Contraception xx (2015) xxx – xxx

Original research article

Comparative contraceptive effectiveness of levonorgestrel-releasing and

copper intrauterine devices: the European Active Surveillance Study for
Intrauterine Devices
Klaas Heinemann⁎, Suzanne Reed, Sabine Moehner, Thai Do Minh
ZEG-Berlin, Chausseestrasse 115, 10115 Berlin, Germany
Received 2 October 2014; revised 8 January 2015; accepted 12 January 2015

Abstract

Objectives: The objective was to measure the rate of unintended pregnancies in women using levonorgestrel-releasing intrauterine systems
(LNG IUSs, releasing 20 mcg LNG daily) and copper intrauterine devices (IUDs) in a typical population of IUD users and to describe
associated complications.
Methods: A multinational, prospective, non-interventional cohort study of new users of LNG IUS and copper IUDs was performed.
Following a baseline survey, study participants and their physicians completed one follow-up questionnaire after 12 months. A multifaceted
four-level follow-up procedure minimized loss to follow-up. Patient-reported outcomes were validated by the treating physicians.
Results: A total of 61,448 women with a newly inserted IUD were enrolled in six European countries between 2006 and 2012. The copper
IUD cohort contained more than 30 different types. Validated 1-year follow-up information for 58,324 users between 18 and 50 years of age
(70% using LNG IUS, 30% using copper IUDs) was collected. A total of 118 contraceptive failures occurred (26 LNG, 92 copper). Both
types of IUD were highly effective, with overall Pearl indices of 0.06 [95% confidence interval (CI): 0.04–0.09] and 0.52 (95% CI:
0.42–0.64) for LNG IUS and copper IUDs, respectively. The adjusted hazard ratio for LNG IUS vs. copper IUDs was 0.16 (95% CI:
0.10–0.25). Tenty-one pregnancies (7 LNG IUS, 14 copper IUD) were ectopic, yielding an adjusted hazard ratio for ectopic pregnancy of
0.26 (95% CI: 0.10–0.66).
Conclusions: The contraceptive failure rate was low with both IUDs. However, the LNG IUS was associated with a significantly lower risk
of pregnancy, including ectopic pregnancy, than the copper IUDs.
Implications: To our knowledge, this is the first large-scale, multinational, prospective epidemiological study to measure and compare the
contraceptive effectiveness of LNG IUSs and copper IUDs during routine clinical practice. Clinicians and patients should be aware of
differences in rates of unintended pregnancies and associated complications in relation to IUD us.
© 2015 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).

Keywords: Intrauterine device (IUD); Levonorgestrel (LNG); Copper; Ectopic pregnancy; EURAS IUD

1. Introduction respective ectopic pregnancy rates range from 0.02–0.2 in

LNG IUS users and from 0.1–0.8 in copper IUD users
Contraceptive failure rates for both copper intrauterine [3,4,8,11].
devices (IUDs) and levonorgestrel-releasing intrauterine Results from large-scale prospective studies on contra-
systems (LNG IUSs) are among the lowest for reversible ceptive effectiveness and pregnancy-related outcomes,
methods of contraception [1,2], with studies of various types including ectopic pregnancies, have not previously been
showing rates of around 0.1–2.2 per 100 WY for copper available for LNG IUS or for LNG IUS in comparison to
IUDs and 0.1–0.6 per 100 WY for LNG IUSs [3–10]. The copper IUDs. This article presents the final results on
⁎ Corresponding author at: ZEG-Berlin Center for Epidemiology and
pregnancy-related outcomes from the European Active
Health Research Invalidenstrasse 115 10115 Berlin Germany. Tel.: + 49 30 Surveillance Study for Intrauterine Devices (EURAS IUD).
945 101 24, +49 1717 554 556 (mobile); fax: +49 30 945 101 26. Results on uterine perforation from the EURAS IUD study
E-mail address: [email protected] (K. Heinemann). are reported separately [12].
http://dx.doi.org/10.1016/j.contraception.2015.01.011
0010-7824/© 2015 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
2 K. Heinemann et al. / Contraception xx (2015) xxx–xxx

