
Patch Adhesion. Patch replacement information was used to assess patch adhesion. The
percentage of patches replaced for the reason of "fell off" was summarized
as patches that completely detached due to lack of adhesion. The percentage
of patches replaced because of partial detachment was also summarized.
Safety. Adverse events, both those reported by participants and those observed
by study center personnel, were documented throughout the study. Clinical
laboratory tests (hematology and serum chemistry), vital signs (blood pressure,
pulse rate, and temperature), and physical and gynecologic examinations were
performed prestudy and at the final visit.
Pregnancy rates were estimated by Pearl Index and life-table analyses
and were based on all pregnancies occurring during therapy and the total number
of cycles during therapy through cycle 13, with data censored after the cycle
of conception. The Pearl Index was calculated as [(number of pregnancies ×
1300)/number of cycles during therapy)]. Under a modified intent-to-treat
analysis, all participants who received the study drug for at least 1 day
and were not pregnant at the start of cycle 1 were included in the evaluation
of efficacy. The overall Pearl Index includes all pregnancies in the modified
intent-to-treat population. The method-failure Pearl Index, which included
only those pregnancies in which there was not a concurrent dosing error, was
also calculated. Kaplan-Meier estimates of the probability of pregnancy at
cycles 6 and 13 were used to summarize the results by treatment group. Estimates
were obtained at cycle 6 since all participants were to complete this cycle.
Participants who had not become pregnant were censored at their last cycle
on treatment. Treatment group comparisons were conducted using the log-rank
test. For both Pearl Indexes and life-table estimates of the probability of
pregnancy, 95% CIs were calculated for each treatment group. The 95% CI for
the Pearl Index was derived using a variance obtained by the delta method.16
A key end point for the evaluation of cycle control was the incidence
of breakthrough bleeding and/or spotting during cycle 3. Treatment differences
were evaluated by a χ2 test to compare the percentage of participants
with the event. Supportive cycle control data included an evaluation of breakthrough
spotting, breakthrough bleeding, and amenorrhea over all cycles.
To compare the 2 treatment groups with regard to the proportion of each
participant's cycles in which there was perfect compliance, a t test was used in which the participant was the experimental unit.
Although not directly compared due to the correlation of results across cycles
within an individual, compliance rates across cycles were calculated.
For the most frequently reported adverse events, the Fisher exact test
was used to compare the proportion of participants with the event and the
proportion of participants in which it was a treatment-limiting event.
For vital signs and hematology and chemistry data, the mean changes
at the last visit from baseline were analyzed using the t test.
All participants who took the study drug were evaluable for safety.
Safety was assessed by the following: adverse events, mean changes from baseline
in hematology and serum chemistry parameters and vital signs, and changes
in physical and gynecologic examinations.
The demographics of the 2 treatment groups were comparable (Table 1). Approximately 22% of the participants
had not used OCs within the past 2 months, and 55% switched directly from
OCs to their assigned study treatment (ie, were using OCs in the cycle prior
to study start); 8% switched indirectly (ie, within 2 months, but not in the
cycle immediately prior to start) from OCs; and for 15%, data were not provided. Figure 1 shows the allocation of participants
following randomization and includes the number distributed to 6 and 13 cycles
of study drug use. There were 812 participants treated with the patch, and
70% completed the study; 605 participants were treated with the OC, and 76%
completed the study. One participant in the patch group was excluded from
the efficacy analyses because she had a pretreatment pregnancy and began using
the study drug in violation of the protocol. All 1417 participants treated
were included in the safety analyses.
The overall and method-failure Pearl Indexes were numerically lower
in the patch group (1.24 and 0.99, respectively) than in the OC group (2.18
and 1.25, respectively), although the differences between the treatments were
not statistically significant (P = .57 and P = .80, respectively) (Table 2). In the patch group, 4 method-failure pregnancies and 1
user-failure pregnancy occurred among 811 women treated for 5240 cycles. In
the OC group, 4 method-failure and 3 user-failure pregnancies occurred among
605 women treated for 4167 cycles.
The life-table analyses indicated that the probability of pregnancy
through 6 or 13 cycles was also lower with the patch than with the OC (Table 2).
Table 3 provides a profile
of each of the 12 pregnancies that occurred during therapy.
The incidence of breakthrough bleeding is presented for representative
cycles in Figure 2A. There were
no statistically significant differences between the patch and the OC with
respect to breakthrough bleeding at any cycle. The incidence of breakthrough
bleeding and/or spotting is presented for representative cycles in Figure 2B. For cycles 1 and 2, the patch
group had significantly higher rates of breakthrough bleeding and/or spotting,
but in subsequent cycles the 2 groups did not differ signicantly. Amenorrhea
occurred in 0.1% of patch users and 0.2% of OC users.
