|
Use of Copper Intrauterine Devices and the Risk
of Tubal Infertility among Nulligravid Women
David Hubacher, Ph.D., Roger Lara-Ricalde, M.D., Douglas J.
Taylor, Ph.D., Fernando Guerra-Infante, Ph.D., and Raymundo Guzman-Rodriguez,
M.D.
|
ABSTRACT
Background Previous
studies of intrauterine devices (IUDs), many of which are no longer
in use, suggested that they might cause tubal infertility. The
concern that IUDs that contain copper — currently the most commonly
used type — may increase the risk of infertility in nulligravid
women has limited the use of this highly effective method of birth
control.
Methods We conducted
a case–control study of 1895 women recruited between 1997 and 1999.
We enrolled 358 women with primary infertility who had tubal
occlusion documented by hysterosalpingography, as well as 953 women
with primary infertility who did not have tubal occlusion (infertile
controls) and 584 primigravid women (pregnant controls). We
collected information on the women's past use of contraceptives,
including copper IUDs, previous sexual relationships, and history of
genital tract infections. Each woman's blood was tested for
antibodies to Chlamydia trachomatis.
We used stratified analyses and logistic regression to assess the
association between the previous use of a copper IUD and tubal occlusion.
Results In analyses
involving the women with tubal occlusion and the infertile controls,
the odds ratio for tubal occlusion associated with the previous use
of a copper IUD was 1.0 (95 percent confidence interval, 0.6 to
1.7). When the primigravid women served as the controls, the
corresponding odds ratio was 0.9 (95 percent confidence interval,
0.5 to 1.6). Tubal infertility was not associated with the duration
of IUD use, the reason for the removal of the IUD, or the presence
or absence of gynecologic problems related to its use. The presence
of antibodies to chlamydia was associated with infertility.
Conclusions The previous
use of a copper IUD is not associated with an increased risk of
tubal occlusion among nulligravid women, whereas infection with C. trachomatis is.
Intrauterine devices (IUDs) have long been
believed to cause pelvic inflammatory disease and subsequent tubal
infertility. Many IUDs were withdrawn from the market in the United
States and other countries because of concern about safety, and the
use of one — the Dalkon Shield — was eventually shown to be
strongly associated with pelvic inflammatory disease. Copper-containing
IUDs were first approved for use in the United States in 1976 and
are still being marketed. Lingering concern about the potential
risks of IUDs has discouraged women — particularly those who have
never been pregnant — from using even copper devices.
Research on IUDs in the 1970s and 1980s was marked by
confusion and controversy. Two early reports from the Women's Health
Study in the United States showed that the use of an IUD increased
the risk of pelvic inflammatory disease by at least 60 percent,1,2
although subsequent reanalyses suggested a less marked increase in
risk.3,4,5 The
Oxford Family Planning Association contraceptive study6,7
initially reported that IUD use increased the risk of pelvic
inflammatory disease by a factor of 10 ; however, a refined analysis
using better comparison groups and separating results according to
the type of device found no significant increase in the risk of
pelvic inflammatory disease with medicated devices such as copper
IUDs.8
More recent studies by the World Health Organization9 and
by a team of researchers in the United States10
found that the incidence of pelvic inflammatory disease among IUD
users is less than 2 episodes per 1000 years of use, consistent with
conservative estimates of the incidence of pelvic inflammatory
disease in the general population.11 A
recent meta-analysis of 36 studies concluded that the use of any IUD
is positively associated with pelvic inflammatory disease.12
Good evidence suggests that the increase in the risk of pelvic
inflammatory disease associated with IUD use is related only to the
process of inserting the device and that after the first month of
use, the risk of infection is not significantly higher than that
in women without IUDs.13
The recognized association between pelvic inflammatory
disease and tubal infertility14
has aroused some concern that the use of an IUD may lead to this
complication. Two case–control studies in the United States
published more than 15 years ago15,16
reported positive associations between IUD use and tubal infertility.
