Hormonal contraception requires the use of a progestin, and many formulations also contain ethinyl estradiol (E2). These hormones can be delivered by several routes, with vaginal administration and even vaginal administration of an oral contraceptive (OC) being effective and often reducing hormonal side effects.1 Because daily administration can reduce method compliance, placement of a vaginal ring for 3 weeks will allow a constant diffusion of the contraceptive steroids from the reservoir of the ring into the bloodstream. NuvaRing (Organon USA Inc., Roseland, NJ) is a vaginal ring approved for contraceptive use in the United States. This flexible ethylene vinyl acetate copolymer ring is 54 mm in diameter and 4 mm in thickness and releases 120 μg etonogestrel and 15 μg ethinyl E2 daily.2 The bioavailability of the etonogestrel with ring use was greater than an oral contraceptive,2 and this ring produced ovulation suppression in 100% of subjects studied.3 Systemic ethinyl E2 exposure was approximately 50% less than a 30-μg ethinyl E2 oral contraceptive,2 and although vaginal ethinyl E2 concentrations are unknown, they are possibly greater than the systemic levels.
Combination contraceptive vaginal rings can reduce the irregular bleeding and vaginal thinning seen with progestin-only vaginal rings.4,5 During the evolution of the contraceptive ring, ring design changed to smaller, more flexible models, resulting in a decrease in the incidence of genital complaints and foreign body sensation.6,7 The use of a ring to deliver estrogen replacement locally in menopausal women is known to be effective in improving vaginal epithelial maturation and thickness and lubrication and has high patient acceptability.8 Colposcopy studies have not detected more than an occasional area of erythema with the use of combined contraceptive or estrogen-only rings and no evidence of vaginal injury.8–10
Despite evidence for the positive effects of estrogen-releasing vaginal rings and lack of evidence that they may cause injury to vaginal epithelium, several studies list vulvovaginal symptoms as frequent side effects.4,6 In one study, the most common adverse event, with 21% to 27% of subjects reporting it, was vaginal discharge occurring regardless of ring diameter.6 In a large contraceptive trial4 with NuvaRing, of the possible treatment-related adverse events, 5.6% were vaginitis and 4.8% were leukorrhea. No laboratory confirmation or more specific diagnoses were given to these events.
It is not clear whether it is the contraceptive vaginal ring or the use of contraceptive hormones in a sexually active population not using a barrier method that is responsible for causing these symptoms. We examined the genital symptoms and examination and laboratory findings with ring and oral contraceptive use.
MATERIALS AND METHODS
The study was approved by the University of Washington Institutional Review Board, and women were recruited by flyer and newspaper advertisement for participation in a study involving a new type of contraception. Women were screened by telephone for exclusions before clinic visit. The study was conducted at a University-affiliated, metropolitan, dedicated research clinic.
Subjects were excluded for age less than 18 years or greater than 45 years; standard contraindications to combined oral contraceptive use; inability to speak or read English; current use of a contraceptive implant or intrauterine device; contraceptive injection use in the previous 6 months; and current diagnosis of uterine infection, fibroid, or cervical dysplasia. Women were also excluded for having risk factors for developing genital infections, such as diabetes, chronic use of immune suppressors like corticosteroids, autoimmune disorders such as human immunodeficiency virus, a personal history of greater than 3 genital yeast infections treated by a health care provider in the prior 12 months, more than 3 symptomatic genital herpes outbreaks in the prior 12 months, or a diagnosed and treated sexually transmitted infection in the prior 12 months. Abnormal pelvic examination or severely abnormal cervical cytology at baseline was a reason for exclusion or early exit.
Written informed consent was obtained. At the first in-person visit, a complete medical and gynecologic history was taken. A standardized questionnaire was administered to collect demographic, interpersonal, and intrapersonal data, including questions regarding frequency of sexual intercourse, contraception use, genital hygiene habits such as douching, underwear or clothing type, wearing underwear for sleep, bubble baths, scented soap use, and genital waxing or shaving. They were asked by questionnaire, “Do you think you have more vaginal discharge than you should have?” and “Do you think you have more vaginal odor than you should have?”
