595 Effects of Contraceptive Method on the Vaginal Microbial Flora: A Prospective Evaluation Kalpana Gupta,1 Sharon L. Hillier,2 Thomas M. Hooton,1 Pacita L. Roberts,1 and Walter E. Stamm1 1 Department of Medicine, Division of Allergy and Infectious Diseases, University of Washington, Seattle; 2Department of Obstetrics, Gynecology, and Reproductive Sciences, University of Pittsburgh, Magee-Women’s Research Institute, Pittsburgh, Pennsylvania A prospective evaluation of 331 university women who were initiating use of oral contraceptive pills (OCPs), a cervical cap, diaphragm-spermicide, or other spermicidal methods was done to assess the effects of commonly used contraceptive methods on the vaginal flora. Vaginal introital cultures were performed at baseline and then weekly for 1 month. The prevalence of Escherichia coli vaginal colonization and of abnormal vaginal Gram stain scores (Nugent criteria) increased significantly among women using a cervical cap or diaphragmspermicide but not among women using OCPs. Women with E. coli colonization were significantly more likely to have an abnormal Nugent score and an absence of lactobacilli. In a multivariate model, use of spermicidal contraception in the preceding week was associated with an abnormal Nugent score and with colonization with E. coli, Enterococcus species, and anaerobic gram-negative rods. Thus, spermicidal methods of contraception are associated with alterations of the vaginal microflora that consequently may predispose women to genitourinary infections. The normal vaginal ecosystem is increasingly recognized as an important host defense mechanism against exogenous urogenital infection. Alteration of the normally Lactobacillusdominant flora has been associated with an increased risk of bacterial vaginosis (BV), gonorrhea, infection with human immunodeficiency virus, and Escherichia coli vaginal colonization, the critical step preceding urinary tract infection (UTI) [1–5]. The vaginal flora can be disrupted by a variety of factors, including sexual intercourse, antimicrobial use, and douching [1, 6, 7]. The effects of specific contraceptive methods on the vaginal flora, particularly those involving spermicide use, are not clear. Whereas some studies have shown a significant decrease in the prevalence of vaginal lactobacilli among diaphragm-spermicide users [6, 8] or among spermicide users alone [9–11], others have found no such effect [12, 13]. The effects of other birth control methods, such as oral contraceptive pills (OCPs) and the cervical cap, on the vaginal microbial ecology are even less well delineated. To assess the effects of these contraceptives and of sexual activity on the vaginal microflora, we Received 12 July 1999; revised 12 October 1999; electronically published 14 February 2000. Presented in part: annual meeting of the Infectious Diseases Society for Obstetrics and Gynecology, Barrier Creek, Colorado, 1996. Written informed consent was obtained from all participants, and human experimentation guidelines of the University of Washington Human Subjects Review Committee were followed. Grant support: National Institutes of Health (DK 47549 and DK 53369). Reprints or correspondence: Dr. Kalpana Gupta, 1959 NE Pacific St., University of Washington, HSB, BB1221, Box 356523, Seattle, WA 98195 ([email protected]). The Journal of Infectious Diseases 2000; 181:595–601 q 2000 by the Infectious Diseases Society of America. All rights reserved. 0022-1899/2000/18102-0024$02.00 prospectively studied a cohort of 331 healthy, sexually active university women who were initiating a new method of contraception. Methods Study population. Women included in the present study were participants in a larger 2-site prospective cohort study of risk factors for UTI in women [14]. Only women enrolled at 1 of the sites, the student health service at the University of Washington, were included in this aspect of the study. Complete details of the overall study design have been published [14]. Briefly, women between 18–40 years of age who were initiating a new method of birth control, were not pregnant, had <1 UTI in the previous year, had no chronic illnesses, and had no known anatomic or functional abnormalities of the urinary tract were eligible to