In subanalyses evaluating the risk according to the type of IUD
used, copper devices were reported to increase the risk of tubal
infertility in one study16
(though only among women with more than one sexual partner) but not
in the other15;
however, later reanalysis of the data in the second study suggested
an increased risk of infertility associated with the copper IUD.17
Since then, numerous case–control studies18,19,20,21
and cohort studies22,23,24,25
have attempted to clarify the relation, but controversy has
persisted.
Methods
We conducted an unmatched case–control study in three public
hospitals in Mexico City, Mexico (the National Perinatology Institute,
Gynecology and Obstetrics Hospital Number 4 of the Mexican Social
Security Institute, and the Women's Hospital). All consecutive
nulligravid, infertile women 18 years of age or older who were
scheduled for diagnostic hysterosalpingography were invited to
participate. Infertility was defined by the failure to conceive
after one year or more of unprotected intercourse. Criteria for
exclusion included previous pregnancy, tubal sterilization, and
previous diagnostic laparoscopy. After undergoing hysterosalpingography, the
infertile women were classified on the basis of the radiologic evidence
as women with tubal occlusion (case subjects) or as infertile controls.
From the same hospitals, we recruited a second control group
consisting of primigravid women in their first or second trimester.
In face-to-face interviews lasting an average of 20 minutes, all
participants were asked about their past use of contraceptives,
previous sexual relationships, and history of genital tract
infections; the interviews with the infertile women were conducted
before they knew whether they had tubal occlusion. The instruments
for recording the results of hysterosalpingography were adapted from
the recommendations of the American Fertility Society (now the
American Society for Reproductive Medicine).26
Participants also donated a sample of blood to be tested for
antibodies to Chlamydia trachomatis.
The institutional review boards of Family Health International and
the participating hospitals approved the study; written informed
consent was obtained from all enrolled women. Recruitment began in
September 1997 and was completed in October 1999; data analysis was
completed in December 2000.
We recruited 1311 infertile women (358 women with tubal
occlusion and 953 controls) and 584 pregnant controls; fewer than 5
percent of the women who met the eligibility criteria declined to
participate. We designed the study to have 90 percent power to
detect a doubling of the risk of tubal occlusion with IUD use in
analyses involving the infertile controls; the study had 87 percent
power to detect a doubling of the risk in analyses involving the
pregnant controls (two-sided test, 0.05 alpha level).27 If
we set the power at the standard 80 percent level, we had enough
study subjects to detect odds ratios of 1.8 and 1.9 in analyses
involving the infertile controls and the pregnant controls,
respectively.
Before recruitment began, the radiologists met to
standardize their approach to classifying tubal pathology. Tubal
occlusion was diagnosed if a water-based contrast medium failed to
spill from either tube into the peritoneal cavity. Fluoroscopy was
used, and the last films were taken 15 minutes after the contrast medium
had been injected. The radiologists were unaware of the information
collected in the women's interviews.
Serologic tests for detecting antibodies to chlamydia are
accepted measures of past infection.28,29,30 An
indirect fluorescent IgG antibody–staining kit (Hemagen Diagnostics,
Columbia, Md.) was used to process the serum samples. As in previous
studies using these kits, samples that tested positive at dilutions
of 1:256 were considered diagnostic of past exposure to C. trachomatis.31
All serum samples were processed as recommended by the manufacturer.
Our primary exposure variable was the previous use or
nonuse of an IUD containing copper. Other variables that were
considered as possibly predictive of tubal infertility included the
presence or absence of antibodies to C. trachomatis, the number of lifetime sexual
partners, the presence or absence of a history of genital tract
infections, the presence or absence of a history of gynecologic symptoms
suggestive of infection, the past use or nonuse of other methods of
contraception, family income, education, employment status, and the
presence or absence of a history of coitus during the teenage years.
Regarding their most recent sexual partners (up to six), the women
were asked about the length of the relationship and whether they believed
their partners had engaged in concurrent sexual relations with other
women. To adjust for age, we used the age when the infertile women
first suspected they were unable to conceive and the age when the
pregnant women first began attempting to conceive. We excluded from
the analyses exposure that occurred after the onset of infertility.