At enrollment and each subsequent visit, subjects had blood pressure and weight measurements and a complete gynecologic examination. Genital examination findings were recorded on standardized forms and included external and internal genital appearance, absence of vaginal erosions or loss of epithelial integrity, and vaginal discharge assessment (color, consistency, and amount), and cervical ectopy was scored by percent of glandular epithelium estimated to cover the surface of the ectocervix. A vaginal discharge sample was collected from the right vaginal wall with a sterile, single use, amine-controlled swab (Hardwood Products Inc, Guilford, ME). The moist swab was applied to pH paper (ColorpHast indicator strips, EM Science, Gibbstown, NJ), and results detected by a color change were recorded as a numeric pH result. A second swab was used to collect vaginal discharge for saline wet mount preparation to perform microscopy for the detection of trichomonads or clue cells. This swab was then applied to a glass slide with potassium hydroxide (KOH) solution for KOH microscopy for yeast, and a fishy or amine odor detected by whiff of this swab was recorded as a positive result.
A sample from the left vaginal wall was then applied to a glass slide, allowed to air dry, and sent to an off-site research facility for Gram stain and analysis by a single microbiologist who was unaware of study drug assignment. The standardized scoring system used averaged the number of bacterial morphotypes of Lactobacillus, Gardnerella, Bacteroides, and Mobiluncus seen per 400x field.11 A Nugent score of 0–3 was given for normal flora, 4–6 for disrupted and intermediate flora, and 7–10 was consistent with bacterial vaginosis.11 Additionally, the presence of white blood cells was determined and recorded as the mean number of white blood cells counted per 400× field.
For microbiology cultures, 2 sterile single-use Dacron swabs were used to sample discharge from the left vaginal wall, placed in anaerobic transport media (Port-A-Cul tube; BBL Microbiology Systems, Cockeysville, MD) and transported within 24 hours. At the microbiology laboratory, the samples were placed into 1.5 ml of phosphate buffered saline and vortexed, and 100 μl was used to inoculate the following media: Columbia agar with 5% sheep blood, Candida bacille Calmette-Guérin agar (PML Microbiologicals, Wilsonville, OR), human blood bilayer Tween agar (Remel Scientific, Lenexa, KS), and Rogosa selective Lactobacillus agar (prepared in-house). Serial 1:10 dilutions in phosphate buffered saline were carried out, and 100 μl from each dilution was inoculated onto sheep blood agar plates out to 1:10−9 dilution. The blood agar, Candida agar, and human blood bilayer Tween agar were incubated at 37°C in 5% to 10% CO2 for 48 hours. The Rogosa agar was incubated at 37°C in an anaerobic glove box (Sheldon Laboratories, Cornelius, OR) for 5 days. The plates were examined for the presence of Lactobacillus and yeast, with quantification noted. A wet mount was used to confirm the isolation of yeast.
Lactobacilli were identified by colony morphology and characteristic Gram stain.12 All Lactobacilli isolated were tested for production of hydrogen peroxide (H2O2) in a qualitative assay on a tetramethylbenzidine agar plate. After 3 days of incubation in an anaerobic glove box at 37°C, the plates were exposed to ambient air for up to 30 minutes and observed for a blue color from the hydrogen peroxide horseradish reaction.13
After all samples for laboratory testing were collected, a large, preweighed, nonsterile cotton swab (Puritan 8-inch Tippe OB/GYN Applicator, Hardwood Products Company Inc., Guilford, ME) was used to collect all vaginal discharge and moisture from the vaginal apex and fornices and exposed lower vaginal walls, without collection from the ectocervix, but including discharge adherent to the speculum after removal. This “wet” swab was then reweighed to determine vaginal discharge weight. This procedure was performed in a standardized way, repeated in the same manner for each subject, at each visit.
Cervical cytology sampling was performed using smear fixation and routine staining by an outside vendor at enrollment and exit visits only. Urine was collected for gonorrhea and Chlamydia DNA probe testing for all subjects at enrollment and later if indicated. Urine pregnancy testing was also performed if the subject was thought to be at risk for pregnancy by the provider at any study visit.
At the time of baseline pelvic examination, subjects were taught vaginal ring use and their ability to insert the vaginal ring was recorded on a 4-point scale from easily learned with first insertion to could not insert without help. Subjects were also examined with the ring in place during a Valsalva maneuver to assess for pelvic prolapse and ring expulsion.