The objective of this article was to compare the contracep- about any potential complications associated with the IUD,
tive effectiveness and pregnancy-related outcomes, including medical checkups, illnesses, hospitalization and pregnancy
ectopic pregnancies, for users of LNG and copper IUDs. that occurred during the time period since IUD insertion.
All events and pregnancy-related outcomes reported by
study participants, including subjectively perceived symp-
2. Methods toms and the diagnoses as understood by them, were entered
into the study database. These report forms were immedi-
The purpose of the EURAS IUD was to identify and ately passed on to the study center’s medical reviewer or
compare the risks of LNG IUS and copper IUD use in a review group, which contacted the participants for clarifica-
population of typical users of IUDs. The methods are tion if necessary and the treating physicians for validation.
described in detail elsewhere [12]. The primary outcome of The follow-up questionnaire sent to physicians 12 months
interest in EURAS IUD was the incidence of uterine after insertion recorded any pregnancy outcomes in addition
perforation. A further a priori objective of the EURAS IUD to examination dates, IUD position (including unnoticed
study (and the focus of this article) was to compare the expulsion), uterine perforation, complications and patients’
contraceptive effectiveness and pregnancy-related out- medical conditions.
comes for users of LNG versus copper IUDs, including The sample size of this study was chosen to test
ectopic pregnancies. noninferiority of LNG IUS regarding the perforation risk
EURAS IUD was a multinational, prospective, con- (the primary outcome of interest in this study) in comparison
trolled, long-term cohort study with recruitment in six to copper IUDs. Sample size calculations showed that 60,000
European countries (Austria, Finland, Germany, Poland, participants would be sufficient to exclude a 1.7-fold risk.
Sweden, UK) from 2006 to 2012. Its two cohorts consisted These calculations were based on an estimated perforation
of new users of LNG and copper IUDs. rate of 0.5 per 1000 insertions, a one-sided α of 0.025 and a
A non-interference approach was chosen; participating power (1 − β) of 0.80.
health care providers prescribed and provided the method as Crude as well as adjusted hazard ratios (HRs) were
they normally would. The EURAS IUD study was approved calculated. The appropriate confounding variables were built
by the ethical committee of the physicians' association in into the model. Based on the expectation of a small absolute
Berlin, Germany, and the Ethics Committee of Hospital number of unintended pregnancies, the confounding vari-
District of Southwest Finland. The other participating ables were limited to a small number of predefined risk
countries accepted these approvals. The final results of factors (age, BMI, parity) recommended by the independent
this study were reported to the European Medicines Safety Monitoring and Advisory Council. All analyses were
Agency, to the US Food and Drug Administration and performed with the statistical package STATA 9.0.
other health authorities.
Recruitment of study participants was conducted via a
network of health care professionals (HCPs), such as 3. Results
gynecologists and midwives who regularly insert IUDs,
either office based or in specialized clinics. All women with Recruitment started in Germany, Austria and Finland in
a newly inserted IUD were eligible for enrollment. 2006, the United Kingdom in 2008, Sweden in early 2009
Therefore, the copper IUD cohort consists of numerous and Poland in November 2009. Recruitment stopped at the
different types of copper IUDs primarily characterized here end of 2012. In total, 61,448 women with a newly inserted
by their surface area (b 300 mm 2, ≥ 300 mm 2). After the IUD were enrolled by more than 1200 participating HCPs. In
decision to initiate an IUD was made, participating HCPs total, 1235 women, or 2.0% (1.7% for LNG IUS, 2.8% for
asked the women whether they were willing to participate in copper IUD), were lost to follow-up during the 1-year
the study. Because of the non-interference approach, follow-up period. For all analyses regarding unplanned
eligibility criteria were minimal: these included a willingness pregnancies, only women in the age group of 18–50 years
to sign an informed consent form and data privacy form, and were included. This analysis includes validated follow-up
an absence of a language barrier that could prevent the information for a total of 58,324 women in this age group:
patient from completing the questionnaires. 41,001 users of LNG IUS and 17,323 users of copper IUDs,
At the time of IUD insertion, study participants completed resulting in 44,633 and 17,703 WYs of observation,
a baseline questionnaire on which they recorded information respectively. More than 30 types of copper IUDs were
about their state of health and potential risk factors. These included in the copper IUD cohort, the most frequent ones
included medical and lifetime reproductive history, medica- being NovaT (200 or 380) with 37%, T Safe Cu 380 with
tion history, history of contraceptive use, age, body mass 18% and Multiload CU (250 or 375) with 14%. The copper
index (BMI), smoking, alcohol, exercise and lifting of heavy surface areas were less than 300 mm 2, 300 mm 2 and more
objects, and level of education. Study participants and the than 300 mm 2 in 7.8%, 1.6% and 71.3% of the inserted
inserting HCP received a follow-up questionnaire 12 months copper IUDs inserted in this study, respectively. In 19.3%,
after enrollment. The follow-up survey recorded information the surface area was not known.
 K. Heinemann et al. / Contraception xx (2015) xxx–xxx 3
1.0
The women in the LNG IUS and copper IUD cohorts had
0.9
mean ages of 37.4 and 33.3 years, respectively. More than
0.8
40% of the LNG IUS users in this study were over 40 years
0.7
of age, as compared to only 24% in the copper IUD group.

