High prevalence of bacterial vaginosis in adolescent girls in a tropical area of Ecuador
P Hay, Courtyard Clinic, St. George’s Hospital, Blackshaw Road, London, SW17 0QT, UK. Email email@example.com
Please cite this paper as: Vaca M, Guadalupe I, Erazo S, Tinizaray K, Chico M, Cooper P, Hay P. High prevalence of bacterial vaginosis in adolescent girls in a tropical area of Ecuador. BJOG 2010;117:225–228.
Bacterial vaginosis (BV) is a common clinical syndrome, but data are scarce on the BV prevalence in tropical regions among sexually active and virgin adolescents. To estimate the prevalence of BV among adolescent girls in an Ecuadorian coastal town, girls were asked to complete a questionnaire on risk factors for BV and vaginal samples were examined. Bacterial vaginosis was present in 31.5% of 213 girls, and the prevalence was similar in self-reported virgin and sexually active girls (OR 1.06, 95% CI, 0.51–2.21, P = 0.88), although the power of this analysis was limited. The prevalence of BV was high among Ecuadorian adolescent girls, and did not appear to be associated with sexual activity.
Bacterial vaginosis (BV) is a common clinical syndrome characterised by an overgrowth of vaginal anaerobes and depletion of the normal Lactobacillus population. It has been associated with an increased risk of upper genital tract infections, adverse pregnancy outcomes, and an increased risk of acquisition of sexually transmitted infections (STIs), including HIV.1 Bacterial vaginosis is the commonest cause of abnormal vaginal discharge in industrialised countries, and a population-based study estimated a prevalence of 9% in Europe. Risk factors for BV include vaginal douching, smoking and being black. Debate continues as to whether it is an STI, as it has been linked to high-risk sexual behaviours in some studies recently summarised in a meta-analysis.2
The most compelling evidence that it is not an STI is its occurrence in virgin adolescent girls. Bump and Buesching3 looked for clue cells on microscopy of vaginal wet-mount samples as a marker of BV. They detected clue cells in 13% of their subjects, with no difference in prevalence between the 68 sexually active and 52 virginal adolescent girls. Subsequently, a study of older women, female military recruits, reported a similar prevalence in sexually experienced and inexperienced women.4 Such studies rely on an accurate ascertainment of true virgin or sexually experienced status. Bump and Buesching used peer recruiters from the same school class. Sexual activity ‘was obtained by a confidential questionnaire with extensive interview supplementation’, but it is not clear who supplemented the interviews. If a virgin girl was found to have a sexually transmitted disease (STD), she was interviewed by the principal investigator to assess if any sexual activity or sexual abuse had occurred. The paper states that no subjects had their classification altered because of this. The other study used questionnaires.4 In a study published after ours was designed, an Australian group showed that more detail may be required before classifying women as virgins or non-virgins. They took a more detailed history of any sexual activity from their questionnaire, and found that detection of the BV-associated organisms Gardnerella and Atopobium vaginae was associated with non-penetrative sexual activity, such as fingering or non-penetrative genital–genital contact.5
There are few data on the prevalence of BV in adolescent and virgin populations living in the tropics. Jones and colleagues studied STIs in a population of low-income women in coastal Peru aged 18–30.6 They used computer-assisted audio interviews to assess the sexual history. Fourteen percent of 779 women reported no sexual experience. The prevalence of BV was 26.6%, and it was not associated with sexual experience. We wanted to replicate the earlier study of Bump and Buesching,3 which looked for BV in virgin and non-virgin adolescent girls, to estimate the prevalence of BV and identify associated risk factors. We conducted a cross-sectional study in adolescent girls in tropical Ecuador, who were also being screened for helminths.
A sample of 200 girls was considered to provide adequate precision for a point prevalence estimate of approximately 15%, with 95% CIs of 10–20%. The study was conducted in girls of mixed-race attending two secondary schools in La Concordia, a town of approximately 30 000 inhabitants in Esmeraldas Province. Informed written consent was obtained from a parent and the girl signed her assent. The age of legal consent in Ecuador is 18 years. The vaginal smear was collected by the study physician (MV) in the presence of the girl’s mother in a consulting room. The slides and questionnaire were coded with a patient identification number taken from a list of 5-digit random numbers. Two sterile cotton swabs were introduced into the posterior vaginal fornix: the first was smeared onto two microscope slides for Gram staining, and the second was used for fresh smear examination. The slides were dried and flame-fixed. A blood sample was collected for syphilis serology.