An off-site institutional research pharmacy supplied the randomization schedule and allocation verification. A computerized random number generator program with balanced blocks of ten (concealed from clinical study staff performing the enrollment) was used to randomly assign subjects to ring-first or pill-first assignment. Opaque randomization envelopes were labeled with the sequential randomization number and contained study assignment information for dispensing of the assigned study drug. There was no masking following randomization. Subjects were dispensed NuvaRing or Alesse (20 μg ethinyl E2 and 100 μg levonorgestrel; Wyeth-Ayerst Laboratories, St. David, PA) according to the allocation schedule. Subjects were instructed on medication use as labeled in the package insert, which included leaving the ring in place within the vagina at all times for the 21 days of use followed by a ring-free week.
There were 3 return visits planned: the first during cycle 2; the second during cycle 4; and to allow diary completion, the method used during cycle 6 was used for cycle 7, with the last revisit also serving as the exit visit. At all revisits the pelvic examination and laboratory assessments were repeated. The provider performing the pelvic examination was not blinded to method assignment because the ring was left in place during the examination to prevent disruption of vaginal discharge for accurate measurement. If a patient reported symptoms of a genital infection or reported treatment, these events were investigated, and if possible, laboratory testing or documentation was obtained. Symptomatic patients who were found to be positive for bacterial vaginosis by clinical criteria or yeast by KOH preparation at the visit were offered treatment for their condition.
Subjects completed a daily diary for the entire six months, responding to the following questions: use of study drug, vaginal spotting and bleeding, hygiene product use, sexual activity, and six genital symptoms. The subjects scored their genital symptoms on a scale of zero to four, with 0 = “no problem/normal,” 1 = “mild,” 2 = “moderate,” 3 = “worse,” and 4 = “severely abnormal problem” for the following symptoms: “vaginal wetness,” “vaginal odor,” “yellow colored discharge,” “vaginal discomfort/pain inside,” “vulvar itch,” and “vulvar discomfort/pain outside.”
A standardized questionnaire was administered at revisits to collect information regarding satisfaction with the method, willingness to recommend to other women, and general well being. If using the ring method, subjects were specifically queried regarding ring use and if the ring had been expelled or interfered with daily activities such as tampon use, defecation, urination, or if a sensation of slippage of the ring had occurred.
A partner questionnaire was distributed to subjects during ring use if she reported a current sexual partner whom she was willing to contact, and the male partner was asked to complete and return it anonymously in the mail. The questionnaire collected information regarding the amount of coitus with the ring in place, if the male partner could feel the vaginal ring during coitus, and the overall impression of the ring. At the exit visit, subjects were asked about the reason for exit and what contraceptive method they planned to use.
Symptom diary data from cycle days 1 through 28 were used for individual and total daily symptom score comparison by the method assigned. Symptom responses were cycle averages and averaged within subject for each method. If diary data from a 28-day cycle were incomplete then the scores for the data in-hand were extrapolated to a 28-day score. Data from examination and laboratory variables at revisit and exit were compared by subjects with a method in use for 14 or more days before and at the time of the visit.
Data were entered into a statistical spreadsheet. The study variables were entered as continuous variables or dichotomous variables as appropriate. Statistics were calculated using S-PLUS 6.1 (Insightful Corp., Seattle, WA) or SAS 8.02 (SAS Institute Inc., Cary, NC), χ 2 and Fisher exact tests were used to produce significance values for categorical differences and t tests adjusted for unequal variances were used for continuous variables. Estimated fold differences for count and ordinal outcomes were obtained using correlated response regression models (ie, generalized estimating equations with log link).14 Logistic regression was used for correlated binary responses to compute the P values for comparisons of the proportions positive for each subject with outcomes before and after crossover. Regressions for dual response per subject analyses used exchangeable working covariance matrices, and all others used independent covariance matrices. Empirical standard errors were used throughout.
Except where specifically stated otherwise, we used intent-to-treat analyses and included all collected data from all randomly assigned subjects even if the subject exited early or did not take the study drug as directed. All analyses were unadjusted with the exception of the laboratory findings (colony counts), which were adjusted for subject baseline value. The randomization and prospective crossover design allowed us to use the subjects as their own controls for comparison.
