The association of Chlamydia trachomatis, Neisseria

gonorrhoeae, and group B streptococci with preterm rupture

of the membranes and pregnancy outcome
Lindsay S. Alger, MD, Judith C. Lovchik, PhD, John R. Hebel, PhD,
Lillian R. Blackmon, MD, and M. Carlyle Crenshaw, MD
Baltimore, Maryland

There is conflicting evidence regarding a possible causal role for Chlamydia trachomatis in the
development of preterm premature rupture of the membranes. We investigated the relative prevalence of
endocervical infection with C. trachomatis and group B streptococci in patients with preterm premature
rupture of membranes compared with a control group taken from the same obstetric population.
C. trachomatis was isolated from 23/52 (44%) patients with preterm premature rupture of membranes
versus 13/84 (15%) women in the control group (p < 0.001). This association was independent of
infection with group B streptococci or Neisseria gonorrhoeae. Group B streptococci were isolated from
16% of the patients with preterm premature rupture of membranes versus 4% of the control population
(p < 0.05). The risk of preterm premature rupture of membranes associated with group B streptococcal
infection was independent of infection with C. trachomatis and N. gonorrhoeae. Endocervical infection with
C. trachomatis did not significantly affect early maternal complication rates after delivery. (AM J 0BSTET
GYNECOL 1988;159:397·404.)

Key words: Chlamydia trachomatis, group B streptococci, Neisseria gonorrhoeae, preterm

premature rupture of membranes

Although infection is not the only cause of preterm
delivery, several observations support the hypothesis
that maternal genital infection may frequently play a
causal role: ( 1) Similar demographic risk factors such
as youth or low socioeconomic status are associated with
both preterm delivery and an increased incidence of
sexually transmitted infections. (2) Seasonal variations
in coital frequency parallel the variations in amniotic
fluid infection and perinatal mortality. (3) Clinical
and histologically documented chorioamnionitis occurs
more frequently in association with preterm deliveries.
(4) Both mothers and infants are more likely to develop
early onset infectious sequelae after a preterm delivery
compared with delivery at term. (5) Direct sampling of
amniotic fluid by amniocentesis demonstrates patho-
genic microorganisms in a significant proportion of pa-
tients in preterm labor or with preterm rupture of the
membranes. Bacteria may be recovered in 20% to 30%
of such samples, compared with 2% to 4% of fluid
From the Departments of Obstetrics and Gynecology, Pediatrics, and
Epidemiology and Preventive Medicine, University of Maryland
School of Medicine.
ReceivedforpublicationApril27, 1987; revised November 19, 1987;
accepted March 17, 1988.
Reprint requests: Lindsay S. Alger, MD, Department of Obstetrics
and Gynecology, University of Maryland Hospital, 22 S. Greene
St., Baltimore, MD 21201.
specimens from uncomplicated third trimester preg-
nancies. (6) A large number of studies demonstrate an
association between specific organisms and preterm de-
livery. Organisms linked to prematurity include Neis-
seria gonorrhoeae, group B streptococci, Bacteroides sp.
(and perhaps other anaerobes), Trichomonas vaginalis,
and possibly mycoplasmas.
There is conflicting evidence regarding a possible
causal role for Chlamydia trachomatis in the development
of preterm premature rupture of membranes.'·• How-
ever, previous studies demonstrating ( 1) the existence
of a tetracycline- and erythromycin-sensitive organism
that contributes to prematurity,5 • 6 (2) the ability of chla-
mydia to proliferate in human amnion cells with a rep-
lication cycle that obligates cell death, and (3) the as-
sociation of chlamydia with a mucopurulent endocer-
vical inflammatory process make it a likely candidate
on theoretic grounds. Previous studies had either
not investigated the relationship between endocervical
C. trachomatis and preterm premature rupture of mem-
branes while controlling for the two other pathogens
most frequently linked to preterm premature rupture
of membranes, N. gonorrhoeae and group B streptococci,
or had included insufficient patients to draw con-
clusions.
The University of Maryland Hospital serves a pri-
marily indigent population with a high prevalence of