We classified women into one of six mutually exclusive
groups on the basis of use of contraceptive methods: no previous
method (or rhythm or withdrawal), condoms only, vaginal spermicides
only, hormonal methods only (oral or injectable contraceptives), condoms
and hormonal methods, and IUDs (none of the women reported a history
of diaphragm use). If a previous user of vaginal spermicides had
also used one of the other methods, she was assigned to the group
that used that other method. All women who had used an IUD reported
having used a device containing copper. The vast majority of the
copper IUDs used by the women in the study were T-shaped (containing
either 220 mm2 or 380 mm2 of copper surface).
Data were collected on the duration of use of a given method, any
gynecologic problems that occurred during its use, and the reasons
for the discontinuation of its use.
We calculated crude and adjusted odds ratios (with 95
percent confidence intervals) as measures of the association between
IUD use and tubal occlusion. Logistic regression was used to control
for other factors simultaneously.
Results
The infertile women with tubal occlusion (case subjects)
and the infertile controls were similar in terms of level of
education, employment status, family income, and number of months
spent attempting to conceive (Table 1). As
compared with the women with tubal occlusion, the pregnant controls
were younger, better educated, and less likely to work outside the
home and had lower family incomes.
|
The prevalence of the possible risk factors for tubal occlusion was
similar among the women with tubal occlusion and the infertile controls
(Table 2).
However, as compared with the women with tubal occlusion, the
pregnant controls had had more sexual partners, were more likely to
report suspected infidelity by partners, and had lower rates of
previous upper genital tract infection, symptoms of pelvic
inflammatory disease, and positive tests for antibodies to
chlamydia.
|
When no previous contraceptive use was defined as the reference category,
previous use of a copper IUD was not associated with an increased
risk of tubal occlusion either in the analysis including the infertile
controls (odds ratio, 1.0; 95 percent confidence interval, 0.6 to
1.6) or in the analysis including the pregnant controls (odds ratio,
0.7; 95 percent confidence interval, 0.4 to 1.2) (Table 3). In the
latter analysis, women whose sexual partners used condoms had a 50
percent lower risk of tubal occlusion than those who used no
contraception.
|
Similarly, when no previous IUD use was defined as the reference category,
previous use of a copper IUD was not associated with tubal occlusion
in the analyses including either the infertile controls (odds ratio,
1.0; 95 percent confidence interval, 0.6 to 1.7) or the pregnant
controls (odds ratio, 0.9; 95 percent confidence interval, 0.5 to
1.6) (Table 4).
A longer duration of use of a copper IUD, the removal of the IUD
because of side effects, and a history of gynecologic symptoms
during the use of a copper IUD were not associated with increased
odds of tubal occlusion.
|
The presence of antibodies to C. trachomatis
among women who had not used a copper IUD was associated with tubal
occlusion (odds ratio, 2.4; 95 percent confidence interval, 1.7 to
3.2) in the analysis including the pregnant controls (Table 5). Among
women who had used an IUD, there was no significant association between
antibodies to C. trachomatis and
tubal infertility, but there were relatively few women in this group.
|
We considered using a broader definition of a case to include either
tubal occlusion or adhesions, but hysterosalpingography has a
limited ability to identify adhesions.32,33 To
address the concern that our case group was defined too narrowly
because we excluded women who had adhesions only, we performed an
additional analysis. This analysis involved reclassifying as case
subjects the infertile controls who had no occlusion but had
adhesions identified on hysterosalpingography and positive serologic
tests for C. trachomatis.
Previous research suggests that serologic testing for chlamydia may
improve the accuracy of the diagnosis of tubal disease when used in
conjunction with, or even in place of, hysterosalpingography.34,35
After this reclassification, the adjusted odds ratios for tubal
infertility associated with IUD use were 1.2 (95 percent confidence
interval, 0.7 to 1.9) for the analysis including the infertile
controls and 1.0 (95 percent confidence interval, 0.6 to 1.5) for
that including the pregnant controls.