Eighty women were enrolled and randomly assigned during the 6 months ending May of 2002, with data collection complete in December of 2002. Study completion rates by study assignment and reasons for study exclusion, nonrandomization, and early exit are listed on Figure 1. Of the 80 women randomly assigned, 64 (80%) completed the entire 6 months of observation, with no significant difference between study arms and only 1 dropout for ring-related discomfort that occurred after only 3 days of use. Baseline demographics and hygiene habits are shown on Table 1 and Table 2, respectively, compared by study arm assignment, and other than smoking, the two groups were similar. All statistical analyses were evaluated for potential confounding due to smoking, and because it had no marked effect on estimates, confidence intervals, or P values, it is not included as an adjustment to the reported results.
On medical history for the entire group, 7.5% of women reported a history of Chlamydia, 10% bacterial vaginosis, 7.5% genital herpes, 2.5% pelvic inflammatory disease, 55% urinary tract infection, 59% genital yeast infection, and 43% had self-treated at least once with an over-the-counter yeast medication. These rates of genital infection history were similar between the 2 arms and not significantly different. At baseline, among the 35 subjects who reported having coitus at least once per week, 10 (28.6%) were positive for yeast by culture compared with 2 (4.4%) of the 45 women having coitus less than once per week (P = .004). Prior OC use and baseline hygiene habits, individually and collectively, did not significantly predict positive yeast cultures at baseline.
The laboratory and examination findings, compared by baseline and method used for 14 or more days at revisit are presented in Table 3. Laboratory comparisons include all revisit 1 (precrossover) and exit visit (postcrossover) data. Revisit 2 observations were excluded because 68% occurred within 14 days of crossover. The results are not materially altered by either the inclusion of revisit 2 or the removal of the baseline adjustment. The average number of days using the study method for revisit 1 were 39 ± 6 and for the exit visit 93 ± 17.
At baseline, 15% (12/80) of women had yeast on culture; during ring use 18.8% (13/69) and with OC use 22.5% (16/71) of samples were positive for yeast by culture (P = .12). At baseline, 83.8% (67/80) of women were positive for any Lactobacillus by culture, and at subsequent visits this percentage was similar and no different by method (P = .28). However, the concentration of Lactobacillus colony-forming units positive for hydrogen peroxide (H202) production increased during ring use (fold difference 2.67, confidence interval 1.49, 4.78; P < .001) and increased over baseline values. All other examination and laboratory findings were not significantly different, including vaginal pH values and discharge weight.
There were no cases of Trichomonas encountered in this study. Bacterial vaginosis by Nugent's and Amsel's criteria scores were not significantly different between methods; however, positive tests were infrequent in both groups. During the study there were only 3 symptomatic women who were treated for bacterial vaginosis and all during OC use. There were also 10 urinary tract infections treated, with only 2 occurring during ring use. Yeast vulvovaginitis was the most common clinical genital diagnosis, and 24 women were either treated by the study or self-medicated; however, only 14 of these were confirmed by KOH or culture, and 7 occurred during ring use.
Symptom diary scores for all subjects who reported diary data are averaged over the 3 cycles by assigned method in Table 4. The study flow chart (Fig. 1) shows the numbers of subjects who contributed diary data, although some subjects would complete visits but not return a diary. Seventy-eight subjects reported diary data for at least 1 ring cycle, and 75 subjects reported diary data for at least one OC cycle (Table 4). Ninety-six percent of diaries contained exactly 28 days of symptom reports, whereas 12 (less than 3%) contained less than 28 days, and 5 diaries contained between 31 to 35 days, due to accidental extended ring use.
Most symptoms were scored as mild, and a severe score was very rare and sporadic. Vaginal wetness had the highest symptom score among ring users, but the scores were still quite low considering they represent 3-cycle averages of 28-day totals (Table 4). There was a nonsignificant trend over the entire study of decreased reporting of symptoms, with on average 20% fewer total symptoms reported after crossover (P = .28), perhaps reflecting improved method tolerance over time for both methods. However, both the ring-first and pill-first subjects reported on average more vaginal wetness during ring use than during pill use (fold difference 2.74, 95% confidence interval 1.80, 4.18; P < .001). There was also a carry-over effect for total symptoms, with ring-first subjects reporting more symptoms with subsequent pill use than pill-first subjects (P = .045).