Hazard Ratio
0.6
Another difference between cohorts with potential impact on
0.5
contraceptive failure included the prevalence of breastfeed-
0.4
ing at time of insertion (9.2% vs. 14.6%). In the LNG IUS
0.3
cohort, 19.8% of users had given birth during the 12 months
0.2
preceding insertion, compared to 28.7% of copper IUD
0.1
users. Some 93.0% of the LNG IUS users and 88.0% of the
0.0
copper IUD users had ever been pregnant. The proportion of Contraceptive Failure Ectopic Pregnancies
smokers in both cohorts was comparable. The educational
level of the copper IUD users and LNG IUS users was * adjusted for age, body mass index and parity
comparable: 55.6% of the copper IUD users had at least
Fig. 1. Relative risk of contraceptive failure and of ectopic pregnancy in
university entrance level, compared to 49.6% of the LNG
LNG IUS compared to copper IUD users: Adjusted* HR (95% CI).
IUS users. More detailed information about baseline
characteristics has been reported separately [12].
One hundred eighteen validated, unintended pregnan-
cies occurred during the 1-year follow-up period: 26 in stratified by age groups (18–b30 years, 30–b40 years,
LNG IUS users and 92 in copper IUD users. Twenty-nine of 40–50 years) (Table 1). The respective estimate for LNG
these unintended pregnancies (13 LNG IUS, 16 copper IUS and copper users in the age group of 18–35 years was
IUD) occurred after an unrecognized IUD expulsion and 0.15 and 0.97, respectively. The risk of contraceptive failure
were therefore considered to have resulted from a failure of in LNG IUS users compared to copper IUD users was
the contraceptive method. They were therefore included in statistically significantly lower in all age groups except for
this analysis. women between 40 and 50 years of age.
In 27 pregnant women (1 LNG IUS, 26 copper IUD), the Stratification by country also showed statistically signif-
IUD was found dislocated (not in the uterine cavity, but icantly lower risks for LNG IUS users compared to copper
still in the body). While it is possible to attribute IUD users in all countries except for Poland, where the
dislocation of the IUD to the mechanical forces of the sample size was too small. In Poland, no pregnancies with
pregnancy itself, over 80% of these cases were diagnosed LNG IUS occurred, and there were four pregnancies with
in the very early stages of pregnancy (6 weeks or less after copper IUDs (data not shown).
conception), most likely before the position of the IUD Cox regression analysis that included the predefined
could have been affected. prognostic factors age, BMI, and parity yielded an adjusted
In three cases of unintended pregnancy (three copper HR of 0.16 (95% CI: 0.10–0.25) for LNG vs. copper IUDs
IUD) the IUD was perforated. The remaining unintended (Fig. 1). The same comparison excluding those pregnancy
pregnancies occurred with the IUD entirely within the uterus. cases with an unnoticed expulsion of the IUD yielded an
Contraceptive failure rates differed markedly between adjusted HR of 0.10 (95% CI: 0.05–0.18).
cohorts. The overall Pearl index (PI; pregnancies per 100 Significant differences between cohorts were evident in
WYs) in the LNG IUS cohort was 0.06 [95% confidence the ectopic pregnancy rates. Seven women with LNG IUS
interval (CI): 0.04–0.09); the respective PI for the copper and 14 women with copper IUDs had an ectopic pregnancy,
IUD users was 0.52 (95% CI: 0.42–0.64). The life-table resulting in incidence rates of 0.02 per 100 WY (95% CI:
estimates of the rate of contraceptive failure for the first year 0.01–0.03) and 0.08 per 100 WY (95% CI: 0.04–0.13),
of use were 0.07% and 0.63% for LNG IUS and copper IUD, respectively. The proportion of ectopic pregnancies among
respectively. all contraceptive failure pregnancies was higher in LNG IUS
This resulted in a crude risk ratio for LNG IUS use users compared to copper IUD users (27% vs. 15%, p = .16),
compared to copper IUD use of 0.11 (95% CI: 0.07–0.17). but due to the substantially lower risk of contraceptive failure
To control for potential differences in contraceptive failure in LNG IUS users, the overall risk for ectopic pregnancies
rates associated with different surface areas in the copper was significantly lower in LNG IUS users compared to
IUD cohort, stratified analyses were conducted using copper IUD users [HR 0.20 (95% CI: 0.08–0.48)]. The HR
300 mm 2 as the cutoff point (b 300 mm 2, ≥ 300 mm 2). did not change substantially after adjustment for age, BMI
The pearl indices for copper IUD with less than 300 mm 2 and parity [HR 0.26 (95% CI: 0.10–0.66)] (Fig. 1).
was 0.56 (95% CI: 0.24–1.09), and 0.62 (95% CI: Two of the seven ectopic pregnancies in LNG IUS users
0.50–0.78) for 300 mm 2 or greater. and 10 of the 14 ectopic pregnancies in copper IUD users
Age distribution differed markedly between cohorts, so resulted in a salpingectomy. In addition, two of the
the life-table estimates of the rate of contraceptive failure for extrauterine pregnancies in copper IUD users led to removal
the first year of use for the cohorts were therefore calculated of an ovary.
4 K. Heinemann et al. / Contraception xx (2015) xxx–xxx

