Incidence of infections associated with the intrauterine contraceptive device in an isolated community PHILIP
B.
JACKSON JOHN
MEAD, B.
VAN
Burlington,
M.D.,
BEECHAM, S.
MAECK,
F.A.C.O.G. M.D. M.D.,
F.A.C.0.G
Vermont
In an attempt to estimate more precisely the frequency of infections associated with the intrauterine contraceptive device (IUD), all gynecologic morbidity resulting from infection and occurring during a tuto-year period in an isolated community was reviewed. Ten septic abortions occurred, and all but one were associated with IUD use. In 26 gynecologic inpatients (41 per cent of all admissions for acute pelvic inflammatory disease), pelvic infection was associated with IUD use. In contradistinction to the septic abortion data, implicating only the Dalkon Shield, the gynecologic infections were associated with various types of devices.
SEVERAL RECENT article? have documented the occurrence of serious, occasionally fatal, pelvic infections associated with intrauterine contraceptive device (IUD) use. These discussions have fostered the emergence of “numerator” theories regarding incidence figures, as exemplified by the recent report of the Center for Disease Control concering IUD morbidity.5 In point of fact, an accurate incidence of these infections cannot be determined from such anecdotal case reports and casual surveys because the total number of insertions (corresponding to the “denominator”) is not known. Indirect attempts to deter-
mine the incidence of IUD-associated infections, with the use of questionnaires and sales distribution information, have yielded inconclusive data and have led to what Mishell and associate@ have described as a discipline of “pyramiding assumptions.” We have attempted to evaluate the magnitude of this problem by collecting all gynecologic morbidity data relating to infections in women in the childbearing years which occurred during a two-year period in an isolated community.
Material and methods Chittenden County (Burlington, Vermont) has a total population of 105,700, with 24,795 women in the age range 15 to 44 years old (estimate for July 1, 1973, from 1970 census data). This is a geographically isolated area, and essentially all seriously ill patients from the county are admitted to the Medical Center Hospital of Vermont. It is unlikely that patients with serious infections are cared for elsewhere. Furthermore, venereal disease is relatively uncommon in
From the Department of Obstetrics and Gynecology, Medical Center Hospital of Vermont, Universib of Vwmont College of Medicine. Received
for publication
Rw’sed
May
Accepted
30,
March
26, 1975.
1975.
#lay 30, 1975.
Reprint requests: Dr. Philip B. Mead, Department Obstetrics and Gynecology, University of Vermont of Medicine, Burlington, Vermont 05401.
of College
79
80
Mead, Beecham, and Maeck
Table
I. Summary
of IUD-associated
septic abortions
Insertionznjection interval
Parity
(mQnths)
Type of IUD
Dalkon Shield
NO
Dalkon
Shield
YKS
14 13% 13
Dalkon Dalkon
Shield Shield
No
Dalkon Shield
No
8
8
Dalkon
Shield
Yes
Unknown Unknown 4
18
Dalkon
Shield
Yes
14 13
Dalkon Shield
Yes
Dalkon
Yes
12 16
23 33
o-0-1-0 3-o-o-3
27 28 26
1 -o- I- 1 3-o-o-3 2-o-o-2
30 36 Unknown
21
o-o-o-o
33 22 21
2-o-o-2 2-o- l-2 1-0-1-l
18 18
Vermont so that there are few patients with gonorrhea1 infections to confuse the picture. (Only 168 cases of gonorrhea in women were documented in this county during the two-year study period.) Finally, legal abortion services in this area are highly developed, and abortions are easily and inexpensively obtained. For this reason, it is unlikely that many, if any, criminal or self-induced abortions are performed. Because of these unique local factors, we feel that our experience with IUD-associated infections in the last several years provides an indication of the magnitude of the problem. All admissions to the Medical Center Hospital of Vermont from July 1, 1972, through June 30, 1974, were reviewed. The data presented below represent all admissions for gynecologic infection-both IUDassociated and otherwise-during this two-year period. Since the study was concerned only with infections serious enough to require hospitalization, no attempt was made to collect information regarding infections in women managed as outpatients.