Women who reported more vaginal wetness did not differ on laboratory findings from women who did not report this symptom (data not shown). On examination, these women more frequently had observed ring descent during the baseline visit with bearing down; however this observation was not significant (P = .28). Of the 64 women who completed the study, 46 women reported that the ring never slipped and 6 reported ring slippage at least weekly or more. Having this complaint of “ring slippage” was not associated with an increase in vaginal wetness scores by diary or any examination or laboratory finding. Eleven women with 50% or greater of their ectocervix covered by glandular epithelium or cervical ectopy at baseline examination were more likely to report vaginal wetness in their diary and have yellow discharge seen at examination (P = .03). At revisit and exit examinations, this incidence of cervical ectopy remained unchanged over the 6-month study.
Of the 76 women who used the ring for at least one cycle, 54 had a partner and agreed to give the male questionnaire to a current partner, and 39 questionnaires were returned completed (72%). Seventy-two percent of men (28/39) never or rarely felt the ring during coitus, 92% (36/39) reported no change in sensation during intercourse, and 87% (34/39) did not feel the ring move during intercourse. Of the men who reported 10 or more coital events with ring use, 16% (4/24) reported that the ring would sometimes come out during coitus and 2 of these men thought that it interrupted lovemaking. Ninety-three percent of partners would definitely (19/39) or possibly (17/39) recommend the ring to other couples. Only three men responded that they would not recommend the ring as a method for other couples.
Women easily learned to use the ring. At the baseline visit, 84% of women could insert and remove the ring on their first attempt. None of the women were deemed unable to insert or remove the ring. On bearing down, none of the subjects were able to expel the device; however, almost all of the subjects had not had a vaginal delivery.
Of the 76 women who used the ring during the study, 26 reported noncompliance with the recommended schedule for ring use. Eight women reported missing more than two pills in a cycle and one subject became pregnant after missing three pills following the pill-free week. There were no other pregnancies during the study. We administered emergency contraceptive pills to two subjects who reported discovering their ring had been displaced without their knowledge during coitus. We then instructed all women to check for ring placement after coitus. Overall, of the 80 subjects, 67 reported at least 1 adverse event during the study. There were similar rates of headaches, gastrointestinal symptoms, respiratory symptoms, breast tenderness, and mood complaints during use of the 2 methods. There were no serious adverse events, although one pill-first subject was exited early after the diagnosis of symptomatic gallstones. Blood pressure and weight did not vary significantly between groups at baseline or over the study.
During ring use, there were fewer total bleeding days, and the number of intermenstrual bleeding days were an average of 0.2 days compared with 1.4 days with OC use (P < .001). Average levels of satisfaction reported at exit visit, on a scale of 1 to 5, where 1 = dissatisfied and 5 = best method, was 4.3 ± 0.9 for ring use and 3.6 ± 1.0 for OC use (P < .001). At the end of the study, 32 of the 64 women (50%) who completed the study planned to use the ring, compared with 18 women (28%) who planned to use pills, 4 women chose another method (6%), and 10 (16%) who chose no contraception. Of the 18 women who completed the entire study and had reported condom use as their primary method of contraception at study entry, 12 (66%) exited the study with a prescription for the ring.
We randomly assigned women to either cyclic use of the 20 μg ethinyl E2/100 μg levonorgestrel OC or the 15 μg ethinyl E2/120 μg etonogestrel contraceptive vaginal ring for 3 cycles, with a crossover to the medication that was not initially assigned for another 3 cycles. We recruited and studied primarily college-educated nulliparous women who were interested in a new contraceptive alternative, and many of our subjects were using the male condom for contraception at entry (Table 1). Over the 6-month study, we had good study retention (Figure 1) and a total of 4 pelvic examinations and daily symptom diary data for many of the subjects. Our study was small, and it was not powered to detect any specific outcome because for many genital conditions, like bacterial vaginosis, many asymptomatic women would need to be studied to reliably measure an effect. We were unable to correlate any hygiene habit to baseline genital pathology, such as yeast detected by culture (Table 2).
Women were more likely to report mild to moderate vaginal wetness as a symptom on the daily diary during ring use, compared with during OC use (Table 4). There were no differences in examination or laboratory findings, except for increased cervical ectopy at baseline, to predict the occurrence of this symptom and the finding of leukorrhea on examination. We did not detect any differences in vaginal discharge amount even with measurement by weight, and there were also no changes in Nugent Gram stain score or vaginal white cell number compared by method in use (Table 3).