participate. Women were evaluated at enrollment and then weekly for 1 month after initiation of their new contraceptive method. Vaginal introital cultures and detailed interviews were obtained at each of these visits. In addition, women maintained a daily diary, which was provided at enrollment to record occurrences of sexual intercourse, contraceptive use, antimicrobial use, menses, and genitourinary symptoms. Laboratory methods. Vaginal cultures were collected by inserting a sterile cotton swab ∼1 inch into the vagina and were transported to the laboratory in Amies transport medium. Semiquantitative cultures for aerobic gram-negative rods and cocci, anaerobic gram-negative rods, and lactobacilli were performed by using standard laboratory methods [15, 16]. Results were categorized semiquantitatively as 0 (no growth), 11 (colonies in the first streak zone), 21 (colonies in the second streak zone), 31 (colonies in the third streak zone), and 41 (colonies in the fourth streak zone). Vaginal smears were prepared by rolling a second vaginal 596 Gupta et al. swab across a microscope slide at the patient’s bedside and allowing it to air-dry. The smears were then Gram stained, and the flora were scored as normal (Nugent scores 0–3), intermediate (Nugent scores 4–6), or consistent with BV (Nugent scores 7–10), by use of criteria and methods described elsewhere [16, 17]. Statistical methods. Changes in vaginal flora were assessed by comparing the baseline visit culture results to the week 1 and week 4 culture results by use of McNemar’s test. Only women who completed the baseline visit and 4 weekly follow-up visits were included in this analysis. Women starting their birth control method before the baseline culture, reporting no episodes of vaginal intercourse during the study period, or using an alternative method of birth control in addition to their primary method were excluded. Condom use in conjunction with OCP use was allowed if the frequency of spermicide use was less than once per week. A second analysis examining the temporal relationship between vaginal microbial flora and vaginal intercourse and contraceptive use was performed by using logistic regression analysis. For the present analysis, each visit was analyzed separately and was correlated with sexual activity and contraceptive use in the week before the visit as ascertained from the patients’ diaries. Visits for which diary information in the preceding week was not available were excluded from this analysis. The analyses were adjusted for multiple observations per woman using the method of generalized estimating equations [18]. Results Of the 348 university women enrolled in the larger study, 331 (95%) met criteria for inclusion in the present analysis. A total of 1644 visits that yielded vaginal culture data were available from these women. The majority of the women were white (79%), were never married (85%), were secretors of blood group antigens (75%), and had no previous history of UTIs (72%). The mean age was 23 years, and all women were sexually active or initiating sexual activity. More detailed specific characteristics of the study group have been reported [14]. Changes in vaginal flora after initiating a new contraceptive method. All 5 visits were completed by 213 women, resulting in 1065 visits with vaginal cultures for inclusion in this analysis. The new contraceptive methods used by these women were OCPs (103 women), diaphragm-spermicide (75 women), and cervical cap (35 women). The proportions of women colonized with specific organisms at baseline and at 1 week and 4 weeks after initiating the new contraceptive methods are shown in figure 1. The prevalence of each organism at baseline was similar among women in the 3 groups, except for Enterococcus species, which were present in 16% and 17% of women in the OCP and diaphragm-spermicide groups, respectively, and in only 3% of women in the cervical cap group. Most organisms were present in <20% of women at baseline, except for lactobacilli, which were present in 85%–94% of women. The prevalence of women with lactobacilli did not change significantly 1 week and 4 weeks after initiating any of the 3 contraceptive methods. In contrast, the prevalence of women with E. coli JID 2000;181 (February) increased significantly in the cervical cap group, from 17% at baseline to 40% at weeks 1 and 4 (P = .03 and P = .03, respectively), and in the diaphragm-spermicide group, from 15% at baseline to 48% at week 1 and to 32% at week 4 (P = .001 and P = .005, respectively). There was a small decrement in the prevalence of E. coli in the OCP group, from 22% at baseline to 12% at week 4 (P = .04). Women without vaginal E. coli at baseline were more likely to be colonized with E. coli at any follow-up visit if they were in the diaphragm-spermicide (60%) and cervical cap groups (49%) than if they were in the OCP group (27%). Conversely, women with vaginal E. coli at baseline were more likely to subsequently become negative in the OCP group (11%) than in the diaphragm-spermicide (5%) or cervical cap groups (0%). Like E. coli, the prevalence of women with vaginal Enterococcus species also increased significantly at weeks 1 and 4 among the cervical cap group (P = .008 and P = .014, respectively) and at week 1 among the diaphragm-spermicide group (P = .001). There were only minor changes in enterococcal colonization in the OCP group. The prevalence of women with Candida species decreased slightly from 16% at baseline to 5% at week 4 (P ! .01) among those in the OCP group. None of the other organisms isolated, including Staphylococcus aureus, anaerobic gram-negative rods, group B streptococci, and Gardnerella vaginalis, changed significantly among the 3 groups of women over the 1-month observation period. Increases in the prevalence of anaerobic gram-negative rods that approached statistical significance were seen at the week 1 follow-up visit in the cervical cap (14%–29%) and diaphragm-spermicide (20%–32%) groups but not in the OCP group. By using semiquantitative cultures, we further determined whether initiation of these contraceptive methods altered the quantities of specific vaginal microbes (data not shown). The majority of women in each contraceptive group had either 21 or 31 lactobacilli throughout the study period, with little evident quantitative change after initiation of their contraceptive method. In contrast, although the majority of women in each contraceptive group had an absence of E. coli throughout the study period, there were significant increases in the percentages of women with 21, 31, and 41 E. coli among the diaphragmspermicide and cervical cap groups over the study period. These changes were not observed in the OCP group. Changes in Nugent score after initiating a new contraceptive method. Nugent scores were evaluated at each visit. At baseline, a score consistent with normal flora (1–3) was present in 86% (cervical cap group), 67% (diaphragm-spermicide group), and 65% (OCP group) of the women (figure 2). At week 1, a significant reduction in the prevalence of women with Nugent scores of 1–3 was observed in the cervical cap group (54%, P = .002) and in the diaphragm-spermicide group (44%, P = .001), although little change occurred in the OCP group (62%, P = .55). At week 4, the percentage of women with normal scores remained significantly reduced, as compared with base- JID 2000;181 (February) Contraceptive Method and Vaginal Flora 597 Figure 1. Changes in the prevalence of vaginal microflora stratified by contraceptive method. Bars represent the percentage of women with a vaginal culture positive for 111 of each microorganism at baseline (black bar); week 1 (light gray bar); and week 4 (dark gray bar). Bacteroides includes organisms belonging to the genera Bacteroides, Prevotella, and Porphyromonas. Statistically significant differences between baseline and follow-up weeks denoted by * (P ! .05) and † (P ! .01). CNS, coagulase-negative staphylococci; GBS, Group B streptococci. line rates in the cervical cap group (53%, P = .001), although smaller nonsignificant differences were seen in the diaphragmspermicide (57%) and the OCP (56%) groups, as compared with baseline scores. A majority of these score changes represented shifts from normal flora to intermediate flora rather than to BV. Relationship between Nugent score and vaginal colonization with E. coli and enterococci. The association between Nugent scores and vaginal colonization with E. coli and enterococci were also