The European Journal of Contraception & Reproductive

Health Care

ISSN: 1362-5187 (Print) 1473-0782 (Online) Journal homepage: http://www.tandfonline.com/loi/iejc20

Perforation risk and intra-uterine devices: results

of the EURAS-IUD 5-year extension study

Clare Barnett, Sabine Moehner, Thai Do Minh & Klaas Heinemann

To cite this article: Clare Barnett, Sabine Moehner, Thai Do Minh & Klaas Heinemann
(2018): Perforation risk and intra-uterine devices: results of the EURAS-IUD 5-year
extension study, The European Journal of Contraception & Reproductive Health Care, DOI:
10.1080/13625187.2017.1412427

To link to this article: https://doi.org/10.1080/13625187.2017.1412427

Published online: 11 Jan 2018.

Submit your article to this journal

Article views: 35

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=iejc20
THE EUROPEAN JOURNAL OF CONTRACEPTION & REPRODUCTIVE HEALTH CARE, 2018
https://doi.org/10.1080/13625187.2017.1412427

ORIGINAL RESEARCH ARTICLE

Perforation risk and intra-uterine devices: results of the EURAS-IUD 5-year

extension study
Clare Barnett, Sabine Moehner, Thai Do Minh and Klaas Heinemann
ZEG Berlin – Berlin Center for Epidemiology and Health Research, Berlin, Germany

ABSTRACT ARTICLE HISTORY

Objectives: The objective of this analysis was to identify intra-uterine devices (IUD) perforations Received 24 November 2017
detected from 12 to 60 months following IUD insertion, and to combine this information with (our Accepted 28 November 2017
previously published) data about perforations detected in the first 12 months to calculate cumula- Published online 11 January
tive perforation rates. 2018
Methods and materials: Prospective, non-interventional cohort study with new users of levonor- KEYWORDS
gestrel-releasing intra-uterine systems (LNG-IUS) and copper-IUD. The original cohort included IUD; breastfeeding;
61,448 women followed for 12 months. Of these, we had sufficient resources to perform an add- prospective cohort study;
itional follow-up and analysis at 60 months in 39,009 women. Inclusion criteria for this analysis was active surveillance;
insertion prior to 31 July 2010. All potential cases were validated via the health care professional or levonorgestrel; childbirth;
medical records. Crude and adjusted relative risks were calculated using a logistic regression contraception
model.
Results: We identified 23 additional perforations (19 LNG-IUS and 4 copper-IUD) more than 12
months after insertion. Added to perforations detected at 12 months, the overall perforation rate
was 2.1 per 1000 insertions (95% CI: 1.6–2.8) for LNG-IUS users (40 þ 19 perforations/27,630 inser-
tions) and 1.6 per 1000 insertions (95% CI: 0.9–2.5) for copper-IUD users (14 þ 4 perforations/11,379
insertions). LNG-IUS users had a borderline higher risk of perforation compared with copper-IUD
users (ORadj 1.7; 95% CI: 1.0–2.8). Forty-five (58%) of the 77 perforations were associated with sus-
pected risk factors. Breastfeeding (RR 4.9, 95% CI: 3.0–7.8) and time since delivery (RR 3.0, CI:
1.5–5.4) remained significant risk factors in perforations detected after 12 months. No perforations
resulted in serious injury to intra-abdominal or pelvic structures.
Conclusions: The incidence of uterine perforations in this study was low, although higher than the
commonly reported rate. Approximately one third of perforations are detected 12 months after
insertion. Clinical sequalae of perforations are generally mild and associated with a very low risk of
injury to intra-abdominal and pelvic structures.

IMPLICATIONS
 Uterine perforation is a rare risk associated with intra-uterine device use. Late diagnosed perfora-
tions can occur, although women can be reassured that the morbidity associated with detection
and removal is low.

