Trichomoniasis in Pregnancy and Mental Retardation in Children

JOSHUA R. MANN, SUZANNE MCDERMOTT, TIMOTHY L. BARNES, JAMES HARDIN,

HAIKUN BAO, AND LI ZHOU

PURPOSE: Trichomoniasis is a highly prevalent sexually transmitted infection and is associated with
premature rupture of membranes, preterm birth, and low birth weight. This study examines the association
between maternal trichomoniasis and intellectual disability (ID) in children.
METHODS: This study utilized linked maternal, infant, and child records for 134,596 Medicaid-insured
singleton births in South Carolina from 1996 through 2002. Data were obtained from Medicaid billing
records, birth certificates, and administrative data from the South Carolina Department of Education
(DOE) and the Department of Disabilities and Special Needs (DDSN). Pregnancies during which women
were diagnosed with urinary tract infection, chlamydia, gonorrhea, or vulvovaginal candidiasis were
excluded, as were children diagnosed with a known cause of mental retardation. Odds of diagnosed ID
in children were modeled using population averaged generalized estimating equation models.
RESULTS: Controlling for potential confounders, women with trichomoniasis were significantly more
likely to have a child with ID (hazard ratio [HR] Z 1.28; 95% confidence interval [CI], 1.12–1.46). The
association was stronger for moderate to severe ID documented by the school system or DDSN
(HR Z 1.84; 95% CI, 1.35–2.51). Second-trimester trichomoniasis was associated with more than
a three-fold increase in the odds a child was identified as trainable mentally handicapped or profoundly
mentally handicapped in the public school system, or was receiving ID services from DDSN. There was
not a significant difference in the risk of ID in children of women with treated versus untreated
trichomoniasis.
CONCLUSION: Maternal trichomoniasis may be a preventable risk factor for ID.
Ann Epidemiol 2009;19:891–899. Ó 2009 Elsevier Inc. All rights reserved.
KEY WORDS: Mental Retardation, Trichomonas Vaginalis, Pregnancy, Cohort Study.

cytomegalovirus, syphilis, and toxoplasmosis (7). Urinary

INTRODUCTION
Trichomoniasis is the most common non-viral sexually
transmitted infection in the United States, with an
estimated 7.4 million new cases annually (1, 2). A nationally
representative study of reproductive-age American women
found that trichomonal infection was present in 3.1% of
the study population; the prevalence was higher (13.3%)
among African Americans (3). The Vaginal Infections in
Pregnancy Study detected trichomoniasis in almost 13% of
pregnant women screened; the prevalence was over 20%
among pregnant African American women (4). Trichomo-
niasis during pregnancy has been shown to produce both
a vaginal and systemic immune response and is associated
with adverse outcomes, including premature rupture of the
membranes, preterm birth, and low birth weight (1, 5, 6).
Intellectual disability (ID), synonymous with mental
retardation, is known to be associated with maternal infec-
tions including group B streptococcus, herpes simplex virus,
From the School of Medicine (J.R.M, S.M.) and Arnold School of Public
Health (T.L.B., J.H., H.B., L.Z.), University of South Carolina, Columbia.
Address correspondence to: Joshua R. Mann, MD, MPH, University of
South Carolina School of Medicine, Department of Family and Preventive
Medicine, 3209 Colonial Dr., Columbia, SC 29203. Tel: (803) 434-4575.
Fax: (803) 434-8374. E-mail: Joshua.mann@sc.edu.
Received February 4, 2009; accepted August 12, 2009.
tract infections (UTIs) during pregnancy have also been
associated with increased risk of ID, especially if the
infection is left untreated (8). Since preterm birth is an
established risk factor for neurocognitive deficits such
as ID (9), it is reasonable to hypothesize that infections
associated with preterm birth may also be associated with
increased risk of ID. While a substantial amount of research
has established maternal infection (particularly intra-
amniotic infection) as a risk factor for cerebral palsy (10),
studies of a potential association between maternal
reproductive tract infection and ID are lacking.
METHODS
This project began as a broader study of maternal genitouri-
nary infections and pregnancy and child outcomes,
including ID. It was granted exempt status by the University
of South Carolina Institutional Review Board. We utilized
a retrospective cohort study design, wherein we obtained
de-identified South Carolina Medicaid billing records
for pregnancies and deliveries that occurred from 1996
through 2002 (Medicaid purchases health services for
low-income pregnant women and children). We also
obtained linked files to birth certificates, neonatal and fetal
death certificates, Medicaid billing records for children, and