Table 1
Life-table estimates for contraceptive failure in first year of use by IUD cohort by age group.
Age group LNG IUD Copper IUD
No. of pregnancies LT est. (95% CI) No. of pregnancies LT est. (95% CI)
18–b30 years 7 0.16 (0.03–0.28) 54 1.35 (0.95–1.76)
N = 30–b40 years 16 0.12 (0.06–0.18) 36 0.57 (0.37–0.77)
N = 40–b51 years 3 0.01 (0–0.03) 2 0.05 (0–0.12)

4. Discussion This epidemiological study investigated routine clinical

practice in an international environment and probably allows
The contraceptive effectiveness was very high in both extrapolation to other populations. The greater effectiveness of the
cohorts in this study. However, use of LNG IUS in this study LNG IUS was evident for all age groups of women and is unlikely
was associated with one eighth the incidence of pregnancy to be due to bias, confounding or chance. Clinicians and women
during the first year of use, including a marked reduction in seeking contraception need to be aware of these differences in the
ectopic pregnancy, compared to use of a copper IUD. A contraceptive effectiveness in contemporary IUDs.
discrepancy this large is unlikely to be explained by effects
of selection bias or information bias [13–16]. Differential References
losses to follow-up are also unlikely to have caused the
disparity: the study achieved a low overall loss to follow-up [1] Winner B, Peipert JF, Zhao Q, Buckel C, Madden T, Allsworth JE, et al.
with 1.7% and 2.8% in the LNG IUS and copper IUD users, Effectiveness of long-acting reversible contraception. N Engl J Med
respectively. Our analysis controlled for important potential 2012;366(21):1998–2007.
[2] Trussell J. Contraceptive failure in the United States. Contraception
confounding factors, and the results were unchanged. The
2011;83(5):397–404.
significantly lower risk for ectopic pregnancy in LNG IUS [3] Andersson K, Odlind V, Rybo G. Levonorgestrel-releasing and
users compared to copper IUD users is not likely to be due to copper-releasing (Nova T) IUDs during five years of use: a randomized
chance. Hence, we believe these findings are robust. comparative trial. Contraception 1994;49(1):56–72.
A wide range of pearl indices for LNG IUS has been [4] Backman T, Rauramo I, Huhtala S, Koskenvuo M. Pregnancy during
the use of levonorgestrel intrauterine system. Am J Obstet Gynecol
reported in the scientific literature. A cross-sectional study
2004;190(1):50–4.
by Backman in more than 17,000 women, for example, [5] Kulier R, O'Brien PA, Helmerhorst FM, Usher-Patel M, D'Arcangues C.
showed a PI of 0.11 [4]. Similar results of 0.1–0.2 were Copper containing, framed intra-uterine devices for contraception.
found in reviews of randomized clinical trials [7–9]. Other Cochrane Database Syst Rev 2007;4:1-152 [CD005347].
authors have found higher rates of contraceptive failure. [6] Cox M, Blacksell S. Clinical performance of the levonorgestrel intra-
uterine system in routine use by the UK Family Planning and