Introduction that up to one quarter of all IUD perforation cases were

Levonorgestrel-releasing intrauterine systems (LNG-IUSs) are
an effective and safe contraceptive with additional non-
contraceptive benefits including treatment for menorrhagia,
endometriosis and endometrial hyperplasia. Concerns
remain regarding the overall safety of intra-uterine devices
(IUDs), particularly regarding potential serious complications
associated with uterine perforation. Until recently, the rela-
tive risk of uterine perforation with LNG-IUS compared to
copper-IUD was poorly studied. We found a relative risk of
perforation in users of LNG-IUS compared to copper-IUD of
1.61 (95% CI: 0.96–2.70) at 12 months after insertion when
adjusted for age, body mass index (BMI), breastfeeding at
time of insertion and parity [1]. Although low overall, per-
foration rates were significantly higher among women who
were breastfeeding at the time of insertion compared to
those not breastfeeding, irrespective of the temporal rela-
tionship to delivery.
In 2011, during the recruitment phase of the 12-month
study, van Grootheest et al. [2] published data suggesting
diagnosed more than 1 year after insertion. Therefore, it
was decided to prolong the follow-up phase of EURAS-IUD
for a sub-cohort of women to capture IUD perforation cases
detected between 12 and 60 months following insertion
and identify whether the safety and efficacy of LNG-IUS
and copper-IUDs change when late-perforation cases are
included.
Methods
EURAS IUD (NCT00461175) was a multinational, prospective,
controlled cohort study that followed women for 12
months following an IUD insertion to measure incidence
and risk factors associated with uterine perforation.
Recruitment was based on an all-comer approach and the
final proportion of LNG-IUS and copper-IUD users was
determined by prescribing practices. Further details on
the study methodology can be found in previous publica-
tions [1,3].