examined. As the prevalence of women with abnormal flora as measured by Nugent score increased, the prevalence and semiquantitative counts of E. coli and enterococcal vaginal colonization also increased (figure 3). A score consistent with intermediate or BV flora was present among 62% of women with 41 E. coli but only 31% of women with no E. coli (P ! .001). Similarly, such scores were present among 64% of women with 41 enterococci and only 35% of women with no enterococci (P = .01). Figure 2. Changes in Nugent score stratified by contraceptive method. Bars represent the percentage of women within each contraceptive group with a Nugent score of 1–3 (dark gray bar); 4–6 (light gray bar); and 7–10 (black bar) at each visit. *P ! .05 for comparison of baseline versus follow-up week. JID 2000;181 (February) Contraceptive Method and Vaginal Flora 599 users of the cervical cap and trended toward a decrease among diaphragm-spermicide users. The results of a second multivariate model that combined all 3 spermicidal contraceptive methods into 1 independent variable are also presented in table 1. In this analysis, the risk of having an abnormal Nugent score or colonization with E. coli, Enterococcus species, or anaerobic gram-negative rods was increased among spermicide users, whereas the risk of Candida species colonization was decreased. Discussion Figure 3. Association between abnormal Nugent score and vaginal colonization with Escherichia coli and enterococci. Significant increase was observed in the percentage of women who had an abnormal Nugent score (defined as a score >4) with increasing semiquantitative amounts of E. coli and enterococci (P < .01). Relationship between vaginal colonization with E. coli and lactobacilli. There was an inverse association between vaginal lactobacilli and E. coli colonization. As shown in figure 4, as the quantity of lactobacilli per vaginal culture increased from 0 to 41, the prevalence of E. coli vaginal colonization decreased from 34% to 18%. Furthermore, after controlling for intercourse, antimicrobial use, and birth control method in the preceding week, the risk of E. coli colonization at any visit was 2-fold greater among women without lactobacilli, as compared with women with lactobacilli (P = .01). Multivariate model assessing the risk of vaginal colonization with normal and abnormal flora by contraceptive method. To determine the relative risk of vaginal colonization with each organism among women using a diaphragm-spermicide, cervical cap, or other spermicidal method in the preceding 7 days, a multivariate model with these exposures as covariates was constructed (table 1). Sexual intercourse and antimicrobial use in the preceding 7 days were also included in the same model. After controlling for sexual activity and antimicrobial use, the relative risk of being colonized with lactobacilli, coagulase-negative staphylococci, group B streptococci, or G. vaginalis among diaphragm-spermicide, cervical cap, and spermicide users was not statistically significant. Conversely, there was a significantly increased risk of E. coli vaginal colonization of ∼2-fold among spermicide users, 3-fold among cervical cap users, and 4-fold among diaphragm-spermicide users. Intercourse also increased the risk of E. coli vaginal colonization by 3.5-fold. As with E. coli, the risk of vaginal colonization with Enterococcus species was also significantly increased in association with intercourse or diaphragm-spermicide use in the previous week but was not significantly elevated among users of the cervical cap or spermicide alone. The risk of vaginal colonization with anaerobic gram-negative rods and the risk of having an abnormal Nugent score were both significantly increased among users of a diaphragm-spermicide in the preceding week, whereas the risk of Candida species colonization was significantly decreased among The effects of contraceptive choice on the vaginal microbial flora have important implications for women who suffer from infections associated with disruptions in the vaginal ecology, such as BV and UTIs. The present prospective study of women who were initiating a new method of contraception (and, in some cases, sexual activity) showed that OCPs have little effect on the