Although hysterosalpingography is the standard method for
evaluating tubal patency,36 it
has some diagnostic limitations that laparoscopy does not have. Of
the 1311 infertile women in our study, only 321 (24 percent)
underwent laparoscopy. Those with abnormal results on
hysterosalpingography were twice as likely as women with negative
results to undergo laparoscopy. Using the laparoscopy reports, we
divided this subgroup into 185 women with any evidence of tubal
disease, including adhesions, and 136 infertile controls with no
evidence of tubal disease. The adjusted odds ratio for tubal
infertility associated with previous IUD use was 1.5 (95 percent
confidence interval, 0.7 to 3.5) in analyses including the infertile
controls and 1.9 (95 percent confidence interval, 1.0 to 3.5) in
analyses using the pregnant controls.
Discussion
Our finding that the use of a copper IUD was not a risk
factor for tubal occlusion among nulligravid women contradicts some
previous reports that aroused concern about future fertility in
women who use copper IUDs.15,16
Given the media attention to the problems with the Dalkon Shield and
the associated litigation in U.S. courts during the 1970s and 1980s,
women who were infertile and had previously used an IUD may have
been more inclined to investigate the cause of their condition than
infertile women who had never used an IUD. Consequently, women with
a history of IUD use may have been disproportionately represented in
previous studies in the United States of women with tubal
infertility. In contrast, our study was conducted in Mexico, where
IUD use is well accepted and where such bias is unlikely to occur.
Past research on this topic used only primigravid women as
controls. We included an infertile control group for several
reasons. First, we wanted to ensure that the women with tubal
occlusion came from the same population as the controls. Second, this
approach minimizes bias due to the differential recall of IUD use
according to diagnosis. We also included a control group of pregnant
women to address the association between IUD use and the inability
to conceive.
Exposure to C.
trachomatis has been cited as an important cause of tubal
infertility.21,37 We
found higher rates of positive tests for antibodies to chlamydia, a
validated marker of past exposure,28,29,30
among women with tubal occlusion and among infertile controls than
among pregnant women. A weakness of the antibody test is that it
does not indicate whether exposure to C. trachomatis preceded the onset of tubal disease, although
it is likely that it did. The fact that the prevalence of antibodies
was similar among women with tubal occlusion and infertile controls is
not surprising, since the infertile controls may have had other
evidence of disease attributable to chlamydia, such as adhesions,
which are not readily detectable by hysterosalpingography.32,33
Thus, it is possible that previous studies found an increased risk
of tubal infertility associated with the use of a copper IUD because
of the unmeasured confounding effect of exposure to sexually
transmitted disease — specifically, C.
trachomatis.
Laparoscopy is another diagnostic procedure for women with
infertility. The decision about whether to undergo laparoscopy is an
individual one that depends on clinical findings (including those
from hysterosalpingography) and other factors, such as the presence
or absence of a history of pelvic pain. In our study, only one quarter
of the women underwent laparoscopy, including a disproportionate number
of those with abnormal results on hysterosalpingography. The
adjusted odds ratios for tubal infertility according to the analysis
of the women who underwent laparoscopy, although higher than those
calculated on the basis of all the women who underwent
hysterosalpingography, were not inconsistent with the results of the
primary analysis, and the differences between these ratios may
reflect selection bias. An unbiased study in which laparoscopy was
required for the identification of cases would not be feasible
because of its cost, the time it would consume, and the ethical
problems it would raise. Our analysis, based on
hysterosalpingographic results, focused on damage to the lumen of
the fallopian tubes; it is unlikely that we found no increased risk
associated with IUD use simply because copper IUDs affect only
structures exterior to the fallopian tubes.
More than 100 million women worldwide use IUDs. Asia
accounts for the majority of use, but IUD use is also common among
married women of reproductive age in Scandinavian countries
(prevalence, 18 percent) and in other European countries (7
percent).38
In contrast, only 1 percent of women in the United States use the
IUD.39
This low rate is thought to reflect the widespread concern about
health risks associated with the method. In lieu of using an IUD,
women may prematurely request sterilization (and may regret it later40,41),
choose less effective or less convenient methods, or risk an
unwanted pregnancy.