CONTACT Klaas Heinemann k.heinemann@zeg-berlin.de ZEG-Berlin – Berlin Center for Epidemiology and Health Research, Invalidenstr. 115, 10115
Berlin, Germany
ß 2018 The European Society of Contraception and Reproductive Health
2 C. BARNETT ET AL.
Baseline and 12-month follow-up data from all women
recruited into EURAS IUD before 31 July 2010 – in total
39,009 – formed the 12–60 month extension study
cohort. Of these, all women with an IUD in situ at 12
months received a follow-up questionnaire 5 years after
IUD insertion (30,144 women). Women who had had their
IUDs removed at 12 months due to perforation, expulsion
or personal preference were not sent 5-year follow-up
questionnaires, however, their data were included in the
analyses of total incidence rates and risk factors for
perforation.
Sample size was based on published data from four
national pharmacovigilance centres showing that approxi-
mately 25% of perforation cases were diagnosed more than
12 months after insertion [2]. Using this estimate, we calcu-
lated that approximately 30,000 women whose IUD was still
in place at 12 months, would be required to estimate the
anticipated proportion of additionally diagnosed perfora-
tions with acceptable precision. This calculation assumed a
response rate of 2/3.
The primary outcome for the 12-month and 60-month
analysis was the incidence of uterine perforation, including
the proportion of uterine perforations associated with ser-
ious clinical complications, the interval between IUD inser-
tion and diagnosis of uterine perforation and the impact of
IUD insertion in close proximity to childbirth on the uterine
perforation rate. For each of these outcome measures, com-
parisons were made between LNG and copper-IUDs. All
events reported by the participating health care providers
(HCPs) and/or study participants where any part of the
device crossed the endometrium and entered the myome-
trium was considered a perforation. Previously performed
sensitivity analysis conducted on the 12-month data
showed that limiting events to perforations completely
crossing the myometrium did not significantly alter risk
ratios [4].
Baseline data collected from study participants included
information on medical and gynaecological history, medica-
tion, contraceptive use, age, BMI, lifestyle factors and edu-
cation [1,3]. Specific questions were asked on breastfeeding
and time since last delivery. Physicians provided informa-
tion on the IUD insertion procedure and documented any
IUD perforations identified during or immediately following
insertion. At 12 and 60 months, women were contacted
and information obtained on IUD use and complications
associated with IUD use. Physicians were also sent ques-
tionnaires for each woman they enrolled to capture add-
itional perforation-related information. All perforation
reports were validated via medical chart review. Contact
data were managed in compliance with data protection
regulations.
Table 1. Descriptive statistics by user cohorts, with comparison to 12-month study cohorts.
LNG-IUS
Study 5-year study
N (%) N ¼ 27,630 (70.8)
Age, mean (SD) 37.3 (7.5)
Weight, mean (SD) 69.9 (14.5)
BMI, mean (SD) 25.3 (5.1)
Current smoker N ¼ 6630 (24.0)
First-ever users of IUD (%) N ¼ 12,677 (45.9)
Nulligravid N ¼ 1787 (6.5)
Birth in last 12 months N ¼ 5516 (20.0)
Breastfeeding at time of insertion N ¼ 2660 (9.6)
Crude, as well as adjusted, relative risks were calculated
using a logistic regression model. Based on the expectation
of a small absolute number of perforations, confounding
variables were limited to a predefined set of risk factors as
recommended by the independent Safety Monitoring and
Advisory Council. In addition, sensitivity analyses were per-
formed with other potential risk factors to verify the robust-
ness of the primary analysis including the use of the
backward stepwise approach, where all prognostic factors
that did not change the point estimate of the risk ratio by
more than 10% or that had no statistically significant
impact (p > .10) were removed. Data on time since delivery
were dichotomised at 36 weeks using a data-driven cut-
point. All analyses were performed using the STATA 11.0
(College Station, TX, USA) and SAS 9.1 (Cary, NC, USA). This
report complies with the STROBE guidelines for observa-
tions studies [5].
The study was approved by the ethical committee of
the physicians’ association in Berlin, Germany and the
Ethics Committee of Hospital District of Southwest Finland.
Results
EURAS-IUD included 61,448 women with a newly inserted
IUD [1]. The 5-year sub-cohort included all women recruited
into the study prior to 31 July 2010 (39,009 women, 27,630
LNG-IUS users and 11,379 copper-IUD users). At the time of
recruitment, only one LNG-IUS containing 52 mg of LNG
was prescribed in Europe. The exact number of different
copper-IUDs is unknown since the standard varieties are
marketed under different names in different countries.
However, approximately 50 copper-IUDs of differing size,
shape and metal content were included.
A total of 30,114 women were sent 5-year follow-up ques-
tionnaires, of which we received 23,896 valid datasets. The
non-response rate for the 5-year questionnaire was 20.7%;
506 women actively dropped out or withdrew consent for
study participation and 5742 were lost to follow-up. A
detailed explanation of the recruitment of study participants
into the 5-year sub-cohort can be found in Appendix 1.
The study population of the 5-year sub-cohort was com-
parable to the study population at 12 months [1]. Overall
copper-IUD users were younger and more likely to be nulli-
gravid or to have given birth in the 12 months prior to IUD
insertion (see Table 1).
At 5 years, data on the IUD status were missing for
approximately 27% of the study population (24.8% of LNG-
IUS users and 33.4% of copper-IUD users). In total 29.2% of
LNG-IUS users and 18.9% of copper-IUD users still had their
IUDs in situ. The remaining 46.0% of LNG-IUS users and
Copper-IUD
12-month study 5-year study 12-month study
N ¼ 43,078 (70.1) N ¼ 11,379 (29.2) N ¼ 18,370 (29.9)
37.4 (7.5) 33.5 (7.7) 33.3 (7.6)
70.1 (14.7) 68.3 (14.0) 68.1 (13.8)
25.4 (5.1) 24.9 (5.0) 24.9 (4.9)
N ¼ 10,013 (23.2) N ¼ 2802 (24.6) N ¼ 4448 (24.2)
N ¼ 19,932 (46.3) N ¼ 5643 (49.6) N ¼ 9559 (52.0)
N ¼ 3036 (7.0) N ¼ 1190 (10.5) N ¼ 2209 (12.0)
N ¼ 12,423 (21.9) N ¼ 3,267 (28.7) N ¼ 5291 (34.7)
N ¼ 3968 (9.2) N ¼ 1700 (14.9) N ¼ 2687 (14.63)
 THE EUROPEAN JOURNAL OF CONTRACEPTION & REPRODUCTIVE HEALTH CARE 3