After dressing, the mother was asked to leave the room and the girl remained in the room in the presence of the study physician. Data on risk factors for BV were obtained by questionnaire. Before filling in the questionnaire, the study physician explained to the girl that the questionnaire information was completely confidential and could not be linked to her name. The questionnaire was filled in by the girl in the consulting room, where the study physician was the only other person present and was available to answer any questions that might arise.
The Gram-stained slides were subsequently read by an experienced observer (PH) using the Hay–Ison classification.7 This is the method recommended for use in UK genitourinary medicine (GUM) clinics, and may produce a slightly higher prevalence rate than the Nugent criteria, as some samples that score 6 with Nugent scoring are interpreted as BV with Hay–Ison scoring. The reader was blinded to the questionnaire results when reading the slides. Treatment for BV was provided if symptoms were elicited. The study protocol was approved by the Ethics Committee of the Hospital Pedro Vicente Maldonado.
Samples were collected from 215 girls (mean 14.9 years, range 13–17 years). Vaginal sex ever was reported by 19.1% of girls, and 7.0% had current sexual partners. Almost all girls had started menstruation (94.4%), at a mean age of 12 years (range 9–16 years). Washing of the external genitalia at least once daily was practiced by 189 of the 195 (96.9%) girls that responded to the question, vaginal douching by 28 (13.1%) of the responding 214 girls, and the use of external protective pads by 119 of the 215 (53.3%) girls. The majority reported having had a vaginal discharge ever (205/215 or 95.3%) and within the previous 2 weeks (194/215 or 90.2%). The proportions with girls with different grades of BV according to the Hay–Ison classification6 were: grade 0, 1.4%; grade 1, 61.5%; grade 2, 5.6%; grade 3, 30.1%; grade 4, 1.4%. Two smears were not classifiable because of inadequate samples, and were excluded from the analysis.
A total of 67 (31.5%) of the 213 girls had evidence of BV (grade 3). Fresh vaginal smears showed the presence of Trichomonas vaginalis trophozoites in one (0.5%) and Candida albicans in 51 (23.7%). None of the serum samples were positive for syphilis by Venereal Disease Research Laboratory (VDRL) test. Risk factors for BV are provided in Table 1. The proportion of girls reporting vaginal sex ever, with a current sexual partner, and the proportion of girls denying vaginal sex among those with BV compared with those without BV were 19.5 versus 17.2% and 7.4 versus 9.4% and 80.5 versus 82.8%, respectively. Vaginal sex ever was not associated with BV risk, and too few girls reported current sexual activity to find any statistically significant association. These analyses were limited by low power: a difference of 20% in the prevalence of the exposure (vaginal sex ever) would be required to detect a significant difference at P < 0.05 with 80% power. More frequent washing of the external genitalia was significantly associated with an increased BV risk (≤1 wash per day versus >1 wash per day, OR 1.93, 95% CI 1.04–3.58, P = 0.04), and there was a trend of increased risk in smokers (OR 2.00, 95% CI 0.69–5.76, P = 0.20). A total of 15 (7.0%) girls were using some form of contraception, of which four (1.9%) were using hormonal methods.