This study suggests that the use of copper IUDs is much
safer than was previously thought. Nulligravid women who are not at
risk for a sexually transmitted disease are appropriate candidates for
the copper IUD. Contemporary copper IUDs may be among the safest,
most effective, and least expensive reversible contraceptives available.42,43
Supported by the U.S. Agency for International Development
(USAID) through a contract with Family Health International
(contract CCP-A-00-95-00022-02) and by the National Institute for
Child Health and Human Development, National Institutes of Health,
through an Interagency Agreement (Y1-HD-7230-01) with USAID. This
article does not necessarily reflect the views or policies of the
Department of Health and Human Services or of USAID, nor does
mention of trade names, commercial products, or organizations imply
endorsement by the U.S. government.
We are indebted to the participants for making this
research possible and to Rocío Dávila-Mendoza, Dr. Zigor Campos-Goenaga,
Dr. Maria del Carmen Tavera-Hernández, Maria Elena Guevara-Reyes,
Enimia Zárate-Aragón, Bersabe Bautista-García, Deborah Cousins, Dr.
Jaroslav Hulka, Cathy Dudnanski, Carmen Cardenas-Lopez, Dr. David
Grimes, Dr. Ken Schulz, Dr. Julio de la Jara, Dr. Esteben Garcia,
Dr. Alonso Garcia-Luna, Marie McLeod, and Research Triangle
Institute.
Source Information
From Family Health International, Research Triangle Park,
N.C. (D.H., D.J.T.); and the National Perinatology Institute, Mexico City,
Mexico (R.L.-R., F.G.-I., R.G.-R.).
Address reprint requests to Dr. Hubacher at Family Health
International, P.O. Box 13950, Research Triangle Park, NC 27709, or at [log in to unmask].
References
|
Time to Pardon the IUD?
|
The
report in this issue of the Journal
on the effect of copper intrauterine contraceptive devices on the
risk of infertility in nulliparous women deserves the attention of
American women and their clinicians.1
Despite the efficacy of the intrauterine device (IUD) in preventing
pregnancy, concern that the use of an IUD may cause pelvic infection
and infertility has made it a contraceptive option that is rarely
used in the United States.
Hubacher et al. report the results of a case–control study
involving nearly 2000 women in Mexico City, Mexico.1 The
study compared nulligravid infertile women who had tubal occlusion
with infertile women who did not have tubal occlusion, as well as
with pregnant women, in terms of the previous use of IUDs and other
potential risk factors for infertility. They also tested the women
for the presence of antibodies to Chlamydia
trachomatis, the most important contributor to infertility.
The authors found no increase in the risk of tubal occlusion associated
with the previous use of copper IUDs; in analyses involving the
infertile controls, the odds ratio associated with the use of a
copper IUD was 1.0 (95 percent confidence interval, 0.6 to 1.7), and
similar results were found in analyses involving the pregnant
controls. By contrast, there was a strong association between
previous chlamydial infection and infertility in these women. This
important study could not have been conducted in the United States,
because nulliparous women rarely use intrauterine contraception
here, and even women in this country who have already borne a child
have tended to shun these devices, but the results should change
current opinion about copper IUDs.
The use of intrauterine contraception has declined in the
United States even as it has increased in the rest of the world.
Thirty years ago, nearly 10 percent of women in the United States
who used contraception chose IUDs, but now less than 1 percent do
so. In contrast, more than one in five women in Denmark and Germany
use IUDs. Reluctance to use IUDs and other highly effective methods
of contraception is one reason why, by the end of their childbearing
years, nearly half of the women in the United States (46 percent)
have had at least one elective abortion, and 27 percent have
undergone surgery for sterilization. These rates are about three times
as high as those in western Europe.
It is ironic that the modern IUD, now largely rejected
here, was developed by gynecologists in the United States, and its
use was a routine part of gynecologic practice in this country 30
years ago. What changed? Research to create IUDs that caused less
menstrual pain and bleeding, the principal reasons why women had the
devices removed, led to the development of the infamous Dalkon
Shield, a poorly designed and poorly tested device designed to
accommodate nulliparous women. Introduced just at the time the
danger of venous thrombosis associated with oral contraceptives was
becoming widely publicized, the Dalkon Shield appealed to sexually
active young women.
