Content uploaded by Emily Harville
Author content
All content in this area was uploaded by Emily Harville on Sep 18, 2018
Content may be subject to copyright.
JOURNAL OF WOMEN’S HEALTH
Volume 14, Number 7, 2005
© Mary Ann Liebert, Inc.
Perceived Life Stress and Bacterial Vaginosis
EMILY W. HARVILLE, M.S.P.H.,
1
MAUREEN C. HATCH, Ph.D.,
2,3
and JUN ZHANG, M.D., Ph.D.
2,4
ABSTRACT
Background: Bacterial vaginosis (BV) is a common vaginal condition produced by overgrowth
of anaerobic bacteria. Consequences of the condition may include preterm birth and pelvic
inflammatory disease (PID). Because stress can suppress immune function, increased stress
might increase the risk of BV. Our objective was to determine whether life stress was asso-
ciated with risk of bacterial vaginosis in a cohort of nonpregnant women.
Methods: A total of 411 African American women receiving routine gynecological care were
recruited from two New York City hospitals. They were asked to rate the pressure they felt
over the last week as a result of change, relationships, sickness, and finances using the Global
Assessment of Recent Stress scale. An overall measure of stress was created by summing the
responses over the categories. Stress was categorized into low, intermediate, and high tertiles.
BV was diagnosed by gram stain score.
Results: In almost all domains of life stress, women with high stress were more likely to
have BV than those with low stress; however, none of the differences reached statistical sig-
nificance. Thirty-four percent of women with BV had high overall stress as opposed to 26%
of women without BV, giving an adjusted relative risk (RR) of 1.4 (95% confidence interval,
0.95, 2.1).
Conclusions: In a cohort of African American women in New York City, perceived life stress
showed no clear association with BV. Because of the prevalence of both the exposure and the
disease, further study is warranted.
627
INTRODUCTION
B
ACTERIAL VAGINOSIS
(BV)
IS A COMMON
condi-
tion among women of reproductive age.
1
It is
caused by an overgrowth of naturally occurring
anaerobic bacteria, particularly Mobiluncus spp.
and Gardnerella vaginalis, and a corresponding re-
duction in Lactobacillus spp., normally present in
high numbers. BV is very common, usually de-
tected in 10%–40% of women worldwide. How-
ever, a majority of women with the bacterial
profile are asymptomatic.
2,3
Prevalence of BV is
reported to be higher among women who have
multiple sexual partners, who have low incomes
1
Department of Epidemiology, University of North Carolina Chapel Hill, Chapel Hill, North Carolina.
2
Department of Community and Preventive Medicine, Mount Sinai School of Medicine, New York, New York.
3
Present address: Division of Cancer Epidemiology and Genetics, National Cancer Institute, National Institutes of
Health, DHHS, Rockville, Maryland.
4
Present address: Epidemiology Branch, National Institute of Child Health and Human Development, National In-
stitutes of Health, DHHS, Rockville, Maryland.
This study was funded by the National Institute of Allergy and Infectious Disease, National Institutes of Health
(grant R01AI41036). E.W.H. is a Howard Hughes Predoctoral Fellow.
5791_08_p627-633 9/7/05 8:55 AM Page 627
or lower levels of education, who smoke, and
who do not use hormonal contraception.
2–4
As-
sociations with douching have been studied,
but the evidence for causality is mixed.
5
BV has
been associated with serious reproductive conse-
quences: women with BV are 1.4–3 times more
likely to deliver preterm
1,3,6
and approximately 3
times more likely to develop pelvic inflammatory
disease (PID),
7,8
although not every study has
confirmed this.
9
Psychological stress can downregulate im-
mune responses by disrupting neuroendocrino-
logical pathways, and many immune cells re-
spond to signals from stress hormones.
10,11
Stress
has also been associated with decreased defenses
in the mucosal lining of the gastrointestinal
tract.
10
Alterations in the vaginal mucosal im-
mune system have been linked to BV in some
studies.
12
We hypothesize that higher levels of
stress may be associated with a greater preva-
lence of BV.