47.7% of copper-IUD users had had their IUD removed at
the time of answering the 5-year follow-up questionnaire –
of these, approximately 1.5% (1.4% LNG-IUS and 1.6% cop-
per-IUD) reported a spontaneous IUD expulsion.
We found 23 additional perforations (19 LNG-IUS and 4
copper-IUD) identified more than 12 months after insertion.
Added to the perforations detected at 12 months, a total
of 77 perforations were identified, of which 59 (40 þ 19)
occurred during LNG-IUS use and 18 (14 þ 4) during cop-
per-IUD use. One perforation in each cohort occurred prior
to insertion of the IUD (during sounding). These cases do
not contribute to the IUD perforation rates. The overall per-
foration rate was 2.1 per 1000 insertions (95% CI: 1.6–2.8)
for LNG-IUS users (39 þ 19 perforations/27,630 insertions)
and 1.6 per 1000 insertions (95% CI: 0.9–2.5) for copper-IUD
users (4 þ 13 perforations/11,379 insertions). The crude rela-
tive risk for LNG-IUS versus copper-IUDs was 1.4 (95% CI:
0.8–2.3).
A logistic regression model containing the predefined
prognostic factors age, BMI, breastfeeding at time of
100
90
80
% of diagnosed perforaons
insertion and parity (preferred model) yielded an adjusted
risk ratio of 1.7 (95% CI: 1.0–2.8). ‘Model 2’ containing the
variables age, BMI, time since last delivery and HCP experi-
ence did not substantially change the risk ratio (RR 1.6;
95% CI: 1.0–2.8). Therefore, our results cannot exclude a
two-fold higher risk of uterine perforation with LNG-IUS use
compared to copper-IUD use. Sensitivity analyses using a
backwards stepwise procedure yielded almost identical
results (results not shown).
The time interval between insertions and perforation
diagnosis did not differ substantially between cohorts
(Figure 1). A small proportion of perforations were diag-
nosed during or immediately after insertion (2% of LNG-IUS
and 17% of copper-IUD). Approximately 69% of perfora-
tions were diagnosed within 12 months of insertion, with
the remaining perforations diagnosed between 1 and 5
years. In two LNG-IUS users, the perforation was diagnosed
after the women became pregnant.
Most early perforations were detected following symp-
toms of pain and/or bleeding. However, in 29% of LNG-IUS
and 17% of copper-IUD perforations, the perforation was
asymptomatic (Table 2). This figure increases substantially
when data are limited to perforations diagnosed after 12
months, where 58% of LNG-IUS users and 75% of copper-

IUD users were asymptomatic at the time of diagnosis.

70 Proportions did not change when stratified by breastfeed-
60 ing status (data not shown). A small number (n ¼ 2, 3.4%)
50 of the LNG-IUS perforations were diagnosed only at the
time of planned IUD removal.
40
No perforation caused serious sequelae, such as bowel
30
or bladder injury, septicaemia, or peritonitis. Most IUDs
20 that perforated the uterus were removed laparoscopically
LNG-IUS
10 (66% of LNG-IUS and 59% of copper-IUD cases).
Copper IUD Approximately one quarter of IUDs could be removed vagi-
0
0 5 10 15 20 25 30 35 40 45 50 55 60 65 nally via hysteroscopy or direct visualisation and removal
Months aer IUD inseron during clinical examination. Laparotomy was a rare proced-
ure (see Table 3).
Figure 1. Kaplan–Meier plot of time to IUD perforation diagnosis.

Table 2. Perforations: circumstances leading to diagnosis at 1 and 5 years of follow-up, stratified by IUS.
LNG-IUS
1-y FU
(40 perforations)
Category n %
Diagnosed at insertion 1 2.5
Pains and/or bleeding 27 67.5
Strings not felt by patient 0 0.0
Pregnancy 1 2.5
Incidental finding (e.g., during diagnostic procedure) 1 2.5
Routine check of IUD position 10 25.0
At planned removal 0 0.0
Total 40 100.0
Percent of perforations.
5-y FU only
(19 perforations)
n %
0 0.0
8 42.1
0 0.0
2 10.5
0 0.0
7 36.8
2 10.5
19 100.0
Copper-IUD
1-y FU 5-y FU only
(14 perforations) (4 perforations)
n % n %
3 21.4 0 0.0
7 50.0 1 25.0
1 7.1 1 25.0
1 7.1 0 0.0
1 7.1 0 0.0
1 7.1 2 50.0
0 0.0 0 0.0
14 100.0 4 100.0

Table 3. Perforation reported and confirmed during follow-up, by removal technique.