Results Obstetric 3,640
patients
cases. During were
Septicemia
the two-year
delivered
of
their
study period, infants
in
hospital. During the same period, 10 patients were admitted with the diagnosis of septic abortion. Of these 10 septic abortions, only one was not associated with the use of an IUD, while the remaining nine patients all had a Dalkon Shield in place at the time the diagnosis was made. The nine patients admitted with septic abortion associated with an IUD ranged in age from 21 to 33 years old. All were white and all but one was married.
Shield
NO
Cervix: alpha streptococcus Blood, cervix: Eubactcria and Peptostreptoctmus Cervix: E. m/i Cultures lost
Cervix: Bnctemzdet ,j7>li.,, B. m,elnlzino~eni~lc,\ , and Peptococcus Blood: Aacteroirlrv,f~nffilu and E. co/i Cervix: Bacteroifi43 fm&i, Blood, cervix: E. roli _’ Blood. cervix: Enteroc occ i Blood: betahemohric streptococci (not Group A)
Only one patient was a primigravida, and Jrone had more than one previous abortion. The insertioninfection interval ranged from four months to three years. Eight of these nine septic abortions occurred in the second trimester. The seriousness of these infections is underscored by the fact that septicemia was documented in five of the nine cases. Responsible pathogens included streptococci, enterococci, anaerobes, and E. coli (Table I). ,411 patients Marc managed with prompt uterine evacuation, appropriate anand supportive care, anti all nlatle tibiotic therapy, uneventful recoveries. Gynecologic cases. The gynecologic data arc prtasented in Tables II through VII. During the two-year study period, there were 2,352 admissions to the Gynecologic Service; 63 of these patients were admitted with a diagnosis of some type of acute pelvic inHammatory disease. Twenty-six of these patients (41 per cent of all patients admitted for acute pelvic infammatory disease) were wearing an intrauterine device. The specific diagnoses and their association with IUD use are presented in Table Il. In contradistinction to the septic abortiou data presented above, implicating only the Dalkon Shield, these gynecologic infections were associated with various types of devices (Table III). it has been estimated that the Dalkon Shield accounted for approximately 39 per cent of all IUD’s in use in the United States during the period of this study.? Interestingly, 38 per cent (10/26) of our patients with an IUD-associated gynecologic infection had a. Dalkon Shield in place. The insertion-infection interval ranged from onr month to 42 months, with half of who infectior,s
Infections and the IUD 81
Volume Number
125 I
Table
II. Admissions
for
acute
pelvic
Table
inflammatory
disease
V. Cultures
from
IUD-associated
gynecologic
infections IUDassociated
No IUD
16 4 3 1 0 1
23 0 2 0 11 1
- 1 26 (41%)
- 1 37 (59%)
Diagn0si.s Acute salpingitis Endomyometritis Tuboovarian abscess Ovarian abscess Pelvic abscess Fitz-Hugh-Curtis syndrome Pelvic cellulitis Total
patients
Culture
results
No.
Predominant pathogen(s) Normal vaginal flora Not taken
isolated
Total
Table
26
VI. Pathogens
IUD-associated
isolated
from
gynecologic
III.
Devices
associates
with
Table
No.
I
Ualkon Shield hlajzlin Spring Copper T Lippes Loop Not recorded
10 4 4 3 5
~I‘otal
26
IV. Insertion-infection
interval
Table
(gynecologic
infections)
VII.
Major
)
surgical
IUD-associated Irwrtion-infection interval (months)
No. of isolations
procedures
gynecologic
necessitated
No. of patients
1
3 1 1 2 2 1 2 2 6 1 5
.Total
adnexectomy adnexectomy
associated
In
of
infection. of resolution with medical
these
seven
believed
each
case,
the
indication
and (Table
adnexal
unilateral,
No. of patients 2 2 3
of tenderness management patients,
to be
described
26
1
Unilateral adnexectomy Hysterectomy, unilateral Hysterectomy, bilateral
failure masses
by
infections
Procedure
3 10 12 18 21 24 30 36 42 Unknown
with
At-robes Enterococci Pseudomonas E. coli N. gonowhoeae Beta hemolytic streptococci (not Group A) Anaerobes Bacteroides,fragilis Peptococcus Peptostreptococcus Fusobacteria
gynecologic
infections Device
15 patients
infections
Microorganism
Table
15 5 6
was
inflammatory VII). In four
involvement
a curious
finding
was previously
by Ledger.8
Comment occurring
two
Cultures with
years
were
or more obtained
an
IUD-associated
instances, transcervical
predominant cultures
(Table
V).