I and many other gynecologists inserted the new Dalkon Shield,
with its spiny margins and braided tail, in our young patients because
we thought it would be better tolerated than earlier IUDs by a
uterus that had never sustained a pregnancy and because we had been
deceived about its efficacy.2 We
did not worry about health risks because IUDs were considered safe
on the basis of early research.3,4 The
Food and Drug Administration (FDA) was not yet responsible for
evaluating the safety of such devices in clinical trials.
Our complacency was disturbed in 1973 when the Family Planning
Evaluation Division at the Centers for Disease Control reported several
deaths due to pelvic infection related to the use of IUDs.5,6
Subsequent retrospective case–control studies linked IUDs with
sexually transmitted infections, which had become a major cause of
infertility.7,8
Further research found a direct relation between IUD use and
infertility due to tubal occlusion.9,10
These reports formed the foundation for litigation, first against
the sellers of the Dalkon Shield and then against other companies
that manufactured IUDs. Although the FDA forced only the Dalkon
Shield from the market, the other U.S. manufacturers abandoned sales
in order to avoid legal costs. As a result, no IUDs were sold in the
United States between 1983 and 1988, and many women, frightened by
reports of infection and infertility, had their IUDs removed. Soon,
teaching about contraception condemned IUDs, and the few physicians
who still trusted intrauterine contraception had neither devices to
insert nor patients who wanted them.
In the rest of the world, however, intrauterine
contraceptive devices continued to be used, and ongoing research
resulted in better devices. One, the copper T 380 (with 380 mm2
of copper surface), was introduced to an empty U.S. market in 1988
by a company too small to attract litigation or, initially, many
customers. This device (which goes under the trade name ParaGard) is
the only copper IUD available in the United States but is one of
several used in Mexico and other countries. Although few IUDs were
provided to women in the United States during the past 20 years,
reanalyses of previously collected data and new retrospective
case–control studies have helped to identify the characteristics of
IUDs and their users that were associated with pelvic infection and
infertility.11
The use of the Dalkon Shield was associated with a much higher
relative risk of pelvic infection than any of the others (eight
times as high), indicating that its inclusion in previous studies
had inflated the risk associated with IUDs in general.12
Other results demonstrated that women in relationships that were
more likely to be exclusive (marriage or cohabitation) were not at
increased risk,13
supporting the contention that "men, not IUDs," cause
pelvic infection.
Because pelvic infection can occlude the oviducts and cause
infertility, epidemiologists in Seattle and Boston examined the
relation between previous IUD use and subsequent tubal infertility.9,10
It is with these two case–control studies, published in the Journal 16 years ago, that the report in
this week's issue of the Journal
should be compared. Neither found a significantly increased risk of
tubal infertility associated with the use of a copper IUD (relative
risk, 1.5 [95 percent confidence interval, 0.8 to 3.0] and 1.3 [95
percent confidence interval, 0.6 to 3.0], respectively), and both
determined that IUDs made of plastic did increase the risk. Although
the 95 percent confidence intervals span 1.0 (indicating no increased
risk), the point estimates were used to indict copper IUDs along
with all other types.
The finding of the present study that exposure to chlamydia
is associated with infertility provides an explanation for previous reports
of an increased risk of pelvic infection among women who were not
involved in exclusive relationships and an increased risk associated
with the insertion, but not the prolonged use, of IUDs.14 At
the time of these earlier studies,9,10 C. trachomatis was not
recognized as an important cause of tubal infertility.
No matter what form of contraception a woman chooses, the
increasing prevalence of chlamydia and other sexually transmitted
pathogens, including the human immunodeficiency virus, means that
condoms should be used with high-risk partners. But the findings
reported by Hubacher et al. should reassure clinicians and women
alike that copper IUDs, which are by far the most common type of IUD
used in the United States and around the world, are not a threat to
the health or future fertility of the women who use them, including
those without children.
Philip D. Darney, M.D.
University of California, San Francisco
San Francisco, CA 94110
References
Edward E.
Rylander, M.D.
Diplomat American
Board of Family Practice.
Diplomat American
Board of Palliative Medicine.