397
398 Alger et al.
cervical colonization with C. trachomatis. We selected this
population to investigate the relative incidence of cer-
vical colonization with C. trachomatis, N. gonorrhoeae, and
group B streptococci in patients with preterm prema-
ture rupture of membranes compared with a similar
control group.
Material and methods
Women entering the Labor and Delivery Suite of The
University of Maryland Hospital from january 1983 to
March 1984 with spontaneous rupture of membranes
preceding the onset of contractions at <37 and >20
completed weeks of gestation were admitted to the
study. Preterm premature rupture of membranes was
confirmed by both the presence of alkaline fluid in the
vagina and a fern arborization pattern on microscopic
examination of this fluid. Gestational age was con-
firmed by menstrual history, clinical examination, and
ultrasound evaluation. Patients with uterine anomalies,
antibiotic use within 4 weeks before presentation, or
diabetes mellitus were excluded. Fifty-two patients with
preterm premature rupture of membranes were en-
rolled in the study.
Microbiologic specimens were obtained sequentially
for N. gonorrhoeae, group B streptococci, and C. tra-
chomatis cultures immediately on presentation, at the
initial sterile speculum examination. Because we were
interested only in organisms present adjacent to the
membranes, all specimens were obtained from the en-
docervical canal after the ectocervix had been cleared
of secretions by cotton-tipped swabs. Swab specimens
for N. gonorrhoeae and group B streptococci were trans-
ported to the microbiology laboratory within 2 hours
and plated on Martin-Lewis agar, chocolate agar, and
trypticase soy agar containing 5% sheep blood imme-
diately on receipt. Cultures were incubated at 37° C in
5% carbon dioxide and examined at 24, 48, and 72
hours. Group B streptococci were identified by serotype
agglutination reagents (Micro Scan, Lexington, Ky.).
N. gonorrhoeae was identified by fermentation pattern
by the RIM series (Austin Biological Labs, Austin,
Tex.). Specimens for C. trachomatis isolation were pro-
cessed as previously described. 7
In the absence of any clinical signs of infection, pa-
tients with preterm premature rupture of membranes
between 26 and 34 weeks' gestation were routinely
given betamethasone and, if indicated by the presence
of contractions, tocolytic agents in an attempt to delay
delivery for at least 48 hours and promote fetal lung
maturation. Patients did not receive antibiotic ther-
apy for C. trachomatis infection until after delivery.
Pregnancy outcome was assessed for interval from
membrane rupture to delivery, infectious maternal
complications as represented by chorioamnionitis or
endometritis, and route of delivery. Neonates were
August 1988
Am J Obstet Gynecol
evaluated for birth weight, gestational age as deter-
mined by standard examination, Apgar scores, neo-
natal respiratory distress (defined as any respiratory
symptoms persisting beyond the first several hours of
life), neonatal infections, length of hospital stay, and
neonatal survival.
Eighty-four patients attending the University of
Maryland Hospital public and private prenatal clinics
were selected as controls. In choosing a control pop-
ulation, we attempted to select two comparable patients
for each case of preterm premature rupture of mem-
branes regarding age ( ± 1 year), parity (nulliparous
versus parous), and gestational age ( ± 2 weeks). For 20
patients with preterm premature rupture of mem-
branes it was possible to find only one comparable con-
trol woman. Endocervical specimens for N. gonorrhoeae,
group B streptococci, and C. trachomatis cultures were
taken from control patients receiving prenatal care at
the gestational age corresponding to the matched pa-
tient and inoculated within 3 hours of collection. Both
patients with preterm premature rupture of mem-
branes and the controls were screened for N. gonor-
rhoeae infection at their initial prenatal visit. Patients in
both groups who had positive cultures were treated
with intramuscular penicillin G, 4.8 X 106 U. Neither
group of patients was screened or treated for group B
streptococci or C. trachomatis infection before study cul-
tures were obtained.
The association between antenatal C. trachomatis and
group B streptococcal infections and preterm prema-
ture rupture of membranes was analyzed statistically
by logistic regression analysis to assess relative risks
while controlling for age, parity, gestational age, and
other potential confounding variables. 8 Estimates of the
odds ratios for preterm premature rupture of mem-
branes with either or both such infections were deter-
mined from the regression coefficients as described by
Schlesselman. 9 Odds ratios are indicative of the relative
risk of an outcome (such as preterm premature rupture
of membranes) for individuals with a specified risk fac-
tor (such as C. trachomatis infection). Three different
logistic regression models were used for the analysis.
In each the preterm premature rupture of membranes
case-control status of the women defined the depen-