Research directly linking stress and BV is very
limited. Culhane et al.
13,14
assessed stress in a co-
hort of innercity pregnant women using the Co-
hen Perceived Stress Scale, as well as neighbor-
hood-level variables. They found that higher
perceived stress was associated with an elevated
risk of BV (odds ratio [OR] 1.3, 1.0-1.6); the mean
score on the stress scale was 24.6 in women with
BV compared with 22.2 in women without BV.
Ruiz et al.
15
studied a group of pregnant women
in central Texas, also measuring stress with the
Cohen Perceived Stress Scale, and found no as-
sociation with either BV or chlamydia. No previ-
ous study has attempted to measure the associa-
tion between BV and stress in nonpregnant
women. We examined this association in an in-
nercity African American population.
MATERIALS AND METHODS
African American women seeking outpatient
gynecological care and family planning at Mount
Sinai Medical Center and North General Hospi-
tal in New York City from 1999 to 2001 were re-
cruited for a study of bacterial vaginosis and
douching. All women scheduled for appoint-
ments at the two recruitment sites were ap-
proached in the waiting rooms of the respective
clinics by a study interviewer. Prospective sub-
jects were asked to answer a brief screening form
to determine eligibility. All eligible women were
then invited to participate in the study. Women
were eligible if they were between 18 and 45 years
old, had not had a hysterectomy or bilateral
oophorectomy, were not menopausal, pregnant,
or immediately postpartum, and did not have
certain chronic conditions (diabetes, HIV, auto-
immune disorders). We excluded women with
chronic vaginitis, defined as having been treated
with antibiotics for the same type of vaginitis
within the previous 3 months, as well as women
who had used antibiotics or other medicines with
immunosuppressive effects in the last 3 months.
The protocol entailed a brief (10–20 minute) in-
terview administered immediately after the sub-
ject’s agreement to participate and the collection
of additional vaginal swabs during the subject’s
physical examination.
Five hundred eighty-five women were ap-
proached to be screened to determine eligibility.
Sixteen (2.8%) refused to be screened. Of those
who were screened, 9 women (1.6%) refused to
participate, and 134 (23.5%) proved to be ineligi-
ble. The most common reasons for ineligibility
were having diabetes and being HIV positive (21
women for each), as well as being pregnant (15).
Four hundred eleven (96.9%) of the 426 women
who were eligible agreed to participate. Of those,
400 who had complete data on stress and BV com-
prise the sample for analysis.
The structured interview included questions on
sociodemographic characteristics, lifestyle, femi-
nine hygiene, sexual behavior, reproductive his-
tory, and employment. Stress was measured by
the Global Assessment of Recent Stress (GARS)
scale, an instrument designed to evaluate per-
ceived short-term stress, defined as “a feeling of
pressure.”
16
To help the individual assess overall
feelings of stress, participants were first asked to
evaluate stress during the past week in seven dif-
ferent areas of life (work/job/school, interper-
sonal relationships, changes in relationships, sick-
ness or injury, financial, unusual happenings, and
change or lack of change in daily routine). Each
of the eight items in the scale is scored from 0
(none) to 9 (extreme), and examples are given to
help guide the respondents in selecting the
amount of perceived stress or “pressure.” The
GARS scale has been found to be reproducible
(test-retest correlation between r0.69 and r
0.92) and to correlate well with clinical judgments
and with scores from the Holmes and Rahe mod-
HARVILLE ET AL.
628
5791_08_p627-633 9/7/05 8:55 AM Page 628
ified Social Readjustment Rating Scale,
16
which
measures specific life events rather than global re-
sponses. The scale also is significantly associated
with such psychological markers as depression and
anxiety.
16
Significant, although moderate, correla-
tions have also been shown with physiological pa-
rameters, including blood pressure, response to
cold pressor, and changes in immune function.
16
Gynecologists were asked to collect extra vagi-
nal swabs during the pelvic examination. A swab
was placed in a test tube and sent to the Clinical
Microbiology Laboratories at Mount Sinai Med-
ical Center for gram stain analysis, using the sys-
tem of Nugent et al.