LNG IUD Copper-IUD
Category 0–12 months n (%) 1–5 years n (%) Total n (%) 0–12 months n (%) 1–5 years n (%) Total n (%)
Total 39 (100.0) 19 (100.0) 58 (100.0) 13 (100.0) 4 (100.0) 17 (100.0)
Laparoscopy 29 (74.4) 9 (47.4) 38 (65.6) 8 (61.5) 2 (50.0) 10 (58.8)
Laparotomy 1 (2.6) 0 (0.0) 1 (1.7) 1 (7.7) 0 (0.0) 1 (5.9)
Hysteroscopy 4 (10.3) 6 (31.6) 10 (17.2) 1 (7.7) 0 (0.0) 1 (5.9)
Pulled out by the strings by the vaginal route 2 (5.1) 2 (10.5) 4 (6.9) 3 (23.1) 1 (25.0) 4 (23.5)
Not removed 1 (2.6) 1 (5.3) 2 (3.4) 0 (0.0) 0 (0.0) 0 (0.0)
Unknown 2 (5.1) 1 (5.3) 3 (5.2) 0 (0.0) 1 (25.0) 1 (5.9)
Perforations that occurred before IUD were introduced excluded (n ¼ 2).
4 C. BARNETT ET AL.
Perforations per 1,000 Insertions 14.0
12.0
10.0
8.0
6.0
4.0
2.0
0.0
Breast- Not
feeding Breast-
feeding
LNG-IUS
Figure 2. Perforations per 1000 IUD insertions stratified by cohorts and
breastfeeding at time of insertion (95% CI) (N of perforations/N of women).
Perforation risk was higher for first-time IUD users
(unAdj IR 3.55, 95% CI 2.59–4.75) for LNG-IUS and unAdj IR
2.13, 95% CI 1.10–3.71) for copper-IUD. Breastfeeding and
time since delivery remained independently associated with
perforation risk. The relative perforation risk for women
breastfeeding versus not breastfeeding was 4.9 (95% CI:
3.0–7.8). Women who had delivered 36 weeks and were
breastfeeding at the time of insertion had a relative risk of
6.8 (95% CI: 4.5–9.9) per 1000 insertions compared to 5.1
(95% CI: 0.6–18.4) for women who were breastfeeding and
gave birth >36 weeks prior to insertion (Figure 2). Among
women who were not breastfeeding at the time of inser-
tion, the corresponding figures were 3.0 (95% CI: 1.5–5.4)
and 1.2 (95% CI: 0.8–1.7), respectively.
Discussion
There was no significant difference in the relative risk esti-
mate at 5 years in comparison to results at 12 months. At
5 years the relative risk of uterine perforation with an LNG-
IUS compared to a copper-IUD was 1.65 (95% CI:
0.96–2.82). At 12 months, we found a relative risk of 1.61
(95% CI: 0.96–2.70) [1].
The proportion of LNG-IUS perforations diagnosed at
least 1 year after insertion was 32.7%. This is slightly higher
than reported by van Grootheest et al. [2] (25.6%). No per-
foration caused serious sequalae. Only one IUD required
laparotomy for removal and there were no cases of perfo-
rated bowel or peritonitis.
While determining the temporal relationship between
perforation and diagnosis in an observational study of this
nature remain challenging, our assumption is that most
perforations occurred at the time, or soon after, the date of
insertion. Breastfeeding at the time of insertion and time
since delivery remained independently associated with the
risk of uterine perforation at 5 years. Moreover, most perfo-
rations diagnosed after 12 months were associated with no,
or relatively few, symptoms. It is probable that due to their
low morbidity, these ‘late-diagnosed’ perforations went
unnoticed, in some cases, for several years.
A reduced likelihood of perforations was seen in women
who had used an IUD before (repeat/consecutive users).
The most likely explanation is the “healthy-user” effect [6].
Women have inherently higher perforation risks due to ana-
tomical factors such as cervical stenosis or unusual angula-
tion of the uterus would self-select ‘out’ of the pool of
potential IUD users after their initial, difficult, or failed IUD
insertion attempt leaving the consecutive user cohort with
a lower overall risk.
Some asymptomatic perforations may have been missed
in women whose IUD remained in situ at the 5-year follow-
up. Most of these women would have had a gynaecological
examination and IUD position check in the intervening 5
years. Nevertheless, a slight upward adjustment of the
overall perforation rate is possible.
EURAS-IUD was a large, epidemiological study. Although
Breast- Not
feeding Breast- relative risk differences between LNG-IUS and Copper-IUD
feeding
Copper IUD
were observed, for an association of this magnitude, it is
not possible to discriminate among bias, confounding,
causation, and chance as alternative explanations [7–11].
When designing EURAS-IUD, we tried to address some
issues commonly seen in observational research to more
accurately approximate risks. The study was a prospective,
comparative cohort design in a study population of real
users. Co-variates were collected at baseline and outcomes
of interest were validated with the treating physician. Given
the many potential threats to validity, it may be difficult to
conclude that LNG-IUS has a higher perforation rate com-
pared to copper-IUD.
Our non-response rate for the 5-year extension study
was approximately 21%. In theory, a disproportionately
high percentage of perforations could have occurred in
patients who were lost to follow-up, however, given that
the 5-year study results reflect both our results at 12
months and van Grootheest results, we believe this is
unlikely to be the case [2].
Conclusion
No clinically relevant difference in uterine perforation risk
was evident between LNG-IUS and copper-IUDs during rou-
tine clinical use at 12 months or at 5 years of follow-up.
We found that approximately one third of IUD perforations
were identified “late” (i.e., more than 12 months after the
initial insertion date). Women with uterine perforations
commonly present to their HCP with abdominal pain and
irregular bleeding, however, in many cases they are discov-
ered during routine gynaecological check-ups. From a pub-
lic health perspective, the rarity of perforation and the
absence of serious sequelae are reassuring.
Disclosure statement
The study was supervised by an independent Safety Monitoring and
Advisory Council with full authority over the study (including study
protocol, protocol amendments, data analysis, and stopping the
study).
Funding
The study was funded by a manufacturer of hormonal contraceptives.
Unconditional grant from Bayer AG, Germany. The funder had no
access to the source data and did not participate in designing the
study, analysing the data, or preparing this publication.
References
[1] Heinemann K, Reed S, Moehner S, et al. Risk of uterine perfor-
ation with levonorgestrel-releasing and copper intrauterine
devices in the European Active Surveillance Study on
Intrauterine Devices. Contraception. 2015;91:274–279.
 THE EUROPEAN JOURNAL OF CONTRACEPTION & REPRODUCTIVE HEALTH CARE 5