dent variable. The first model included positivity for a
given pathogen, C. trachomatis or group B streptococci,
as the only independent variable. This enabled the de-
termination of the crude relative risk for preterm pre-
mature rupture of membranes with that pathogen. The
second model, in addition to positivity for the patho-
gen, included several other potential preterm prema-
ture rupture of membranes risk factors as independent
variables. These extraneous factors were maternal age,
gestational age, parity, race, marital status, care pay-
ment type, source of prenatal care, and postscreening
Volume 159
Number 2
Table I. Demographic characteristics of patients with preterm premature rupture of membranes
compared with controls
Patients with
preterm premature
Age (yr)
Parity*
Gestational age (wk)
Race (% black)
Married (% single)
Medical assistance
(%yes)
Prenatal care
(% clinic or none)
Data are mean ± SD.
*Term plus preterm deliveries.
N. gonorrhoeae infection. The second model enabled the
determination of the relative risk for preterm prema-
ture rupture of membranes with the pathogen, ad-
justed for the extraneous factors listed above. A third
model included C. trachomatis positivity, group B strep-
tococci positivity, and their interaction as indepen-
dent variables. This model enabled determination
of whether the effects of the two pathogens are ex-
plained by confounding between them and, further-
more, whether their effects are synergistic.
Socioeconomic status was assessed by two parame-
ters: (1) the method of payment for medical costs and
(2) the site of prenatal care. For method of payment,
patients were classified as (1) having medical insurance
or (2) having no insurance or receiving government
assistance (Medicaid). For site of care, patients were
classified as receiving prenatal care from (1) a private
physician or (2) a public clinic. Three patients who re-
ceived no prenatal care were put in the latter group.
Results
Population. The demographic characteristics of the
patients with preterm premature rupture of mem-
branes and the controls are listed in Table I. The pa-
tients with preterm premature rupture of membranes
were young, predominantly black, unmarried dinic pa-
tients receiving governmental assistance. Controls were
similar to patients in age, parity, gestational age, race,
marital status, socioeconomic status, and prenatal care.
No statistically significant differences between C. tra-
chomatis-positive and -negative patients with preterm
premature rupture of membranes with regard to race,
marital status, or socioeconomic status were observed
in this relatively small and homogeneous population.
However, a trend toward an association between C. tra-
chomatis colonization and black ethnicity, single marital
status, and low socioeconomic status was noted. Because
of the association of youth with C. trachomatis infection,
C. trachomatis and preterm rupture of membranes 399
rupture of Control
membranes (n = 84)
22.8 ± 5.7 22.4 ± 5.5
1.5 ± 1.5 1.2 ± 1.3
30.0 ± 4.6 29.9 ± 4.5
71% 80%
83% 81%
71% 68%
90% 90%
there was also a trend toward decreased parity m
C. trachomatis-positive patients.
Prevalence of infection. There was a gr~ater prev-
alence of cervical C. trachomatis infection in patients with
preterm premature rupture of membrG~nes than in con-
trols (p < 0.001; Table II). Although group B strep-
tococcal infection was far less prevalent in our popu-
lation than was C. trachomatis infection; group B strep-
tococci were isolated significantly more often in patients
with preterm premature rupture of membranes as
compared with controls (p < 0.05; Table III). Despite
similar management for both groups of patients, which
included routine screening and treatment for genital
gonorrhea, N. gonorrhoeae was isolated from 6 of the
45 patients with preterm premature rupture of mem-
branes (13%). No control was infected with N. gonor-
rhoeae at the time of C. trachomatis .culture; however,
three were infected with N. gonorrhoeae earlier in preg-
nancy. This is consistent with the <5% prevalence of
N. gonorrhoeae infection in our clinic population.
Infections with more than one pathogen were found
in four patients with preterm premature rupture of
membranes but none of the controls. Of the 48 patients
for whom all culture results were available, only 20
(41.7%) of the preterm premature rupture of mem-
branes group were not infected with C. trachomatis or
group B streptococci. Of these 20 patients, four were
infected with N. gonorrhoeae at delivery, leaving only 16
(33.3%) uninfected with one of these genital pathogens
compared with 81.0% of the control group.
The relative risk of preterm premature rupture of
membranes associated with C. trachomatis and group B
streptococcal infections is shown in Table IV. The in-
creased risk of preterm premature rupture of mem-
branes associated with C. trachomatis infection was sta-
tistically significant (p < 0.001) with and without ad-
justment for other potential risk factors. This was also
true for group B streptococcal infection (p = 0.002).
400 Alger et al. August 1988
Am J Obstet Gynecol

