See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/233931571

Recurrence of Bacterial Vaginosis Is Significantly Associated With

Posttreatment Sexual Activities and Hormonal Contraceptive Use

Article  in  Clinical Infectious Diseases · December 2012

DOI: 10.1093/cid/cis1030 · Source: PubMed

CITATIONS READS

57 84

9 authors, including:

Catriona S Bradshaw Lenka A Vodstrcil

Monash University (Australia) University of Melbourne
319 PUBLICATIONS   4,606 CITATIONS    95 PUBLICATIONS   846 CITATIONS   

SEE PROFILE SEE PROFILE

Jane Hocking Suzanne M Garland

University of Melbourne Royal Women's Hospital in Victoria
424 PUBLICATIONS   7,130 CITATIONS    854 PUBLICATIONS   22,078 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Vaccination for HPV among MSM View project

VACCINE study View project

All content following this page was uploaded by Catriona S Bradshaw on 26 October 2016.

The user has requested enhancement of the downloaded file.

 MAJOR ARTICLE

Recurrence of Bacterial Vaginosis Is

Significantly Associated With Posttreatment
Sexual Activities and Hormonal Contraceptive
Use
Catriona S. Bradshaw,1,5,6,a Lenka A. Vodstrcil,1,a Jane S. Hocking,1,2 Matthew Law,9 Marie Pirotta,3
Suzanne M. Garland,4,7,8,10 Deborah De Guingand,1 Anna N. Morton,5 and Christopher K. Fairley1,5
1
Melbourne School of Population Health, 2Center for Women's Health, Gender and Society, and Departments of 3General Practice and 4Obstetrics and
Gynaecology, University of Melbourne, 5Melbourne Sexual Health Centre, The Alfred Hospital, 6Department of Epidemiology and Preventive Medicine,
Monash University, 7Department of Microbiology and Infectious Diseases, The Royal Women’s Hospital, and 8Department of Microbiology, The Royal

Downloaded from http://cid.oxfordjournals.org/ by guest on October 26, 2016

Children’s Hospital, Melbourne; 9The Kirby Institute, University of New South Wales, Darlinghurst; and 10Murdoch Children’s Research Institute,
Parkville, Australia

(See the Editorial Commentary by Marrazzo on pages 787–9.)

Background. Bacterial vaginosis (BV) recurrence posttreatment is common. Our aim was to determine if be-
haviors were associated with BV recurrence in women in a randomized controlled trial (RCT).
Methods. Symptomatic 18- to 50-year-old females with BV (≥3 Amsel criteria and Nugent score
[NS] = 4–10) were enrolled in a 3-arm randomized double-blind RCT Melbourne Sexual Health Centre, Australia,
in 2009–2010. All 450 participants received oral metronidazole (7 days) and were equally randomized to vaginal
clindamycin, lactobacillus-vaginal probiotic or vaginal placebo. At 1, 2, 3, and 6 months, participants self-collected
vaginal smears and completed questionnaires. Primary endpoint was NS = 7–10. Cox regression was used to esti-
mate hazard ratios (HRs) for risk of BV recurrence associated with baseline and longitudinal characteristics.
Results. Four hundred four (90%) women with postrandomization data contributed to analyses. Cumulative
6-month BV recurrence was 28% (95% confidence interval [CI], 24%–33%) and not associated with treatment.
After stratifying for treatment and adjusting for age and sex frequency, recurrence was associated with having the
same pre-/posttreatment sexual partner (adjusted HR [AHR] = 1.9; 95% CI, 1.2–3.0), inconsistent condom use
(AHR = 1.9; 95% CI, 1.0–3.3), and being non-Australian (AHR = 1.5; 95% CI, 1.0–2.1), and halved with use of
an estrogen-containing contraceptive (AHR = 0.5; 95% CI, .3–.8).
Conclusions. Risk of BV recurrence was increased with the same pre-/posttreatment sexual partner and in-
consistent condom use, and halved with use of estrogen-containing contraceptives. Behavioral and contraceptive
practices may modify the effectiveness of BV treatment.
Clinical Trials Registration. ACTRN12607000350426.
Keywords. bacterial vaginosis; contraceptive use; estrogen; sexual partner; condom use.

Received 31 July 2012; accepted 5 October 2012; electronically published 12

December 2012.
Bacterial vaginosis (BV) is the commonest cause of ab-
a
C. S. B. and L. A. V. contributed equally to this work. normal vaginal discharge in reproductive-age women,
Correspondence: Catriona S. Bradshaw, MD, PhD, Melbourne Sexual Health
Centre, 580 Swanston St, Carlton, VIC 3053, Australia (cbradshaw@mshc.org.au).
affecting 29% of 14- to 49-year-old North American
Clinical Infectious Diseases 2013;56(6):777–86
women in the National Health and Nutrition Survey [1],
© The Author 2012. Published by Oxford University Press on behalf of the Infectious and 12% of 17- to 28-year-old Australian women
Diseases Society of America. All rights reserved. For Permissions, please e-mail:
journals.permissions@oup.com.
attending general and sexual health practices [2]. Al-
DOI: 10.1093/cid/cis1030 though the etiology of BV is unclear, next-generation

