Bacterial vaginosis (BV) is the most common cause of vaginal symptoms prompting women to seek medical care and is widely prevalent, affecting 29% of women in the United States.1 Normal vaginal flora is predominated by hydrogen peroxide-producing lactobacillus species.2 BV is characterized by a decline in vaginal lactobacilli with an increase in the number of Gram-negative anaerobes such as Gardnerella vaginalis, Atopobium vaginae, and Prevotella species.3,4 Risk factors for BV include black race, sexual activity, douching, and cigarette smoking.1,5,6 BV has been associated with adverse reproductive health outcomes including an increased risk of sexually transmitted infections (STIs), pelvic inflammatory disease, and gynecologic postoperative infections.7–12
BV is diagnosed clinically by the presence of 3 of the 4 following criteria described by Amsel et al: homogenous vaginal discharge, vaginal pH >4.5, amine odor upon addition of potassium hydroxide, and the presence of clue cells on saline microscopy >20% per high powered field.13 In general, in research studies, the Nugent score based on Gram stain of a vaginal smear is used rather than clinical criteria; a score of 0 to 3 is consistent with normal flora, 4 to 6 intermediate flora, and ≥7 is diagnostic of BV.14 Self-obtained vaginal swabs (SOVS) have been shown to be acceptable to women, and slide preparation by study participants has been shown to be reliable for subsequent Gram stain and Nugent score.15
Previous studies have shown that combined oral contraceptives (COC) and progestin-only contraceptives, as well as condom use, are protective against BV.16–18 The relationship between BV and IUD use is less clear; some studies have shown an increase in the risk of BV among IUD users,19,20 whereas other studies have not found any increase in risk.18,21,22 A longitudinal study comparing the incidence of BV in IUD users to COC users found an almost 3-fold increase in BV among IUD users.23 Limitations of this study include no description of the type(s) of IUDs used and comparison of new IUD users to established COC users. Possible explanations for an association between IUD use and BV acquisition include increased anaerobic bacteria in the vagina because of the presence of a string in the vagina or a foreign body in the uterus, or increased volume and duration of menstrual flow in users of the copper IUD.23,24 We hypothesized that among women initiating contraception, the incidence of BV would be higher among new users of the IUD users compared with new users of COC, ring, and patch.
MATERIALS AND METHODS
We conducted a longitudinal study of BV acquisition over the initial 6-months of contraceptive use in women who selected the intrauterine device (IUD) compared with women who selected COC, the vaginal ring, or the contraceptive patch. This study was conducted as a substudy of a larger, ongoing contraceptive study, the Contraceptive CHOICE Project (CHOICE). This is a prospective cohort study designed to promote the use of long-acting reversible methods of contraception and to evaluate use, satisfaction, and continuation of reversible contraceptive methods. The methods of CHOICE have been described in detail elsewhere.24 Separate approval was obtained from the Washington University School of Medicine Human Research Protection Office before participant recruitment.
Women were eligible to participate in this study if they were between the ages of 18 and 45 years, enrolling into CHOICE and initiating a copper or levonorgestrel IUD, COC, the contraceptive vaginal ring, or the transdermal contraceptive patch. Exclusion criteria included current pregnancy, history of a vaginal infection in the past 30 days, or BV at the time of enrollment by Amsel criteria or Nugent score. Other inclusion criteria included English-speaking, willing and able to provide informed consent, and willing and able to comply with the study protocol. All CHOICE participants were asked to collect the SOVS as part of the CHOICE enrollment process. We used the SOVS to screen eligible women for BV using modified Amsel criteria in which a clinical diagnosis of BV was made if 2 out of 3 of the following criteria were present: vaginal pH >4.5, positive “whiff test” which is a release of amines from vaginal fluid upon the additional of potassium hydroxide, and presence of >20% clue cells per high powered field on saline microscopy.25 If the participant was negative for BV, she was then approached for participation in the study by a member of the research team. A baseline vaginal smear was also obtained for Gram stain and Nugent score. Women who were enrolled and subsequently found to have a BV by Nugent score (score ≥7) were excluded from the analysis.