Table II. Chlamydia infection in patients with Table IV. Association of preterm premature
preterm premature rupture of membranes rupture of membranes with chlamydia and
and controls group B streptococci
Culture + Culture - Total
Pathogen Adjustment factors p value
No., % No., % No., %
C. trachomatis None 4.33 <0.001
Patients with preterm 23 44.2 29 55.8 52 100 PPROM risk factors* 7.73 <0.001
premature rupture Risk factors + GBS 8.69 <0.001
of membranes Group B streptococci None 4.73 0.031
Controls 13 15.5 71' 84.5 84 100 PPROM risk factors 8.04 0.007
Risk factors + CT 13.29 0.002

PPROM, Preterm premature rupture of membranes; GBS,

group B streptococci; CT, C. trachomatis.
Table III. Group B streptococci in patients *Maternal age, gestational age, parity, race, marital status,
with preterm premature rupture of care payment type, source of prenatal care, and postscreening
membranes and controls N. gonorrhoeae infection.

Culture + Culture - Total

No.I % No. J % No.] %
Patients with preterm 7 15.6 38 84.4 45
premature rupture
of membranes
Controls 3 3.8 77 96.2 80
The risks of pre term premature rupture of membranes
attributed to C. trachomatis and to group B streptococci
were found to be mutually independent and indepen-
dent of infection with N. gonorrhoeae. Additional anal-
yses indicated that there were no significant interactions
between C. trachomatis and group B streptococcal in-
fections relating to the risk of preterm premature rup-
ture of membranes.
The frequency of C. trachoma/is isolation, unlike that
of group B streptococci, in our population was age
dependent, with younger patients more likely to be
infected than older patients (Tables V and VI). The
mean age of C. trachomatis-positive patients was 20.7
years, compared with 23.2 years for C. trachomatis-
negative patients (p < 0.05). The risk of preterm pre-
mature rupture of membranes associated with both
C. trachomatis and group B streptococcal infections was
also markedly increased in patients younger than the
median age of 21 years. Even in the older age group,
however, we observed an increased prevalence of
C. trachomatis infection among patients as compared
with controls (Table V), although the difference was
not significant.
Pregnancy outcome. There were no significant dif-
ferences in gestational age at membrane rupture, in-
terval to delivery, and delivery route between patients
with preterm premature rupture of membranes in-
fected with C. trachomatis and those with negative cer-
vical cultures. The mean interval to delivery was 2.8
days for patients with endocervical C. trachomatis and
3.3 days for C. trachomatis-negative women. There also
was no increase in intrapartum or early postpartum
infectious morbidity associated with C. trachomatis. Five
100 patients in each group experienced infectious compli-
cations.
Because we were investigating the prevalence of
100
C. trachomatis infection and not obstetric outcome in the
controls, the majority of C. trachomatis-infected control
women were treated with erythromycin as soon as cul-
ture results were available. Nonetheless, the incidence
of preterm delivery associated with positive versus neg-
ative C. trachomatis cultures in the control group was
16.0% (2113) versus 7.0% (5/71). The two-culture-
positive women who delivered prematurely were
among three controls who had not been treated with
erythromycin after the discovery of a positive culture.
This suggests that treatment of antenatal chlamydial
infection may reduce the incidence of preterm delivery.
Infant outcome. The 52 women with preterm pre-
mature rupture of membranes delivered 53 infants
(one set of twins was born to a C. trachomatis-negative
mother). There was an increased incidence of fetal
deaths in utero, all occurring in previable gestations in
the C. trachomatis-positive mothers compared with the
C. trachomatis-negative mothers (13.0% versus 3.3%).
Perinatal mortality for the C. trachomatis-positive moth-
ers was twice that of the culture-negative mothers
(21.7% versus 10.0%). Neither difference was statisti-
cally significant, however.
There were no significant differences in the birth
weights, gestational ages, or Apgar scores at delivery
between the two groups, whether including or exclud-
ing the previable births. Length of hospital stay was
independent of the presence or absence of a positive
maternal culture for C. trachomatis but was influenced
primarily by birth weight, maturity, and the degree of
respiratory illness. There were no significant differ-
ences in the incidence of respiratory distress or the
requirement for ventilatory support beyond 24 hours
in infants born to C. trachomatis-positive mothers.
Volume 159 C. trachomatis and pr~term rupture of membranes 401
Number 2

Table V. C. tracho'TIUJ,tis in patients with preterm premature rupture of membranes versus controls
categorized by age
Culture + Culture -
Total Odds
Age (yr) Group No.
I % No.
I % no. ratio p value

<21 Patients 16 69.6 7 30.4 23

32 78.0 8.13 <0.05
Controls 9 22.0 41
2:21 Patients 7 24.1 22 75.9 29
Controls 4 9.3 39 90.7 43 3.10 >0.05

Table VI. Group B streptococci in patients with preterm premature rupture of membranes versus
controls categorized by age
Culture + Culture -
Total Odds p
Age (yr) Group No. I % No. I % no. ratio value

<21 Patients 3 14.3 18 85.7 21

6.17 >0.05
Controls 1 2.6 37 97.4 38
Patients 4 16.7 20 83.3 24
4.00 >0.05
Controls 2 4.8 40 95.2 42