Behaviors Associated With BV Recurrence • CID 2013:56 (15 March) • 777

sequencing methods have advanced our understanding of this
complex polymicrobial condition, with high bacterial species
diversity reported in women with BV compared to normal
flora, and increasing numbers of specific noncultivable BV-
associated bacteria (BVAB) being described [3]. Two of the com-
monest BVAB, Gardnerella vaginalis and Atopobium vaginae,
contribute to a vaginal biofilm associated with BV, which
rapidly reaccumulates following antibiotic therapy [4, 5].
Epidemiologic studies consistently report an association
between BV and sexual activity. In a study of young women,
BV was strongly associated with penile-vaginal sex, was rare in
women only engaging in noncoital practices, and was absent
in women reporting no sexual contact with others [6]. BV was
associated with increased numbers of sexual partners and
inconsistent condom use in a meta-analysis [7], and BVAB
and BV-associated biofilms have been detected in the male
coronal sulcus and urine [8–10]. Most researchers concur that
BV is associated with sexual activity, but whether this is
because of transmission of BVAB between partners or because
sexual activity adversely impacts on colonization with protec-
tive Lactobacillus species is unknown [11].
One of the challenges facing clinicians is provision of effec-
tive BV treatment, as long-term efficacy of recommended anti-
biotics is poor and relapse is common [12, 13]. Treatment trials
have yielded mixed results regarding the contribution of post-
treatment sexual activity to BV recurrence. In a study of 121
women following oral metronidazole treatment, 58% recurred
within 12 months. Recurrence was associated with having a
regular sexual partner (RSP) or a female sexual partner (FSP),
not using hormonal contraception, and history of BV [12].
Sanchez et al found that interim unprotected intercourse was
associated with BV recurrence [14], and Schwebke et al reported
that abstaining from sex/consistent condom use reduced recur-
rence by 50% [15]. In contrast, in women who have sex with
women (WSW), Marrazzo et al found that recurrence was not
associated with posttreatment sexual activity, but with baseline
and posttreatment detection of specific BVAB [16].
Current approaches to the management of BV are clearly
suboptimal, with reliance on antibiotics that cure less
than half the treated population. We sought to determine the
behavioral and contraceptive factors associated with BV recur-
rence in an analysis of a double-blind, placebo-controlled
treatment trial, in which combination oral and vaginal thera-
pies did not improve recurrence rates over metronidazole
alone [17]. Our hypothesis was that BV recurrence is signifi-
cantly influenced by sexual and behavioral practices.
METHODS
This was a double-blind, placebo-controlled, parallel-group
trial with balanced randomization (3 arms, 1:1:1) conducted
778 • CID 2013:56 (15 March) • Bradshaw et al
in 450 women from December 2007 through May 2010 at
Melbourne Sexual Health Centre, Australia [17]. Participants
received 400 mg of oral metronidazole twice daily for 7 days
and were randomized to 1 of 3 vaginal interventions: placebo,
2% clindamycin cream, or a commercially available estriol-
and Lactobacillus-containing probiotic. This study has been
described in detail elsewhere [17]. In brief, 18- to 50-year-old
women were eligible if they had abnormal vaginal discharge or
odor with ≥3 Amsel criteria and a Nugent score (NS) = 4–10.
Women were ineligible if positive for human immunodeficien-
cy virus (HIV), pregnant, breastfeeding, or unable to use
condoms/abstain from sex during vaginal treatment.
At baseline, participants were examined by a clinician,
Nugent and Amsel scores were documented, and paper-based
questionnaires were completed recording demographic, behav-
ioral, clinical, and contraceptive data. After treatment, women
completed questionnaires and self-collected vaginal smears and
returned specimens by mail at 1, 2, 3, and 6 months or until
Downloaded from http://cid.oxfordjournals.org/ by guest on October 26, 2016
BV recurrence. The primary endpoint was BV recurrence,
defined as an NS of 7–10 at any interval after treatment, at
which point participants were offered retreatment. Behavioral
data were recorded for the interval preceding each NS. “Same
pre-/posttreatment RSP” was defined as having the same sexual
partner pretreatment and in the 2 months after treatment.
Smears were scored by 1 of 3 blinded microbiologists experi-
enced in the Nugent method [17]. Women providing ≥2 sets of
microbiological and behavioral data (follow-up data on ≥1 oc-
casions) were included in longitudinal analyses. Specimens not
returned by interval midpoint, or at 6 months if >210 days after
enrollment, were considered missing, and participants who did
not return specimens were deemed lost to follow-up. Data were
censored when participants experienced BV recurrence, reached
210 days, or were lost to follow-up.
Statistical Methods and Ethical Approval
Frequencies and 95% confidence intervals (CIs) of baseline
demographic, clinical, and behavioral data were analyzed
using Stata software 12.0 (StataCorp LP). The rate per 100
person-years of BV recurrence for selected variables was calcu-
lated along with Poisson 95% CIs. Cox regression estimated
hazard ratios (HRs) for the risk of BV recurrence associated
with baseline and longitudinal characteristics. Characteristics/
exposures that changed during the study period were recon-
structed based on questionnaire data from each study interval
and examined in a number of ways including “present” or
“absent” and “current/recent” or “not recent” exposure. Multi-
variate cox regression models were used to assess whether
these characteristics were predictors of BV as time-dependent
covariates. Models were built sequentially starting with the
characteristic most strongly associated with BV by univariate
analysis and continued until no other variable reached
Table 1. Clinical and Laboratory Results of Study Population Table 2. Demographic and Clinical Characteristics of Study
(N = 404) Population (N = 404) at Baseline

Characteristic No. (%) Characteristic No. (%a)

Pretreatment Nugent score Age, y, median (range) 27 (17–49)
4–6a 64 (15.8) Country of birth
7–10 340 (84.2) Australia/New Zealand 251 (62.3)
Duration of symptomsb Otherb 152 (37.7)
≤21 d 211 (52.2) Educational level
>21 d 194 (47.8) Primary/secondary 140 (35.6)
Rating of abnormal vaginal dischargeb Tertiary/postgraduate 253 (64.4)
None 21 (5.3) History of BV
Mild 75 (19.0) No 205 (51.5)
Moderate/heavy 299 (75.7) Yes 193 (48.5)
Rating of abnormal vaginal odorb Current smoker
None 26 (6.6) No 228 (56.7)
Mild 45 (11.5) Yes 174 (43.3)
Moderate/heavy 322 (81.9) No. of cigarettes smoked per day
Effect of vaginal symptoms on quality of lifeb None 232 (57.4)

Downloaded from http://cid.oxfordjournals.org/ by guest on October 26, 2016

No effect 30 (7.6) <10 84 (20.8)
Somewhat concerned 65 (16.5) ≥10 88 (21.8)
Concerned/very distressed 300 (75.9) Currently douchesc
Effect of vaginal symptoms on sexual satisfactionb No 293 (73.6)
No effect 45 (11.4) Yes 105 (26.4)
Mild 56 (14.2) Douching product (n = 102)
Moderate/major 294 (74.4) Water 50 (49.0)
BV recurrence over 6 mo Water based 23 (22.6)
No BV recurrence (NS 0–6) 289 (71.5) Vinegar/hygiene product/other 29 (28.4)
BV recurrence (NS 7–10) 115 (28.5) Douching frequency
Abbreviation: BV, bacterial vaginosis; NS, Nugent score.
None 293 (74.0)
a
All had 3–4 Amsel criteria. Weekly or less often 57 (14.4)
b
Self-reported pretreatment symptoms and their effects. Daily 46 (85.6)
Current contraceptive method
Condoms only 171 (43.5)
Estrogen/progesterone 110 (28.0)
significance. Univariate and multivariate analyses were strati-
Progesterone only 33 (8.4)
fied for treatment group. As condom use and frequency of sex
No contraception/withdrawal/ 79 (20.1)
were moderately correlated, we built 3 multivariate models: 2 PCI/spermicide
containing one of the correlated variables, the third with both. History of chlamydia
Sensitivity analyses were performed adjusting for treatment No 282 (71.0)
group, and with and without age. A level of P < .05 was con- Yes 115 (29.0)
sidered statistically significant. History of HSV
This trial was approved by the Human Research and Ethics No 342 (86.2)
Committees of the Alfred Hospital and Monash University, Yes 55 (13.8)
History of abnormal Pap result
Melbourne, Australia, with written informed consent obtained
No 290 (73.1)
from participants.
Yes 107 (26.9)

Data are presented as No. (%) unless otherwise specified.

RESULTS
Of the 450 volunteers, 46 provided no postrandomization data
and were excluded. Postrandomization NS and behavioral data
were available for 404 (90%) women, who contributed 152
person-years of follow-up (median, 103 days; no difference
Abbreviations: BV, bacterial vaginosis; HSV, herpes simplex virus; PCI,
postcoital intervention.
a
Up to 3% of participants may have missing data for some variables;
therefore, proportions are calculated using available data.
b
China and Southeast Asia (35%), Britain and Ireland (20%), Eastern and
Western Europe (17%), and North America (16%).
c
Douching defined as flushing or washing the vagina out with a solution or water.

Behaviors Associated With BV Recurrence • CID 2013:56 (15 March) • 779

Table 3. Sexual Behavioral Characteristics of Study Population NS = 4–6 and ≥3 Amsel criteria (Table 1). Over the course of
(N = 404) at Baseline 6 months, 115 women (28%) experienced BV recurrence;
there was no difference in recurrence of BV or abnormal flora
Characteristic No. (%a) between treatment arms [17].
No. of male partners last 12 mo
0–1 205 (51.0) Demographic, Behavioral, and Clinical Characteristics of the
≥2 197 (49.0) Study Population
No. of female partners last 12 mo The median age of participants was 27 years (range, 17–49
None 314 (78.9)
years; Table 2). Most women had a tertiary education (64%)
≥1 84 (21.1)
and were born in Australia or New Zealand (62%), with the
Any penile-vaginal sex in the last 12 mo
remainder born in China or Southeast Asia (35%), Britain or
No 38 (9.4)
Yes 366 (90.6)
Ireland (20%), Eastern or Western Europe (17%), and North
Any penile-anal sex in the last 12 mo America (16%). A history of BV was reported by 193 (49%)
No 273 (68.4) women, 174 (43%) were current smokers, and 105 (26%)
Yes 126 (31.6) douched. Condoms were the main contraceptive method for
Any receptive oral sex in the last 12 mo 171 (43%) women, and 110 (28%) used an estrogen-contain-
No 40 (9.9) ing method of contraception (ECC). A history of chlamydia
Yes 363 (90.1) (29%), genital warts (18%), herpes (14%), and abnormal cervi-

Downloaded from http://cid.oxfordjournals.org/ by guest on October 26, 2016

Condom use for any vaginal sex in last 3 mo cal cytology (27%) was common, whereas gonorrhoea was
Alwaysb 66 (16.4) rare (2%).
Not always 336 (83.6)
In the prior 12 months, 197 (49%) women reported having
Condom use for any anal sex in last 3 mo (n = 126)
≥2 male sexual partners (MSPs) and 84 (21%) reported
Always 293 (73.8)
having ≥1 FSP; the majority reported penile-vaginal (91%)
Not always 104 (26.2)
Current sex work
and receptive oral (90%) sex (Table 3). Two hundred seventy-
No 341 (84.8) two women (68%) had an RSP at randomization; of these, 252
Yes 61 (15.2) (93%) were MSPs and 20 (7%) were FSPs. Among women
Practices with RSP with an RSP, 225 (83%) had ongoing vaginal and/or receptive
Current RSP at time of randomization oral sex with the same pretreatment RSP in the 2 months after
No 131 (32.5) treatment; 211 (94%) with MSPs and 14 (6%) with FSPs. Few
Yes 272 (67.5) (16%) participants reported consistent condom use for penile-
Sex of RSP at baseline vaginal sex in the 3 months prior to enrollment. Sixty-one
Male 252 (92.7) participants (15%) were sex workers.
Female 20 (7.3)
Practices with FSP
Demographic and Behavioral Factors Associated With BV
Any receptive oral sex from an FSP in last 12 mo
Recurrence
No 335 (83.8)
Yes 65 (16.2)
Univariate analysis of baseline and longitudinal characteristics
Any unwashed shared toy use with an FSP last 3 mo (n = 82) associated with BV recurrence, stratified by treatment group,
No 65 (79.3) are described in Table 4. Baseline characteristics associated
Yes 17 (20.7) with recurrence included being born outside Australia/New
Zealand (HR = 1.48; 95% CI, 1.02–2.15), having a baseline
Abbreviations: FSP, female sexual partner; RSP, regular sexual partner.
a
Up to 2% of participants may have missing data for some variables; RSP (HR = 1.72; 95% CI, 1.12–2.64), and ECC use (HR = 0.57;
therefore, proportions are calculated using available data. 95% CI, .37–.87). Age, education, treatment, and history of
b
Or no vaginal sex in last 3 months. BV were not associated with recurrence. No other baseline be-
haviors were associated with recurrence, including number of
MSPs or FSPs; vaginal, anal, or receptive oral sex in the last 12
months; or inconsistent condom use for vaginal or anal sex in
between treatment groups, P = .90). Fifty-seven participants the last 3 months.
(13%) without BV recurrence did not complete 6 months of Longitudinal univariate posttreatment practices and behav-
follow-up, but contributed person-time to survival analyses as iors associated with BVrecurrence (Table 4) included having
non recurrent cases based on their last NS. At baseline, the penile-vaginal sex (HR = 1.88; 95% CI, 1.15–3.07), inconsistent
majority of women had an NS = 7–10 (84%); 64 (16%) had an condom use for penile-vaginal sex (HR = 1.96; 95% CI, 1.29–

780 • CID 2013:56 (15 March) • Bradshaw et al

Table 4. Baseline Demographics and Baseline and Longitudinal Behavioral Factors Associated With Recurrence, Stratified for Treat-
ment Group

BV Recurrence Rate per

Characteristic (N = 404) 100 py (95% CI) Unadjusted HR (95% CI) P Value
Demographics, treatment allocation, and conditions
Age, y
17–27 76 (59–98) 1
>27 75 (58–98) 0.99 (.39–1.42) .935
Country of birth
Australia/New Zealand 63 (49–81) 1
Other 97 (74–127) 1.48 (1.02–2.15) .038
Educational level
Primary/secondary 94 (71–124) 1
Tertiary 66 (51–84) 0.71 (.49–1.03) .069
Vaginal treatment group
Placebo 74 (54–103) 1
Probiotic 75 (54–103) 0.99 (.63–1.58) .978
Clindamycin 78 (58–106) 1.06 (.68–1.65) .791

Downloaded from http://cid.oxfordjournals.org/ by guest on October 26, 2016

History of BV
No 72 (55–93) 1
Yes 79 (61–103) 1.12 (.78–1.62) .539
Use of ECC at enrollment
No 89 (72–110) 1
Yes 52 (32–76) 0.57 (.37–.87) .010
Baseline sexual behavior
No. of male partners last 12 mo
0–2 82 (64–106) 1
≥3 70 (54–92) 0.84 (.58–1.20) .340
No. of female partners last 12 mo
0 73 (59–90) 1
≥1 84 (58–123) 1.16 (.76–1.77) .482
Any penile-vaginal sex in the last 12 mo
No 62 (32–120) 1
Yes 77 (64–93) 1.24 (.64–2.45) .517
Any penile-anal sex in the last 12 mo
No 75 (60–94) 1
Yes 77 (56–107) 1.02 (.69–1.51) .918
Any receptive oral sex in the last 12 mo
No 90 (52–155) 1
Yes 74 (61–90) 0.82 (.46–1.47) .510
Condom use for any vaginal sex in last 3 mo
Always (or not that 78 (64–95) 1
practice)
Not always 63 (39–103) 0.79 (.48–1.32) .370
Condom use for any anal sex in last 3 mo
Always (or not that 79 (55–112) 1
practice)
Not always 75 (61–93) 0.96 (.64–1.43) .828
RSP at enrollment
No 52 (36–76) 1
Yes 88 (72–109) 1.72 (1.12–2.64) .014

Behaviors Associated With BV Recurrence • CID 2013:56 (15 March) • 781

Table 4 continued.

BV Recurrence Rate per

Characteristic (N = 404) 100 py (95% CI) Unadjusted HR (95% CI) P Value
a
Longitudinal behaviors
Smoker
Never 69 (54–90) 1
Ever during observation 83 (64–108) 1.18 (.82–1.69) .378
No. of cigarettes smoked per day
None 70 (55–89) 1
1–9 86 (58–128) 1.24 (.79–1.95) .340
≥10 84 (56–126) 1.13 (.73–1.76) .587
Currently douching
No 70 (56–87) 1
Yes 94 (66–135) 1.28 (.84–1.96) .246
Current/recent use of ECC
No 87 (70–107) 1
Yes 55 (38–80) 0.62 (.40–.95) .028
Current sex work

Downloaded from http://cid.oxfordjournals.org/ by guest on October 26, 2016

No 74 (60–90) 1
Yes 91 (56–148) 1.17 (.69–1.99) .565
Any penile-vaginal sex
No 46 (29–72) 1
Yes 85 (70–105) 1.88 (1.15–3.07) .012
Any penile-anal sex
No 74 (60–90) 1
Yes 83 (49–140) 1.09 (.63–1.89) .754
Any receptive oral sexb
No 76 (57–100) 1
Yes 72 (56–92) 0.93 (.64–1.36) .707
Condom use for any penile-vaginal sex
Alwaysc 47 (33–68) 1
Not always 93 (75–115) 1.96 (1.29–2.99) .002
Any new male partner
None 73 (59–90) 1
≥1 84 (57–124) 1.16 (.76–1.77) .491
Any female partner
None 74 (61–90) 1
≥1 118 (59–236) 1.59 (.77–3.29) .213
Same pre-/posttreatment RSPd
No 53 (38–73) 1
Yes 95 (76–118) 1.84 (1.25–2.73) .002
Frequency of sex with any sexual partnerse
No sex 41 (24–69) 1
1–4 times per mo 87 (65–116) 2.18 (1.19–4.01) .012
≥5 times per mo 82 (62–107) 1.96 (1.09–3.52) .025

Unadjusted hazard ratios are stratified by treatment group, bolded text indicates significant associations at the level p < 0.05.
Abbreviations: BV, bacterial vaginosis; CI, confidence interval; ECC, estrogen-containing contraceptive; HR, hazard ratio; py, person-years; RSP, regular sexual
partner.
a
Each variable is comprised of behaviors reported longitudinally by participants at each study interval.
b
Receptive oral sex is defined as being given oral sex.
c
Or no penile-vaginal sex.
d
Same pre-/posttreatment RSP, defined as same sexual partner pretreatment and in the 2 months posttreatment.
e
Sex with females is having received oral sex, sex with males is penile-vaginal sex.

782 • CID 2013:56 (15 March) • Bradshaw et al

Table 5. Demographic and Behavioral Factors Associated With Recurrence of Bacterial Vaginosis, Multivariate Model

BV Model 1a Model 2a Model 3a

Recurrence
Rate per
Characteristic 100 py Unadjusted HR P Adjusted HR P Adjusted HR P Adjusted HR P
(N = 404) (95% CI) (95% CI) Value (95% CI) Value (95% CI) Value (95% CI) Value
Age, y
17–27 76 (59–98) 1 1 1 1
>27 75 (58–98) 0.98 (.68–1.41) .902 0.98 (.68–1.43) .931 0.97 (.66–1.41) .863 0.95 (.66–1.39) .812
Country of birth
Australia/New Zealand 63 (49–81) 1 1 1 1
Other 97 (74–127) 1.54 (1.06–2.22) .022 1.49 (1.03–2.17) .034 1.47 (1.01–2.14) .042 1.46 (1.01–2.13) .045
Current/recent use of ECC
No 87 (70–107) 1 1 1 1
Yes 55 (38–80) 0.63 (.41–.96) .031 0.51 (.33–.78) .002 0.56 (.36–.87) .010 0.52 (.34–.81) .004
Same pre-/posttreatment RSPb
No 53 (38–73) 1 1 1 1
Yes 95 (76–118) 1.84 (1.25–2.73) .002 1.77 (1.14–2.73) .010 1.99 (1.25–3.18) .004 1.91 (1.19–3.07) .007
Condom use for any penile-vaginal sex

Downloaded from http://cid.oxfordjournals.org/ by guest on October 26, 2016

Alwaysc 47 (33–68) 1 1 1
Not always 93 (75–115) 1.96 (1.28–3.00) .002 1.74 (1.12–2.70) .014 1.85 (1.04–3.29) .037
Frequency of sex with any sexual partnersd
No sex 41 (24–69) 1 1 1
1–4 times per mo 87 (65–116) 2.17 (1.18–3.99) .013 1.74 (.92–3.29) .090 1.09 (.49–2.40) .826
≥5 times per mo 82 (62–107) 1.97 (1.09–3.56) .025 1.41 (.74–2.69) .300 0.85 (.36–1.97) .700
Bolded text indicates significant associations at the level p < 0.05.
Abbreviations: BV, bacterial vaginosis; CI, confidence interval; ECC, estrogen-containing contraceptive; HR, hazard ratio; py, person-years; RSP, regular sexual
partner.
a
All models are stratified for treatment.
b
Same pre-/posttreatment RSP defined as same sexual partner pretreatment and in the 2 months posttreatment.
c
Or no penile-vaginal sex.
d
Sex with females is defined as having received oral sex, sex with males is defined as penile-vaginal sex.

2.99), and having the same pre-/posttreatment RSP (male or
female) (HR = 1.84; 95% CI, 1.25–2.73). Current or recent use
of an ECC (oral contraceptive pill = 108, NuvaRing = 2) was
protective against BV recurrence (HR = 0.62; 95% CI, .40–.95).
Of the 110 women using an ECC at baseline, 80 (72%) had
continuous ECC use, and 30 (28%) inconsistent ECC use
during follow-up. Because ovulation can be delayed for several
months after ceasing ECC, ECC use was analyzed longitudi-
nally as “current/recent” or “not recent.” Few women (n = 33)
used progesterone-only methods of contraception, which
limited our power to examine their association with recur-
rence; however, use of any hormonal contraception was not
significantly protective against recurrence (HR = 0.83; 95% CI,
.56–1.24).
We explored the association between BV recurrence and
frequency of sexual activity with partners (male and female)
and then with frequency of penile-vaginal sex alone. Com-
pared with no sexual activity, sexual contact with partners ≤4
times per month or >4 times per month was associated with
recurrence (HR = 2.18; 95% CI, 1.19–4.01 and HR = 1.96; 95%
CI, 1.09–3.52, respectively); however, there was no dose re-
sponse between increased frequency of sexual activity and risk
of recurrence. Frequency of penile-vaginal sex and/or any
sexual activity with MSPs and FSPs were examined as contin-
uous variables and at a number of cut points: there was no
significant association with increased sexual activity and in-
creased risk of recurrence using these methods (data not
shown). No association was found between BV recurrence and
a new sexual partner, FSPs, smoking, douching, or engaging
in sex work.
Factors associated with BV recurrence by univariate analysis
were included in multivariate analyses and stratified for treat-
ment group. This included country of birth, ECC use, same
pre-/posttreatment RSP, and condom use for penile-vaginal
sex (Table 5). Age was included owing to reported associations
with BV and sexual/contraceptive practices. Frequency of
sexual activity was included to determine whether it exerted
an independent effect on recurrence after adjusting for RSP.