Ó 2009 Elsevier Inc. All rights reserved. 1047-2797/09/$–see front matter

360 Park Avenue South, New York, NY 10010 doi:10.1016/j.annepidem.2009.08.004
892 Mann et al.
TRICHOMONIASIS IN PREGNANCY AND MENTAL RETARDATION IN CHILDREN
Selected Abbreviations and Acronyms
ID Z intellectual disability
UTI Z urinary tract infection
DOE Z Department of Education
DDSN Z Department of Disabilities and Special Needs
ICD Z International Classification of Diseases
MR Z mental retardation
EMH Z Educable Mentally Handicapped
TMH Z Trainable Mentally Handicapped
PMH Z Profoundly Mentally Handicapped
IQ Z intelligence quotient
administrative data for children receiving services from the
South Carolina Department of Education (DOE) and the
South Carolina Department of Disabilities and Special
Needs (DDSN).
Case Definition of Intellectual Disability
Because the term ‘‘intellectual disability (ID)’’ has replaced
mental retardation (MR) in most classification schemes,
‘‘ID’’ will be used even in cases when MR was the actual
diagnosis. Children with ID were identified based on three
data sources, updated through spring of 2008. The first
source was the Medicaid file which includes International
Classification of Diseases, Ninth Revision (ICD-9) diagnosis
codes in clinical billing records. The ICD-9 codes for MR
are 317 (mild MR), 318 (moderate, severe, or profound
MR, depending on the specific fourth digit), and 319
(unspecified or unknown severity). The second source of
diagnosis was a data file from the DOE. Three categories
of ID are used in public school records: educable mentally
handicapped (EMH), trainable mentally handicapped
(TMH), and profoundly mentally handicapped (PMH).
EMH corresponds to mild ID, TMH with moderate to
severe, and PMH with profound. The third source of ID
diagnoses was enrollment in services from DDSN. DDSN
provides support services to individuals with ID and their
families. Eligibility for ID services through DDSN requires
a psychological assessment, including IQ and adaptive func-
tion testing, and medical evaluation.
Based on the criteria utilized by the three sources, the
level of certainty of a diagnosis of ID is greatest for cases
identified based on (1) TMH or PMH placement or (2)
receipt of DDSN services for ID. Other diagnoses of ID
(MR in Medicaid records or EMH classification in school)
should be considered possible cases of ID, because of
(1) our inability to confirm the criteria for diagnoses made
by community physicians and (2) the potential for non-
biological factors (i.e., social deprivation) to cause develop-
mental delays that produce ‘false positive’ cases of mild ID
(EMH) in DOE records.
We initially modeled the outcome ‘‘any ID’’ (MR of any
severity, in any data source); then we modeled each of the
AEP Vol. 19, No. 12
December 2009: 891–899
more restrictive outcomes (TMH/PMH and DDSN MR).
Finally, we modeled a combined outcome of TMH/PMD
school placement or receipt of ID-related services from
DDSN, since the latter group has a more severe level of
ID that is less likely to be diagnosed in error; that is, they
almost certainly represent true cases of ID.
Case Definition for Infection
Maternal genitourinary infections were identified using
ICD-9 codes in the Medicaid billing data. Cases of tricho-
moniasis were identified on the basis of ICD-9 code 131.0
or 131.9. The other specific infections included were chla-
mydia/non-gonococcal urethritis, gonorrhea, vulvovaginal
candidiasis, and urinary tract infection. The definition of
genitourinary infection also included the following nonspe-
cific conditions: vaginitis, cervicitis, ascending reproductive
tract infection (pelvic inflammatory disease or chorioam-
nionitis), and unspecified genitourinary infection (ICD-9
codes for all these conditions are available from the
authors). Initial analyses showed that trichomoniasis was
a more important predictor of MR than other infections,
so we focused the additional analyses on trichomoniasis.
Exclusions
We excluded children who were not singleton births, who died
during the first 28 days of life, or who were diagnosed
with a known or likely cause of ID. The known/likely causes
removed were Down syndrome, Edwards syndrome, Patau
syndrome, fragile X syndrome, fetal alcohol syndrome,
Prader-Willi syndrome, congenital syphilis, congenital brain
anomaly, cerebral degeneration, ‘‘shaken baby’’ syndrome,
child abuse, intracranial injury, meningitis, encephalitis,
phenylketonuria, toxoplasmosis, congenital rubella, or
congenital hypothyroidism. By excluding children with these
known or likely causes of ID, we were able to evaluate maternal
infection as a potential risk factor for unexplained ID.
Statistical Modeling
Differential length of follow-up for children in our cohort
could bias the results of the study. Therefore, we utilized
survival analysis (multivariable Cox proportional hazards
models using PROC PHREG in SAS v. 9.1) to model the
risk of acquiring a diagnosis of ID over time. Data on precise
length of follow-up in the various data sources were not
available. However, we obtained information describing
whether a child was enrolled in South Carolina Medicaid
or public school in each year following birth. We also
obtained information on the year in which children were
first diagnosed with ID in the Medicaid data, or first enrolled
in special education in the public schools, or enrolled in an
SC DDSN program. We calculated the length of follow-up
by subtracting the year of birth from the most recent year
AEP Vol. 19, No. 12
December 2009: 891–899 TRICHOMONIASIS IN PREGNANCY AND MENTAL RETARDATION IN CHILDREN
of Medicaid or public school enrollment. Children with ID
were censored at the time when they were first identified as
having ID (calculated by subtracting the year of birth from
the year of initial diagnosis in Medicaid, entrance into
special education, or enrollment in DDSN services).
We considered using the total number of years of follow-
up as the time variable for the survival analysis (that is,
summing the number of years in which a child was enrolled
in Medicaid or the public school system). However, this
approach would have assumed that a year of follow-up has
equal weight no matter when it occurs; that is, a child
followed up for 2 years after birth would have been treated
the same as a child who moved out of state after birth but
then returned when he or she was 10 years old and was
enrolled in public schools for the next 2 years. Clearly, the
child in the second example would be far more likely to be
identified as having ID than the child who left the state
when he or she was 2 years old, since ID tends to become
more apparent as a child ages and is expected to be able to
perform more complex mental tasks. For that reason, we
chose to utilize the child’s age in the most recent year of
eligibility rather than the total number of years of follow-up.
Because some women (approximately 20%) had more
than one child during the study period, observations were
not completely independent. We utilized the COVS option
in PROC PHREG, which utilizes a sandwich estimate for the
covariance matrix and yields a robust standard error for the
parameter estimates, permitting accurate calculation of p
values in the context of repeat pregnancies (correlated data).
We tested the validity of the proportional hazards
assumption of the Cox models by creating time-dependent
variables (defined as the multiplicative interaction between
the independent variable and the log of follow-up time). If
a time-dependent variable was statistically significant
(statistically significant interaction between the variable
and the log of time), this indicated that the hazards were
not proportional over time. These variables were removed
from the model, but were still adjusted for by stratification
(using the STRATA option of PROC PHREG). We verified
adherence with the proportional hazards assumption by
calculating Schoenfeld residuals for each variable in the
model and examining the Pearson correlation coefficient
between the Schoenfeld residuals and ranked ‘failure times.’
Statistically significant correlations indicate violations of
the proportional hazards assumption. Variables that failed
either test were removed from the model, but they were still
adjusted for by stratification (using the STRATA option of
PROC PHREG).
Treatment for Trichomoniasis
The Centers for Disease Control and Prevention recom-
mends treatment for symptomatic trichomoniasis using
Mann et al. 893
a single 2-g dose of metronidazole (11). We wanted to
evaluate the possibility that treatment with metronidazole
positively or negatively modified the association between
trichomoniasis and MR. Treatment status was determined
by using Medicaid outpatient pharmacy billing records,
which provided drug names and dates dispensed. Women
with trichomoniasis were considered treated if they filled
a prescription for oral metronidazole within 14 days of the
first diagnosis of trichomoniasis. All other women were
considered untreated. This analysis was limited to women
who were diagnosed with trichomoniasis at least 30 days
prior to delivery, to permit adequate time for provision of
treatment and for treatment effects.
Because this is an observational study, metronidazole
treatment for trichomoniasis was not a controlled variable.
Therefore, estimates of the impact of treatment could be
biased by the propensity for particular women to be treated.
Propensity score methods address estimating models with
such variables. We estimated the probability of receiving
treatment for the infection based on whether the mother
was younger than 21 years, whether the mother was white,
whether the mother used alcohol, and whether the mother
used tobacco. We calculated weights equal to the inverse
probability of receiving treatment. Weights for women
without trichomoniasis were set to one, multiple weights
for the same mother were replaced with the per-mother
mean weight, and all weights were scaled to sum to the
number of observations in the data set. Utilizing these
weights, we estimated a Cox proportional hazards regression
model where mothers were the independent unit of analysis.
The modeling was limited to women diagnosed with tricho-
moniasis at least 30 days and accounted for the same cova-
riates as the other models.
RESULTS
The full cohort included 144,837 infants, of whom 3,542
were non-singleton births and another 377 died in the first
28 days of life. An additional 5,426 children were removed
because they were diagnosed with a genetic or chromosomal
condition or congenital anomaly known to cause ID, or with
a traumatic brain injury, central nervous system infection, or
other high-risk condition likely to substantially increase the
risk of ID. Of the children excluded because of high-risk
diagnoses, 766 (14.1%) had MR compared to a rate of
4.0% in the remaining children (prevalence was 4.4% for
the sample prior to any exclusions). There were 891 children
who were excluded because they lacked follow-up informa-
tion in the Medicaid file and did not enroll in public schools.
Finally, five observations were deleted because they had
missing information for one or more covariates.
894 Mann et al. AEP Vol. 19, No. 12
TRICHOMONIASIS IN PREGNANCY AND MENTAL RETARDATION IN CHILDREN December 2009: 891–899