The small difference in symptoms reported did not lead to method discontinuation, and at study exit more women chose to continue or restart the ring. Women also gave a higher method satisfaction score to the ring over the OC. Our daily diary system for collecting genital symptom data was able to measure differences between women and method in use; however, we would recommend in the future a pretreatment cycle be included to confirm the groups are similar at baseline. Women recorded fewer and less intense symptoms later in the study and this decline could represent recording fatigue or perhaps, with the ongoing use of a hormonal contraceptive, genital symptoms became less noticeable.
From the comparison of examination and laboratory findings by method, there is a single interesting and significant finding. Women who were using the ring had a significant 2.7-fold increase in the concentration of H2O2-producing Lactobacilli (Table 3). A converse finding of a reduction in H2O2-producing Lactobacilli has been reported with depo-medroxyprogesterone acetate use, and this was correlated with very low systemic E2 levels.15 The contraceptive vaginal ring use results in low systemic E2 exposure, but it is possible that there is a “uterine-first-pass” or some sort of preferential delivery of vaginally administered hormones to the genital tissues.16,17
It had been thought that perhaps the symptom of vaginal discharge with contraceptive vaginal ring use was caused by inflammation, and one early study did report a slight increase in vaginal white blood cells compared with pill use, although the bacterial flora did not change.18 Another study of vaginal flora found no difference between OC and combination cyclic vaginal ring users in numbers of bacteria.19 A later study using an etonogestrel ring similar to the one used in our study found that even with extended use there was no increase in vaginal white blood cells or changes in vaginal flora.20 These findings are similar to ours, although these previous studies did not measure H2O2 production in Lactobacillus populations. Therefore, a comparison with this finding is not possible. There has been no evidence that the use of the contraceptive vaginal ring induces bacterial vaginosis, and in fact, one abstract suggests that the Gram stain score may normalize for some women with use of the ring that we studied.9,10
The current use of an OC has been linked to symptomatic genital yeast infection.21,22 The biologic theory holds that estrogen promotes a glycogen-rich vaginal environment where Candida or yeast species can thrive on this substrate and give rise to symptomatic infection. Conversely, it is known that a hypoestrogenic state such as that induced by the use of depomedroxyprogesterone can lead to a loss of epithelial glycogen and a marked reduction in Candida.15,23 We did not see a significant increase in the percentage of women with yeast by culture, but we did see a nonsignificant increase in the estimated average concentrations of yeast with pill use (P = .08). Perhaps the oral contraceptive with enterohepatic circulation stimulates vaginal glycogen production more than vaginal absorption of an estrogen? The mechanism for the increase in H2O2-producing Lactobacillus population with ring use is not known. Is it moderated by an estrogen receptor responding to increased local concentrations of estrogen and not from an increase in substrate? But then what is the reason for the increase in wetness if it is not an increase in glycogen and the associated water? Only further study can elucidate this paradox.
Our study is limited by the small number of women studied and the short duration of use. It is important to note that the ability to generalize the findings is limited by our demographics. In particular, we chose to study only women without risks for or a history of frequent genital infections. This study provides some reassurance that an increase in vaginal wetness reported by a woman using the combination vaginal ring most likely does not represent pathology. Our finding of an increase in H2O2-producing Lactobacillus concentrations may even indicate a favorable effect on vaginal health.24 Increasing the concentration of H2O2-producing Lactobacillus has been shown to prevent viral and bacterial infections25 and is inversely associated with bacterial vaginosis.13 This increase in H2O2-producing Lactobacillus concentrations with combination vaginal ring use should be further explored.
1. Ziaei S, Rajaei L, Faghihzadeh S, Lamyian M. Comparative study and evaluation of side effects of low-dose contraceptive pills administered by the oral and vaginal route. Contraception 2002;65:329–31.
2. Timmer CJ, Mulders TM. Pharmacokinetics of etonogestrel and ethinylestradiol released from a combined contraceptive vaginal ring. Clin Pharmacokinet 2000;39:233–42.
3. Mulders TM, Dieben TO. Use of the novel combined contraceptive vaginal ring NuvaRing for ovulation inhibition. Fertil Steril 2001;75:865–70.