Behaviors Associated With BV Recurrence • CID 2013:56 (15 March) • 783

As frequency of sex and condom use were moderately correlat-
ed (r = 0.66, P < .001), 3 multivariate analyses were conducted;
2 contained one of the correlated variables, and one included
both variables. Multivariate models were stratified for treat-
ment group, but sensitivity analyses were also performed ad-
justing for treatment group, and with and without age,
yielding similar results (data not shown).
Model 1 includes the variables most strongly associated
with BV recurrence, and demonstrates that having the same
pre-/posttreatment RSP (adjusted HR [AHR] = 1.77; 95% CI,
1.14–2.73) and inconsistent condom use for penile-vaginal sex
with any MSP (AHR = 1.74; 95% CI, 1.12–2.70) were associat-
ed with BV recurrence, whereas ECC use was protective
(AHR = 0.51; 95% CI, .33–.78). Being born outside Australia/
New Zealand showed a modest association with recurrence
(AHR = 1.49; 95% CI, 1.03–2.17). The second model included
frequency of sex and omitted the correlated “condom use”
variable. This analysis showed that frequency of sexual activity
was not associated with recurrence, exposure to the same pre-/
posttreatment RSP conferred an increased risk (AHR = 1.99;
95% CI, 1.25–3.18), and ECC use halved the risk (AHR = 0.56;
95% CI, .36–.87) of recurrence. The final model included all
variables and confirmed that after adjusting for frequency of
sexual activity, having the same pre-/posttreatment RSP
(AHR = 1.91; 95% CI, 1.19–3.07), inconsistent condom use for
penile-vaginal sex (AHR = 1.85; 95% CI, 1.04–3.29) and being
born outside Australia/New Zealand (AHR = 1.46; 95% CI,
1.01–2.13) were associated with BV recurrence, whereas ECC
use was protective (AHR = 0.52; 95% CI, .34–.81).
DISCUSSION
In this trial, posttreatment sexual and contraceptive behaviors
were strongly associated with BV recurrence. Importantly, ex-
posure to the same pre-/posttreatment RSP, and inconsistent
condom use for penile-vaginal sex, nearly doubled the risk of
BV recurrence, whereas the use of an estrogen-containing con-
traceptive method halved the risk. These data provide compel-
ling evidence that behavioral and contraceptive practices play
a significant role in modifying the effectiveness of current an-
tibiotics in the treatment of BV.
The association between exposure to an ongoing RSP and
recurrence is supported by a previous study in which women
who remained with their pretreatment RSP had a 3-fold in-
creased risk of BV recurrence [12]. The robustness of the asso-
ciation between recurrence and exposure to an RSP across 3
multivariate models, and the fact that it remained after adjust-
ing for frequency of sexual activity, condom use, and other po-
tentially confounding behavioral factors, indicates that ongoing
exposure to an RSP significantly contributes to BV recurrence
in this cohort, and importantly, that more frequent sexual
784 • CID 2013:56 (15 March) • Bradshaw et al
activity does not explain this association. Increased frequency of
sexual activity has been proposed as a mechanism by which
normal vaginal flora is disrupted. Studies have found an associ-
ation between frequency of penile-vaginal, digital-vaginal, toy-
vaginal, or receptive oral sex, and unfavorable vaginal states
including unstable vaginal flora [18], BV [19], reduction in
hydrogen peroxide–producing lactobacilli [11, 20, 21], and in-
creased colonization with G. vaginalis [11]; however, others
report that higher frequency of penile-vaginal sex increases con-
centrations of protective hydrogen peroxide–producing lactoba-
cilli [11].
Inconsistent condom use for penile-vaginal sex was strongly
associated with BV recurrence, and remained so after control-
ling for an RSP and other behaviors. This finding is consistent
with previous studies [14, 15], and a meta-analysis showing a
20% protective effect of consistent condom use against BV [7].
Other data support the concept that exposure to male genitalia
contributes to the development of BV in heterosexual women.
Downloaded from http://cid.oxfordjournals.org/ by guest on October 26, 2016
Wives of circumcised males in a Ugandan circumcision trial
had a significantly reduced risk of BV (adjusted prevalence
risk ratio = 0.60; 95% CI, .38–.94) compared to wives of uncir-
cumcised males [22]. Furthermore, male circumcision has
been associated with a significant reduction in penile anaero-
bic microbial flora, including BV-associated genera Clostri-
diales and Prevotellaceae [9]. Pyrosequencing of the
microbiota of the coronal sulcus and distal urethra in adoles-
cents [23] showed these sites to be colonized by BVAB, with
the composition of the coronal sulcus microbiota influenced
by circumcision and sexual activity [23]. Male carriage of
G. vaginalis, an organism considered to be integral in the
development of BV, is commonly reported [10, 24, 25].
This trial was conducted in a population of sexually
active women who predominantly had sex with men.
Numbers of posttreatment FSPs were too small to examine
the contribution of female partnerships to recurrence. Im-
portantly, there is increasing evidence to support exchange
of vaginal flora and BVAB between WSW. WSW have been
shown to have high levels of concordant vaginal flora with
their FSP [26, 27], and Lactobacillus species are shared
between FSPs [28]. We previously reported that women
with a posttreatment FSP had a 3-fold increased risk of BV
recurrence [12]. Although Marrazzo et al found that BV re-
currence was not associated with posttreatment sexual
activity in WSW [16], this group recently reported that
vaginal-toy use was associated with higher likelihood of col-
onization with G. vaginalis [11] and digital-vaginal and
oral-vaginal sex with reduced load of L. crispatus [21]. They
hypothesized that some protective commensal species could
be particularly sensitive to the effects of sexual activity.
Estrogen-containing contraception, predominantly the oral
contraceptive pill in this cohort, showed a significantly
protective effect against BV recurrence after adjusting for con-
founding factors including condom use and RSP. A number of
cross-sectional studies have reported hormonal contraceptives,
mainly combined, to be protective against prevalent [29–31],
incident [32, 33], and recurrent [12, 31] BV, with a some