TABLE 1. Cohort characteristics

Without ID (n Z 129,208) With ID (n Z 5,388)
Mean SD Mean SD p Value

Gestational age (wk) 38.75 2.00 Gestational age (wk) 37.30 4.04 !0.0001
Birth weight (g) 3213.07 562.16 Birth weight (g) 2881.17 884.82 !0.0001
Maternal age (yr) 22.93 5.23 Maternal age (yr) 23.52 5.66 !0.0001
Oldest age in Medicaid 6.94 3.29 Oldest age in Medicaid 8.62 2.47 !0.0001
Oldest age in DOE file Oldest age in DOE file 9.13 1.99 !0.0001
Category No. % Category No. %

Race White 61,441 47.55 Race White 1,991 36.95 !0.0001

Other 67,767 52.45 Other 3,397 63.05
O12 years education* Yes 83,915 65.95 O12 years education* Yes 3,141 58.30 !0.0001
No 45,293 34.05 No 2,247 41.70
Child’s sex Female 64,218 49.70 Child’s sex Female 1,768 32.80 !0.0001
Male 64,990 50.30 Male 3,620 67.20
Alcohol use Yes 1,048 0.81 Alcohol use Yes 68 1.26 0.0003
No 128,160 99.19 No 5,320 98.74
Tobacco use Yes 26,080 20.18 Tobacco use Yes 1,089 20.21 0.961
No 103,128 79.82 No 4,299 79.79
Trichomoniasis Yes 3,789 2.93 Trichomoniasis Yes 234 4.34 !0.0001
Other specific infection Yes 27,961 21.64 Other specific infection Yes 1,366 25.35 !0.0001
(chlamydia, gonorrhea, (chlamydia, gonorrhea,
candidiasis, UTI) candidiasis, UTI)
ID Z intellectual disability; SD Z standard deviation; DOE Z Department of Education; UTI Z urinary tract infection.
*There were 5,668 women for whom education information was missing. We assumed these women had at least 12 years of education.