[2] van Grootheest K, Sachs B, Harrison-Woolrych M, et al. Uterine
perforation with the levonorgestrel-releasing intrauterine
device: analysis of reports from four national pharmacovigilance
centres. Drug Saf. 2011;34:83–88.
[3] Heinemann K, Reed S, Moehner S, et al. Comparative contracep-
tive effectiveness of levonorgestrel-releasing and copper intra-
uterine devices: the European Active Surveillance Study for
Intrauterine Devices. Contraception. 2015;91:280–283.
[4] Heinemann K, Barnett C, Reed S, et al. IUD use among parous
women and risk of uterine perforation: a secondary analysis.
Contraception. 2017;95:605–607.
[5] von Elm E, Altman DG, Egger M, et al. Strengthening the
Reporting of Observational Studies in Epidemiology (STROBE)
statement: guidelines for reporting observational studies. BMJ.
2007;335:806–808.
[6] Shrank WH, Patrick AR, Brookhart MA. Healthy user and
related biases in observational studies of preventive interven-
tions: a primer for physicians. J Gen Intern Med. 2011;26:
546–550.
[7] Grimes DA, Schulz KF. False alarms and pseudo-epidemics: the
limitations of observational epidemiology. Obstet Gynecol.
2012;120:920–927.
[8] Khoury MJ, James LM, Flanders WD, et al. Interpretation of
recurring weak associations obtained from epidemiologic stud-
ies of suspected human teratogens. Teratology. 1992;46:69–77.
[9] Shapiro S. Bias in the evaluation of low-magnitude associations:
an empirical perspective. Am J Epidemiol. 2000;151:939–945.
[10] Shapiro S. Causation, bias and confounding: a hitchhiker's guide
to the epidemiological galaxy Part 2. Principles of causality in
epidemiological research: confounding, effect modification and
strength of association. J Fam Plann Reprod Health Care.
2008;34:185–190.
[11] Shapiro S. Causation, bias and confounding: a hitchhiker's guide
to the epidemiological galaxy. Part 3: principles of causality in
epidemiological research: statistical stability, dose- and dur-
ation-response effects, internal and external consistency, ana-
logy and biological plausibility. J Fam Plann Reprod Health
Care. 2008;34:261–264.