All women who agreed to participate in the study provided written informed consent and completed a baseline questionnaire collecting demographic information as well as sexual history, hygiene, and infection history. We gave participants detailed instructions on how to perform SOVS and to make a smear on a glass slide for subsequent Gram stain and Nugent score. Each participant was provided with a study kit containing a calendar, swabs, slides, instructions, and prepaid mailing materials.
Each month, for the next 5 months, participants collected a SOVS and prepared a slide at home. They also completed a questionnaire about vaginal symptoms in the preceding month, douching habits, recent treatment for STIs or symptomatic vaginal discharge, vaginal bleeding, and new sexual partners. The slides and questionnaire were then mailed back to the research site. Participants also completed a final 6-month in-person visit at our university-based research clinic where they collected a final SOVS, prepared a slide for Gram stain, and completed a final written questionnaire. Participants were provided with an incentive for participation.
After receiving the participant slides, we performed Gram stain and microscopy to determine the Nugent score for each slide as has been previously described.14 All the slides were Gram-stained and scored by a single member of the research team. Ten percent of the slides were double-read and scored by a second member of the research team. There was high concordance between the 2 readers with a κ of 0.81.
The Nugent score was not used for clinical care and women were not treated for BV diagnosed during the study on the basis of their Nugent score alone. If during the study a participant had complaints of vaginal symptoms, she was referred for gynecologic care.
On the basis of results of published studies, we anticipated a BV incidence of 10% among COC, ring, and patch users and 30% among IUD users.23 We calculated that a total of 144 women (72 in each group) would be required to find a statistically significant difference in the rate of BV between the 2 groups with an α of 0.05 and 80% power. Assuming a 20% loss-to-follow-up, we planned to enroll 84 women into each group for a total of 168 participants.
Demographic characteristics were described using frequencies, percentages, and means. Baseline differences between the IUD users and COC, ring, and patch users were analyzed using chi-square and t-test and Fisher exact tests as appropriate. In addition, on the basis of previous studies, we hypothesized that several variables would be associated with BV acquisition including age, race, condom use, and douching. We planned to include these variables in our final model regardless of statistical significance. For condom use and douching, we collected information about baseline use. For variables where we collected the information at multiple time points, we created a second variable to evaluate use over the 6-month study period on the basis of whether the participant reported the behavior on any follow-up survey. We hypothesized that women with intermediate flora at baseline would also be more likely to acquire BV over the study period. Cox Proportional Hazards regression was performed using these variables to estimate the hazards ratio (HR) and 95% confidence interval (CI) for the association between demographic and behavioral characteristics and contraceptive method use and acquisition of BV. Proportional hazards assumptions were tested and found to be appropriate. Interaction terms between covariates significant in the univariate model were tested and found not to be statistically significant. All analyses were performed using in Stata 11 (StataCorp LP, College Station, TX).
We recruited and enrolled 157 women into the study. There were 153 women who were negative for BV at baseline by both clinical criteria and Nugent score. Of these, 90 (58.8%) were new IUD users and 63 (41.2%) were new users of COC (16.3%), ring (20.3%), or patch (4.6%). Among the 90 IUD users, 59 (65.6%) chose the levonorgestrel intrauterine system (LNG-IUS) and 31 (34.4%) chose the copper IUD. As a result of slower-than-expected recruitment, we were unable to reach our target enrollment of 168 women. In addition, because of the high rate of IUD uptake in CHOICE (>50%), there was a greater proportion of IUD users resulting in unequal distribution between the 2 groups.
Over the course of the study, each woman could submit a maximum of 6 slides. The mean number of slides submitted was 3.7 (SD: 2.1). There were 19 (12.4%) women who did not submit any slides or complete any questionnaires over the study period; 4 of these women withdrew from the study (3 women cited lack of time and 1 woman withdrew from the parent study). An additional participant became pregnant after 4 months of participation and was withdrawn from the study at this time as pregnancy was a criterion for exclusion. At least 2 slides were available for 134 (87.6%) women. There were 120 (78.4%) women who returned 2 or more slides and 49 (32.0%) women who submitted all 6 slides. One hundred and fourteen (74.0%) women returned for the 6-month in-person visit, and Gram stain results were available for 104 women (68.0%). Ten slides were not scored due to discarded or damaged slides or an inadequate sample.