Descriptive data for the 134,596 mother-child pairs,
available for analysis after exclusions, are shown in Table 1.
The mothers were roughly evenly split between white and
African American women, with only a small proportion of
‘other’ races. There were 5,668 women for whom education
information was missing; we initially assumed that these
women had at least 12 years of education. We also tested
the effect (in our modeling) of assuming that women with
missing education data had less than 12 years of education.
The change had no effect on the key findings, and therefore
only the first set of models is reported below.
Trichomoniasis was diagnosed in 4,023 women (3%).
Chlamydia, gonorrhea, urinary tract infections, or vulvova-
ginal candidiasis was diagnosed in 29,327 women (22%).
There were 5,388 children (4%) identified as having ID in
at least one of the three data sources. There were 102,890
children who enrolled in the public school children. Among
these children, those without ID who entered the school
system remained enrolled to a mean age of 8.9 years, and
those with ID remained to 9.1 years of age.
The most common source for diagnoses of ID was
Medicaid (4,857). There were 1,520 children identified as
having ‘‘educable mental handicap’’ (this equates to mild
ID) in the DOE records, compared to 345 children with
‘‘trainable mental handicap’’ or ‘‘profound mental
disability’’ (moderate to severe ID). There were 565 children
receiving services from DDSN for ‘‘mental retardation’’; of
these children, 160 were identified in DOE as TMH/PMH
and were also receiving DDSN services.
In our initial multivariable proportional hazards model,
controlling for maternal age and stratifying on race, educa-
tion level, alcohol use, tobacco use, and child’s sex, women
with a genitourinary infection were significantly more likely
to have a child with MR (adjusted HR Z 1.19, 95% CI:
1.13–1.26). When we re-estimated the model with the
specific infection diagnoses as independent variables, chla-
mydial and urinary tract infection violated the assumption
of proportional hazards, as did maternal age, race, education
level, and child’s sex. Trichomoniasis was significantly asso-
ciated with ID and appeared to meet proportional hazards
criteria. Therefore we stratified the model on maternal age
(!20, 20–30, O30), race and education level, child’s sex,
and an indicator variable for the diagnosis of at least one
TABLE 2. Multivariable predictors of intellectual disability,
any source*
HR 95% CI p Value
Trichomoniasis 1.28 1.12–1.46 0.0003
Alcohol use 1.30 1.02–1.66 0.035
Tobacco use 1.06 0.99–1.15 0.095
HR Z hazard ratio; CI Z confidence interval.
*Stratified on maternal age, education and race, child’s sex, and diagnosis with
another genitourinary infection. Excluding children with genetic and other high-
risk conditions.
AEP Vol. 19, No. 12 Mann et al. 895
December 2009: 891–899 TRICHOMONIASIS IN PREGNANCY AND MENTAL RETARDATION IN CHILDREN

other specific genitourinary infection besides trichomoniasis

(chlamydia, gonorrhea, urinary tract infection, or candidi-

1
asis). The results are shown in Table 2. Trichomoniasis
(HR Z 1.28, 95% CI: 1.12–1.46) remained significantly
associated with increased risk of ID. Trichomoniasis met

.995
criteria for proportional hazards, and we are unaware of

Survival
any previous studies demonstrating an association between
trichomoniasis during pregnancy and child ID. Therefore

.9 9
we chose to focus the remainder of the analyses on
describing the association between trichomoniasis and ID.
We estimated the same multivariable model predicting

.985
any ID, limiting the sample to 109,052 women who did not 0 5 10 15
have another specific genitourinary (GU) infection Age

diagnosed. We examined the effect of limiting the model to trichomonas no trichomonas

women who did not have another specific genitourinary

(GU) infection, and then to women who did have another
infection. The adjusted association between trichomoniasis
and ID was similar in children of women with (HR Z 1.24;
95% CI: 1.01–1.51) and without (HR Z 1.31; 95% CI:
1.10–1.57) another infection.
Fig. 1 displays the survival curve for a diagnosis with any
ID, by trichomoniasis status. The curve shows that the prob-
ability of not being identified as having ID was noticeably
lower for children of women with trichomoniasis than for
women without trichomoniasis, for every age.
We repeated the primary model, retaining women
regardless of whether they had another infection. We
stratified on low birth weight (!2500 g) and preterm birth
(!37 weeks) in addition to other infections, child’s sex,
maternal race, and maternal education; stratifying by
preterm birth and low birth weight only changed the point
estimate slightly (HR Z 1.23; 95% CI: 1.08–1.41).
We continued our analyses, focusing on children identi-
fied as having moderate to severe ID in the DOE data or
receiving services from DDSN. This decision was made
1
FIGURE 2. Survival function for moderate/severe ID, by Tricho-
monas status.
because these children represent a more severe level of ID,
with which children without ID are very unlikely to be
mislabeled. It also represents a particularly important
outcome, because of the high level of impairment associated
with moderate to severe ID. Fig. 2 displays the survival func-
tion for moderate to severe ID, by trichomoniasis status.
With the exception of very young children (whom health-
care providers and educators are likely hesitant to label
with a high level of intellectual impairment), the cumula-
tive probability of having been identified as having
moderate to severe ID was substantially greater for children
of women with trichomoniasis at every age.
Table 3 displays the results of the primary multivariable
model predicting moderate to severe ID (TMH or PMH)
in the public school system, or receipt of services from the
DDSN because of ‘‘mental retardation.’’ The model was
limited to children who had enrolled in public schools or
were receiving DDSN services. Maternal tobacco use and
child’s sex violated the proportional hazards assumption,
so the model was stratified on these variables. We also

stratified on the presence of one of the other four specific

.9 8

TABLE 3. Multivariable predictors of moderate to severe

intellectual disability*,y
Survival
.96

HR 95% CI p Value

Trichomoniasis 1.83 1.34–2.48 0.0001

.94

White race 0.70 0.59–0.82 !0.0001

Maternal age 1.04 1.03–1.06 !0.0001
O12 years education 0.70 0.59–0.82 !0.0001
.92