Compliance with the dosing schedule of the patch was better than that
of the OC. The mean proportion of each participant's cycles that demonstrated
perfect compliance was 88.2% (811 total participants, 5141 total cycles) with
the patch and 77.7% (605 total participants, 4134 total cycles) with the OC
(P<.001). Overall, the percentage of cycles in
which there was perfect compliance was 88.7% (4558/5141) with the patch and
79.2% (3276/4134) with the OC.
A total of 4.6% of all patches were replaced for either complete (1.8%
[300/16 673]) or partial (2.8% [470/16 673]) detachment.
There were no unexpected adverse events with either treatment. The most
frequent adverse events and the percentage of participants who discontinued
therapy for each adverse event are listed in Table 4. With the exception of mild-to-moderate application site
reactions in the patch group, the types of adverse events were similar in
the 2 treatment groups. The frequency of application site reactions did not
increase over time. The overall occurrence of breast discomfort was higher
for the patch than for the OC. This difference was significant only in cycles
1 and 2 (15.4% vs 3.5% in cycle 1 [P<.001] and
6.6% vs 1.5% in cycle 2 [P<.001]). For cycles
3 to 13, the by-cycle incidence of breast discomfort was not significantly
different between treatments (P≥.10 for cycles 3-13; 3.2% to 0% vs 1.7%
to 0%, respectively). For those participants with breast discomfort, 85% rated
the discomfort as mild-to-moderate in severity. Dysmenorrhea was also more
frequent in the patch group. The most common adverse events causing treatment
discontinuation were application site reactions, nausea, headache, dysmenorrhea,
and breast discomfort. Headache and dysmenorrhea, although infrequent causes
of discontinuation, were significantly more likely to be treatment limiting
in the patch group. For both the patch and OC, approximately 2% of participants
(patch [16/812], OC [11/605]) reported a serious adverse event, and these
events were generally similar between treatments. The serious adverse events
that occurred with the patch included injury (3 participants), cholecystitis
(2), abdominal pain (3), pyelonephritis (1), nausea/vomiting/pharyngitis (1),
pain (1), migraine (1), dehydration (1), sleep disorder (1), diabetes mellitus
(1), and manic-depressive psychosis (1). The serious adverse events that occurred
with the OC included injury (3 participants), cholecystitis (1), abdominal
pain (1), pyelonephritis (2), pharyngitis (1), pelvic inflammatory disease
(1), intracranial hypertension (1), and depression/suicide (1). Three serious
adverse events were considered possibly or likely related to the use of the
patch, including 1 case of pain and paraesthesia in the left arm, 1 case of
migraine, and 1 case of cholecystitis. All 3 of these events resolved, although
the cholecystitis required cholecystectomy. Two serious adverse events were
considered possibly related to the use of the OC, and they included increased
intracranial pressure, which did not resolve, and severe depression leading
There were no clinically meaningful changes in laboratory parameters,
vital signs, or physical or gynecologic examination findings, although total
cholesterol concentrations increased from baseline to the end of therapy by
a mean of 15.8 mg/dL (0.41 mmol/L) and 8.1 mg/dL (0.21 mmol/L) in the patch
and OC groups, respectively (P < .001). Triglyceride
levels increased more in the patch group (9.7 mg/dL [0.11 mmol/L]) than in
the OC group (0.9 mg/dL [0.01 mmol/L]) (P = .008).
In both treatment groups, mean body weight increased from baseline to the
end of therapy by 0.41 kg.
Compliance problems with OCs have been well documented. For example,
in a compliance study by Rosenberg and colleagues,7
almost 50% of OC users reported missing 1 or more pills per cycle, with 22%
missing 2 or more pills. In another compliance study, an electronic device
that monitored the time and date of pill removal from the container indicated
that 30% to 51% of women skipped 3 or more dosing days per cycle.9 While many factors contribute to less than perfect
compliance with daily OC dosing, women who do not have a regular pill-taking
routine and those who do not read and understand the dosing instructions are
at greatest risk for poor compliance.8 Contraceptive
compliance is clearly related to contraceptive efficacy. United States prescribing
information for contraceptive products summarizes failure rates for all methods
of contraception from National Survey of Family Growth data. Long-acting hormonal
contraceptives that require infrequent patient activity toward compliance
(eg, progestin implants, 3-month progestin injectables) have reduced contraceptive
failures associated with noncompliance. This can be seen by the small difference
between the "perfect-use" (method) failure rate and "typical-use" (method
failure plus user failure) failure rate.17
The contraceptive efficacy of the patch in this randomized clinical
trial was compared with that of an OC. The "overall" Pearl Index and probability
of pregnancy through 13 cycles were numerically lower for the contraceptive
patch compared with the OC, but the sample size was relatively small and this
difference was not statistically significant. The numerically lower overall
failure rate for the contraceptive patch may be due to the better compliance
shown with this once-weekly dosing regimen vs the daily dosing for the OC.