Cox, for example, found a PI of 0.6 in an open-label,
Reproductive Health Research Network: 12-month report. Br J Fam
single-group study in almost 600 women [6]. Plann 2000;26(3):143–7.
As in our study, PIs for various copper IUD systems are [7] Sivin I, el Mahgoub S, McCarthy T, Mishell Jr DR, Shoupe D, Alvarez F,
generally found to be higher than those for LNG IUS, although et al. Long-term contraception with the levonorgestrel 20 mcg/day (LNg
not necessarily to the extent that we found here. A Cochrane 20) and the copper T 380Ag intrauterine devices: a five-year randomized
study. Contraception 1990;42(4):361–78.
review of randomized controlled trials reported PIs of 0.5–2.2
[8] Sivin I. Dose- and age-dependent ectopic pregnancy risks with
for copper IUDs with a surface area of less than 300 mm 2, and intrauterine contraception. Obstet Gynecol 1991;78(2):291–8.
of 0.1–1.0 for greater than 300 mm 2 [5]. Sivin et al. and [9] Stoddard A, McNicholas C, Peipert JF. Efficacy and safety of long-
Mansour et al. reviewed several randomized clinical trials and acting reversible contraception. Drugs 2011;71(8):969–80.
found results in the same range [8,10]. Two additional [10] Mansour D, Inki P, Gemzell-Danielsson K. Efficacy of contraceptive
randomized clinical trials in more than 2000 women, methods: a review of the literature. Eur J Contracept Reprod Health
Care 2010;15(1):4–6.
respectively, showed similar PIs of 0.3–1.3 [3,7]. [11] Ganacharya S, Bhattoa HP, Batár I. Ectopic pregnancy among non-
These results on ectopic pregnancies are consistent with medicated and copper-containing intrauterine device users: a 10-year
findings from other studies. Previously reported ectopic follow-up. Eur J Obstet Gynecol Reprod Biol 2003;111(1):78–82.
pregnancy rates for LNG IUS users ranged from 0.02 to 0.2 [12] Heinemann K, Reed SR, Moehner S, Thai DM. Risk of uterine
perforation with levonorgestrel-releasing and copper intrauterine devices
per 100 WY, and rates for copper IUD users were from 0.1 to
in the European Active Surveillance Study on Intrauterine Devices; 2015.
0.8 per 100 WY [3,4,8,11]. [13] Taubes G. Epidemiology faces its limits. Science 1995;269(5221):164–9.
In this study, we collected no information on concurrent [14] Hill AB. The environment and disease: association or causation? Proc
use of other contraceptive methods. However, use of R Soc Med 1965;58:295–300.
additional contraceptive methods is probably more prevalent [15] Grimes DA, Schulz KF. False alarms and pseudo-epidemics:
in younger women not yet living in stable partnerships. As the limitations of observational epidemiology. Obstet Gynecol
2012;120(4):920–7.
the copper IUD cohort was markedly younger, this would [16] Khoury MJ, James LM, Flanders WD, Erickson JD. Interpretation of
probably have led to an underestimation of the risk ratio recurring weak associations obtained from epidemiologic studies of
rather than an overestimation. suspected human teratogens. Teratology 1992;46(1):69–77.