Individual infants at risk developed chlamydial or
gonococcal conjunctivitis or clinical presentations com-
patible with the diagnosis of early onset group B strep-
tococcal pneumonitis, but no infant in either group had
culture-proved bacterial sepsis in the first week of life.
Comment
Preterm delivery has been associated with maternal
genital infections, most commonly with N. gonorrhoeae 10
and group B streptococci.'' The role of C. trachomatis
has only recently been examined, since the techniques
for diagnosing C. trachomatis infections have become
more widely available.
However, in 1971 Elder et al. 5 reported an interesting
but unexplained observation. By treating asymptom-
atic, nonbacteriuric pregnant patients empirically with
tetracycline, they reduced the incidence of premature
delivery. Similarly, McGregor et al., 6 treating patients
in preterm labor at <34 weeks' gestation empirically
with erythromycin base, noted a trend toward prolon-
gation of pregnancy. Because C. trachomatis is sensitive
to both of these antibiotics, this suggested to us that
this agent might play a role in causing preterm delivery.
There are strong theoretical reasons to suggest that
C. trachomatis could play a causal role in the initiation
of both preterm labor and preterm premature rupture
of membranes. The hallmark of chlamydial cervicitis is
a mucopurulent discharge characterized by an in-
creased concentration of polymorphonuclear leuko-
cytes. This inflammatory process is capable of altering
the local tissue pH and thus perturbing lysosomal mem-
branes in the adjacent chorioamnion. These lysosomes
contain the enzyme phospholipase A 2 , which initiates
cleavage of arachidonic acid from the phospholipid
components of the chorioamnion. Arachidonic acid in
turn is capable of initiating the cascade of reactions that
results in prostaglandin synthesis and the onset of uter-
ine contractions. The concomitant release of other ly-
sosomal enzymes might be expected to produce a direct
cytotoxic effect on chorioamniotic membrane cells, and
hence membrane rupture. Granulocyte elastase pref-
erentially degrades type III collagen, which has been
shown to be deficient at the site of rupture. 12
Chlamydia might also contribute to early membrane
rupture more directly; the life cycle of C. tracho'TIUJ,tis,
which has been shown to replicate in human amnion
cells, requires cellular death as the organism is released
from the infected cell to spread to others. This cyto-
pathic effect could weaken membranes directly.
Previous reports implicating group B streptococci,
N. gonorrhoeae, or C. trachomatis as causative factors in
preterm delivery had failed to examine the target pop-
ulation simultaneously for these three agents. Our Bal-
timore obstetric population, with a C. trachomatis prev-
alence two to three times that found in other studies,
seemed well suited for examining a possible relation-
ship between C. trachomatis infection and prematurity.
To more clearly define the relative roles of N. gonor-
rhoeae, group B streptococci, and C. trachomatis in pre-
term premature rupture of membranes, we cultured
for all three organisms. The risk of preterm premature
rupture of membranes associated with C. trachomatis
infection was highly significant (p < 0.001). Although
group B streptococcal infection was far less prevalent
in our population than was C. trachomatis infection, a
similar association was seen between preterm prema-
ture rupture of membranes and group B streptococci,
as has been reported: u However, the risks of preterm
402 Alger et al.
premature rupture of membranes attributed to C. tra-
chomatis and to group B streptococci were mutually
independent and independent ·of infection with N. gon-
orrhoeae. The importance of infection in causing pre-
term premature rupture of membranes is supported
by these data; only 33.3% of patients with preterm pre-
mature rupture of membranes were uninfected with
one of these pathogens compared with 81.0% of the
control group.
We found an unusually high C. trachomatis infection
rate of 44.2% among patients with preterm premature
rupture of membranes. The risk of preterm premature
rupture of membranes associated with both C. tracho-
matis and group B streptococcal infections was mark-
edly increased in the age group <21 years. The 69.6%
prevalence among our patients <21 years old is the
highest reported in any population to date. Demo-
graphic parameters known to be associated with C. tra-
chomatis genital infection include age, socioeconomic
status, and marital status. Our results are consistent
with these findings. We did not find any association
between C. trachomatis infection and parity (controlling
for age) or gestational age.
Although previous studies have shown that the in-
terval between membrane rupture and delivery is re-
duced in patients infected with various other mi-
croorganisms, we found no such reduction in our
C. trachomatis-positive patients. A possible explanation
is that the replication cycle of C. trachomatis is sufficiently
long that once membrane rupture allows communica-
tion between the vaginal flora and amniotic cavity, the
effects of rapid replication of endogenous microorgan-
isms are expressed before any chlamydia! effects can
be detected. The long replication cycle of C. trachomatis
may also explain why there was no difference in intra-
partum or early onset postpartum infectious morbidity
between the C. trachomatis-positive and -negative pa-
tients or their infants. Histologic examination of 16 of
the 52 placentas from the patients with preterm pre-
mature rupture of membranes showed evidence of
chorioamnionitis in most cases, whether or not the
patients were infected with C. trachomatis. Because the
C. trachomatis-positive patients were treated with ap-
propriate antibiotics upon receipt of culture results
within a few days of delivery, increased late postpartum
endometritis as reported by others 13 was not seen in
our study.
Our results demonstrating an association between the
presence of endocervical C. trachomatis and preterm
premature rupture of membranes are consistent with
those reported by Martin et aP Although they did not
specificially address the incidence of preterm prema-
ture rupture of membranes, they found a significant
increase in preterm deliveries among their patients who
August 1988
Am J Obstet Gynecol
were culture positive at < 19 weeks' gestation. Patients
with positive cultures were at even greater risk for de-
livering before 30 weeks than for delivery before 37
weeks' gestation when compared with controls. They
also noted an increase in perinatal deaths, especially
stillbirths.
Harrison et al. 3 did not find an increased incidence
of preterm premature rupture of membranes in their
pregnant patients who were culture positive for C. tra-
chomatis. However, the subgroup of culture-positive pa-
tients who were also IgM seropositive did have an in-
creased prevalence of premature rupture of mem-
branes. The meaning of an IgM-positive infection in
pregnancy is not clear. It may reflect primary infection,
greater antigenic load, recent spread, or more inva-
sive disease. It is possible that the majority of our
C. trachomatis-positive patients would have been IgM
seropositive had they been tested.
After the completion of our study, Sweet et alY re-
ported the reults of a large prospective study of preg-
nancy outcome in women with endocervical C. tracho-
matis infection compared with uninfected patients. The
authors found no significant differences between the