Appendix 1
The EURAS-IUD study was conducted in six European countries (Austria, Finland, Germany, Poland, Sweden and the United Kingdom). The study
was implemented initially in Austria, Finland and Germany in 2006 and expanded to the United Kingdom, Sweden and Poland in the fall of 2009
(Figure A1). A total of 61,448 women were enrolled at baseline. There were no set inclusion criteria and recruitment was based on an all-comer
approach. All women with a newly inserted IUD were eligible for study inclusion with no limitation on age, parity or type of IUD fitted. A ‘non-
interference’ approach was used to provide standardised, comprehensive, reliable information under routine medical conditions. During the
recruitment period (2006–2012) the only levonorgestrel-IUD available in Europe contained 52 mg of levonorgestrel. Copper-IUDs are widely dis-
tributed and over 50 different devices formed the copper-IUD cohort.
Baseline data from all women recruited in EURAS-IUD prior to 30 July 2010 were included in the 5-year extension study (27,630 LNG-IUS users
and 11,379 copper-IUD users). Exclusion criteria was used to prevent women, where an additional questionnaire would not provide new data,
from receiving a 5-year follow-up questionnaire. Women were excluded if they dropped out, or withdrew their information consent during the
first 12 months of the study (433 women), if no follow-up information was available at 12 months (826 women), if only physician follow-up infor-
mation was available (2886 women), if there was missing data on IUD use (326 women) and finally, if the IUD was not in situ at 12 months
(4394). Of the 4394 women who had no IUD at 12 months, the IUD had been removed for personal reasons in 3831 cases. For 399 women, the
IUD had been expelled, 54 women had a perforation, and for 80 women the 12-month questionnaire stated that the IUD was no longer in place,
however, further details were missing (i.e., unclear whether the IUD was removed or expelled). These women were not sent 5-year follow-up
questionnaires, however, their data were included in the analysis of perforation incidence.
A total of 30,114 women were sent 5-year follow-up questionnaires from which we received 23,896 valid datasets. The non-response rate for
the 5-year questionnaire was 20.7%, of which 506 women actively dropped out or withdrew consent for study participation and 5742 were lost
to follow-up.

Baseline Recruited after

61.448 22.439
participants August 2010

Inclusion
Recruited prior
criteria for 5-year
39.009 to 1st August 8.865 4.394 No IUD at 12 months
questionnaire not
2010
fulfilled
No patient or HCP
826
follow-up available
Only 12 month HCP
2.886
follow-up available
Dropout (withdrawal
433
of informed consent)
Missing data on IUD
326
use at 12 months

Non-Responder: Dropout
5-year follow-up
30.144 6.248 dropout or Loss to 506 (Withdrawal IC, going
initiated
follow-up to live abroad)

5.742 Loss-to-follow-up

5-year follow-up
23.896
data obtained
Figure A1. Study flow in the 5-year extension study.