17
Briefly, each gram-stained
smear was evaluated for morphotypes: Lacto-
bacillus, G. vaginalis, Bacteroides, Mobiluncus. Each
morphotype except Mobiluncus (score 0–2) was
quantitated from 1 to 4, with Lactobacillus
scored inversely to their quantity. The scoring cri-
terion is a weighted sum of these scores, with BV
defined as a gram stain score 7.
This project was approved by the Institutional
Review Boards at the Mount Sinai School of Med-
icine and North General Hospital.
To avoid small cell counts and to provide a
more convenient summary of the data, a three-
level stress measure (low 0–3, intermediate 4–6,
high 7–9) was created from each stress domain.
A composite measure was created by summing
the categorized responses on all the stress do-
mains. Bivariate analyses were performed, using
stratified analysis to determine the relative risk
of BV for each category of stress, with the lowest
level as a referent. Risk factors for BV were in-
vestigated by examining associations with de-
mographic, sexual, and hygiene factors: age, par-
ity, income, education, recent unprotected sex
(defined as vaginal intercourse without a condom
within the last 24 hours), a new sex partner within
the last 3 months, number of sex partners, mari-
tal status, body mass index (BMI), smoking, use
of hormonal contraception, and frequency of
douching. Chi-square tests were used to deter-
mine statistical significance. Covariates were
modeled in the form listed in Tables 1 and 2.
Potential confounders were assessed for their
association with both exposure and outcome, us-
ing the criteria of relative risk (RR) 2 or p0.15
for association. Variables screened were age, par-
ity, recent new sex partner, number of sex part-
ners, marital status, poverty, education, and use
of hormonal contraception. Douching frequency
was only weakly associated with BV in these data
and so was not considered.
18
Only hormonal con-
traception fit these criteria, so it was included in
the regression model. Each stress variable was
modeled separately. Binomial log-linear regres-
sion was chosen over logistic modeling because
the outcome is common, making RR a better ef-
fect measure than an OR.
RESULTS
The study population comprised African Amer-
ican women of whom 73% were never married
and 82% were parous (Table 1). All except 5 wo-
STRESS AND BACTERIAL VAGINOSIS 629
T
ABLE
1. C
HARACTERISTICS OF
411 N
EW
Y
ORK
C
ITY
A
FRICAN
A
MERICAN
W
OMEN
Characteristic n%
Age, years
a
18–21 28 7
21–25 81 20
25–30 100 25
30–35 78 19
35–40 55 14
40 59 15
Education
b
High school 272 67
High school 137 34
Household income (month)
a
$500 105 26
$500–800 122 30
$800 174 43
Marital status
b
Unmarried 297 73
Married 57 14
Separated, widowed, or divorced 55 13
Smoked in the past 3 months
Yes 137 33
No 274 67
Body mass index (kg/m
2
)
c
20 30 7
20–25 122 30
25–30 103 25
30 152 37
Parity
d
07418
1 149 37
2 185 45
Douching frequency in the last 3 months
e
Never 216 53
Once a month or less 132 32
More than once a month 61 15
a
Missing data on 10 women.
b
Missing data on 2 women.
c
Missing data on 4 women.
d
Missing data on 3 women.
e
Missing data on 2 women.
5791_08_p627-633 9/7/05 8:55 AM Page 629
men were sexually active at some point, and 87%
were sexually active in the last year. One hundred
eight (27%) had BV, defined as a gram score 7.
One hundred nineteen (30%) had a gram score of
4–6 (intermediate), and the remaining 173 had a
gram score 4.
Several social, hygiene, and sexual factors
were assessed in bivariate analyses for their as-
sociation with risk for BV (Table 2). Use of hor-
monal contraception was inversely related to
risk of BV (21% vs. 31%, p0.03). Being mar-
ried and of higher income were modestly asso-
ciated with a reduced risk of BV, as was having
recent, unprotected sex, possibly explained by
an association with marital status. Women who
smoked were more likely to report high overall
stress (41% vs. 33% at highest level, p0.03),
as were women who used hormonal contracep-
tion (39% vs. 33%, p0.11). Women with more
children also reported more stress (p0.12), as
did women with more sex partners (52% of wo-
men with two or more partners reported over-
all high stress vs. 38% of those with no partner,
p0.20). Income was not significantly associ-
ated with overall stress, although there was a
trend of increased reported financial pressure
with decreased income.