two groups in either their rates of premature rupture
of membranes or preterm delivery. However, similar
to the results of Harrison et al./ the subgroup of pa-
tients who were IgM seropositive against C. trachomatis
were at increased risk for premature rupture of mem-
branes and for preterm delivery. This increased risk
was no longer present when a multivariate logistic
regression analysis that controlled for concurrent in-
fection with group B streptococci, Mycoplasma hominis,
and Ureaplasma urealyticum was performed. However,
because U. urealyticum was recovered from 90% of IgM-
seropositive patients, the authors concluded that inter-
pretation of these results was difficult and perhaps
larger numbers were needed. Intriguingly, premature
rupture of membranes with preterm delivery was twice
as prevalent in their C. trachomatis-infected patients
compared with uninfected patients, a difference sig-
nificant by univariate analysis (no multivariate analyses
were cited in this case.)
Several points must be considered when interpreting
the results of these studies. As with group B strepto-
coccal infection, most patients with C. trachomatis infec-
tion do not develop premature rupture of membranes;
therefore, prospective studies such as those of Harrison
et aP and Sweet et al. 14 require large numbers of pa-
tients to achieve st_atistical significance. However, ~~­
though the majority of patients with cervical C. tra-
chomatis infection will not deliver prematurely, of those
patients who do develop preterm premature rupture
of membranes, a disproportionate number will be in-
fected with C. trachomatis. If indeed IgM seropositivity
Volume 159
Number 2
is associated with premature rupture of membranes,
one would expect a higher prevalence of IgM sero-
positivity in our patients with preterm premature rup-
ture of membranes than in controls had they been
tested. Thus the results of their studies and ours are
not necessarily inconsistent. Additionally, population
differences between studies such as prevalence of en-
docervical C. trachomatis, race, or number of sexual part-
ners may influence findings.
Hardy et al! investigated an adolescent population
demographically similar to ours and did not find
C. trachomatis infection to be a risk factor for preterm
delivery. However, because the mean gestational age at
time of initial culture was 34 weeks, patients with
C. trachomatis who developed premature rupture of
membranes before that time would not have been en-
tered into the study. They did observe that patients
with both C. trachomatis and T. vagina/is had a preva-
lence of low-birth-weight infants four times that of un-
infected adolescents.
Hardy et al. 4 also reported a very low prevalence of
endocervical N. gonorrhoeae infection among clinic pa-
tients during the second half of pregnancy, in agree-
ment with our findings. None of our 52 control women
were infected with N. gonorrhoeae; furthermore, in the
presence of a strikingly high prevalence of C. trachomatis
in their study, only one of their 115 adolescents was
N. gonorrhoeae-positive. Because pregnant patients in
our clinics are routinely screened for N. gonorrhoeae at
their first prenatal visit and treated if positive, we could
not assess the risk of preterm premature rupture of
membranes due toN. gonorrhoeae on a prospective basis.
However, given similar management plans, six patients
with preterm premature rupture of membranes were
found to be N. gonorrhoeae positive at delivery. Thus it
appears that the few patients who are unsuccessfully
treated or contract N. gonorrhoeae after the initial
screening are at particular risk for developing preterm
premature rupture of membranes. Handsfield et al. 15
reported a 75% incidence of prematurely ruptured
membranes and a 67% incidence of premature delivery
in 12 women with positive N. gonorrhoeae cultures at
delivery. In our study, the risk of preterm premature
rupture of membranes associated with C. trachomatis
and group B streptococcal infections was independent
of infection with N. gonorrhoeae, however.
Other agents implicated in prematurity include
M. hominis, U. urealyticum, and bacterial vaginosis (Gard-
nerella vagina/is in association with increased concen-
tration of vaginal anaerobes). 16 Four groups of inves-
tigators have failed to find an association between ma-
ternal mycoplasma colonization and risk of premature
rupture of membranes.'· 4 ' 14' 17 It is possible that only a
subgroup of patients, such as those from whom my-
C. trachomatis and preterm rupture of membranes 403
coplasma can be recovered from the amniotic cavity, is
at risk. In our clinic populations, 70% of patients have
M hominis recoverable from the endocervix and 86%
have cervical U. urealyticum (unpublished observations),
which is similar to prevalences reported by Hardy's
group studying pregnant adolescents from the same
geographic area. Because both organisms are so prev-
alent, it would be difficult to demonstrate a significant
difference in prevalence between patients with preterm
premature rupture of membranes and controls. We did
not specifically look for G. vagina/is colonization because
this organism had not yet been implicated at the time
the study was initiated.
In conclusion, we have found an association between
cervical infection with C. trachomatis and risk of preterm
premature rupture of membranes that is independent
of other risk factors including infection with group B
streptococci and N. gonorrhoeae. Cervical infection with
C. trachomatis did not significantly affect early maternal
or neonatal complication rates after delivery. It is im-
portant to determine whether antenatal treatment of
C. trachomatis infections can reduce the incidence of
prematurity.
REFERENCES
1. Frommell GT, Rothenberg R, Wang S, Mcintosh K. Chla-
mydia! infection of mothers and their infants. J Pediatr
1979;95:29.
2. Martin DH, Koutsky L, Eschenbach DA, et al. Prematurity
and perinatal mortality in pregnancies complicated by
maternal Chlamydia trachomatis infections. JAMA 1982;
247:1585.
3. Harrison HR, Alexander ER, Weinstein L, et al. Cervical
Chlamydia trachomatis and mycoplasmal infections in preg-
nancy. Epidemiology and outcomes. JAMA 1983;250:
1721.
4. Hardy PH, Hardy JB, Nell EE, Graham DA, et al. Prev-
alence of six sexually transmitted disease agents among
pregnant inner-city adolescents and pregnancy outcome.
Lancet 1984;2:333.
5. Elder HA, Santamaria BAG, SmithS, Kass EH. The nat-
ural history of asymptomatic bacteriuria during preg-
nancy: the effect of tetracycline on the clinical course and
the outcome of pregnancy. AM J 0BSTET GYNECOL
1971; 111:441.
6. McGregor J, French J, Spencer N, Reller LB. Random-
ized, prospective double-blinded trial of erythromycin
base in idiopathic preterm labor [Abstracts]. In: Proceed-
ings of the Interscience Conference on Antimicrobial
Agents and Chemotherapy. 1984.
7. Chacko MR, Lovchik JC. Chlamydia trachomatis infection
in sexually active adolescents. Prevalence and risk factors.
Pediatrics 1984;73:836.
8. Walker SH, Duncan DB. Estimation of the probability of
an event as a function of several independent variables.
Biometrika 1967;54:167.
9. Schlesselman JJ. Case-control studies: design, conduct,
analysis. New York: Oxford University Press, 1982.
10. Edwards LE, Barrada Ml, Hamann AA, Hakanson EY.
Gonorrhea in pregnancy. AM J 0BSTET GYNECOL 1978;
132:637.
11. ReganJA, Chao S,James LS. Premature rupture of mem-
branes, preterm delivery, and group B streptococcal col-
Alger et al. August 1988
Am J Obstet Gynecol