Alcohol use 1.18 0.58–2.37 0.651

0 5 10 15
age HR Z hazard ratio; CI Z confidence interval.
y
trichomonas no trichomonas Stratified on other infection, maternal tobacco use, and child’s sex; excluding chil-
dren with genetic and other high-risk conditions.
*Trainable mental handicapped or profoundly mental handicapped in public school
FIGURE 1. Survival function for any ID, by Trichomonas status. records or receipt of South Carolina Department of Disability and Special Needs
services for ‘‘mental retardation.’’
896 Mann et al. AEP Vol. 19, No. 12
TRICHOMONIASIS IN PREGNANCY AND MENTAL RETARDATION IN CHILDREN December 2009: 891–899

TABLE 4. Trichomoniasis and moderate-severe intellectual disability, sensitivity analysis

Outcome Inclusion/Exclusion Stratified on: HR 95% CI p Value

Any moderate-severe ID 1. Present in DOE or DDSN records 2.00 1.34–3.00 0.0008

2. No high-risk diagnoses in child
3. No other infection
Any moderate-severe ID 1. Present in DOE or DDSN records 1.66 1.04–2.66 0.033
2. No high-risk diagnoses in child
3. With other infection
Any moderate-severe ID 1. Present in DOE or DDSN records 1. Other infection 1.74 1.28–2.36 0.0004
2. No high-risk diagnoses in child 2. Preterm
3. Low birth weight
Any moderate-severe ID 1. Present in DOE or DDSN records 1. Other infection 1.33 1.01–1.75 0.045
2. Including children with and without 2. Preterm
high-risk diagnoses 3. Low birth weight
4. Child diagnosis of
medical exclusion
Moderate-severe ID in school records 1. Present in DOE records 1. Other infection 1.82 1.15–2.89 0.011
2. No high-risk diagnoses in child
3. Excluding children with DDSN ID only
Moderate-severe ID in school records 1. Present in DOE records 1. Other infection 1.75 1.10–2.79 0.018
2. No high-risk diagnoses in child 2. Preterm
3. Excluding children with DDSN ID only 3. Low birth weight
Receipt of DDSN services for ID 1. No high-risk diagnoses in child 1. Other infection 1.87 1.33–2.63 0.0003
2. Excluding children with TMH/PMH only
Receipt of DDSN services for ID 1. No high-risk diagnoses in child 1. Other infection 1.78 1.26–2.48 0.001
2. Excluding children with TMH/PMH only 2. Preterm
3. Low birth weight
Any moderate-severe ID 1. Present in DOE or DDSN records 1. Other infection 1.82 1.34–2.47 0.0001
2. No high-risk diagnoses in child
3. Assuming missing maternal education data
means !12 years
Any moderate-severe ID 1. Present in DOE or DDSN records 1. Other infection 1.77 1.25–2.50 0.001
2. Excluding women diagnosed with
trichomoniasis !30 days prior to delivery
HR Z hazard ratio; CI Z confidence interval; DOE Z Department of Education; DDSN Z Department of Disabilities and Special Needs; TMH Z trainable mentally hand-
icapped; PMH Z profoundly mentally handicapped.
Note: Each model excludes non-singleton births; is adjusted for maternal age, race, education, and alcohol use; and is stratified on child’s sex and maternal tobacco use.