A significant difference in compliance with the patch vs an OC has been reported
in another randomized clinical trial.18 Evidence
that better compliance may have played a role in the lower overall pregnancy
rate for the contraceptive patch can be found by comparing the overall and
method failure rates for the patch and the OC, similar to comparing the typical-use
(overall) and perfect-use (method) failure rates for long-acting hormonal
methods. The method failure rate, which excludes pregnancies associated with
noncompliance, is similar for the contraceptive patch and the OC (0.99 vs
1.25). A contraceptive with high compliance would be associated with an overall
failure rate (method failure plus user failure) that is very similar to the
method failure rate alone, as seen with the contraceptive patch (1.24 vs 0.99).
A contraceptive with lower compliance would have a greater difference between
the overall failure rate and the method failure rate due to more user failures,
as seen with the OC (2.18 vs 1.25). Additional evidence that better compliance
contributed to the low probability of pregnancy with the contraceptive patch
is seen from analysis of 6-cycle data, a treatment duration that all randomized
participants were to complete. After 6 cycles, the overall probability of
pregnancy with the new transdermal contraceptive was half that of the OC (0.6%
vs 1.2%) while the method failure probability of pregnancy was nearly the
same (0.4% vs 0.6%); however, neither of these differences was statistically
significant. Given that at least 63% (891/1417) of the participants in this
trial (direct plus indirect switchers combined) were experienced OC users,
the significantly higher compliance shown for the transdermal contraceptive
method (even in experienced OC users) could result in lower typical-use failure
rates (as reported by the National Survey of Family Growth6)
than seen with the OC comparator.
The incidence of unexpected vaginal bleeding with a hormonal contraceptive
is commonly reported as a reason for discontinuation of a particular method.8 Currently, the only available hormonal methods that
require less frequent dosing than OCs contain only progestin, and progestin-only
methods are associated with frequent episodes of unexpected bleeding when
compared with combination estrogen-progestin methods.19,20
In this trial, the contraceptive patch had an incidence of breakthrough bleeding
that was comparable to the OC at all cycles. The incidence of breakthrough
bleeding and/or spotting was significantly higher among patch users only during
the first 2 cycles.
The rate of adhesion of a contraceptive patch will also contribute to
contraceptive efficacy, cycle control, and patient satisfaction. A total of
4.6% of all patches in this study were replaced for either complete (1.8%
[300/16673]) or partial (2.8% [470/16673]) detachment.
Mild-to-moderate application site reactions occurred in the patch group,
and 2.6% of participants withdrew from the study as a result. Several dermal
studies have been conducted with the contraceptive patch.21,22
In these dermal studies, the patch was not associated with either phototoxicity
or photoallergy and demonstrated only mild skin irritation in a minority of
participants. No other new or unexpected adverse reactions were observed.
The incidence of mild-to-moderate breast discomfort was higher with the patch
than the OC, only in the first 2 cycles, but the discontinuation rate for
breast discomfort was not significantly different between the patch and the
OC. Both the contraceptive patch and the OC led to a mean increase in total
cholesterol level. In another study, the patch did not significantly change
the low-density lipoprotein/high-density lipoprotein ratio when compared with
The current study design has potential limitations. A double-dummy design
to blind the treatments to the participants and to the investigators was considered
but not used, since the use of back-up contraception following placebo dosing
errors would compromise the assessment of efficacy in the study. The study
was not designed to detect differences in contraceptive efficacy. There was
a slight imbalance in the percentage of participants who withdrew or were
lost to follow-up (29.7% in the patch group and 24.3% in the OC group).
The transdermal combination hormonal contraceptive patch uses weekly
dosing to complete a 21-day regimen followed by 1 dose-free week. The weekly
dosing was associated with significantly better compliance than is observed
with daily dosing regimens. The speculation that the improved compliance will
result in lower typical-use contraceptive failure rates will need to be confirmed
in future studies.
Create a personal account or sign in to: