Contraception xxx (2018) xxx–xxx

Contents lists available at ScienceDirect

Contraception

journal homepage: www.elsevier.com/locate/con

Review article

The relationship between progestin hormonal contraception and depression:

a systematic review
Brett L. Worly a,b,⁎, Tamar L. Gur a,b,c,d, Jonathan Schaffir a
a
Department of Obstetrics & Gynecology, The Ohio State University, Columbus, OH
b
Dept of Psychiatry & Behavioral Health, The Ohio State University, Columbus, OH
c
Dept of Neuroscience, The Ohio State University, Columbus, OH
d
Institute for Behavioral Medicine Research, Wexner Medical Center at The Ohio State University, Columbus, OH

a r t i c l e i n f o a b s t r a c t

Article history: Objective: We performed a systematic review to look for an association between progestin-only contraception
Received 23 May 2017 and depression.
Received in revised form 18 January 2018 Methods: We searched PubMed, Ovid and Web of Science for English-language articles including progestin-only
Accepted 22 January 2018 contraception and depression from database inception to September 2016. We evaluated study quality with the
Available online xxxx
procedures guiding reviews for the United States Preventive Services Task Force and the Cochrane Risk of Bias
Tools. We included studies that evaluated progestin-only contraception and depression, focusing on externally
Keywords:
Progestin
validated depression measures. We excluded case studies, review articles and other psychiatric disorders.
Hormonal contraception Results: We identified 26 studies that met the inclusion criteria, including 5 randomized controlled trials, 11 co-
Depression hort studies and 10 cross-sectional studies. We found minimal association between progestin-only methods and
Adolescents depression. No correlation with depression was found in five low-quality, high-risk-of-bias progestin subdermal
Postpartum depression implant studies and four out of five varying-quality and medium-risk-of-bias levonorgestrel intrauterine device
studies. Three medroxyprogesterone acetate intramuscular injection trials with varying levels of quality and bias
show no difference in depression. Two progestin-only contraceptive pill studies with varying levels of quality and
bias indicate no increase in depression scores, while one good-quality, medium-bias study shows an association
between progestin-only pills, the intrauterine device and depression.
Conclusion: Despite perceptions in the community of increased depression following the initiation of progestin
contraceptives, the preponderance of evidence does not support an association based on validated measures
(mostly level II-1 evidence, moderate quality, low risk of bias).
© 2018 Published by Elsevier Inc.

1. Introduction due to female population prevalence. The major depression lifetime

Sexually active women of reproductive age who want to prevent
pregnancy need reliable contraceptive options. Decisions about which
method to choose may involve factors such as efficacy, medical prob-
lems, previous experiences with side effects or failures, or concerns
about imperfect compliance. Due to the risks or side effects of estrogen,
many women choose a hormonal contraceptive that is formulated only
with progestin. Even with limited progestin exposure, concerns exist
about side effects including weight gain, acne, mood changes and de-
pression. Depression side effects of progestin contraception have been
stressed in the lay press and are common patient concerns [1,2]. Depres-
sion is also a concern for women considering hormonal contraception
⁎ Corresponding author. Tel.: +1 614 293 9899; fax: +1 614 366 0894.
E-mail address: Brett.worly@osumc.edu (B.L. Worly).
prevalence for US women is 7.4/100, twice that of men [3]. Concerns
about progestin's influence on mood are based on clinical experience
and basic science research [4,5].
Concerns about progestin-related depression effects also arise from
early clinical data on depot medroxyprogesterone acetate (DMPA),
which was approved for use as a long-acting contraceptive in 1992 by
the Food and Drug Administration (FDA). The package labeling stated
that women with depression should be observed carefully, and addi-
tional administration should not be performed if depression recurred.
This concern was based on FDA clinical data showing that 1.5% of
4200 users reported depression and 0.5% discontinued their use of
DMPA for this reason [6]. Since then, many studies have sought an
answer to whether DMPA may cause depression.
The relationship between progestin-only contraceptives and depres-
sion remains unclear. This study aims to systematically review the med-
ical literature regarding this relationship, additionally including
generalized and postpartum depression, as well as adolescents.

https://doi.org/10.1016/j.contraception.2018.01.010
0010-7824/© 2018 Published by Elsevier Inc.

Please cite this article as: Worly BL, et al, The relationship between progestin hormonal contraception and depression: a systematic review,
Contraception (2018), https://doi.org/10.1016/j.contraception.2018.01.010
2 B.L. Worly et al. / Contraception xxx (2018) xxx–xxx
2. Methods
We conducted this systematic review using Preferred Reporting
Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines
[7]. Prior to conducting our literature review, we identified the follow-
ing questions to guide our search:
1. Is there an association or causative link between progestin-only
hormonal contraception and depression effects in human females?
2. Does the type of progestin or route of administration influence
such an association?
3. Are there certain populations (such as adolescents, postpartum
patients or women with a history of depression) in which this as-
sociation exists?
2.1. Literature search
We searched PubMed, Ovid and Web of Science databases for English-
language peer-reviewed articles published from database inception until
September 2016 to identify studies examining depression effects of pro-
gestin contraceptives. We used PRISMA guidelines to report the data.
PROSPERO Registration number is CRD42017059302. We used several
combinations of search terms in order to address our three key questions
(Appendix A). In addition to our electronic search, we cross-referenced re-
view and other articles identified by our search, compared searches and
discussed abstract presentations presented at national meetings over the
past 10 years to find articles that may have been excluded based on publi-
cation bias due to null hypothesis findings or unfavorable results.
We searched clinical trials and abstracts with MESH terms and sepa-
rate title and abstract searches with direction and assistance from a librar-
ian. We excluded review articles, nonhuman studies, articles in languages
other than English and case reports. We used search terms “progestin-
only contraception,” “medroxyprogesterone depression,” “Levonorgestrel
IUD,” “levonorgestrel depression,” “Subdermal rod,” “etonogestrel de-
pression,” “norethindrone depression,” “Progestin [AND] depression
[AND] contraception,” “progestin [AND] postpartum depression,” “Con-
traception [AND] levonorgestrel [OR] medroxyprogesterone acetate
[OR] [AND] mood.” “Mood swings” and other psychiatric disorders were
not the focus of this project, as they are mediated by different mecha-
nisms, and were excluded. We reviewed studies for contraceptive meth-
od, population studied, measurements used and significance of effect. We
compiled these data to look for patterns in response to similar medica-
tions and populations. We assessed each study and followed procedures
guiding reviews for the United States Preventive Services Task Force
and rated studies as “good,” “fair” or “poor” [8]. We gave most weight to
randomized controlled trials (RCT) and externally validated depression
measures. Limited amounts of evidence from prospective trials led the au-
thors to also consider observational studies. Had more trial and compara-
tive cohort data existed, the review team would have focused on a higher
level of evidence. The principal summary measure was risk of depression
related to progestin-only contraception, using odds ratios (ORs) and dif-
ference in means as reported.
We considered performing a meta-analysis, but this was inappropri-
ate due to clinical and statistical heterogeneity of measures. We conduct-
ed a qualitative and narrative synthesis, highlighting effect consistency
areas and findings from studies with the lowest risk of bias, identifying
where data are lacking or insufficient to draw conclusions. We performed
an assessment of bias using the Cochrane Risk of Bias Tools, rating articles
with “low,” “medium” or “high” risk of bias [9] (Tables 1A-7).
3. Results
3.1. Study selection
Our search used multiple title and abstract searches plus PubMed
MESH search terms including “medroxyprogesterone depression”
Please cite this article as: Worly BL, et al, The relationship between progestin hormonal contraception and depression: a systematic review,
Contraception (2018), https://doi.org/10.1016/j.contraception.2018.01.010
with 155 abstracts, “levonorgestrel depression” with 80 abstracts, “nor-
ethindrone depression” with 126 abstracts, “etonorgestrel depression”
with 8 abstracts, “Progestin [AND] depression [AND] contraception”
with 190 abstracts and “progestin [AND] postpartum depression” with
78 abstracts. This initially yielded 2305 citations and, once exclusion
criteria were applied, yielded 696 abstracts. We identified 41 articles
for possible inclusion after cross-referencing and further examining ab-
stracts. All study authors reviewed the abstracts, and we excluded those
ruled ineligible based on the above criteria. After careful review, the au-
thors agreed on the inclusion of 26 articles that met all criteria (Fig. 1).
3.2. Depression association by type of contraceptive method
3.2.1. Injectable medroxyprogesterone acetate
Since the FDA warning on DMPA in 1992, providers have expressed
concerns about whether this highly effective contraceptive method is
associated with depression. No RCTs evaluating depression risk for
DMPA users exist in the general population.
In a multicentered prospective study of 495 women starting DMPA
[10], Westhoff et al. administered the Mental Health Inventory (MHI) at
initiation and 1 year. Depression scores dropped in those continuing the
medication, from 7.4 to 6.7 (p=.03), a statistically significant decrease.
In addition, women with the highest depression scores at baseline im-
proved, suggesting that DMPA does not make depression worse. This
study did not compare DMPA users versus the general population.
Berenson et al. [11] recruited 608 US women ages 16–33 years old and
allowed them to choose barrier methods of contraception, DMPA or com-
bined oral contraceptive pills (COCP). The authors evaluated participants
every 6 months for 2 years with a symptom checklist, Beck Depression In-
ventory (BDI), Zung Anxiety instrument and Positive and Negative Affect
Scale (PANAS). While 355 people dropped out of the study, the DMPA pa-
tients had lower scores on the BDI, with no increased risk of depression
[OR 1.08, 95% confidence interval (CI) 0.71–1.62] compared to the barrier
method group. The reference group baseline depression rate from the
self-reported symptom checklist was 34%, with an additional 31% in the
nonhormonal group developing new-onset depression symptoms over
24 months, based on the checklist. BDI scores were 1.4 units lower for
the DMPA group compared to the nonhormonal group at 24 months
(pb.05). Although these studies were not randomized, large sample
sizes and use of validated scales provide convincing evidence that
DMPA does not increase the risk of depression.
Two prospective cohort studies both used the same validated scale,
the Community Epidemiology Depression Scale (CESD), to assess
mood. Civic et al. [12] evaluated 183 DMPA users and 274 nonusers at
6-month intervals for 3 years. In their study, significantly more
women who discontinued DMPA had scores over 10 at 3 months indi-
cating more depression symptoms (36.4%) before (OR=2.30, 95% CI
1.42–3.70) and after (OR=2.46, 95% CI 1.46–4.14) discontinuation com-
pared to women who remained users (21.1%). The OR for developing
depression symptoms (based on the CESD) in DMPA users was 1.44
(95% CI 1.00–2.07), and the OR for DMPA discontinuers was 1.60 (95%
CI 1.03–2.48) compared with nonusers. Depression symptoms in
nonusers were found to occur in 11.8%–17.5% of this population at the
various study visits based on CESD scores. A smaller prospective cohort
study by Westhoff et al. [13] looked at 80 DMPA users and 26 nonusers
with the CESD 4 weeks after DMPA injection and immediately prior to
the next injection. The full CESD was used with a range of 0–60, with
scores over 16 suggesting clinical depression for adults and scores
over 22 indicating depression in adolescents. The mean CESD score for
nonusers was 14.4, while mean score for DMPA users was 15.6 (not sig-
nificant). A total of 20%–41% of subjects were noted to have depression
scores indicating increased risk of clinical depression during the study,
including DMPA users and nonusers. In the Civic study, sample size
was smaller, depression cutoff values were lower, the population was
on average 5 years older, and discontinuation rate was higher, making
 B.L. Worly et al. / Contraception xxx (2018) xxx–xxx 3