Approximately a quarter of the women re-
ported high levels of stress on any one of the
domains covered by the GARS scale, with fi-
nancial pressures cited most frequently (31% in
the upper tertile) (Table 3). Pressure due to rou-
tine (17%) and sickness (16%) were least often
reported. Thirty-six percent of the women re-
ported high overall stress.
The overall stress score was not associated with
BV (Table 3). Increased reported pressure was as-
sociated with increased rates of BV in the fol-
lowing domains: pressure from routine, change
in relationships, financial pressure, unusual hap-
penings. The summed score of all domains was
also associated with BV. The prevalence ratios
were 1.26 (95% confidence interval [CI] 0.84, 1.88)
and 1.44 (95% CI 0.98, 2.11) for the moderate and
high scores, respectively, compared with that for
the low score. These findings changed very little
after controlling for hormonal contraceptive use,
the one variable identified as a confounding fac-
tor. Results using the original continuous mea-
sure of stress and the gram score as a continuous
variable were similar to the results from the cat-
egorical analysis (data not shown).
DISCUSSION
In this study of African American women in
New York, we did not find any clear association
between stress and BV. For some domains (pres-
sure from routine, changes in relationships, and un-
usual happenings), as well as the sum of all the sub-
scores, the prevalence of BV increased with higher
stress level. In spite of the fact that pressure in sev-
eral domains showed an association with BV and
the sum of the domain subscores also showed an
association, the subjects’ assessment of their over-
all level of pressure was unrelated to the risk of
bacterial vaginosis. This may indicate that respon-
dents had an easier time rating the level of stress
in specific areas of life than the level of pressure
they were feeling overall. Indeed, the overall esti-
mate of pressure was only moderately correlated
with the sum of the domain subscores.
Women in the highest third of the summed
stress score had a 45% higher prevalence of BV
compared with those in the lowest tertile. This is
lower than in one previous study, which reported
an OR of 2.2 (1.1, 4.2) for the most highly stressed
women,
14
although similar to the OR of 1.3 (1.0,
1.6) in a study comparing those above vs. below
the median in perceived stress.
13
Because a high
percentage of African American women are at an
elevated stress level and BV is such a common
disorder, even a modest association could have
significant public health implications.
In this population, BV was inversely associated
with using hormonal contraception, similar to
what has been found in several previous stud-
ies.
2,19,20
Also consistent with other studies is the
inverse association with income,
1,2
although this
was not statistically significant. We did not see
any associations with other suspected risk factors,
such as age, education, or smoking, although the
trends with marital status were in the predicted
direction. This may be in part because we con-
ducted the study in a gynecological clinic instead
of a sexually transmitted disease (STD) or prena-
tal care clinic and limited the study to African
Americans. A strength of the study is the focus
on the population at risk for BV, unselected for
pregnancy or risk of STDs. Another strength is
the excellent participation rate. We used a vali-
dated stress instrument, the GARS, that has been
associated with alterations in immune parame-
ters,
21
including reduced immune response to de-
layed hypersensitivity skin tests.
16
HARVILLE ET AL.
630
5791_08_p627-633 9/7/05 8:55 AM Page 630
STRESS AND BACTERIAL VAGINOSIS 631
T
ABLE
2. A
SSOCIATIONS BETWEEN
B
ACTERIAL
V
AGINOSIS AND
C
OVARIATES
IN
411 A
FRICAN
A
MERICAN
W
OMEN IN
N
EW
Y
ORK
C
ITY
Bacterial vaginosis by gram score
a
Yes No
Characteristic n%n%p
Age, years
b
18–21 4 3.7 24 8.5 0.65
21–25 21 19.6 57 20.1
25–30 28 26.2 66 23.3
30–35 24 22.4 53 18.7
35–40 14 13.1 40 14.1
40 16 15.0 43 15.2
Education
c
High school 67 62.6 196 67.4 0.38
High school 40 37.4 95 32.7
Household income per month
b
$500 27 25.5 74 26.0 0.17
$500–800 40 37.7 81 28.4
$800 39 36.8 130 45.6
Marital status
c
Never married 83 77.6 204 70.1 0.23
Married 10 9.4 46 15.8
Separated, widowed, or 14 13.1 41 14.1
divorced
Smoked within the last 3 months
Yes 37 34.3 97 33.2 0.84
No 71 65.7 195 66.8
Body mass index (kg/m
2
)
d
20 10 9.4 20 6.9 0.81
20–25 29 27.4 88 30.3
25–30 26 24.5 74 25.5
30 41 38.7 108 37.2
Parity
e
0 19 17.9 52 17.9 0.58
1 35 32.7 110 37.9
2 53 49.5 128 44.1
New sex partner in last 30 days
c
Yes 6 5.6 8 2.8 0.22
No 101 94.4 283 97.3
Number of sex partners in last 3 months
e
2 6 5.6 11 3.8 0.72
1 79 73.8 216 74.5
0 22 20.6 63 21.7
Recent unprotected sexual intercourse
c
Yes 11 10.2 45 15.5 0.17
No 97 89.8 245 84.5
Used hormonal contraception in last 3 months
f
Yes 37 34.6 136 46.6 0.03
No 70 65.4 156 53.4
Douching frequency in the last 3 months
c
Never 60 55.6 152 52.4 0.83
Once a month or less 34 31.5 95 32.8
More than once a month 14 13.0 43 14.8
a
Missing data on 11 women.
b
Missing data on 10 women.
c
Missing data on 2 women.
d
Missing data on 4 women.
e
Missing data on 3 women.
f
Missing data on 1 women.
5791_08_p627-633 9/7/05 8:55 AM Page 631
On the other hand, stress is a difficult concept to
measure, and no single scale can capture every-
thing. There was no measure of specific life events,
coping styles, social support, or biomarkers of
stress in this study. It is possible that a more de-
tailed stress measurement would reveal stronger
associations or would provide a more definitive
finding of no association. Our sample size was
somewhat limited; we had approximately 80%
power to detect an RR of 2, and the observed as-
sociations were more modest. Only one variable
was found to have a potential confounding effect,
and adjusting made no material difference to the
results. However, there is the potential for unmea-
sured confounding, perhaps by health habits or
day of menstrual cycle. The sample is one of con-
venience, and results cannot be generalized to wo-
men who do not use gynecological services.
Especially valuable in further research would
be a prospective cohort study of women who are
BV free at the start. In view of the modest risks
observed, larger samples will be needed to en-
sure adequate statistical power. As a cross-sec-
tional study, this research cannot distinguish be-
tween factors that predict incidence of BV and
factors that predict duration or between cause
HARVILLE ET AL.