vaginal microbial flora, whereas diaphragm-spermicide use is associated with vaginal colonization with E. coli, Enterococcus species, and anaerobic gram-negative rods, as well as with an elevated Nugent score. Cervical cap use and other spermicide use were also each associated with similar alterations in flora, particularly E. coli colonization. Conversely, neither intercourse alone, nor use of any of the contraceptive methods, significantly affected the overall prevalence of vaginal lactobacilli. Our data also suggest that these alterations in vaginal flora do not occur independently of one another. Even after controlling for intercourse, spermicide use, and antimicrobial use in the previous week, the inverse association between lactobacilli and vaginal E. coli colonization was highly statistically significant. Although we did not specifically assay hydrogen peroxide (H2O2) production by lactobacilli strains in the present study, it is likely that a majority of lactobacilli isolated did produce H2O2, because these cultures were taken from healthy premenopausal women, a population in which the prevalence of H2O2-producing lactobacilli approaches 95% [19]. An inverse Figure 4. Association between vaginal lactobacilli and Escherichia coli colonization. A significant decrease was observed in the percentage of women who had 111 vaginal E. coli colonization with increasing semiquantitative amounts of lactobacilli (P < .01). 600 Gupta et al. Table 1. JID 2000;181 (February) Adjusted risk of vaginal colonization by contraceptive method. a b Adjusted odds ratio (95% CI) for colonization by exposure in the previous week Microbial organism Intercourse Lactobacilli Escherichia coli Enterococcus species Candida species Anaerobic gram-negative rods Abnormal Nugent score 0.8 3.5 1.8 1.3 0.9 1.3 (0.5–1.2) (2–6) (1.1–2.8) (0.7–2.4) (0.6–1.3) (0.9–1.7) Diaphragm or spermicide 1 (0.6–1.8) 3.8 (2.5–5.6) 2.5 (1.7–3.8) 0.5 (0.2–1.1) 1.7 (1.2–2.3) 1.5 (1.1–2.1) Cervical cap 1.5 2.7 1.4 0.3 1.5 1.1 (0.7–3.4) (1.6–4.4) (0.8–2.6) (0.1–0.8) (0.9–2.3) (0.7–1.8) Other spermicidal method 1.4 (0.8–2.5) 1.7 (1.1–2.6) 1.6 (1–2.7) 0.9 (0.4–1.7) 1.1 (0.7–1.7) 1 (0.7–1.5) Any spermicidal c method 1.3 2.8 1.8 0.5 1.5 1.3 (0.8–2.1) (2–4) (1.3–2.7) (0.3–0.9) (1.1–2.1) (1–1.7) a The same multivariate model was applied to each organism. Group B streptococci, coagulase-negative staphylococci, and Gardnerella vaginalis were not significantly associated with any of the exposures. b Defined as the presence of the organism in a semiquantitative concentration of 111. c All 3 spermicidal contraceptive methods were combined into 1 variable in this second model. relationship between H2O2-producing lactobacilli and vaginal E. coli has been reported in women with a history of recurrent UTI [5]. Our data suggest that this phenomenon may also be important in the pathogenesis of acute E. coli cystitis. In addition, women who had an abnormal Nugent score, suggesting an overgrowth of anaerobic flora, were also more likely to have E. coli and enterococcal colonization. The prevalence and semiquantitative amount of these aerobic bacteria increased as the prevalence of abnormal Nugent scores increased, suggesting a linear relationship between these perturbations in vaginal ecology. Recent appreciation that the loss of H2O2-producing lactobacilli may play a role in the pathogenesis of both BV [1] and E. coli [5] vaginal colonization provides a plausible common pathway for these events, and further evidence that these events are not entirely independent [1, 5]. Thus, exposures that lead to a loss of H2O2-producing lactobacilli likely predispose a woman to vaginal colonization with both anaerobic and aerobic bacteria. The extent and type of microbial alteration, and the resulting clinical syndrome, such as BV or UTI, likely depends on a combination of additional factors, such as intercourse frequency, spermicide use, douching, a new sex partner, or new sexually acquired pathogens. Additional host factors, such as secretor status or estrogen state, may also be important. The association between BV and E. coli vaginal colonization has been reported among women with acute UTI symptoms who use diaphragm-spermicide [20] but was not found among asymptomatic pregnant