GU infection diagnoses. Maternal age, race, education and
tobacco use were included as covariates. Trichomoniasis was
strongly associated with increased risk of moderate to severe
ID (HR Z 1.83; 95% CI: 1.34–2.48).
Table 4 displays the results of additional analyses to test
the robustness of the association between trichomoniasis
and ID. The association was similar for moderate to severe
ID identified in the school system (adjusted HR Z 1.82;
95% CI: 1.15–2.89) and for receipt of DDSN services for
ID (adjusted HR Z 1.87; 95% CI: 1.33–2.63) when
modeled separately. The increase in risk of moderate to
severe ID for children of women with trichomoniasis was
similar in those without one of the other four specific infec-
tions (HR Z 2.00; 95% CI: 1.34–3.00) as in those with
another infection (HR Z 1.66; 95% CI: 1.04–2.66).
When we included women with and without other infec-
tions and stratified by low birth weight and preterm birth in
addition to other infections, tobacco use, and child’s sex, the
association between trichomoniasis and moderate to severe
ID remained strongly significant (HR Z 1.74; 95% CI:
1.28–2.36).
We were concerned that some clinicians may test for
trichomonal infection due to complications of pregnancy
such as preterm labor, premature rupture of membranes,
or neurologic complications with the newborn noted
at delivery. We re-estimated the models predicting
moderate to severe ID, excluding women diagnosed with
trichomoniasis within 30 days of delivery. Trichomoniasis
remained significantly associated with increased risk
(adjusted HR Z 1.79; 95% CI: 1.26–2.53).
We investigated the effect of infection timing by catego-
rizing diagnoses of trichomoniasis by trimester. Second-
trimester trichomoniasis was significantly associated
with increased risk of moderate to severe ID (adjusted
HR Z 1.98; 95% CI: 1.24–3.16), while first trimester
(adjusted HR Z 1.22; 95% CI: 0.64–2.32) and third
trimester (adjusted HR Z 1.51; 95% CI: 0.94–2.44) were
not. Stratifying on low birth weight and preterm birth did
AEP Vol. 19, No. 12
December 2009: 891–899 TRICHOMONIASIS IN PREGNANCY AND MENTAL RETARDATION IN CHILDREN
not change this pattern (adjusted HR Z 1.94; 95%
CI: 1.22–3.08) for second-semester trichomoniasis, not
significant for first- and third-trimester trichomoniasis).
Additional sensitivity analyses, shown in Table 4,
demonstrate the overall robustness of the association. Not
shown in Table 4, we modeled the association of trichomo-
niasis with ID in children who were preterm or had low birth
weight, and in children who were neither preterm nor
had low birth weight. The HRs were similar for the two
groups (HR Z 1.72; 95% CI: 1.04–2.84 in preterm or
low-birth-weight children; HR Z 1.79; 95% CI: 1.21–2.64
in full-term, normal-birth-weight children).
In the inverse propensity-score–weighted models exam-
ining the association of oral metronidazole treatment with
ID, we did not find a statistically significant effect of treatment
on the risk of any ID (HR Z 1.07; 95% CI: 0.79–1.44) or
moderate-severe ID (HR Z 0.72; 95% CI: 0.35–1.46).
DISCUSSION
We believe this is the first study to report an association
between maternal trichomoniasis and ID in children. While
our initial hypothesis did not relate specifically to trichomo-
niasis, and therefore the analysis must be considered explor-
atory, the association between trichomoniasis and ID was
robust and highly statistically significant. It was particularly
strong for more severe/definitive classifications of ID. We
believe this pattern makes the results very compelling,
despite the exploratory nature of the analysis.
It is established that mild ID typically does not have
a known biological cause and is much more common in
children whose parents have low income and/or little educa-
tion; therefore mild ID is believed to include a substantial
proportion of children with poor cognitive performance
due to social deprivation, or to being at the ‘‘low end of
the bell curve’’ of intelligence (12, 13). Inclusion of these
children in the analysis would be expected to result in
non-differential outcome misclassification and a reduction
in the observed association with trichomoniasis. More
severe cases of ID, on the other hand, are frequently due
to biologic insults and are relatively evenly distributed by
social class. While it is possible for a child with an IQ at
the lower end of the mild ID range to be misclassified as
having moderate ID, it is almost inconceivable that a child
with normal intelligence would be erroneously labeled as
having moderate to severe ID. Though some children with
moderate to severe ID may be placed in a less severe category
of ID initially (to avoid adverse consequences of labeling),
these children will almost certainly need more intensive
educational and other services than children who truly
have mild ID, which should lead to their ultimate categori-
zation in the moderate-severe ID range. Therefore we
believe misclassification of the moderate-severe ID outcome
Mann et al. 897
in either direction should be rare, which permits more
accurate measurement of the effect of trichomoniasis than
the ‘‘any ID’’ outcome.
Children with TMH/PMH, by definition, have moderate
to severe ID (IQ !55). All children with confirmed ID are
eligible for DDSN services, but those who are enrolled in
these services tend to be those with more severe disabilities.
A review of IQ scores for 1,009 children under the age of
12 years, receiving DDSN services related to ID in 2008,
indicates 70% of the children had IQ scores below 55.
Both school-based services for TMH or PMH and receipt
of services from DDSN require a complete battery of
individual testing; therefore they represent rigorous case
definitions of ID. The fact that trichomoniasis was strongly
associated with such a severe level of impairment is compel-
ling, and calls for further research on the effects of maternal
trichomoniasis on child outcomes.
The most significant limitation of this study is our
reliance upon billing data to identify maternal infections.
We cannot be certain that all the women diagnosed with
trichomoniasis were diagnosed accurately. It is almost
certain that some women without diagnosed trichomoniasis
actually had Trichomonas vaginalis infection, since tricho-