Table 1A
Depot medroxyprogesterone and depression randomized control and prospective cohort trials (postpartum women, adolescent women and depressed women are included in Tables 5, 6
and 7, respectively)

Author (year), Study design Screening Population Results Strengths Weaknesses Quality/bias
sources of support tool [8,9]

Westhoff (1998) Prospective CES-D 80 women using CES-D 1 month after injection is Prospective study, No control group, Poor, high
[10], no funding descriptive DMPA 15.3 and 16.0 before first externally validated descriptive, not
listed study Ages 13–43 injection or immediately before measure randomized, small
New York Followed up for 3 90 days postinjection (p=.36) sample size
1993–1994 months Adjusted for parity, number of
No matching Excluded women injections
with doctors outside
this clinic
Westhoff (1998) Prospective MHI 2007 women, 495 MHI scores no different in Prospective trial, externally Small subgroup Fair,
[16], NICHD and cohort study DMPA, 910 implantable rod patients from validated instrument, US analysis, not medium
Wyeth-Ayerst Texas, subdermal rod, 314 baseline to 6 months in those population, few exclusion randomized, no
Laboratories Pennsylvania, COCP, 288 continuers (n=279, 7.9 to 7.7, criteria, baseline depression cohort comparison
New York sterilization p=.59), discontinuers (n=287, symptom assessment, group
1993–1994 Ages 15–45 8.8 to 9.0, p=.62), or lost to 24-month follow-up
No matching Followed up for 2 follow-up (n=136, 8.4 to 9.0,
years p=.30)
Excluded patients Adjusted for baseline depression,
less than 15 years relationship satisfaction, partner
old, not starting a approval of method
new method, no
contraception visit in
clinic in past 3
months
Civic (2000) [12], Prospective CES-D 183 women DMPA DMPA depression at 3 years 1.44 Externally validated study, Women dropped Good,
National Institute cohort study users and 274 (95% CI 1.00–2.07), nonuser 1.00 prospective follow-up for 3 from follow-up once medium
of Child Health Washington nonusers (ref), Discontinuer 1.60 years, exclusion criteria, pregnant, not able to
and Human 1994–1999 Followed up for 3 (1.03–2.48), appropriate adjustments in assess women before
Development at DMPA and years Depression symptoms nonuser analysis DMPA initiation, not
the National bone density Ages 18–39 1.00 (ref), before randomized
Institutes of study Excluded for discontinuation 2.30
Health Matched for pregnancy, lactation, (1.42–3.70), for discontinuers at
age and clinic attempting 1st visit 2.46 (1.46–4.14), 2nd
pregnancy, condition visit 1.52 (0.83–2.79), 3rd visit
affecting bone 1.63 (0.83–3.20), 4th–6th visit
density, planned 1.17 (0.65–2.10)
move Adjusted for race, education, age,
prior history of depression, and
repeated measures for the same
subject

direct comparisons challenging due to ascertainment and identification
bias [12,13].
Although studies featuring self-report of symptoms suggest a high
discontinuation rate due to depression symptoms [14,15], studies
using validated measures of depressed mood show no significant detri-
mental DMPA effect, while one indicates depression scores decrease
[10–13]. In Civic's study, the most common time for patients to report
depression leading to discontinuation was 3 months postinjection;
however, low rates (3%–36%) of depression persist that do not compro-
mise compliance. These rates suggest that screening for depression may
be appropriate for women who discontinue DMPA (Tables 1A and B).
3.2.2. Subdermal progestin
Another systemic route of administration for progestin contracep-
tion is the subdermal implant. The early version of this method was
Norplant, composed of six rods containing the more androgenic proges-
tin levonorgestrel (LNG). Over time, this device has been simplified, and
the most recent version consists of a single etonogestrel implantable
subdermal rod that lasts for 3 years.
In a multicentered prospective study [16], 910 women received the
six-rod LNG implant, and then participants were interviewed at 6 and
24 months postinsertion for self-reported depression effects with the
MHI. Depression scores did not appreciably change for implant users
or nonusers from baseline during the first 6 months of use (7.9 vs. 7.7,
p=.59; 8.8 vs. 9.0, p=.62). Depression scores increased significantly in
the 293 who continued the implants for 2 years (7.9 vs. 8.8, p=.01),
with relationship dissatisfaction being the biggest predictor of
depression. No statistical assessment is provided comparing LNG con-
tinuers and discontinuers at 24 months. While these changes are statis-
tically significant, the clinical significance of these small differences in
score is questionable. This study, though hampered by a high dropout
rate, suggests that evolving changes in depression over time may have
more to do with social stressors than with medication itself. The steril-
ization group's MHI data were not reported, but this would have been
the ideal group for comparison at 2 years, as increasing prevalence of
depression over time would be expected of both groups.
Unfortunately, there are no other studies that use validated instru-
ments to screen for depression symptoms in progestin implant users.
Other existing studies inquire about depression or depression side ef-
fects and rely on self-report only, and do not show a significant relation-
ship between subdermal implants and depression [17–25]. Although
some studies suggest that a subset of women indicates depression as a
reason for discontinuation, they cannot objectively demonstrate an ef-
fect of the medication itself. Additional studies are required that com-
pare validated depression scales in women who do or do not use
progestin subdermal implants (Table 2).
3.2.3. The LNG intrauterine device (IUD)
The LNG IUD is a long-acting reversible contraceptive option that
may have advantages over other hormonal contraceptives in terms of
compliance and efficacy. Because the hormone in this device acts locally
rather than systemically, it might also be expected to have fewer sys-
temic effects, including depression. Few studies have specifically ad-
dressed these issues in IUD users.