Baseline characteristics of women who contributed 2 or more slides for Gram stain are shown in Table 1 by contraceptive method. Women in the IUD group were slighter older (mean age, 26.7 years) compared with COC, ring, and patch users (mean age, 24.4 years; P = 0.01). Race, ethnicity, educational level, marital status, insurance coverage, and low SES status were similar between groups. The groups did not differ in previously described risk factors for BV such as number of mean number of lifetime sexual partners, history of douching ever or in the 6 months before enrollment, current smoker, tampon use, or history of BV or a STI. IUD users were almost twice as likely to have intermediate flora at baseline (18.4% vs. 10.3%, P = 0.19), although this was not a statistically significant difference.
COC, ring, and patch users were more likely to report condom use at baseline and over the study period compared with women using IUDs. At baseline, 58.6% of COC, ring, and patch users reported “always” or “almost always” using condoms compared with 36.8% of IUD users (P = 0.05). We also measured condom use longitudinally; over the study period, 24.4% of IUD users and 54.0% of COC, ring, and patch users reported “always” or “almost always” using condoms at one or more time points (P = 0.001) and only 3.3% of IUD users and 27.0% of COC, ring, and patch users reported “always” or “almost always” using condoms at every time point (P < 0.001). Self-report of condom use at every time point was associated with a reduction in incident BV, although this was not a statistically significant finding (HR: 0.69, 95% CI 0.24–1.96).
Over the 6-month study period, there were 35 new cases of BV with an incidence of 29.2%. The incidence of BV was higher among IUD users at 37.0% compared with 19.3% in COC, ring, and patch users (P = 0.03). The incidence of BV did not differ significantly among COC, ring and patch users (15.0% vs. 16.7% vs. 22.4%, respectively, P = 0.24). There was a trend towards increased BV incidence among LNG-IUS users (41.8% compared with 28.0% of copper IUD users); however, this was not a statistically significant finding (P = 0.23). The majority of cases of BV were asymptomatic; only 23.6% of women with a Nugent score ≥7 reported vaginal symptoms at the time of slide collection.
Table 2 shows the results of the Cox proportional hazards model; black race, intermediate vaginal flora at baseline, irregular bleeding over the study period, and IUD use were associated with an increased acquisition of BV in the univariate analysis. Condom use, douching, smoking, or new sexual partner over the 6-month study period were not associated with BV acquisition in the proportional hazards model.
In our multivariable model (also shown in Table 2), we adjusted for the above covariates as well as age, baseline condom use, and douching in the 6 months before enrollment. In the adjusted model, neither IUD use nor race remained significantly associated with BV. Women with intermediate vaginal flora at baseline were more than 3 times as likely to develop BV (HR: 3.15, 95% CI: 1.46–6.80) and women who reported any irregular bleeding during the study period were more than twice as likely to acquire BV (HR: 2.60, 95% CI: 1.04–6.50).
At baseline, women in the IUD group were more likely to have intermediate flora than women in the COC, ring, and patch group. Women with intermediate flora were more likely to acquire BV regardless of contraceptive method chosen; 60% of IUD users with intermediate flora acquired BV over the study period compared with 67% of COC, ring, and patch users, which is shown in Table 3. Women in the IUD group were also more likely to report irregular bleeding over the study period than women in the COC, ring, and patch group (73.8% vs. 42.5%; P < 0.01). Irregular bleeding was more common among LNG-IUS users compared with copper IUD users (41.9% vs. 30.5%), although this was not a statistically significant difference (P = 0.28). After stratifying by irregular bleeding, we found that, in the absence of irregular bleeding, there was no increase in the risk of BV among IUD users (HR: 01.02, 95% CI: 0.20–5.24) compared with COC, ring, and patch users. However, IUD users who reported irregular bleeding had an increased risk of BV (HR: 1.83, 95% CI: 0.69–4.81), although this difference was not statistically significant because of the small sample size in each IUD subgroup.