monal infection is asymptomatic or subclinical in approxi-
mately 50% of infected women, and since the most
common method (wet mount) used by clinicians for diag-
nosing trichomoniasis is less than 70% sensitive for detect-
ing trichomonal infection (14). Therefore the women in our
cohort with diagnosed trichomoniasis represent only a subset
of all women with trichomoniasis. We do not know whether
the association between diagnosed trichomoniasis and
ID extends to women with undiagnosed/subclinical tricho-
monal infection.
It is possible that our findings are impacted by unmea-
sured confounders (such as generally poorer health or higher
risk lifestyles) that are associated with an increased likeli-
hood of acquiring trichomonal infection or with increased
testing for trichomoniasis by clinicians and with adverse
pregnant outcomes. Such a situation could bias the associa-
tion between trichomoniasis and ID away from the null. Our
analyses accounted for the presence of another GU infec-
tion; maternal age, race, and education level; child’s sex,
gestational age, and birth weight; and maternal alcohol
and tobacco use as reported on birth certificates. We were
unable to control for other factors that may be important,
such as maternal nutritional status and use of illegal drugs.
Furthermore, it is likely that alcohol and tobacco use were
underreported by women because of the stigma of drinking
and smoking while pregnant. Access to actual clinical
records (which we did not have) would be necessary to
better control for these potential confounders.
Our cohort was limited to women who were enrolled in
the South Carolina Medicaid program. Medicaid primarily
898 Mann et al.
TRICHOMONIASIS IN PREGNANCY AND MENTAL RETARDATION IN CHILDREN
insures low-income women, with an income eligibility limit
of 185% of poverty for pregnant women in South Carolina.
Approximately 4% of children in the cohort were diagnosed
with ID, which is higher than expected for the general
population (1% to 2%) (15), but it is not surprising for the
prevalence to be greater in a low-income population (12,
16). Although low-income children are more likely to be
diagnosed with ID, we do not believe there is reason to doubt
that the association between ID and trichomoniasis would
be generalizable to higher income children as well.
However, additional study with a wider range of socioeco-
nomic status is probably warranted.
With respect to possible mechanisms for an association
between maternal trichomoniasis and ID, it is noteworthy
that the association was independent of preterm birth and
low birth weight. We believe the most likely mechanism
is neurological insult to the developing brain due to inflam-
mation. A substantial amount of evidence has accumulated
to support the hypothesis that inflammatory cytokines may
contribute to fetal brain insults that lead to cerebral palsy
(17, 18). Though this hypothesis has not been discussed at
great length with respect to ID, it is worth noting that
children with cerebral palsy often have ID as well. It is not
unreasonable to hypothesize that inflammatory cytokines
may also play a role in the development of ID.
Trichomoniasis has been shown to produce a significant
intravaginal and systemic immune response (5, 6), and
though T. vaginalis is not generally thought to invade the
placenta, amniotic fluid, or fetus, it has been linked to
clinical pelvic inflammatory disease and histologic endome-
tritis in non-pregnant women (19–21). It is unknown
whether T. vaginalis may cause pelvic inflammatory disease
directly or facilitates ascending infection by other microbes
(e.g., by decreasing the integrity of the cervical mucus
plug) (19).
It is also possible that trichomoniasis is a marker for one
or more other infections that accompany trichomonal infec-
tion but are not typically diagnosed clinically. An example
would be Mycoplasma hominis, which has been identified as
a coinfection in more than 50% of women with trichomoni-
asis (22–24) and is known to be a potential cause of intra-
uterine infection (25–27). This possibility could not be
evaluated using the administrative data available for this
study. Additional research investigating the effects of
trichomonal infection on the intrauterine environment is
needed to illuminate possible mechanisms whereby tricho-
moniasis may impact fetal brain development.
We did not identify significant differences in the risk of
ID based on treatment for trichomoniasis. This analysis
utilized outpatient pharmacy billing data to identify women
who were treated within 14 days of a diagnosis of trichomo-
niasis. It is possible that some ‘‘untreated’’ women may have
received metronidazole purchased out of pocket or by
AEP Vol. 19, No. 12
December 2009: 891–899
another payer. Additional research using more inclusive
data sources is needed to evaluate the effects of treatment
with oral metronidazole on the risk of ID. A recent random-
ized controlled trial found that women treated with oral
metronidazole for asymptomatic trichomoniasis had
increased risk of preterm delivery (28). To our knowledge,
the impact of treatment on long-term child outcomes has
not been investigated previously.
Even if metronidazole treatment during pregnancy does
not reduce the risk of ID, this study has important public
health implications. Because trichomoniasis has tradition-
ally been considered a benign, self limited condition, public
health efforts to combat the infection have generally been
lacking. More recently, trichomoniasis has emerged as
a risk factor for HIV transmission, particularly in
low-income nations where HIV is highly prevalent (29).
Despite this development, there continues to be less
emphasis on trichomoniasis prevention than on other sexu-
ally transmitted infections such as syphilis, gonorrhea, and
chlamydia (15). Behavioral interventions (promotion of
abstinence, monogamy, and/or condom use) and routine
screening of the reproductive age population have the
potential to reduce the prevalence of trichomoniasis in
pregnant women. If future studies confirm that trichomoni-
asis is in fact a risk factor for adverse child outcomes, such
public health efforts may be warranted.
We would like to thank Pete Bailey and Heather Kirby in
the South Carolina Office of Research and Statistics for
their work in preparing, cleaning and providing the data