Please cite this article as: Worly BL, et al, The relationship between progestin hormonal contraception and depression: a systematic review,
Contraception (2018), https://doi.org/10.1016/j.contraception.2018.01.010
4 B.L. Worly et al. / Contraception xxx (2018) xxx–xxx

Table 1B
Depot medroxyprogesterone and depression retrospective, cross-sectional and self-report trials (postpartum women, adolescent women, and depressed women are included in Tables 5, 6
and 7, respectively)

Author (year), sources Study design Screening Population Results Strengths Weaknesses Quality/bias
of support tool [8,9]

Berenson (2008) [11], Prospective Self-reported219 women on DMPA, 218 Self-reported depressive Prospective Not randomized, no Poor,
National Institutes of cohort study depression on COCP, and 171 with symptoms in cohort study with placebos, recall bias for medium
Child Health & Texas symptoms, nonhormonal nonhormonal externally retrospectively collected
Human 2001–2004 BDI, PANAS, contraception contraception patients validated surveys, data, no data on
Development, Bone mineral Zung Ages 16–35 OR=1.00 (Ref), DMPA comparison symptom severity, 197
National Center for density survey Anxiety Followed up for 2 years OR=1.08 (95% CI between women excluded or lost
Research Resources Controlled for instrument Excluded patients with 0.71–1.62), COCP nonhormonal to follow-up. Many
age and current pregnancy, OR=0.91 (0.59–1.40). COCP, and DMPA exclusion criteria.
reproductive breastfeeding, planned BDI score 1.4 units lower methods
method pregnancy inb3 years, use for DMPA compared to
of DMPA in past 6 months, nonhormonal group at
use of COCP in past 3 24-month follow-up
months, current use of (pb.05). Positive affect
hormonal IUD, scores 1.2 units lower for
contraindication to DMPA users compared to
hormonal contraception, nonhormonal group
lack of menses N3 months (pb.05).
in past year, bilateral Adjusted by race, age,
oophorectomy, follow-up visit number,
over-the-counter and baseline status of
estrogen, high dietary symptom.
isoflavone intake, illness
effecting BMD, eating
disorder, strict vegetarian
diet
Sangi-Haghpeykar Cross-sectional Self-reported 536 first-time DMPA users 8.9% depression side effect Prospective No externally validated Poor, high
(1996) [14], funding prospective symptoms Followed up for 1 year at 3 months, 5.6% at 6 12-month surveys, no control
not listed descriptive Ages 13–46 months, 6.7% at 9 months follow-up group, 1-year
study Depression noted in
No exclusion criteria listed descriptive study, continuation rate was
Texas continuers and no patients 28.6%
1993–1994 discontinuers at 3 excluded
No control months: 6.5% and 14.5%
group (pb.001), 6 months: 4.9%
and 8.6%, and 9 months:
5.6% and 11.4% (others not
statistically significant)
Paul (1997) [15], Cross-sectional Self-reported 252 New Zealand women 47% discontinued by 12 Cross-sectional No control group, no Poor, high
Medical Research descriptive symptoms on DMPA months, 5.8% (n=18) descriptive study externally validated
Council of New study Ages 25–54 discontinued due to with nurse surveys, strong recall
Zealand, Develop- New Zealand Randomly excluded half of depression, with 11.8% interviews bias, not randomized
ment and Research Hormonal subjects under 35 years to (n=8) discontinuing by 3
Raining in Human contraception better approximate age months due to depression,
Reproduction from and breast distribution and 5.2% (n=5)
the World Health cancer study discontinuing due to
Organization 1983–1987 depression at 6–21
months

One RCT did evaluate depression scores, but it focused on patients
whose indication for IUD was heavy menstrual bleeding [26]. The
study involved 236 women randomly assigned to receive either LNG
IUD or hysterectomy. They completed the BDI at 6 and 12 months to as-
sess depression and saw the level of depression scores in the IUD group
drop from a mean baseline score of 5.2 to a mean 6-month follow-up
score of 3.7 (statistical significance of these values was not presented
in the study), with 37% of all participants having mild depression at
baseline and 27% with depression at 6-month follow-up based on BDI
score. Additional statistical analysis and follow-up were not provided
for this portion of the evaluation. The IUD patients who also had depres-
sion were more likely to have a hysterectomy at 6 months (OR 3.70, 95%
CI 1.55–8.82). Although depression decreased for the IUD group, indi-
viduals who reported depression symptoms at 6 months postinsertion
were more likely to discontinue the IUD (OR 3.70, 95% CI 1.55–8.82).
The rate of depression in the control group was not reported.
Another study evaluating women who received LNG IUD to treat
heavy menstrual bleeding also used the BDI but did not have a control
group [27]. This study of 120 women showed no change in BDI scores
from baseline to 6 months (p=.375), though women did report an im-
provement in quality of life. The indication of heavy menstrual bleeding
for these studies may make the conclusions irrelevant for women using
the device strictly for contraception. Overall rates of depression pre- and
posttreatment were not reported.
A large prospective study examined depression using standardized
scales in women using the LNG IUD specifically for contraception [28].
It compared the mental health of patients who did or did not use the
LNG IUD with the BDI, General Health Questionnaire 12 (GHQ-12) and
Composite International Diagnostic Interview. Compared with women
using other contraception, this study showed no difference with linear
regression analysis in psychiatric depression diagnosis for those
women currently using the LNG IUD (B 0.580, 95% CI 0.280–1.201), al-
though BDI (B −0.988, 95% CI −1.917 to −0.059) and GHQ-12 (B −
0.285, 95% CI −0.684 to 0.113) scores were lower. For patients using
the LNG IUD, 4.5% had major depressive disorder compared to 7.8% of
the general population, while 4.5% of IUD users had a major depressive
episode compared to 8.1% of the general population. This study suggests
that the LNG IUD does not have an adverse impact on depression.
Another RCT more specific to the question of LNG IUD dosage did not
use any validated instrument to measure depressive effects. Nilsson
et al. [29] performed a prospective RCT for Finnish women receiving dif-
ferent forms of IUD. A total of 164 subjects receiving a low-dose LNG IUD

Please cite this article as: Worly BL, et al, The relationship between progestin hormonal contraception and depression: a systematic review,
Contraception (2018), https://doi.org/10.1016/j.contraception.2018.01.010
 B.L. Worly et al. / Contraception xxx (2018) xxx–xxx 5

Table 2
Progestin subdermal implant and depression studies (postpartum women, adolescent women and depressed women are included in Tables 5, 6 and 7, respectively)

Author (year), Study design Screening Population Results Strengths Weaknesses Quality/bias
sources of support tool [8,9]

Sivin (1998) [18], Prospective Self-reported 594 women with two-rod LNG Eight women (1.3%) 8-year Side effects from patient Poor, high
Wyeth-Ayerst descriptive side effects subdermal implants discontinued due to prospective study, self-report, no cohort
Research, United study Ages 18–40 depression focus on side group comparison, 70 per
Nations Population US and Excluded not cancer, abnormal symptoms, 3 (0.5%) effects related to 100 women continuation
Fund, US Agency Dominican genital bleeding, women discontinued removal, multisite rate
for International Republic hyperprolactinemia or bloody due to mental illness clinical trial
Development 1990–1998 nipple discharge,
Unmatched hyperlipidemia, severe
cardiovascular problems, mental
illness, diabetes mellitus,
epilepsy, severe headaches,
pelvic inflammatory disease,
ectopic pregnancy
Kozlowski (1995) Retrospective Self-reported 110 LNG implantable rod 1/110 patients noted Clinic data on a No control group, no Poor, high
[19], no funding descriptive side effects patients a side effect of special population externally validated
source listed study in chart Ages 13–21 depression over a year measure, no exclusion
Arkansas review No exclusion criteria listed No adjustments criteria or adjustments for
1991–1992 listed depression factors,
retrospective nature, small
sample size
Sihvo (1995) [17], Descriptive Self-reported 207 LNG implantable rod 53% of women with Population-based Not randomized, small Poor, high
Academy of study, side effects patients, insertions 1–2 years self-reported sample with no sample size, no externally
Finland Finland fami- prior to survey disruptive exclusions validated instrument, no
ly planning Average age=31 depression (n=28) adjustments for other
clinics No exclusion criteria listed had rod removed depression factors
June 1992 No adjustments
No control listed
group
Suman (1998) [25], Retrospective Chart review 144 LNG subdermal implant 3/83 women Electronic medical Depression symptoms Poor, high
funding source not descriptive patients requested rod record review, were self-reported as
listed study Ages 14–21 removal due to evaluation of noted in electronic medical
Minnesota Excluded patients living outside depression adolescent record, no control group,
1990–1993 catchment area, patients lost to symptoms, 11/130 patients retrospective, no externally
No control follow-up patients reported validated measure
group depression
symptoms
Adjusted for timing
of insertion, prior
pregnancies,
unplanned
pregnancies, prior
delivery, marital
status
Inoue (2016) [24], Descriptive Structured 2 women currently using and 8 1/10 women who Structured No externally validated Poor, high
Australian Re- qualitative clinical who had used etonogestrel discontinued the interviews of surveys, small sample size,
search Council, study interview subdermal implantable rod etonorgestrel cited patients who had concern for recall bias
Family Planning Australia Ages 20–50 “depression/anxiety” used the
New South Wales, 2012–2013 No exclusion criteria listed in a structured implantable rod
University of No control interview or who were
Sydney group currently using
Data rod
collected
from CUE
study

(containing LNG 43 mg, releasing 20 mcg daily) were compared with
163 women with a higher-dose LNG IUD (containing LNG 56 mg, releas-
ing 30 mcg daily) and 156 women with a copper IUD. They found no dif-
ference between groups in self-reported depressive effects. Overall
depression rates were not reported.
Another study that merits consideration due to its large size neither
was prospective nor utilized validated measurements. A large
population-based study evaluated new diagnoses of depression or anti-
depressant use in 1,061,997 contraceptive users in the Danish health
registry database. In this study, progestin-only contraceptive users
showed an increased risk of first depression diagnosis [relative risk
(RR) 1.2, 95% CI 1.04–1.31] relative to users of barrier methods, and
LNG IUD users showed a 1.4 RR (95% CI 1.22–1.50). This study benefited
from a large sample size, but given the retrospective nature and weak
degree of association, the evidence is unimpressive. If progestins were
causative of such an effect, one would expect a dose-related response;
however, this analysis showed a higher risk for LNG IUD users than for
women exposed to higher levels of progestin, for example, progestin-
only pill (POP) users [30].
In summary, most studies indicate that the LNG IUD does not carry an
association with depression. For those studies that suggested any adverse
effect, one was a very weak association [30], and the other was in
discontinuers for whom the indication was abnormal bleeding, who
may have had other reasons to experience depressive symptoms. Con-
versely, this study showed that depression scores were reduced 6 months
after starting the LNG IUD, making biological plausibility of these results
unclear because bleeding can still be irregular for 3–6 months after inser-
tion, and the authors did not comment on this association [26]. Thus, risk
for depression effects is much lower, or nonexistent, in LNG IUD users
than with systemic exposure to progestins (Table 3).