632
T
ABLE
3. A
SSOCIATIONS BETWEEN
R
EPORTED
P
RESSURE AND
B
ACTERIAL
V
AGINOSIS
IN A
G
ROUP OF
411 A
FRICAN
A
MERICAN
W
OMEN IN
N
EW
Y
ORK
C
ITY
Bacterial vaginosis by gram score
a
Yes No Unadjusted Adjusted
b
Domain n%n%RRCIRRCI
Pressure from routine
0–3 49 45.4 150 51.4 1 1
3–6 37 34.3 96 32.9 1.13 (0.78, 1.63) 1.07 (0.74, 1.55)
7–9 22 20.4 46 15.8 1.31 (0.86, 2.00) 1.24 (0.81, 1.89)
Pressure from change in relationships
0–3 43 39.8 138 47.3 1 1
3–6 34 31.5 83 28.4 1.22 (0.83, 1.80) 1.19 (0.80, 1.75)
7–9 31 28.7 71 24.3 1.28 (0.86, 1.90) 1.28 (0.87, 1.90)
Financial pressure
0–3 38 35.2 102 34.9 1 1
3–6 32 29.6 107 36.6 0.85 (0.56, 1.27) 0.83 (0.55, 1.25)
7–9 38 35.2 83 28.4 1.16 (0.79, 1.69) 1.15 (0.78, 1.67)
Pressure from unusual happenings
0–3 52 48.2 162 55.5 1 1
3–6 33 30.6 86 29.5 1.14 (0.79, 1.66) 1.09 (0.75, 1.60)
7–9 23 21.3 44 15.1 1.41 (0.94, 2.12) 1.38 (0.92, 2.07)
Relationship pressure
0–3 37 34.3 115 39.4 1 1
3–6 43 39.8 100 34.3 1.24 (0.85, 1.80) 1.24 (0.85, 1.81)
7–9 28 25.9 77 26.4 1.10 (0.72, 1.67) 1.11 (0.73, 1.70)
Sickness/injury
0–3 58 53.7 168 57.5 1 1
3–6 32 29.6 76 26.0 1.15 (0.80, 1.66) 1.09 (0.75, 1.58)
7–9 18 16.7 48 16.4 1.06 (0.68, 1.67) 0.99 (0.63, 1.56)
Pressure from work
0–3 40 37.0 117 40.1 1 1
3–6 47 43.5 113 38.7 1.15 (0.80, 1.65) 1.14 (0.80, 1.64)
7–9 21 19.4 62 21.2 0.99 (0.63, 1.57) 1.00 (0.63, 1.58)
Overall estimate of pressure
0–3 34 31.5 83 28.4 1 1
3–6 37 34.3 106 36.3 0.89 (0.60, 1.32) 0.85 (0.57, 1.26)
7–9 37 34.3 103 35.3 0.91 (0.61, 1.35) 0.93 (0.62, 1.37)
Sum of reported stress
7–11 39 36.1 135 45.9 1 1
11–14 32 29.6 81 27.7 1.26 (0.84, 1.88) 1.30 (0.87, 1.95)
1437 34.3 77 26.4 1.44 (0.98, 2.11) 1.39 (0.95, 2.05)
a
Missing data on 10 women for bacterial vaginosis and on 1 woman for stress measurements.
b
Adjusted for use of hormonal contraception in the last 3 months.
5791_08_p627-633 9/7/05 8:55 AM Page 632
and effect. It may be argued that women with BV
might be more inclined to report stress due to the
BV symptoms, but in our data, the women with
and without BV reported a similar stress level due
to “sickness/injury.”
African American women are at higher risk of
preterm birth, have a higher prevalence of BV, and
are under greater stress in many ways than women
of other racial/ethnic groups.
22
Stress has been as-
sociated with preterm birth in some studies, al-
though the association is not found consistently.
23
A connection between stress and BV is plausible
biologically and would provide an important link
between the social and medical epidemiology of
preterm birth.
24
Further research, using a larger
sample and more extensive measurements of
stress, should attempt to pursue this line of inquiry.
ACKNOWLEDGMENTS
We thank Dr. Tonja Nansel for her helpful com-
ments on the manuscript.
REFERENCES
1. Nelson DB, Macones G. Bacterial vaginosis in preg-
nancy: Current findings and future directions. Epi-
demiol Rev 2002;24:102.
2. Klebanoff MA, Schwebke JR, Zhang J, et al. Vulvo-
vaginal symptoms in women with bacterial vaginosis.
Obstet Gynecol 2004;104:267.
3. French JL, McGregor JA. Bacterial vaginosis: History,
epidemiology, microbiology, sequelae, diagnosis, and
treatment. In: Borchadt KA, Noble MA, eds. Sexually
transmitted diseases: Epidemiology, pathology, diag-
nosis, and treatment. Boca Raton, FL: CRC Press, 1997:3.
4. Schwebke JR, Desmond RA, Oh MK. Predictors of
bacterial vaginosis in adolescent women who douche.
Sex Transm Dis 2004;31:433.