onization of mothers. AM 1 0BSTET GYNECOL 1981;
141:184.
12. Kanayama N, Terao T, Kawashima Y, eta!. Collagen types
in normal and prematurely ruptured amniotic mem-
branes. AM 1 0BSTET GYNECOL 1985;153:899.
13. Wager GP, Martin DH, Koutsky L, eta!. Puerperal infec-
tious morbidity: Relationship to route of delivery and to
antepartum Chlamydia trachomatis infection. AM 1 0BSTET
GYNECOL 1980;138:1028.
14. Sweet RL, Landers DV, Walker C, Schacter]. Chlamydia
trachomatis infection and pregnancy outcome. AM 1 DB-
STET GYNECOL 1987;156:824.
15. Handsfield HH, Hodson WA, Holmes KK. Neonatal
gonococcal infection. Orogastric contamination with Neis-
seria gonorrhoeae. 1AMA 1973;225:697.
16. Gravett MG, Nelson HP, DeRouen T, eta!. Independent
associations of bacterial vaginosis and Chlamydia tracho-
matis infection with adverse pregnancy outcome. 1AMA
1986;256: 1899-903.
17. Minkoff H, Grunebaum AN, Schwarz RH, et a!. Risk fac-
tors for prematurity and premature rupture of mem-
branes: A prospective study of the vaginal flora in preg-
nancy. AM 1 0BSTET GYNECOL 1984; 150:965.