Please cite this article as: Worly BL, et al, The relationship between progestin hormonal contraception and depression: a systematic review,
Contraception (2018), https://doi.org/10.1016/j.contraception.2018.01.010

Table 3
Contraception (2018), https://doi.org/10.1016/j.contraception.2018.01.010
Please cite this article as: Worly BL, et al, The relationship between progestin hormonal contraception and depression: a systematic review,
Levonorgestrel IUD and depression studies (postpartum women, adolescent women and depressed women are included in Tables 5, 6 and 7, respectively)
Author (year), sources
of support
Nilsson (1982) [29], International
Committee for Contraception
Research and the Population
Council, International
Development Research Center of
Canada, Ford Foundation, Rocke-
feller Foundation
Elovainio (2007) [26], National
Development Centre for Welfare
and Health
Toffol (2011) [28], Research
Foundation of the University of
Helsinki and Helsinki University
Central Hospital Research Funds
Pekin (2014) [27], Selcuk University
Hospital
Skovlund (2016) [30], Department
of Gynecology, Rigshospitalet,
University of Copenhagen,
Lundbeck Foundation
6
Study design Screening Population Results Strengths Weaknesses Quality/bias
tool [8,9]
Randomized Self-reported 164 women LNG-IUD A (43 mg LNG, 20 No change in self-reported Randomized control trial Depression assessed with Fair, low
control trial symptoms mcg daily) depression increase or decrease Multiple sites self-report
Finland and 163 women LNG-IUD B (56 mg LNG, 30 for the three groups. Double blind Finland and Brazil only
Brazil mcg daily) IUD A increase 11.2%, decrease 1-year symptoms follow-up No adjustment for life
Study time 156 women Copper IUD 0.6% stressors
period not listed Ages 18–38 IUD B increase 14.4%, decrease
Not matched Followed up N12 months 1.9%
Excluded nulliparous, breastfeeding, Copper IUD increase 7.2%,
currently pregnant, history of ectopic decrease 0.7%
pregnancy No adjustments.
Randomized BDI 236 Finnish women with menorrhagia, age Depressive symptoms on BDI after Externally validated measures, Finnish population only, Fair, low
control trial 35–49 6 months on treatment were randomized control trial, moderate questionable external
Finland Followed up for 6 months related to LNG-IUD sample size validity, no thyroid
1994–1997 119 LNG IUD, 117 hysterectomy discontinuation measures
Women who OR 3.7 (95% CI 1.6–8.2), adjusted
completed for age, marital status, number of
childbearing children, education, # health
B.L. Worly et al. / Contraception xxx (2018) xxx–xxx
services visits for menorrhagia in
6 months, sickness absence days,
menstrual blood loss
Population-based BDI, GHQ, 181 women using OCs compared to 1719 Lower BDI score for LNG-IUD Externally validated BDI, CIDI, and Not randomized, Fair,
cohort CIDI healthy nonusers users compared to healthy GHQ retrospective data medium
Finland 212 using LNG-IUD compared to 1688 controls (r=−0.054; pb.05), Structured interview by trained collection regarding
2000–2001 healthy nonusers LNG-IUD BDI score B=−0.988 nurses (CIDI), nationwide duration of contraception
Health 2000 Ages 18–54 (95% CI −1.9 to −0.059), pb.05 population-based study use, no assessment of
Survey Followed up N12 months Adjusted for age, professional mental status before
Excluded incomplete survey results status, marital status, education, initiation of method
bleeding pattern, psychiatric
diagnosis, duration of
contraception use
Prospective BDI, SF-36 120 premenopausal menorrhagia patients No change after 6 months in BDI Externally validated measures, No control group, no Fair, high
descriptive study (PBAC N75) with LNG-IUD insertion score, from 8 to 7, p=.375, SF-36 prospective. No patients lost to adjustments for
Turkey Ages 18–50 Mental Component Summary follow-up, no early discontinuers. preexisting depression or
2012–2013 Followed up for 6 months Score improvement from 67.7 to stressors
No control group Excluded abnormal paps, cervical disease, 74.9, pb.001.
cavity distortion, uterine/cervical cancer, Adjustments not listed.
PID, liver disease, breast cancer
Population-based National 1,061,997 women Users of POPs with RR 1.34 (95% CI Large population study with Possible detection bias of Good,
prospective database Ages 15–34 1.22–1.25) of first-time use nonselective inclusion, no loss to physicians, not medium
cohort study Mean follow-up 6.4 years antidepressant; LNG IUD RR 1.4 follow-up, no recall bias, multiple randomized, no externally
Denmark Followed up for 5 years (1.31–1.42). sensitivity analyses all with validated surveys
1995–2013 6,832,938 person-years follow-up Users of POPs risk of depression consistent results, linkage of
Danish Sex Excluded for depression diagnosis or diagnosis RR 1.2 (1.04–1.31); LNG Psychiatric and Prescription
Hormone antidepressant use before 15 years old, and IUD RR 1.4 (1.22–1.50). registries
Register Study all women with manic episode, bipolar Adolescent risk of first use of
Linked data from affective disorder, schizophrenia, mental antidepressants and first
National retardation, cancer diagnosis, venous diagnosis of depression for POPs
Prescription thrombosis, infertility treatment RR 2.2 (1.99–2.52) and 1.9
Register and (1.49–2.53); LNG IUD RR 3.1
Psychiatric (2.47–3.84) and 3.2 (2.08–5.03).
Central Research Adjusted for age, calendar year,
Register educational level, polycystic
ovarian syndrome, and
endometriosis
 B.L. Worly et al. / Contraception xxx (2018) xxx–xxx 7

3.2.4. Progestin-only contraceptive pills 3.3. Special populations

Drawing conclusions about studies examining depression in users of
POPs is limited by the use of regimens not available in the United States
and in varying doses. In the only prospective RCT [31], researchers
assigned 150 patients to receive COCP, LNG-only pills or placebo. They
used the BDI at baseline and after 3 months to track depression at the
first day of menses, the premenstrual week, 2 weeks later and the
postmenstrual week. Depression scores were lower in the LNG arm
compared to placebo and COCP groups at months 2 (F=3.2, pb.05), 3
(F=3.6, pb.05) and 4 (F=3.0, pb.05). While this study population is
small, the result is reassuring that low doses of oral progestin are asso-
ciated with lower depression scores.
In the Danish registry study by Skovlund mentioned earlier [30],
POPs were noted to have no increased RR for new depression diagnosis
[norethisterone (RR 1.1, 95% CI 0.88–1.29), LNG (RR 1.5, 95% CI
0.54–3.86)] or slightly increased risk [desogestrel (RR 1.2, 95% CI
1.06–1.42)] compared with barrier contraceptive users. All three of
these medicines were found to be associated with an increased RR of
first use of antidepressants, with norethisterone (RR 1.3, 95% CI
1.18–1.37), LNG (RR 1.7, 95% CI 1.18–2.38) and desogestrel (RR 1.4,
95% CI 1.30–1.46) showing increased risk over nonuse of hormonal con-
traception. Again, this was a retrospective analysis with a healthcare da-
tabase as a means for diagnosing depression, so results may have been
confounded by indications for prescribing. None of these POPs are avail-
able in the United States. The norethindrone-only pills available in the
United States, which are similarly low in dose, are also likely to have lit-
tle adverse effect, though this theory will need to be tested by future re-
search (Table 4).
3.3.1. Effects of progestin contraception on postpartum depression
Some of the literature that examines validated depression scores in
users of progestin-only contraceptives is limited to specific subsets of
women. Postpartum women are a group of particular interest since
this population may frequently consider progestin contraceptives as
an effective method with no adverse effect on breastfeeding.
One such study was a blinded RCT of DMPA and copper IUD in post-
partum patients in London and South Africa [32]. The authors used Edin-
burgh Postnatal Depression Scale (EPDS) and BDI scores at baseline, 1
month and 3 months. EPDS scores were significantly worse for DMPA pa-
tients at 1 month (4 vs. 2, p=.04), and BDI scores were significantly worse
for DMPA users at 3 months (9 vs. 5, p=.002). This result suggests that
DMPA may increase depression scores temporarily in postpartum
women, but the small absolute difference in scores at 1 month suggests
that it may not be clinically significant. However, more DMPA than IUD
users met the criteria for depression at 3 months as determined by the
BDI but not by the EPDS. This conflicting evidence makes it unclear
whether there is truly an adverse effect of DMPA or whether one of
these validated scales may be less reliable in this population.
In a separate RCT by Lawrie et al. [33], 180 postpartum women were
given injectable norethisterone (not available in the United States) or
placebo. Women who received the progestin medication were more
likely to have higher scores on the EPDS (10.6 vs. 7.5, p=.0022). More
patients in the progestin medication group had major depression
based on EPDS scores compared to the placebo at 6 weeks postpartum
(RR 3.035, 95% CI 1.515–6.080, where 45.3% patients tested positive