In this study, we found an association between IUD use and BV in our crude analysis. However, after controlling for confounders, the effect size of the HR was greatly attenuated and was no longer statistically significant. There were 2 factors that emerged as stronger predictors of BV acquisition; intermediate vaginal flora at the time of contraceptive initiation and irregular vaginal bleeding during the first 6-months of contraceptive method use. Stratification by irregular bleeding showed no association between IUD use and BV in the absence of irregular bleeding. However, IUD users with irregular bleeding were almost twice as likely to develop BV. This finding was not statistically significant; however, our study was not powered to find a statistically significant difference in these small subgroups. The difference in the effect size of the HRs for BV acquisition in IUD users with and without irregular bleeding suggests that irregular bleeding may be in the causal pathway for an association between IUD use and BV.
This association between irregular bleeding and BV was an unanticipated finding of our study. There are several potential mechanisms by which irregular bleeding could increase the risk of BV acquisition. Blood has a neutral pH which raises the pH of the normally acidic vagina. Association between menses and BV recurrence has been described26 with an increase in anaerobic bacteria and a decrease in lactobacilli.27 In addition, lactobacilli agglutinate to red blood cells which may result in a decreased vaginal lactobacillus concentration in women with frequent or persistent uterine bleeding.28 Alternatively, the irregular bleeding could be secondary to endometritis resulting from IUD placement in a woman with BV. In our study, we found that the incidence of BV was higher at each time point in women who reported irregular bleeding over the past 30 days than in women without irregular bleeding (data not shown), which suggests that the irregular bleeding may have preceded the change in vaginal flora. As irregular bleeding is a frequent side effect of hormonal contraception and is particularly common during the first 6 months of LNG-IUS use, this finding may be important for patient counseling. The observed association between intermediate flora and BV acquisition may be due to rapid fluctuation of the vaginal microbiota, which has been described in women who acquire BV.29
Our study had several strength, including a prospective longitudinal design which allowed us to evaluate incident BV. In addition, our participants were all new users of their contraceptive method avoiding a comparison between new users and long-term users. Our cohort was a racially and socioeconomically diverse cohort increasing the generalizability of our findings.
It is important to recognize that this study has several limitations. Given our low rate of participant follow-up, we may have underestimated the true incidence of BV in our study population. However, there were no statistically significant differences in the demographics and reproductive characteristics between women lost to follow-up and the other participants (data not shown), suggesting that loss-to-follow-up was not differential. In addition, we had limited power to evaluate our outcomes of interest and to perform a stratified analysis to investigate the effect of irregular bleeding on incident BV. Another limitation is that we compared women using IUDs with women using COC, ring, and patch. Nugent scores have been shown to be similar in users of COC and the vaginal ring25; however, hormonal contraceptives have a protective effect against BV which may have overestimated the effect of IUD use on incident BV. In our study, the incidence of BV was similar among COC, ring, and patch users, suggesting no difference in the effect on vaginal flora between these combined hormonal contraceptive methods. Identification of an appropriate comparison group among women using contraception is challenging because reproductive-age women not using contraception are often seeking pregnancy or are demographically distinct from women using contraception. However, given the high prevalence of COC use in the United States, COC users may be the most appropriate comparison group.30
Irregular bleeding secondary to hormonal contraception usually improves over time; therefore, the association between IUD use and BV may be greatest during the initial use of the method. Further studies with a longer duration of follow-up are needed to investigate whether this association persists overtime as the irregular bleeding secondary to hormonal contraception resolves.
1. Allsworth JE, Peipert JF. Prevalence of bacterial vaginosis: 2001–2004 National Health and Nutrition Examination Survey data. Obstet Gynecol 2007; 109:114–120.
2. Hillier SL. Normal vaginal flora. In: Sexually Transmitted Diseases. 3rd ed. New York, NY: McGraw Hill, 1999.
3. Fredricks DN, Fiedler TL, Marrazzo JM. Molecular identification of bacteria associated with bacterial vaginosis. N Engl J Med 2005; 353:1899–1911.