Short-course antibiotic therapy for the treatment of

chorioamnionitis and postpartum endomyometritis
Thomas G. Stovall, MD, Steven E. Ambrose, MD, Frank W. Ling, MD, and
Garland D. Anderson, MD
Memphis, Tennessee

The development of chorioamnionitis and endomyometritis has traditionally required treatment with
broad-spectrum antibiotic therapy and extended hospitalization. In the past, once parenteral antibiotic
therapy was instituted, it was continued for 5 to 7 days and until the patient remained afebrile for 48 hours.
To shorten the length of hospital stay, the length of parenteral antibiotic administration was reduced and
an oral antibiotic was added, to complete a 7- to 10-day course of therapy. We evaluated the effectiveness
of an even shorter course of parenteral antibiotics without the addition of oral antibiotics. Forty-two patients
with chorioamnionitis and 64 with endomyometritis were enrolled in the study. Antibiotic therapy was
continued until the patient's temperature was <99.5° F for 12 to 24 hours. Of the 106 patients, only two
were readmitted, both as a result of superficial wound separation. No patient had an infectious
complication. A shorter course of parenteral antibiotics without the addition of an oral antibiotic gives
results comparable to the standard extended treatment regimens, but is advantageous with respect to
cost, patient compliance, and hospital stay. (AM J OssTET GYNECOL 1988;159:404-7.)

Key words: Antimicrobial therapy, endomyometritis, chorioamnionitis

Chorioamnionitis develops in 0.5% to 2.0% of all
term pregnancies, but this rate is increased to as much
as 25% when rupture of membranes exceeds 24 hours. 1
It is further increased in patients who have both a pre-
term gestation and prolonged membrane rupture! En-
domyometritis, on the other hand, is the most common
cause of infectious morbidity in the postpartum pa-
tient.3 Several risk factors for the development of en-
domyometritis have been identified, with cesarean de-
livery posing the greatest risk. Cesarean section in-
creases the postpartum infection rate up to 20-fold over
that of patients who have a vaginal birth. Thus as many
From the Department of Obstetrics and Gynecology, University of
Tennessee.
Received for publication December 8, 1987; revised February 22,
1988; accepted April20, 1988.
Reprint requests: Thomas G. Stovall, MD, Department of Obstetrics
and Gynecology, University of Tennessee, 853 Jefferson Ave.,
Memphis, TN 38103.
as 20% to 50% of patients undergoing abdominal de-
livery develop postpartum endomyometritis.' Both
chorioamnionitis and endomyometritis require the ad-
ministration of parenteral antibiotics, usually resulting
in an extended period of hospitalization.
Ten years ago, patients with postpartum endomyo-
metritis were treated with intravenous antibiotics for 7
days and had an average hospital stay of 8 to 9 days. 5
In an attempt to reduce postpartum hospitalization,
outpatient administration of a broad-spectrum oral an-
tibiotic after completion of parenteral therapy has be-
come an accepted practice. The current recommended
regimens consist of administering parenteral antibiotics
until the patient has been afebrile for 24 to 48 hours,
followed by a broad-spectrum oral antibiotic to com-
plete a 7- to 10-day course after discharge from the
hospital. 6•8 Although this course of therapy is recom-
mended in major textbooks and journals, no large-scale
prospective investigation to support the use of oral an-

404