Table 4
Progestin-only contraceptive pills and depression studies (postpartum women, adolescent women and depressed women are included in Tables 5, 6 and 7, respectively)

Author (year), sources of Study design Screening Population Results Strengths Weaknesses Quality/bias
support tool [8,9]

Graham (1995) [31], Randomized BDI 75 Filipino and 75 BDI score decreased in Randomized control Small sample Good, low
Special Programme of control, Scottish women with 50 POP group at 3 months trial, 3-month follow-up, size
Research Development double-blinded assigned to ethinyl (F=3.2, df 2, pb.05) externally validated
and Research Training study estradiol 0.03 mg + LNG compared to placebo and surveys, multisite trial
in Human Scotland and 0.15 mg, 50 received LNG COC groups
Reproduction, World Philippines 0.03 mg, and 50 placebo
health Organization; Years not listed Followed up for 4 months
Schering Health Care, Unmatched No exclusion criteria
Ltd. controls listed
Young (2007) [39], Multisite HamD, 232 women on combined No significant differences Multisite trial, minimal Small sample Fair,
National Institutes of prospective PDSQ, hormonal contraception, were found in major exclusion criteria, size in medium
Mental Health, cohort study QIDS-SR16, 58 on progestin-only depressive episodes, externally validated progestin-only
Bristol-Myers Squibb STAR*D trial IDS-C30, methods, 948 on suicide attempts, surveys group, did not
Company, Forest 2001–2004 SF-12 nonhormonal Hamilton Rating Scale for record reason for
Pharmaceuticals Inc., Unmatched contraception Depression scores or starting
GlaxoSmithKline, King controls Ages 18–40 Inventory for Depressive hormonal
Pharmaceuticals, Followed up for 12 Symptomatology. contraception,
Organon Inc., Pfizer Inc., months Adjusted for age and not randomized
Wyeth-Ayerst Excluded psychotic major years of schooling
Laboratories depressive disorder
patients
Skovlund (2016) [30], Population-based National 1,061,997 women Users of POPs with RR Large population study Possible Good,
Department of prospective database Ages 15–34 1.34 (95% CI 1.22–1.25) of with nonselective detection bias of medium
Gynecology, cohort study Mean follow-up 6.4 years first-time use inclusion, no loss to physicians, not
Rigshospitalet, Denmark Followed up for 5 years antidepressant; follow-up, no recall bias, randomized, no
University of 1995–2013 6,832,938 person-years Users of POPs risk of multiple sensitivity externally
Copenhagen, Lundbeck Danish Sex follow-up depression diagnosis RR analyses all with validated surveys
Foundation Hormone Excluded for depression 1.2 (1.04–1.31). consistent results,
Register Study diagnosis or Adolescent risk of first linkage of Psychiatric
Linked data from antidepressant use before use of antidepressants and Prescription
National 15 years old, and all and first diagnosis of registries
Prescription women with manic depression for POPs RR
Register and episode, bipolar affective 2.2 (1.99–2.52) Adjusted
Psychiatric disorder, schizophrenia, for age, calendar year,
Central Research mental retardation, educational level,
Register cancer diagnosis, venous polycystic ovarian
thrombosis, infertility syndrome, and
treatment endometriosis

Please cite this article as: Worly BL, et al, The relationship between progestin hormonal contraception and depression: a systematic review,
Contraception (2018), https://doi.org/10.1016/j.contraception.2018.01.010
8 B.L. Worly et al. / Contraception xxx (2018) xxx–xxx

Table 5
Progestin-only hormonal contraception and postpartum depression studies

Author (year), sources of Study design Screening Population Results Strengths Weaknesses Quality/bias
support tool [8,9]

Lawrie (1998) [33], Iris Double-blind MADRS 90 women received 6-week Mean MADRS Externally Less than one quarter of Fair, low
Ellen Hodges Trust of the randomized and EPDS norethisterone 200 mg IM scores for norethisterone validated women approached
University of placebo trial injection, 90 patients 8.3 and placebo 4.9 (95% CI surveys, agreed to participate,
Witwatersrand, South South Africa received placebo 0.72–5.47, p=.01), EPDS prospective external validity of South
African Research Council, 1995–1997 Followed up for 3 months for norethisterone 10.6 and randomized African sample
South African Institute for Unmatched Mean age 32 placebo 7.5 (1.01–4.74, design,
Medical Research, controls Excluded women b19 years p=.002); 3-month
Schering Ltd old, N48 h postpartum, 3-month MADRS scores for follow-up
desiring hormonal norethisterone 6.6 and
contraception placebo 6.1 (−1.10–2.10),
EPDS norethisterone 9.3
and placebo 8.5
(−1.18–2.61).
Adjusted for mode of
delivery
Singata-Madliki (2016) Prospective BDI 242 postpartum women 1-month BDI scores DMPA Prospective Powered to detect Good, low
[32], Effective Care randomized within 48 h after childbirth, 8 and IUD 9 (p=.20) RCT, differences in BDI scores
Research Unit of East control single 123 copper IUD and 119 1-month EPDS scores externally but not depression rates.
London Hospital blinded DMPA DMPA 4 and IUD 2 (p=.04) validated Not adjusted for HIV
Complex/University of study East Followed up for 3 months 3-month BDI scores DMPA surveys severity status which may
Fort Hare, South African London and Ages 18–45 9 and IUD 5 (p=.002) have led to bias in favor of
Medical Research Council South Africa Excluded if not willing to use 3-month EPDS scores DMPA arm
2012–2013 DMPA or IUD 48 h after DMPA 2 and IUD 2 (p=.10) Did not assess for
Unmatched childbirth, not able to read Not adjusted postnatal depression risk
controls English or Xhosa, major factors.
depression, no access to Questionable external
telephone, contraindications validity of patient
to DMPA or IUD population.
Stevens-Simon (2000) [35], Prospective CESD 212 US postpartum 6–12-month CESD results Prospective Not randomized, 65 Fair,
Office of Adolescent cohort study adolescents for postpartum women study, patients lost to follow-up, medium
Pregnancy, NIH; Colorado Denver, Ages 13–19 with implant inserted: adolescent US may not be generalizable
Trust; General Clinical Colorado Followed up for 6–12 b2 months: 13.2±10.9 population, to nonpregnant
Research Centers 1992–1993 months after subdermal LNG 2–12 months: 18.0±13.4 externally adolescent population,
Program, NIH (CAMP rod insertion Not inserted: 11.5±7.7; validated did not control for other
study) CESD score increased with measure with depression risk factors
Unmatched 100 inserted rod at b2 number of postpartum CESD
controls months, 72 inserted rod at weeks (r=0.2, p=.01)
2–12 months, 40 did not use Adjusted for adverse
rod pregnancy outcome,
Excluded DMPA users, those multiparity, short interval
without depression pregnancy, difficult
symptoms assessment, and antenatal experience,
65 patients lost to follow-up history of depression,
dissatisfaction with father
of baby, undereducation,
poverty, unmarried,
sexual/physical abuse,
substance abuse
Tsai (2010) [34], no funding Retrospective EPDS 55 postpartum women EPDS 6 weeks postpartum Externally Retrospective, not Fair, high
cohort study receiving DMPA and 192 for DMPA users 5.02, and validated randomized, not powered
Ohio women without hormonal controls 6.17 (p=.16) survey to detect a difference in
2007–2008 contraception after delivery Postpartum depression in postpartum depression
Unmatched Excluded for intrauterine 10.9% of DMPA users and
controls fetal demise or neonatal 14.1% of controls (p=.88)
death, history of depression,
preterm birth

on the EPDS for depression compared to 26.0% patients on placebo).
These differences did not persist at 3 months postpartum. These studies
both suggest that postpartum women may be susceptible to depressive
effects of injectable progestins. The apparent effect of norethisterone
may not be applicable to progestins used for contraception in the
United States because of structural differences.
Evidence to the contrary was seen in a retrospective cohort study
performed by Tsai and Schaffir [34]. They compared 55 postpartum
women receiving DMPA immediately postpartum with 192 postpartum
women using no hormonal contraception. The EPDS was used to mea-
sure depressive symptoms at a 6-week postpartum visit. No statistically
significant difference was seen with 6-week EPDS scores (p=.16). The
baseline depression rate for users and nonusers of DMPA was
comparable. Mean EPDS scores and number of patients diagnosed
with postpartum depression were not significantly different between
DMPA and nonusers for those with previous history of depression.
Another retrospective study evaluated an even narrower segment of
the postpartum population, looking only at adolescents [35]. The study
evaluated 212 postpartum adolescents who received LNG implant
(Norplant) either immediately or 2–12 months after delivery or who
used a different contraceptive method, and utilized the CESD to evaluate
for depression. The group most likely to experience depression based on
CESD measurements was composed of those who received the device
2–12 months after delivery and not those who received it immediately
based on depression rates (CESD scores ≥16) (49% vs. 32%, p=.02) and
CESD depression scores (18.0±13.4 vs. 13.2±10.9, p=.004). These