Table 1. Association between risk factors and BV (Ison–Hay criteria, grades 3 and 4) in 213 adolescents
| <15 years||64 (71.1)||26 (28.9)|| || || |
| ≥15 years||82 (66.7)||41 (33.3)||1.11||0.83–1.49||0.49|
| No||8 (66.7)||4 (33.3)|| || || |
| Yes||135 (68.5)||62 (31.5)||0.92||0.27–3.17||0.89|
|Age of first period|
| <13 years||87 (68.5)||40 (31.5)|| || || |
| ≥13 years||56 (68.3)||26 (31.7)||1.01||0.56–1.83||0.97|
|Vaginal sex ever?|
| No||119 (68.9)||54 (31.2)|| || || |
| Yes||27 (67.5)||13 (32.5)||1.06||0.51–2.21||0.88|
|Number of partners|
| 0||119 (69.8)||54 (31.2)|| || || |
| 1||25 (67.6)||12 (32.4)||1.06||0.49–2.26||0.89|
| ≥2||2 (66.7)||1 (33.3)||1.1||0.10–12.40||0.94|
|Current sexual partner|
| No||136 (69.4)||60 (30.6)|| || || |
| Yes||10 (58.9)||7 (41.2)||1.59||0.58–4.37||0.37|
| No||135 (68.2)||63 (31.8)|| || || |
| Yes||11 (73.3)||4 (26.7)||0.78||0.24–2.54||0.68|
|Frequency washing external genitalia**|
| ≤1 per day||86 (74.4)||30 (25.9)|| || || |
| >1 per day||49 (59.7)||28 (40.3)||1.93||1.04–3.58||0.04|
| No||128 (69.2)||57 (30.8)|| || || |
| Yes||18 (66.7)||9 (33.3)||1.12||0.48–2.65||0.79|
| No||65 (68.4)||30 (31.6)|| || || |
| Yes||81 (68.6)||37 (31.4)||0.99||0.55–1.77||0.97|
|Self-reported discharge in last 2 weeks|
| No||13 (61.9)||8 (38.1)|| || || |
| Yes||133 (69.2)||59 (30.7)||0.72||0.28–1.83||0.49|
|No||137 (69.5)||60 (30.5)||2||0.69–5.76||0.2|
|Yes||8 (53.3)||7 (46.7)|| || || |
Conclusions and discussion
We report a point prevalence of BV of 31.5% (95% CI 25.3–37.7%) among adolescent girls living in a rural town in tropical Ecuador. This prevalence of BV was greater than expected for populations in Europe,1 but was reassuringly similar to the 27% reported in Peru.6 The reason for markedly different prevalence rates in different populations remains unknown. It could relate to genetic factors related to ethnicity, or more likely to behavioural and environmental factors. The proportion of women with BV was almost identical in subjects who reported previous sexual activity (32.5%) and those who did not (31.2%). Although a questionnaire may be subject to bias if participants are not confident of their confidentiality, care was taken to reassure subjects that it would be anonymous. If there is ascertainment bias it is difficult to understand how it would be so systematic that girls with BV would falsely deny sexual activity, and those without BV would falsely declare previous sexual activity. That being the case, at least a trend for an association between BV and sexual activity should be detected if there really is such an association. The 95% CIs are 0.51–2.21, so we lack the statistical power to exclude a two-fold increased risk for any sexual activity. Our data is therefore consistent with the findings of previous studies that have observed a similar rate of BV in sexually active and virgin women.3,4 Future studies should obtain a more detailed history of any digital or genital–genital contact, following the methodology of Tabrizi and colleagues.5
A significant risk factor for BV in this study is frequent (more than once a day) genital washing, which was associated with an increased rate of BV. This is consistent with the literature associating douching and other hygiene measures with BV. A cross-sectional study cannot demonstrate cause and effect, so it is not known if having BV leads to more frequent washing or vice versa. It is interesting that the rate of self-reported vaginal discharge was so high, but did not correlate with a higher rate of BV. We did not correlate this symptom with candidiasis, which is a clinical diagnosis, but improved education about what is normal and abnormal, and about avoiding hygiene methods associated with BV could reduce the rate, and improve overall sexual health.
A limitation of this study was its relatively small size and limited power to detect significant differences for risk factors. Nevertheless, there appears to be little doubt that BV occurs commonly in virgin adolescent girls, making the STI hypothesis less easy to maintain. Current sexual activity might affect the vaginal environment in a way that increases the likelihood of BV, but we had insufficient numbers of such subjects to draw meaningful conclusions.
Bacterial vaginosis is an important public health problem because of its association with the acquisition of STIs, including HIV, and its association with adverse pregnancy outcomes.1 Recurrent symptomatic BV is also a cause of considerable distress to the affected women. The challenge remains how to prevent BV and reduce the prevalence, both for potential reduction in complications and improving the lives of those with the most severe symptoms from recurrent BV.
Disclosure of interests
Contribution to authorship
MV led the trial team of IG, SE, KT and MEC, who were responsible for recruiting subjects, obtaining consent, questionnaires and taking swabs.
PJC and PH conceived the study, wrote the protocol and drafted the paper. PJC performed the statistical analysis. PH read the Gram stained smears.
Details of ethics approval
The study protocol was approved by the Ethics Committee of the Hospital Pedro Vicente Maldonado.
Courtyard Clinic Research Fund.
The authors thank the girls, parents, and teachers at the ‘Colegio Rosa Zarate’ and ‘Colegio Nacional La Concordia’ in La Concordia for their help in conducting the study. The technical advice of Carlos Sandoval of the clinical laboratory at the Hospital Pedro Vicente Maldonado is also gratefully acknowledged.