women in labor [16]. The latter study did find an increased prevalence of enterococcal colonization among women with BV flora, suggesting that aerobic bacteria are associated with the anaerobic overgrowth characterizing BV. The prospective nature of the present study and the use of daily diaries allowed us to elucidate temporal relationships between various exposures and vaginal microbiology. Although diaphragm-spermicide use has been associated with E. coli vaginal colonization [20–22], the individual contributions of intercourse, spermicide use, and contraceptive method have been difficult to sort out. Our data show an incremental increase in the effects on E. coli vaginal colonization of spermicide use alone, cervical cap use, intercourse, and diaphragm-spermicide use. A similar trend is seen with enterococcal colonization. Our data also show a temporal relationship between diaphragmspermicide use and elevated Nugent scores. Most previous studies examining risk factors for BV have involved women attending sexually transmitted disease clinics, a population in which sexual habits, frequency, and number of sex partners is likely different from our study group of asymptomatic young women attending a student health service. Unfortunately, exposures that have been previously associated with BV, such as new sex partner, douching, and history of BV, were too infrequent to assess adequately among our study population. We have previously proposed that spermicidal compounds may predispose women to E. coli vaginal colonization and UTI by virtue of their differential antimicrobial effects on lactobacilli and E. coli [23]. Thus, most lactobacilli (especially H2O2 producers) are susceptible, but nearly all E. coli are highly resistant to concentrations of nonoxynol-9 (N-9) that are likely present during routine use [9]. In the present study, we did observe an inverse relationship between E. coli vaginal colonization and lactobacilli, but the overall prevalence of women with lactobacilli and the quantity of lactobacilli present (as ascertained by our semiquantitative cultures) were not apparently altered after spermicide exposure. It is possible that spermicide exposure exerts a significant but only transient effect on lactobacilli prevalence and concentrations. Thus, a spermicide-mediated antimicrobial effect reducing lactobacilli from 107 colony forming units (cfu) per gram of vaginal fluid to 105 cfu per gram of vaginal fluid (a 99% reduction) might not be detectable 24 h later because of regrowth of lactobacilli. This explanation is supported by a recent study by Watts et al. [24], assessing the effects of N-9 use in the absence of intercourse; the concentration of lactobacilli decreased significantly 4 h after exposure to N-9 but returned to baseline levels by 24 h after exposure. The overall prevalence of lactobacilli was not significantly affected. Alternatively, because H2O2-producing lactobacilli are more sensitive to N-9 [25], perhaps N-9 exposure induces a shift from strains that produce H2O2 to strains that do not. However, a shift of this nature was not observed after a single exposure to N-9 [24]. We did not measure H2O2 production by the strains in the present study and thus cannot JID 2000;181 (February) Contraceptive Method and Vaginal Flora address this issue. Finally, both the abnormal Nugent scores induced by spermicide exposure and the changes in prevalence and concentration of anaerobic gram-negative rods that were observed suggest that spermicides affect the anaerobic vaginal flora in some manner. These effects could also be important in predisposition to E. coli colonization and UTI. In conclusion, barrier methods of contraception, such as the cervical cap and diaphragm-spermicide, clearly alter the normal vaginal ecosystem. Women who develop clinical syndromes associated with altered vaginal flora may choose to avoid these methods of birth control in favor of OCPs, which appear to have minimal effects on the vaginal microbial ecology. Elucidation of additional modifiable factors that disturb vaginal homeostasis is important for furthering our understanding of the pathogenesis of altered vaginal ecology and for providing additional behavioral choices for women. Acknowledgments We gratefully acknowledge Carol Winter, ARNP, and Natalie DeShaw, Research Study Coordinator, for patient enrollment and specimen collection. References 1. Hawes SE, Hillier SL, Benedetti J, et al. Hydrogen peroxide-producing lactobacilli and acquisition of vaginal infections. J Infect Dis 1996; 174: 1058–63. 