Please cite this article as: Worly BL, et al, The relationship between progestin hormonal contraception and depression: a systematic review,
Contraception (2018), https://doi.org/10.1016/j.contraception.2018.01.010
 B.L. Worly et al. / Contraception xxx (2018) xxx–xxx 9

Table 6
Progestin-only contraception and depression studies in adolescents.

Author (year), Study design Screening Population Results Strengths Weaknesses Quality/bias
sources of support tool [8,9]

Gupta (2001) Prospective BDI 39 adolescent girls on DMPA, 24 Mean change in BDI for Prospective Not randomized, 54% lost Good,
[36], New cohort controls without hormonal DMPA group over 12 cohort of to follow-up in DMPA medium
England Massachusetts contraception months was −4.8 adolescent group, 50% lost to
Medical Center 1995–1999 Followed up at 3, 6 and 12 (p=.02), and +0.3 population, follow-up in control group
Research Funds months with BDI and MAACL-R (p=.84) for controls 12-month
Ages 15–21 Dysphoria subscale follow-up,
Excluded women with chronic change for DMPA group externally
illness, physical disabilities, −5.7 (p=.21) and validated
psychiatric illness requiring controls measures
hospitalization, or psychiatric −0.1 (p=.98)
medication, previous DMPA over Positive affect DMPA
last 6 months, OCPs over last 3 group
months −2.1 (p=.46) and
controls 0.1 (p=.98)
Cromer (1994) Prospective Self-reported 75 with OCP, 66 DMPA, 58 LNG No significant changes Prospective study Not randomized, no Poor, high
[37], Roessler 6-month symptoms implantable rod at 3 or 6 months in of special externally validated study,
Foundation, The cohort, Followed up for 6 months self-reported depression population over 6 short follow-up, small
Ohio State adolescent Age 11–20 for OCP 54%, 57%; LNG months sample size
University, health clinic Excluded patients who required implantable rod 51%,
Ortho Ohio confidentiality from 27%; DMPA 55%, 53%
Pharmaceuticals 1993 parents/caregivers Adjusted for headache,
dizziness, depression,
fatigue, nausea
Harel (1995) [38], Prospective Self-reported 36 adolescent women on DMPA Self-reported side Prospective Small study, no externally Poor, high
Maternal and cohort symptoms every 3 months, 27 who got 2nd effects of depression cohort of validated measures, no
Child Health Ohio Depo 6 weeks after first (26%) adolescent comparison data for the
Grant 1993–1994 injection, 15 who switched from population, cohorts, not randomized
Unmatched COCP to DMPA self-reported side
controls Followed up for 9 months effects
Ages 12–20
No exclusion criteria listed
Suman (1998) Retrospective Chart review 144 LNG subdermal implant 3/83 women requested Electronic Depression symptoms Poor, high
[25], funding descriptive patients rod removal due to medical record were self-reported as
source not study Ages 14–21 depression symptoms, review, noted in electronic
listed Minnesota Excluded patients living outside 11/130 patients evaluation of medical record, no control
1990–1993 catchment area, patients lost to reported depression adolescent group, retrospective, no
No control follow-up symptoms patients externally validated
group Adjusted for timing of measure
insertion, prior
pregnancies, unplanned
pregnancies, prior
delivery, marital status

results suggest a reduced depression effect in the adolescent population
receiving the LNG implant immediately after delivery but may not be
generalized to a more mature group of contraceptive users.
The trials mentioned above indicate that postpartum women may
experience higher depression scores if prescribed systemic progestins.
However, the evidence is not sufficiently robust, and there are enough
conflicting data that prescribers should not deprive these women of ef-
fective, reversible, long-acting contraception if they desire it. Discussion
of these possible negative effects could be an important part of postpar-
tum contraception counseling. Screening women for depressive symp-
toms after delivery, regardless of their contraceptive method, is
recommended during this vulnerable period (Table 5).
3.3.2. Adolescents
Adolescents are an important subset to consider in prescribing
progestin-only contraception since this group may be more vulnerable
to depression. Yet reliable contraception is particularly important
since teen pregnancy could have negative effects on education, profes-
sional advancement, salary and familial relationships.
There have been three prospective trials of adolescents using DMPA
looking for a potential association between DMPA and depression.
Gupta et al. [36] evaluated 39 adolescents given DMPA and 24 adoles-
cents who did not use any hormonal contraception. These groups
were administered the BDI and Multiple Affect Adjective Checklist-
Revised (MAACL-R) at baseline and 3, 6 and 12 months. Only 46% of
the DMPA group completed the 1-year study. The BDI scores did not sig-
nificantly change at 3, 6 or 12 months for DMPA users compared to the
control group, but a significant difference was identified between the
DMPA group at baseline and at 12 months (p=.03). Neither of the eval-
uated scores were significantly different for DMPA discontinuers, indi-
cating that depressive symptoms were not likely to be a cause of
discontinuation. The depression rates were not reported in this paper.
Cromer et al. [37] used healthcare provider interviews to assess de-
pression symptoms in teens prescribed DMPA at 3 or 6 months and
showed no significant change in depression diagnosis, but a high base-
line depression rate (53%) was notable. Harel et al. [38] compared ado-
lescents receiving DMPA every 3 months, those receiving DMPA at 6-
week intervals and COCP users who switched to DMPA, and followed
these patients for 9 months with healthcare provider interviews.
These authors hypothesized that this group of patients may benefit
from early administration of DMPA, as it might reduce medication side
effects if there is less fluctuation in dosage. Twenty-six percent of the
adolescents noted depression at the conclusion of the trial, but there
was no statistically significant difference between groups.
The strength of these studies is their prospective nature, but they
share in common small sample sizes, high dropout rates and substantial
prevalence of depression in adolescents that may serve as confounders.
The high dropout suggests that closer follow-up may be indicated in
teens to ensure compliance. As limited as the evidence is, it supports
the assertion that depression symptoms are no more likely in

Please cite this article as: Worly BL, et al, The relationship between progestin hormonal contraception and depression: a systematic review,
Contraception (2018), https://doi.org/10.1016/j.contraception.2018.01.010
10 B.L. Worly et al. / Contraception xxx (2018) xxx–xxx

Table 7
Progestin-only contraception studies in patients with depression

Author (year) sources of Study design Screening Population Results Strengths Weaknesses Quality/bias
support tool [8,9]

Young (2007) [39], Multisite HamD, 232 women on combined No significant differences Multisite trial, Small sample size in Fair,
National Institutes of prospective PDSQ, hormonal contraception, were found in major minimal progestin-only group, medium
Mental Health, cohort study QIDS-SR16, 58 on progestin-only depressive episodes, exclusion criteria, did not record reason for
Bristol-Myers Squibb STAR*D trial IDS-C30, methods, 948 on suicide attempts, externally starting hormonal
Company, Forest 2001–2004 SF-12 nonhormonal Hamilton Rating Scale for validated surveys contraception, not
Pharmaceuticals Inc., Unmatched contraception Depression scores or randomized
GlaxoSmithKline, King controls Ages 18–40 Inventory for Depressive
Pharmaceuticals, Followed up for 12 Symptomatology.
Organon Inc., Pfizer Inc., months Adjusted for age and
Wyeth-Ayerst Excluded psychotic major years of schooling
Laboratories depressive disorder
patients
Berenson (2008) [11], Prospective Self-reported 219 women on DMPA, 218 Self-reported depressive Prospective Not randomized, no Fair,
National Institutes of cohort study depression on COCP, and 171 with symptoms in cohort study with placebos, recall bias for medium
Child Health & Human Texas symptoms, nonhormonal nonhormonal externally retrospectively collected
Development, National 2001–2004 BDI, PANAS, contraception contraception patients validated surveys, data, no data on
Center for Research Bone Zung Ages 16–35 OR=1.00 (Ref), DMPA comparison symptom severity, 197
Resources mineral Anxiety Followed up for 2 years OR=1.08 (95% CI between women excluded or lost
density instrument Excluded patients with 0.71–1.62), COCP nonhormonal to follow-up. Many
survey current pregnancy, OR=0.91 (0.59–1.40). COCP, and DMPA exclusion criteria.
Controlled breastfeeding, planned BDI score 1.4 units lower methods
for age and pregnancy in b3 years, use for DMPA compared to
reproductive of DMPA in past 6 months, nonhormonal group at
method use of COCP in past 3 24-month follow-up
months, current use of (pb.05). Positive affect
hormonal IUD, scores 1.2 units lower for
contraindication to DMPA users compared to
hormonal contraception, nonhormonal group
lack of menses N3 months (pb.05).
in past year, bilateral Adjusted by race, age,
oophorectomy, over the follow-up visit number,
counter estrogen, high and baseline status of
dietary isoflavone intake, symptom.
illness effecting BMD,
eating disorder, strict
vegetarian diet
Cromer (1994) [37], Prospective Self-reported 75 with OCP, 66 DMPA, 58 No significant changes at Prospective study Not randomized, no Poor, high
Roessler Foundation, 6 month symptoms LNG implantable rod 3 or 6 months in of special externally validated
The Ohio State cohort, Followed up for 6 months self-reported depression population over 6 study, short follow-up,
University, Ortho adolescent Age 11–20 for OCP 54%, 57%; LNG months small sample size
Pharmaceuticals health clinic Excluded patients who implantable rod 51%, 27%;
Ohio required confidentiality DMPA 55%, 53%
1993 from parents/caregivers Adjusted for headache,
dizziness, depression,
fatigue, nausea