4. Dieben TO, Roumen FJ, Apter D. Efficacy, cycle control, and user acceptability of a novel combined contraceptive vaginal ring. Obstet Gynecol 2002;100:585–93.
5. Weisberg E, Fraser IS, Lacarra M, Mishell DR Jr., Alvarez F, Brache V, et al. Efficacy, bleeding patterns, and side effects of a 1-year contraceptive vaginal ring. Contraception 1999;59:311–8.
6. Roumen FJ, Dieben TO. Clinical acceptability of an ethylene-vinyl-acetate nonmedicated vaginal ring. Contraception 1999;59:59–62.
7. Weisberg E, Fraser IS, Baker J, Archer D, Landgren BM, Killick S, et al. A randomized comparison of the effects on vaginal and cervical epithelium of a placebo vaginal ring with non-use of a ring. Contraception 2000;62:83–9.
8. Speroff L, for the United States VR Investigator Group. Efficacy and tolerability of a novel estradiol vaginal ring for relief of menopausal symptoms. Obstet Gynecol 2003;102:823–34.
9. Roumen FJ, Boon ME, Velzen D, Dieben TO, Bennink HJ. The cervico-vaginal epithelium during 20 cycles’ use of a combined contraceptive vaginal ring. Hum Reprod 1996;11:2443–8.
10. Archer D, Raine T, Darney P, Alexander NJ. An open label noncomparative study to evaluate the vagina and cervix of NuvaRing users. Fertil Steril 2002;78:S25.
11. 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.
12. Eschenbach DA, Thwin SS, Patton DL, Hooton TM, Stapleton AE, Agnew K, et al. Influence of the normal menstrual cycle on vaginal tissue, discharge, and microflora. Clin Infect Dis 2000;30:901–7.
13. Eschenbach DA, Davick PR, Williams BL, Klebanoff SJ, Young-Smith K, Critchlow CM, et al. Prevalence of hydrogen peroxide-producing Lactobacillus species in normal women and women with bacterial vaginosis. J Clin Microbiol 1989;27:251–6.
14. Diggle P, Liang K-Y, Zeger SL. Analysis of longitudinal data. New York (NY): Oxford University Press;1994.
15. Miller L, Patton DL, Meier A, Thwin SS, Hooton TM, Eschenbach DA. Depomedroxyprogesterone-induced hypoestrogenism and changes in vaginal flora and epithelium. Obstet Gynecol 2000;96:431–9.
16. Cicinelli E, Rubini G, De Ziegler D, Barba B, Pinto V, Di Stefano MG, et al. Absorption and preferential vagina to uterus distribution after vaginal administration of 99m
Tc-pertechnetate in postmenopausal women. Fertil Steril 2001;76:1108–12.
17. Fanchin R, Righini C, Schonauer LM, Olivennes F, Filho JSC, Frydman R. Vaginal versus oral E2 administration: effects on endometrial thickness, uterine perfusion, and contractility. Fertil Steril 2001;76:994–8.
18. Schwan A, Ahren T, Victor A. The effects of contraceptive vaginal ring treatment on vaginal bacteriology and cytology. Contraception 1983;28:341–7.
19. Roy S, Wilkins J, Mishell DR Jr. Effect of a contraceptive vaginal ring and oral contraceptives on the vaginal flora. Contraception 1981;24:481–491.
20. Davies GC, Feng LX, Newton JR. The effects of a combined contraceptive vaginal ring releasing ethinyloestradiol and 3-ketodesogestrel on vaginal flora. Contraception 1992;45:511–8.
21. Geiger AM, Foxman B, Gillepsie BW. The epidemiology of vulvovaginal candidiasis among university students. Am J Pub Health 1995;85:1146–8.
22. Bauters TGM, Dhont MA, Temmerman MIL, Nelis HJ. Prevalence of vulvovaginal candidiasis and susceptibility to fluconazole in women. Am J Obstet Gynecol 2002;187:569–74.
23. Toppozada M, Onsy FA, Fares E, Amir S, Shaala S. The protective influences of progestogen only contraception against vaginal moniliasis. Contraception 1979;20:99–103.
24. Vallor AC, Antonio MAD, 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.
25. Smith SM, Bskin GB, Marx PA. Estrogen protects against vaginal transmission of Simian Immunodeficiency Virus. J Infect Dis 2000;182:708–15.