gynecologic
All
aerobic isolated Seven major
insertion 20
of
gynecologic pathogens or material blood
patients
Responsible
after from
pathogens
these
operations
26
were
women as
a
IV).
in
No
patients In
15
isolated from at operation this
group
of
negative. equally
and anaerobic bacteria. N. on only two occasions (Table of
26
infection. were obtained
cultures
were
(Table the
(27 direct
per result
divided
between
gonowhoeae VI). cent)
was
of
the
IUD-
doubts
that
there
IUD’s
has been
repeatedly
documented
past and present. remain unanswered:
However, (1) What
related
and
infection? vary
used? It accurately
is obvious answered
survey,
sponsored the National
and
pelvic
(2) Does
complications
tive
underwent
one
between
is an occasional
association
infection.
association
the
This in the
two is the relative
depending
on
literature,
related incidence risk the
both questions of IUD
of infectious
type
of
that these questions can by a careful, large-scale such
by the Food Institutes
a study
is currently
and Drug of Health.
being
Administration
device
be most prospecjointly and
82
Mead, Beecham, and Maeck
It is equally obvious that the collection of such data will be time-consuming and cannot be expected to yield useful answers for several years. In the meantime, patients and practicing physicians will demand that some preliminaq estimates of risk-incidence be made. Our experience over a two-year period is presented as a11 example of the kind of “hard” data that are already available. From these data, certain deductions seem inescapable. First, nine of 10 patients admitted with a diagnosis of sep’ic abortion were wearing an IUD. Clearly, the association must be more than coincidence. Second, all septic abortions in our series occurred in association with ow device-the Dalkon Shield. Moreover, in the same study population, gynecologic IUD-associated infections occurred with uariow types of devices with a frequency approximating the frequency of employment of each device. This information inevitably points to the conclusion that there is soule unique association between septic midtrimester abortion and the Dalkon Shield. Recently, we have adopted the policy of immediate removal of an IUD in patients found to be pregnant with a device in situ. Hopefully, this practice will result in a decrease in the ratio of one IUD-related septic abortion for every 400 deliveries observed during the two-year study period.
A third area of concern involves the association between IUD’s and gynecologic infection. In this series. 41 per cent of all women admitted with acute pelvic inflammatory disease wore an IUD. II’ fewer khar~ ,I-1 per cent of the women in OUI- patient populatiolr were using an IUD-as seems probable-one must then consider the possibilit\ of a relationship bctwcen gynecologic infection and IUD’s, Such cannot bt unequivocally established from our figures. HOMCVC~I.. what can be appreciated is the d11wi11i1 of morbidit! resulting from IUD-associated infec.tit)ns--1rtolbictit\ experienced by the patients as incot~vcnietlce. pain, and expense. One of every 90 admissions to the Gynecology Service was prompted In ati IUD-related infection, and 27 per cent of these M OI~CII tln(lerwrnt major operations as a direct result of’ the infer tio11. Of even greater concern is the fact that one third of these women had 110 living childrelr. The potential risk of sterilitv as a secluela of’ pelvic infection ClCi1YlV is ali issue in the choice of contraception for the nullipara. Our report represents the experience in ;I stdl and sornewhat unique community. Nonrthelcss. it is pwsented as 211 example not of speculati\r extrapolation from incomplete data hut rather- a5 obsrrvccl fact derived from patient care.
REFERENCES
‘1Viles, P. J., and Zeiderman, A. M.: Pregnancy complicated by intrauterine contraceptive devices, Obstet. Gynecol. 44: 484, 1974. Hurt, \V. G.: Septic pregnancy associated with Dalkon Shield intrauterine device, Obstet. Gynecol. 44: 491, 1974. Christian, C. D.: Maternal deaths associated with an intrauterine device, .4~ J. OBSTET. GYNECOL. 119: 441, 1974. Zuckerman, J, E., and Stubblefield, P. G.: E. roli septicemia in pregnancy associated with the shield intrauterine contraceptive device, AM. J. OBSTET. GYNECOL. 120: 951, 1974.
5. Center
for Disease Control: IUD safety-Report of