2. Sewankambo N, Gray RH, Wawer MJ, et al. HIV-1 infection associated with abnormal vaginal flora morphology and bacterial vaginosis. Lancet 1997; 350:546–50. 3. Taha TE, Hoover DR, Dallabetta GA, et al. Bacterial vaginosis and disturbances of vaginal flora: association with increased acquisition of HIV. AIDS 1998; 12(13):1699–706. 4. Hillier SL, Krohn MA, Nugent RP, Gibbs RS. Characteristics of 3 vaginal flora patterns assessed by gram stain among pregnant women. Am J Obstet Gynecol 1992; 166:938–44. 5. Gupta K, Stapleton AE, Hooton TM, Fennell CL, Roberts PL, Stamm WE. Inverse association of H2O2-producing lactobacilli and vaginal E. coli colonization in women with recurrent UTI. J Infect Dis 1998; 178:446–50. 6. Hooton TM, Roberts PL, Stamm WE. Effects of recent sexual activity and use of a diaphragm on the vaginal microflora. Clin Infect Dis 1994; 19: 274–8. 7. Reid G, Bruce AW, Cook RL. Effect on urogenital flora of antibiotic therapy for urinary tract infections. Scand J Infect Dis 1990; 22:43–7. 601 8. Soper DE, Brockwell NJ, Dalton HP. Evaluation of the effects of a female condom on the female lower genital tract. Contraception 1991; 44:21–9. 9. Klebanoff SJ. Effects of the spermicidal agent nonoxynol-9 on vaginal microbial flora. J Infect Dis 1992; 165:19–25. 10. Stafford MK, Ward H, Flanagan A, et al. Safety study of nonoxynol-9 as a vaginal microbicide: evidence of adverse effects. J Acquir Immune Defic Syndr Hum Retrovirol 1998; 17:327–31. 11. Rosenstein IJ, Stafford MK, Kitchen VS, Ward H, Weber JN, Taylor-Robinson D. Effect on normal vaginal flora of 3 microbicidal agents potentially active against human immunodeficiency virus type-1. J Infect Dis 1998; 177:1386–90. 12. Hooton TM, Hillier S, Johnson C, Roberts PL, Stamm WE. E. coli bacteriuria and contraceptive method. JAMA 1991; 265:64–9. 13. Richardson BA, Martin HL Jr., Stevens CE, et al. Use of nonoxynol-9 and changes in vaginal lactobacilli. J Infect Dis 1998; 178:441–5. 14. Hooton TM, Scholes D, Hughes JP, et al. A prospective study of risk factors for symptomatic urinary tract infection in young women. N Engl J Med 1996; 335:468–74. 15. Balows A, Hausler WJ Jr, Herrmann KL, Isenberg HD, Shadomy HJ, editors. Manual of Clinical Microbiology. 5th ed. Washington, DC: American Society for Microbiology, 1991. 16. Hillier SL, Krohn MA, Rabe LK, Klebanoff SJ, Eschenbach DA. The normal vaginal flora, H2O2-producing lactobacilli and bacterial vaginosis in pregnant women. Clin Infect Dis 1993; 16 (suppl 4):S273–81. 17. 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. 18. Zeger SL, Liang KY. Longitudinal data analysis for discrete and continuous outcomes. Biometrics 1986; 42:121–30. 19. Eschenbach DA, Davick PR, Williams BL, et al. Prevalence of hydrogen peroxide-producing Lactobacillus species in normal women and women with bacterial vaginosis. J Clin Microbiol 1989; 27:251–6. 20. Hooton TM, Fihn SD, Johnson C, Roberts PL, Stamm WE. Association between bacterial vaginosis and acute cystitis in women using diaphragms. Arch Intern Med 1989; 149:1932–6. 21. Peddie BA, Bishop V, Bailey RR, McGill H. Relationship between contraceptive method and vaginal flora. Aust N Z J Obstet Gynaecol 1984; 24: 217–8. 22. Fihn SD, Latham RH, Roberts PL, Running K, Stamm WE. Association between diaphragm use and urinary tract infection. JAMA 1985; 254: 240–5. 23. Hooton TM, Fennell CL, Clark AM, Stamm WE. Nonoxynol-9: differential antibacterial activity and enhancement of bacterial adherence to vaginal epithelial cells. J Infect Dis 1991; 164:1216–9. 24. Watts DH, Rabe L, Krohn MA, Aura J, Hillier SL. The effects of three nonoxynol-9 preparations on the vaginal flora and epithelium. J Infect Dis 1999; 180:426–37. 25. McGroarty JA, Tomeczek L, Pond DG, Reid G, Bruce AW. Hydrogen peroxide-production by Lactobacillus species: correlation with susceptibility to the spermicidal compound nonoxynol-9. J Infect Dis 1992; 165:1142–4.
© Copyright 2026 Paperzz