studies also reporting that progesterone-only methods may
be associated with a reduced risk for incident [33] and
recurrent [31] BV. One mechanism by which combined con-
traceptives may be protective is that estrogen increases the gly-
cogen-content of epithelial cells, which is a substrate for
Lactobacillus species for the generation of lactic acid, a known
potent inhibitor of BV [34, 35]. Hormonal contraceptives, par-
ticularly progesterone-only, may also reduce heme availability
for anaerobes such as G. vaginalis through reduction in men-
struation. Interestingly, BV is most commonly reported at the
beginning of the menstrual cycle when estradiol levels are
lowest [36, 37], and higher rates of remission occur in preg-
nancy [38]. Contraceptive use may also affect immunological
factors, with Cherpes et al reporting that hormonal contraceptive
use was associated with altered vaginal immunity in BV [39].
This study has a number of limitations. Sexual practices are
often highly correlated, making it difficult to attribute an asso-
ciation to a specific behavior and to infer causality. Even
though the association between exposure to the same RSP and
BV recurrence remained in all 3 models after adjusting for po-
tentially confounding factors, we cannot exclude the effect of
unmeasured confounding. Secondly, we did not have sufficient
power to examine the effect of less common behaviors on re-
currence, such as female partners and progesterone-only con-
traceptives. Although sufficiently common to be assessed,
there was no association between BV recurrence and risk
factors previously associated with BV including history of BV,
smoking, douching, and sex work. Self-sampling was em-
ployed to optimize participant retention, we were therefore
limited in our outcome measure to the use of the Nugent
method, which potentially underestimated BV recurrence by
classifying women with an NS of 4–6 but ≥3 Amsel criteria as
cured. The strengths of our trial include that it was a large
RCT with high retention rates and detailed epidemiologic data
matched to frequent sampling over 6 months. This minimized
participant recall bias and enabled accurate assessment of as-
sociations between recent behaviors and relevant NSs and
longer-term BV recurrence. We applied the gold-standard
Nugent method to ensure generalizability of our findings, and
have an ongoing quality-assurance program to ensure consis-
tency between our experienced microbiologists.
A striking aspect to clinical management of BV is that 1-
month cure rates of recommended therapies are 70%–80%;
however, by 3–12 months, failure rates approach 50% [12, 13].
Our data indicate that the long-term poor performance of rec-
ommended therapies may at least partly be attributed to
Behaviors Associated With BV Recurrence
posttreatment sexual and contraceptive behaviors. Whether
the association with RSP is due to reinfection/transmission of
BVAB from untreated RSPs, another “RSP factor” that hinders
recolonization after treatment with protective Lactobacillus
species, or unmeasured confounding remains unanswered. Al-
though 5 of 6 partner treatment trials failed to reduce BV re-
currence in women, systematic review has shown that these
trials were significantly flawed [40]. These data raise the tanta-
lizing question of whether partner treatment trials in hetero-
sexual and WSW populations need revisiting. Importantly for
clinicians, it appears that potentially modifiable practices, such
as use of condoms and/or estrogen-containing contraceptives,
may provide a degree of protection against recurrence. As
current therapeutic approaches have limited impact on long-
term BV cure, we need to understand the mechanisms by
which these practices modify the risk of BV recurrence, and to
more formally assess such interventions to determine the
degree to which they impact on BV recurrence. In the era of
Downloaded from http://cid.oxfordjournals.org/ by guest on October 26, 2016
combination prevention approaches for HIV infection, it
seems quite possible that treatment of BV could be improved
through integration of modifiable behavioral factors within
more holistic management approaches.
Notes
Acknowledgments. We acknowledge and thank Surbhi Bird, Andrea
Morrow, Sandy Walker, and Eve Urban (Melbourne Sexual Health Centre)
for their assistance with the daily running of the clinical trial; Glenda
Fehler and Leonie Horvath (Melbourne Sexual Health Centre) for assisting
with microscopy for the trial; and Philipp Grob and Valda Prasauskas
from Medinova for providing the probiotic and placebo that were evaluat-
ed in the original trial.
Financial support. This work was supported by the Australian Na-
tional Health and Medical Research Council Project (grant number
APP454644); by the Australian National Health and Medical Research
Council (fellowship numbers APP 456164 and 566576 to C. S. B. and
J. S. H., respectively); and by a Primary Health Care Research Evaluation
and Development Mid-career Fellowship, Department of Health and
Aging (to M. P.).
Potential conflicts of interest. All authors: No reported conflicts.
All authors have submitted the ICMJE Form for Disclosure of Potential
Conflicts of Interest. Conflicts that the editors consider relevant to the
content of the manuscript have been disclosed.
References
1. Allsworth JE, Peipert JF. Prevalence of bacterial vaginosis: 2001–2004
National Health and Nutrition Examination Survey Data. Obstet
Gynecol 2007; 109:114–20.
2. Walker J, Hocking JS, Fairley CK, et al. The prevalence and incidence
of bacterial vaginosis in a cohort of young Australian women. In:
Conference proceedings of the International Society for Sexually
Transmitted Diseases Research. Quebec, Canada, 2011.
3. Fredricks DN, Fiedler TL, Marrazzo JM. Molecular identification of
bacteria associated with bacterial vaginosis. N Engl J Med 2005;
353:1899–911.
4. Swidsinski A, Mendling W, Loening-Baucke V, et al. Adherent bio-
films in bacterial vaginosis. Obstet Gynecol 2005; 106:1013–23.
• CID 2013:56 (15 March) • 785
 5. Swidsinski A, Mendling W, Loening-Baucke V, et al. An adherent
Gardnerella vaginalis biofilm persists on the vaginal epithelium after
standard therapy with oral metronidazole. Am J Obstet Gynecol 2008;
198:97 e91–96.
6. Fethers KA, Fairley CK, Morton A, et al. Early sexual experiences and
risk factors for bacterial vaginosis. J Infect Dis 2009; 200:1662–70.