adolescents who choose DMPA compared to users of other hormonal
contraception options.
Unfortunately, the data on other forms of progestin-only contracep-
tion in adolescents are limited. The only other validated measure of de-
pression in adolescents using long-acting reversible contraception is the
study of postpartum adolescents using LNG implants mentioned earlier
[34]. Again, that study was retrospective and used the CESD to measure
depressive symptoms. The authors identified a high rate of baseline de-
pression (25%), which increased to 49% in girls who had the implant
inserted at 2–12 months.
Given that these studies identify a high rate of depression in adoles-
cents who seek contraception, it would be appropriate to screen adoles-
cent patients for depression regardless of their contraception choice.
However, there is not enough evidence to support withholding proges-
tin contraception from adolescents due to depression concerns, and an
additional retrospective adolescent self-report implant study supports
this concept [25]. This group is at higher risk for unintended pregnancy,
making it important to make reliable contraception available. While risk
of depression exists in all populations, DMPA and the subdermal im-
plantable rod in adolescents do not seem to be associated with an in-
creased risk compared to their adult counterparts (Table 6).
3.3.3. Depressed patients
Literature on prescribing progestin-only contraception to women
with preexisting mental health disorders is scant. Young et al. [39]
looked at contraceptive use in a cohort of women with a history of
major depression who were recruited for a treatment study. The Quick
Inventory of Depressive Symptomatology was used to measure depres-
sion side effects. Out of the 1238 women included in the study, only 58
were using progestin-only methods, primarily DMPA and the LNG im-
plant. Although this group was small, there were no differences in men-
tal health scores between progestin users and nonhormonal
contraceptive users (OR 1.07, p=.73). The progestin-only group was
noted to have lower physical functioning, but this was thought to reflect
the increased medical morbidity of women who were prescribed such
medication relative to healthier women using COCP.
In the Cromer et al. study evaluating adolescents mentioned earlier
[37], it was noted that there was a high baseline depression rate of 53%
in that study population. The participants were given that designation
based on self-report of depression in the prior 6 months, not necessarily
based on medical diagnosis. Nevertheless, in the group followed up at 6
months, the number still reporting depression remains constant for the
DMPA group and drops from 49% to 27% for the LNG implant group.

Please cite this article as: Worly BL, et al, The relationship between progestin hormonal contraception and depression: a systematic review,
Contraception (2018), https://doi.org/10.1016/j.contraception.2018.01.010
 B.L. Worly et al. / Contraception xxx (2018) xxx–xxx
Identification
Records identified through
database searching
(n=2,305)
Records after duplicates removed
(n=1,152)
Screening
Records screened
(n=1,155)
Abstracts assessed for
eligibility
Eligibility
(n=699)
Full text articles
assessed for eligibility
(n=41)
Included
Studies included in
systematic review
(n=26)
Fig. 1. Literature flow diagram in the systematic review.
Clearly, there is limited evidence regarding the use of these medica-
tions in women with a preexisting diagnosis of depression. However,
the World Health Organization supports the use of progestin-only
methods in women with a history of depression and lists depressive ill-
ness as a condition for which there should be no restriction in
recommending any contraceptive methods [40] (Table 7).
4. Conclusions
In summary, it is difficult to draw firm conclusions about the rela-
tionship between progestin-only hormonal contraception and depres-
sion. The data thus far do not support a clear, general relationship
between progestin contraceptives and depression scores or incident de-
pression diagnoses.
Several limitations exist in assessing the link between progestin-
only hormonal contraception and depression. The trials that use validat-
ed measures of depression symptoms use a variety of scales, some of
which evaluate mood or affect but do not specify depression criteria.
In those that have high dropout rates, the authors often do not specify
what, if any, side effects contributed to discontinuation. Retrospective
trials are confounded by many extrinsic influences that women have
in choosing a contraceptive method, many of which may also influence
the incidence of depression symptoms.
Nevertheless, some tentative conclusions may be drawn. The litera-
ture suggests that a minority of progestin-only contraception users may
experience depression symptoms (Level II-3 data), yet the most robust
studies show no association between depression diagnosis and
progestin-only contraception (Level I and II-1 data). Given the variety
of progestins and modes of administration in use, effects may be highly
variable based on receptor specificity and dosage, which make it
Please cite this article as: Worly BL, et al, The relationship between progestin hormonal contraception and depression: a systematic review,
Contraception (2018), https://doi.org/10.1016/j.contraception.2018.01.010
11
Additional records identified
through other sources
(n=3)
Records excluded
(n=460)
Abstracts excluded with
reason
(n=658)
Full text articles
excluded with reasons
(n=15)
difficult to draw conclusions about this entire class of hormonal contra-
ception. The concerns for depression in users of DMPA are unfounded,
as three prospective trials with externally validated surveys indicate
no depression effects, while one indicates a decrease in depression
scores. Statistically weaker, retrospective studies of DMPA users that in-
volve self-reported depression symptoms suggest that depression is
rare but may take 3 to 6 months to manifest, particularly in postpartum
patients. With this in mind, prescribers may find it valuable to schedule
a follow-up visit in this time frame to screen for depression symptoms.
While adverse effects are more commonly seen with systemic proges-
tins, locally acting progestin IUDs are unlikely to have as much effect
and would not qualify for this level of surveillance.
Medical literature offers little guidance on which patients may be
anticipated to have depressive effects on progestins. Additional research
is needed to verify that such concern is warranted with dosage and
timing of progestins commonly used in the United States. Although ad-
olescents may have high rates of depression at baseline, they are not
prone to worsening symptoms by using progestin-containing contra-
ceptives. This effective form of birth control should not be withheld in
this population at high risk of unintended pregnancy. Clearly, more re-
search is necessary before additional recommendations can be firmly
made. In the future, additional research with randomized control trials
and validated measures would be useful to better evaluate the magni-
tude and causality of progestin-induced depression effects.
Funding
This research did not receive any specific grant from funding agen-
cies in the public, commercial or not-for-profit sectors.
12 B.L. Worly et al. / Contraception xxx (2018) xxx–xxx