4. Sobel JD. Bacterial vaginosis. Annu Rev Med 2000; 51:349–356.
5. Cherpes TL, Marrazzo JM, Cosentino LA, et al.. Hormonal contraceptive use modulates the local inflammatory response to bacterial vaginosis. Sex Transm Infect 2008; 84:57–61.
6. Brotman RM, Klebanoff MA, Nansel TR, et al.. A longitudinal study of vaginal douching and bacterial vaginosis—a marginal structural modeling analysis. Am J Epidemiol 2008; 168:188–196.
7. Brotman RM, Klebanoff MA, Nansel TR, et al.. Bacterial vaginosis assessed by gram stain and diminished colonization resistance to incident gonococcal, chlamydial, and trichomonal genital infection. J Infect Dis 2010; 202:1907–1915.
8. Allsworth JE, Peipert JF. Severity of bacterial vaginosis and the risk of sexually transmitted infection. Am J Obstet Gynecol 2011; 205:113.e1–6.
9. Moodley P, Connolly C, Sturm AW. Interrelationships among human immunodeficiency virus type 1 infection, bacterial vaginosis, trichomoniasis, and the presence of yeasts. J Infect Dis 2002; 185:69–73.
10. Haggerty CL, Hillier SL, Bass DC, et al.. Bacterial vaginosis and anaerobic bacteria are associated with endometritis. Clin Infect Dis 2004; 39:990–995.
11. Evans BA, Kell PD, Bond RA, et al.. Predictors of seropositivity to herpes simplex virus type 2 in women. Int J STD AIDS 2003; 14:30–36.
12. Koumans EH, Kendrick JS. Preventing adverse sequelae of bacterial vaginosis: a public health program and research agenda. Sex Transm Dis 2001; 28:292–297.
13. Amsel R, Totten PA, Spiegel CA, et al.. Nonspecific vaginitis. Diagnostic criteria and microbial and epidemiologic associations. Am J Med 1983; 74:14–22.
14. Nugent RP, Krohn MA, Hillier SL. Reliability of diagnosing bacterial vaginosis is improved by a standardized method of gram stain interpretation. J Clin Microbiol 1991; 29:297–301.
15. 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; 24:236–239.
16. Koumans EH, Sternberg M, Bruce C, et al.. The prevalence of bacterial vaginosis in the United States, 2001–2004; Associations with symptoms, sexual behaviors, and reproductive health. Sex Transm Dis 2007; 34:864–869.
17. 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; 34:954–959.
18. Shoubnikova M, Hellberg D, Nilsson S, et al.. Contraceptive use in women with bacterial vaginosis. Contraception 1997; 55:355–358.
19. Moi H. Prevalence of bacterial vaginosis and its association with genital infections, inflammation, and contraceptive methods in women attending sexually transmitted disease and primary health clinics. Int J STD AIDS 1990; 1:86–94.
20. Joesoef MR, Karundeng A, Runtupalit C, et al.. High rate of bacterial vaginosis among women with intrauterine devices in Manado, Indonesia. Contraception 2001; 64:169–172.
21. Lessard T, Simoes JA, Discacciati MG, et al.. Cytological evaluation and investigation of the vaginal flora of long-term users of the levonorgestrel-releasing intrauterine system (LNG-IUS). Contraception 2008; 77:30–33.
22. Donders GG, Berger J, Heuninckx H, Bellen G, et al.. Vaginal flora changes on Pap smears after insertion of levonorgestrel-releasing intrauterine device. Contraception 2011; 83:352–356.
23. Avonts D, Sercu M, Heyerick P, et al.. Incidence of uncomplicated genital infections in women using oral contraception or an intrauterine device: a prospective study. Sex Transm Dis 1990; 17:23–29.
24. Demirezen S, Kucuk A, Beksac MS. The association between copper containing IUCD and bacterial vaginosis. Cent Eur J Public Health 2006; 14:138–140.
25. Veres S, Miller L, Burington B. A comparison between the vaginal ring and oral contraceptives. Obstet Gynecol 2004; 104:555–563.