7. Fethers KA, Fairley CK, Hocking JS, Gurrin LC, Bradshaw CS. Sexual
risk factors and bacterial vaginosis: a systematic review and meta-anal-
ysis. Clin Infect Dis 2008; 47:1426–35.
8. Nelson DE, Van Der Pol B, Dong Q, et al. Characteristic male urine
microbiomes associate with asymptomatic sexually transmitted infec-
tion. PLoS One 2012; 5:e14116.
9. Price LB, Liu CM, Johnson KE, et al. The effects of circumcision on
the penis microbiome. PLoS One 2010; 5:e8422.
10. Schwebke JR, Rivers C, Lee J. Prevalence of Gardnerella vaginalis in
male sexual partners of women with and without bacterial vaginosis.
Sex Transm Dis 2009; 36:92–4.
11. Mitchell C, Manhart LE, Thomas KK, Agnew K, Marrazzo JM. Effect
of sexual activity on vaginal colonization with hydrogen peroxide-
producing lactobacilli and Gardnerella vaginalis. Sex Transm Dis
2011; 38:1137–44.
12. Bradshaw CS, Morton AN, Hocking J, et al. High recurrence rates of
bacterial vaginosis over the course of 12 months after oral metronida-
zole therapy and factors associated with recurrence. J Infect Dis 2006;
193:1478–89.
13. Sobel JD, Schmitt C, Meriwether C. Long-term follow-up of patients
with bacterial vaginosis treated with oral metronidazole and topical
clindamycin. J Infect Dis 1993; 167:783–4.
14. Sanchez S, Garcia PJ, Thomas KK, Catlin M, Holmes KK. Intravaginal
metronidazole gel versus metronidazole plus nystatin ovules for bacte-
rial vaginosis: a randomized controlled trial. Am J Obstet Gynecol
2004; 191:1898–906.
15. Schwebke JR, Desmond RA. A randomized trial of the duration of
therapy with metronidazole plus or minus azithromycin for treat-
ment of symptomatic bacterial vaginosis. Clin Infect Dis 2007;
44:213–9.
16. Marrazzo JM, Thomas KK, Fiedler TL, Ringwood K, Fredricks DN.
Relationship of specific vaginal bacteria and bacterial vaginosis treat-
ment failure in women who have sex with women. Ann Intern Med
2008; 149:20–8.
17. Bradshaw CS, Pirotta M, De Guingand D, et al. Efficacy of oral metro-
nidazole with vaginal clindamycin or vaginal probiotic for bacterial
vaginosis: randomised placebo-controlled double-blind trial. PLoS
One 2012; 7:e34540.
18. Schwebke JR, Richey CM, Weiss HL. Correlation of behaviors with mic-
robiological changes in vaginal flora. J Infect Dis 1999; 180:1632–6.
19. Marrazzo JM, Thomas KK, Fiedler TL, Ringwood K, Fredricks DN.
Risks for acquisition of bacterial vaginosis among women who report
sex with women: a cohort study. PLoS One 2010; 5:e11139.
20. Vallor AC, Antonio MA, Hawes SE, Hillier SL. Factors associated with
acquisition of, or persistent colonization by, vaginal lactobacilli: role of
hydrogen peroxide production. J Infect Dis 2001; 184:1431–6.
21. Mitchell C, Manhart LE, Thomas K, et al. Behavioral predictors of
colonization with Lactobacillus crispatus or Lactobacillus jensenii after
treatment for bacterial vaginosis: a cohort study. Infect Dis Obstet
Gynecol 2012; 2012:706540.
786 • CID 2013:56 (15 March) • Bradshaw et al
View publication stats
22. Gray RH, Kigozi G, Serwadda D, et al. The effects of male circumci-
sion on female partner's genital tract symptoms and vaginal infections
in a randomized trial in Rakai, Uganda. Am J Obstet Gynecol 2009;
200:42.e1–7.
23. Nelson DE, Dong Q, Van Der Pol B, et al. Bacterial communities of
the coronal sulcus and distal urethra of adolescent males. PLoS One
2012; 7:e36298.
24. Bradshaw CS, Tabrizi SN, Read TRH, et al. Etiologies of non-gonococ-
cal urethritis: bacteria, viruses and the association with oro-genital
exposure. J Infect Dis 2006; 193:336–45.
25. Piot P. Distribution of eight serotypes of Ureaplasma urealyticum in
cases of non-gonococcal urethritis and of gonorrhoea, and in healthy
persons. Br J Vener Dis 1976; 52:266–8.
26. Marrazzo JM, Koutsky LA, Eschenbach DA, et al. Characterization of
vaginal flora and bacterial vaginosis in women who have sex with
women. J Infect Dis 2002; 185:1307–13.
27. Bailey JV, Farquhar C, Owen C. Bacterial vaginosis in lesbians and
bisexual women. Sex Transm Dis 2004; 31:691–4.
28. Marrazzo JM, Antonio M, Agnew K, Hillier SL. Distribution of genital
Lactobacillus strains shared by female sex partners. J Infect Dis 2009;
199:680–3.
29. Shoubnikova M, Hellberg D, Nilsson S, Mardh PA. Contraceptive
use in women with bacterial vaginosis. Contraception 1997; 55:355–8.
Downloaded from http://cid.oxfordjournals.org/ by guest on October 26, 2016
30. Calzolari E, Masciangelo R, Milite V, Verteramo R. Bacterial
vaginosis and contraceptive methods. Int J Gynaecol Obstet 2000;
70:341–6.
31. Riggs M, Klebanoff M, Nansel T, et al. Longitudinal association
between hormonal contraceptives and bacterial vaginosis in women of
reproductive age. Sex Transm Dis 2007; 34:954–9.
32. Avonts D, Sercu M, Heyerick P, et al. Incidence of uncomplicated
genital infections in women using oral contraception or an intrauter-
ine device: a prospective study. Sex Transm Dis 1990; 17:23–9.
33. Baeten JM, Nyange PM, Richardson BA, et al. Hormonal contracep-
tion and risk of sexually transmitted disease acquisition: results from a
prospective study. Am J Obstet Gynecol 2001; 185:380–5.
34. O’Hanlon DE, Lanier BR, Moench TR, Cone RA. Cervicovaginal fluid
and semen block the microbicidal activity of hydrogen peroxide pro-
duced by vaginal lactobacilli. BMC Infect Dis 2010; 10:120.
35. O’Hanlon DE, Moench TR, Cone RA. In vaginal fluid, bacteria associ-
ated with bacterial vaginosis can be suppressed with lactic acid but
not hydrogen peroxide. BMC Infect Dis 2011; 11:200.
36. Bradshaw CS, Morton AN, Garland SM, et al. Higher-risk behavioral
practices associated with bacterial vaginosis compared with vaginal
candidiasis. Obstet Gynecol 2005; 106:105–14.
37. Schwebke JR, Morgan SC, Weiss HL. The use of sequential self-
obtained vaginal smears for detecting changes in the vaginal flora. Sex
Transm Dis 1997; 24:236–9.
38. Hay PE, Morgan DJ, Ison CA, et al. A longitudinal study of
bacterial vaginosis during pregnancy. Br J Obstet Gynaecol 1994;
101:1048–53.
39. Cherpes TL, Marrazzo JM, Cosentino LA, et al. Hormonal contracep-
tive use modulates the local inflammatory response to bacterial vagi-
nosis. Sex Transm Infect 2008; 84:57–61.
40. Mehta SD. Systematic review of randomized trials of treatment of
male sexual partners for improved bacterial vaginosis outcomes in
women. Sex Transm Dis 2012; 39:822–30.