used in this study. We would also like to thank Dr. W. David
Hager for his consultation regarding clinical and mecha-
nistic aspects of the paper. This study was funded by Health
Resources and Services Administration (HRSA) in the U.S.
Department of Health and Human Services, HRSA Grant
No. R40MC06636–01–00 ‘‘Maternal Sexual Infections
and Adverse Child Outcomes’’.
The authors have no conflicts of interest related to this research.
REFERENCES
1. Sobel JD. What’s new in bacterial vaginosis and trichomoniasis? Infect Dis
Clin North Am. 2005;19:387–406.
2. Weinstock H, Berman S, Cates W Jr. Sexually transmitted diseases among
American youth: incidence and prevalence estimates, 2000. Perspect Sex
Reprod Health. 2004;36:6–10.
3. Sutton M, Sternberg M, Koumans EH, McQuillan G, Berman S, Markowitz L.
The prevalence of Trichomonas vaginalis infection among reproductive-age
women in the United States, 2001–2004. Clin Infect Dis. 2007;45:
1319–1326.
4. Cotch MF, Pastorek JG, Nugent RP, Hillier SL, Gibbs RS, Martin DH,
et al. Trichomonas vaginalis associated with low birth weight and preterm
delivery. Sex Transm Dis. 1997;24:353–360.
5. Simhan HN, Anderson BL, Krohn MA, Heine RP, Martinez de Tejada B,
Landers DV, et al. Host immune consequences of asymptomatic
AEP Vol. 19, No. 12
December 2009: 891–899 TRICHOMONIASIS IN PREGNANCY AND MENTAL RETARDATION IN CHILDREN
Trichomonas vaginalis infection in pregnancy. Am J Obstet Gynecol.
2007;196:59. e1–5.
6. Anderson BL, Cosentino LA, Simhan HN, Hillier SL. Systemic immune
response to Trichomonas vaginalis infection during pregnancy. Sex
Transm Dis. 2007;34:392–396.
7. Goldenberg RL, Culhane JF, Johnson DC. Maternal infection and adverse
fetal and neonatal outcomes. Clin Perinatol. 2005;32:523–529.
8. McDermott S, Callaghan W, Szwejbka L, Mann H, Daguise V. Urinary
tract infections during pregnancy and mental retardation and develop-
mental delay. Obstet Gynecol. 2000;96:113–119.
9. Johnson S. Cognitive and behavioural outcomes following very preterm
birth. Semin Fetal Neonatal Med. 2007;12:363–373.
10. Hermansen MC, Hermansen MG. Perinatal infections and cerebral palsy.
Clin Perinatol. 2006;33:315–333.
11. Centers for Disease Control and Prevention. Sexually transmitted diseases
treatment guidelines, 2006. MMWR Morb Mortal Wkly Rep. 2006;55(RR-
11):1–93.
12. Durkin MS, Schupf N, Stein ZA, Susser MW. Childhood Cognitive Disability.
In: Wallace RR, Kohatsu N, eds. Maxcy-Rosenau-Last Public Health &
Preventive Medicine. 15th ed. New York: McGraw-Hill; 2008:1173–1183.
13. Stromme P, Magnus P. Correlations between socioeconomic status, IQ and
aetiology in mental retardation: a population-based study of Norwegian
children. Soc Psychiatry Psychiatr Epidemiol. 2000;35:12–18.
14. Hobbs MM, Sena AC, Swygard H, Schwebke JR. Trichomonas vaginalis
and trichomoniasis. In: Holmes KK, Sparling PF, Stamm WE, Piot P,
Wasserheit JN, Corey L, Cohen MS, Watts DH, eds. Sexually trans-
mitted diseases. 4th ed. New York: McGraw-Hill; 2008:771–793.
15. McDermott S, Durkin MS, Schupf N, Stein Z. Epidemiology and etiology
of mental retardation. In: Jacobson JW, Mulick JA, Rojahn J, eds. Hand-
book of intellectual and development disabilities. New York: Springer
Press; 2007:3–40.
16. Andrews H, Goldberg D, Wellen N, Pittman B, Struening E. Prediction of
special education placement from birth certificate data. Am J Prev Med.
1995;11(3 Suppl):55–61.
17. Girard S, Kadhim H, Roy M, Lavoie K, Brochu M, Larouche A, et al. Role of
perinatal inflammation in cerebral palsy. Pediatr Neurol. 2009;40:168–174.
18. Arpino C, D’Argenzio L, Ticconi C, Di Paolo A, Stellin V, Lopez L, et al.
Brain damage in preterm infants: etiological pathways. Ann Ist Super San-
ita. 2005;41:229–237.
Mann et al. 899
19. Moodley P, Wilkinson D, Connolly C, Moodley J, Sturm AW. Trichomonas
vaginalis is associated with pelvic inflammatory disease in women infected
with human inmmunodeficiency virus. Clin Infect Dis. 2002;34:519–522.
20. Paisarntantiwong R, Brockmann S, Clarke L, Landesman S, Feldman J,
Minkoff H. The relationship of vaginal trichomoniasis and pelvic inflam-
matory disease among women colonized with Chlamydia trachomatis. Sex
Transm Dis. 1995;22:344–347.
21. Cherpes TL, Wiesenfeld HC, Melan MA, Kant JA, Cosentino LA, Meyn LA,
et al. The associations between pelvic inflammatory disease, Trichomonas
vaginalis infection, and positive herpes simplex virus type 2 serology. Sex
Transm Dis. 2006;33:747–752.
22. Dessi D, Rappelli P, Diaz N, Cappuccinelli P, Fiori PL. Mycoplasma hom-
inis and Trichomonas vaginalis: a unique case of symbiotic relationship
between two obligate human parasites. Front Biosci. 2006;11:2028–2034.
23. Van Belkum A, van der Schee C, van der Meijden WI, Verbrugh HA,
Sluiters HJ. A clinical study on the association of Trichomonas vaginalis
and Mycoplasma hominis infections in women attending a sexually trans-
mitted disease outpatient clinic. FEMS Immunol Med Microbiol.
2001;32:27–32.
24. Vancini RG, Benchimol M. Entry and intracellular location of Mycoplasma
hominis in Trichomonas vaginalis. Arch Microbiol. 2008;189:7–18.
25. Egawa T, Morioka I, Morisawa T, Yokoyama N, Nakao H, Ohashi M, et al.
Ureaplasma urealyticum and Mycoplasma hominis presence in umbilical
cord is associated with pathogenesis of funisitis. Kobe J Med Sci.
2007;53:241–249.
26. Goldenberg RL, Andrews WW, Goepfert AR, Faye-Petersen O, Cliver SP,
Carlo WA, et al. The Alabama Preterm Birth Study: umbilical cord blood
Ureaplasma urealyticum and Mycoplasma hominis cultures in very preterm
newborn infants. Am J Obstet Gynecol. 2008;198:43. e1–e5.
27. Goldenberg RL, Andrews WW, Faye-Petersen OM, Goepfert AR, Cliver SP,
Hauth JC. The Alabama Preterm Birth Study: intrauterine infection and
placental histologic findings in preterm births of males and females less
than 32 weeks. Am J Obstet Gynecol. 2006;195:1533–1537.
28. Klebanoff MA, Carey JC, Hauth JC, Hillier SL, Nugent RP, Thom EA,
et al. Failure of metronidazole to prevent preterm delivery among pregnant
women with asymptomatic Trichomonas vaginalis infection. N Engl J
Med. 2001;345:487–493.
29. Shafir SC, Sorvillo FJ, Smith L. Current issues and considerations
regarding trichomoniasis and human immunodeficiency virus in African
Americans. Clin Microbiol Rev. 2009;22:37–45.