5. Martino JL, Vermund SH. Vaginal douching: Evi-
dence for risks or benefits to women’s health. Epi-
demiol Rev 2002;24:109.
6. Leitich H, Bodner-Adler B, Brunbauer M, et al. Bac-
terial vaginosis as a risk factor for preterm delivery:
A meta-analysis. Am J Obstet Gynecol 2003;189:139.
7. Hong S, Xin C, Qianhong Y, et al. Pelvic inflamma-
tory disease in the People’s Republic of China: Aeti-
ology and management. Int J STD AIDS 2002;13:568.
8. Peipert JF, Montagno AB, Cooper AS, Sung CJ. Bac-
terial vaginosis as a risk factor for upper genital tract
infection. Am J Obstet Gynecol 1997;177:1184.
9. Ness RB, Hillier SL, Kip KE, et al. Bacterial vaginosis
and risk of pelvic inflammatory disease. Obstet Gy-
necol 2004;104:761.
10. Yang EV, Glaser R. Stress-induced immunomodula-
tion and the implications for health. Int Immuno-
pharmacol 2002;2:315.
11. Lederman RP. Relationship of anxiety, stress, and
psychosocial development to reproductive health. Be-
hav Med 1995;21:101.
12. Cauci S, Monte R, Driussi S, Lanzafame P, Quadri-
foglio F. Impairment of the mucosal immune system:
IgA and IgM cleavage detected in vaginal washings
of a subgroup of patients with bacterial vaginosis. J
Infect Dis 1998;178:1698.
13. Culhane JF, Rauh V, McCollum KF, Elo IT, Hogan V.
Exposure to chronic stress and ethnic differences in
rates of bacterial vaginosis among pregnant women.
Am J Obstet Gynecol 2002;187:1272.
14. Culhane JF, Rauh V, McCollum KF, et al. Maternal
stress is associated with bacterial vaginosis in human
pregnancy. Matern Child Health J 2001;5:127.
15. Ruiz RJ, Fullerton J, Brown CE, Schoolfield J. Rela-
tionships of cortisol, perceived stress, genitourinary
infections, and fetal fibronectin to gestational age at
birth. Biol Res Nurs 2001;3:39.
16. Linn MW. A Global Assessment of Recent Stress
(GARS) scale. Int J Psychiatry Med 1985;15:47.
17. Nugent RP, Krohn MA, Hillier SL. Reliability of di-
agnosing bacterial vaginosis is improved by a stan-
dardized method of gram stain interpretation. J Clin
Microbiol 1991;29:297.
18. Zhang J, Hatch MC, Zhang D, et al. Frequency of
douching and risk of bacterial vaginosis in African-
American women. Obstet Gynecol 2004;104:756.
19. Holzman C, Leventhal JM, Qui H, Jones NM, Wang
J. Factors linked to bacterial vaginosis in nonpregnant
women. Am J Public Health 2001;91:1664.
20. Schwebke JR, Richey CM, Weiss HL. Correlation of
behaviors with microbiological changes in vaginal
flora. J Infect Dis 1999;180:1632.
21. Paik IH, Toh KY, Lee C, Kim JJ, Lee SJ. Psychological
stress may induce increased humoral and decreased
cellular immunity. Behav Med 2000;26:139.
22. Clark VR. The perilous effects of racism on blacks.
Ethn Dis 2001;11:769.
23. Hoffman S, Hatch MC. Stress, social support and
pregnancy outcome: A reassessment based on recent
research. Paediatr Perinat Epidemiol 1996;10:380.
24. Wadhwa PD, Culhane JF, Rauh V, et al. Stress, infec-
tion and preterm birth: A biobehavioural perspective.
Paediatr Perinat Epidemiol 2001;15(Suppl 2):17.
Address reprint requests to:
Emily W. Harville, M.S.P.H.
Department of Epidemiology
University of North Carolina at Chapel Hill
CB 7435
Chapel Hill, NC 27599-7435
E-mail: ewh@unc.edu
STRESS AND BACTERIAL VAGINOSIS 633
5791_08_p627-633 9/7/05 8:55 AM Page 633