Appendix A. Direct evidence search terms. MESH terms and separate [11] Berenson AB, Odom SD, Breitkopf CR, Rahman M. Physiologic and psychologic symp-
toms associated with use of injectable contraception and 20 microgram oral contra-
abstract and title searches were performed ceptive pills. Am J Obstet Gynecol 2008;199:351.e1–351.e12.
[12] Civic D, Scholes D, Ichikawa L, LaCroix AZ, Yoshida CK, Ott SM, et al. Depressive
Progestin-only contraception. symptoms in users and non-users of depot medroxyprogesterone acetate. Contra-
ception 2000;61:385–90.
Depot medroxyprogesterone acetate: “medroxyprogesterone de- [13] Westhoff C, Wieland D, Tiezzi L. Depression in users of depo-medroxyprogesterone
pression”, Levonorgestrel IUD: acetate. Contraception 1995;51:351–4.
“levonorgestrel depression,” [14] Sanghi-Haghpeykar H, Poindexter AN, Bateman L, Ditmore JR. Experiences of inject-
able contraceptive users in an urban setting. Obstet Gynecol 1996;88:227–33.
Subdermal rod: [15] Paul C, Skegg DC, Williams S. Depot medroxyprogesterone acetate: patterns of use
“etonogestrel depression,” was also performed. and reasons for discontinuation. 1997;56:209–14.
Oral pills: [16] Westhoff C, Truman C, Kalmuss D, Cushman L, Rulin M, Heartwell S, et al. Depressive
symptoms and Norplant contraceptive implants. Contraception 1998;57:241–5.
“norethindrone depression”.
[17] Sihvo S, Ollila E, Hemminki E. Perceptions and satisfaction among Norplant users in
Other: Finland. Acta Obstet Gynecol Scand 1995;74:441–5.
“Progestin [AND] depression [AND] contraception,” and “progestin [18] Sivin I, Alvarez F, Mishell DR, Darney P, Wan L, Brache V, et al. Contraception with
[AND] postpartum depression,” two levonorgestrel rod implants. Contraception 1998;58:275–82.
[19] Kozlowski KJ, Rickert VI, Hendon A, Davis P. Adolescents and Norplant: preliminary
“Contraception [AND] levonorgestrel [OR] medroxyprogesterone findings of side effects. J Adolesc Health 1995;16:373–8.
acetate [OR] [AND] mood”. [20] Wong RC, Bell RJ, Thunuguntla K, McNamee K, Voollenhoven B. Implanon users are
less likely to be satisfied with their contraception after 6 months than IUD users.
Contraception 2009;80:452–6.
Appendix B. Abbreviations used in tables [21] Weisberg E, Bateson D, McGeechan K, Mohapatra L. A three-year comparative study
of continuation rates, bleeding patterns and satisfaction in Australian women using
a subdermal contraceptive implant or progestogen releasing-intrauterine system.
BDI: Beck Depression Inventory Eur J Contracept Reprod Health Care 2014;19:5–14.
CESD: Center for Epidemiologic Studies- Depression Scale. [22] Arribas-Mir L, Rueda-Lozano D, Agrela-Cardona M, Cedeno-Benavides T, Olvera-
CI: Confidence Interval. Porcel C, Beuno-Cavanillas A. Insertion and 3-year follow-up experience of 372
etonorgestrel subdermal contraceptive implants by family physicians in Granada,
CIDI: Composite International Diagnostic Interview. Spain. Contraception 2009;80:457–62.
COCP: Combined oral contraceptive pills. [23] Harvey C, Seib C, Lucke J. Continuation rates and reasons for removal among
DMPA: Depot medroxyprogesterone acetate. Implanon users accessing two family planning clinics in Queensland, Australia. Con-
traception 2009;80:527–32.
EPDS: Edinburgh Postnatal Depression Scale.
[24] Inoue K, Kelly M, Barratt A, Bateson D, Rutherford A, Black K, et al. Australian
GHQ: General Health Questionnaire-12. women's experiences of the subdermal implant: a qualitative perspective. Aust
HamD: Hamilton Rating Scale for Depression. Fam Physician 2016;45:734–9.
[25] Suman VJ, Van Winter JT, Evans MP, Simmons PS, Jacobsen SJ, Manolis AL. Levonor-
IDS-C30: Inventory of Depressive Symptomatology.
gestrel contraceptive implants in female patients 14 to 21 years old. Mayo Clin Proc
IUD: Intrauterine device. 1998;73:10–6.
LNG: Levonorgestrel. [26] Elovainio M, Teperi J, Aalto A, Grenman S, Kivela A, Kujansuu E, et al. Depressive
MAACL-R: Multiple Affect Adjective Checklist-Revised. symptoms as predictors of discontinuation of treatment of menorrhagia by
levonorgestrel-releasing intrauterine system. Int J Behav Med 2007;14:70–5.
MADRS: Montgomery–Asberg Depression Rating Scale. [27] Pekin AT, Kerimoglu OS, Kebapcilar AG, Yilmaz SA, Benzer N, Celik C. Depressive
MHI: Mental Health Inventory. symptomatology and quality of life assessment among women using the
PANAS: Positive and Negative Affect Scale. levonorgestrel-releasing intrauterine system: an observational study. Arch Gynecol
Obstet 2014;290:507–11.
PBAC: Pictorial blood loss assessment chart. [28] Toffol E, Heikinheimo O, Koponen P, Luoto R, Partonen T. Hormonal contraception
PDSQ: Psychiatric Diagnostic Screening Questionnaire. and mental health: results of a population-based survey. Hum Reprod 2011;26:
PID: Pelvic inflammatory disease. 3085–93.
[29] Nilssson CG, Luukkainen T, Diaz J, Allonen H. Clinical performance of a new
POP: Progestin-only pills. levonorgestrel-releasing intrauterine device. A randomized comparison with a
QIDS-SR16: Quick Inventory of Depressive Symptomatology-Self- Nova-T-copper device. Contraception 1982;25:345–55.
Report. [30] Skovlund CW, Morch LS, Kessing LV, Lidegaard O. Association of hormonal contra-
ception with depression. JAMA Psychiat 2016;73:1154–62.
RCT: Randomized controlled trial.
[31] Graham CA, Ramos R, Bancroft J, Maglaya C, Farley TM. The effects of steroidal con-
SF-36: Short-Form Health Survey-36. traceptives on the well-being and sexuality of women: a double-blind, placebo-
SF-12: Short Form Health Survey. controlled, two-centre study of combined and progestin-only methods. Contracep-
tion 1995;52:363–9.
[32] Singata-Madliki M, Hofmeyr GJ, Lawrie TA. The effect of depot medroxyprogesterone
References acetate on postnatal depression: a randomized controlled trial. J Fam Plann Reprod
Health Care 2016:1–6.
[1] Bakalar N. “Contraceptives tied to depression risk.” The New York Times. Published [33] Lawrie TA, Hofmeyr GJ, De Jager M, Berk M, Paiker J, Viljoen E. A double blind ran-
9/30/2016. domized placebo controlled trial of postnatal of norethisterone enanthate: the effect
[2] Mettler, K. “It's not in your head: striking new study links birth control to depres- on postnatal depression and serum hormones. Br J Obstet Gynaecol 1998;105:
sion.” The Washington Post. Published 10/5/2016. 1082–90.
[3] Weissman MM, Bland RC, Canino GJ, Faravelli C, Greenwald S, Hwu HG, et al. Cross- [34] Tsai R, Schaffir J. Effect of depot medroxyprogesterone acetate on postpartum de-
national epidemiology of major depression and bipolar disorder. JAMA 1996; pression. Contraception 2010;82:174–7.
276(4):293–9. [35] Stevens-Simon C, Kelly L, Wallis J. The timing of Norplant insertion and postpartum
[4] Kaura V, Ingram CD, Gartside SE, Young AH, Judge SJ. The progesterone metabolite depression in teenagers. J Adolesc Health 2000;26:408–13.
allopregnanolone potentiates GABA(a) receptor-mediated inhibition of 5-HT neuro- [36] Gupta N, O'Brien R, Jacobsen LJ, Davis A, Zuckerman A, Supran S, et al. Mood changes
nal activity. Eur Neuropsychopharmacol 2007;17:108–15. in adolescents using depot-medroxyprogesterone acetate for contraception: a pro-
[5] Li Y, Pehrson AL, Budac DP, Sanchez C, Gulinello M. A rodent model of premenstrual spective study. J Pediatr Adolesc Gynecol 2001;14:71–6.
dysphoria: progesterone withdrawal induces depression-like behavior that is differ- [37] Cromer BA, Smith D, Blair JM, Dwyer J, Brown RT. A prospective study of adolescents
entially sensitive to classes of antidepressants. Behav Brain Res 2012;234:238–47. who chose among levonorgestrel implant (Norplant), medroxyprogesterone acetate
[6] The Upjohn company, NDA 20-246. Application summary. Clinical data summary; 1992. (depo-Provera), or the combined-oral contraceptive pill as contraception. Pediatrics
[7] Mohrer D, Liberait A, Tetzlaff J, Altman DG. Preferred reporting items for systematic 1994;94:687–94.
reviews and meta-analyses: the PRISMA statement. Br Med J 2009;339:b2535. [38] Harel Z, Biro FM, Kollar LM. Depo-Provera in adolescents: effects of early second in-
[8] US Preventive Services Task Force. Procedure manual. https://www. jection or prior oral contraception. J Adolesc Health 1995;16:379–84.
uspreventiveservicestaskforce.org/Page/Name/procedure-manual; 2015, Accessed [39] Young EA, Kornstein SG, Harvey AT, Wisniewski SR, Barkin J, Fava M, et al. Influ-
date: 29 December 2017. ences of hormone-based contraception on depressive symptoms in premeno-
[9] Collaboration TC. In: Green JPHaS, editor. Cochrane handbook for systematic reviews pausal women with major depression. Psychoneuroendocrinology 2007;32:
of interventions: Cochrane book series. A John Wiley & Sons, Ltd.; 2008 843–53.
[10] Westhoff C, Truman C, Kalmuss D, Cushman L, Davidson A, Rulin M, et al. Depressive [40] World Health Organization. Medical eligibility criteria for contraceptive use. Geneva:
symptoms and depo-Provera. Contraception 1998;57:237–40. World Health Organization; 2015.

Please cite this article as: Worly BL, et al, The relationship between progestin hormonal contraception and depression: a systematic review,
Contraception (2018), https://doi.org/10.1016/j.contraception.2018.01.010