Beigi, Richard H. MD, MSc*; Meyn, Leslie A.†; Moore, Donna M.‡; Krohn, Marijane A. PhD†‡; Hillier, Sharon L. PhD†‡
Approximately 75% of reproductive-aged women will experience at least one episode of yeast vaginitis in their lifetimes.1 Candida albicans is responsible for greater than 90% of the cases of symptomatic yeast vaginitis, and this organism is recognized as being part of the normal vaginal flora in reproductive-aged women.2 The precise relationship between yeast colonization and symptomatic yeast vaginitis is not entirely clear; however, it is thought that yeast colonization is likely a necessary precursor to symptomatic vulvovaginal candidiasis. There are few longitudinal epidemiological investigations of yeast colonization. A prospective study of women enrolled in midpregnancy found Candida albicans colonization to be associated with African-American race, unmarried status, and Lactobacillus spp colonization.3 The Human Immunodeficiency Virus (HIV) Epidemiology Research Study (including, also, seronegative controls) found that the only factors predictive of vaginal yeast colonization were HIV seropositivity, recent intravenous drug use, and insulin and/or oral hypoglycemic therapy.4 Small longitudinal studies evaluating women before and after initiation of hormonal methods of contraception have generally found no increase in vaginal colonization by yeast, but their statistical power to detect changes in flora are limited by sample size.5–7
In 1990, antifungal medication for vulvovaginal candidiasis became available over the counter (OTC). The increased availability of antifungal preparations for women has led to widespread use of these products by health care consumers. Seventy-three percent of women with chronic vaginal symptoms have reported OTC yeast medication use, and expenditures on these medications has roughly doubled since their OTC introduction.8,9 Despite this increased use, the accuracy of self-diagnosis has never been validated. Several investigators have questioned the reliability of self-diagnosis and self-treatment.10–12
This study had 3 objectives: 1) to define the cumulative frequency of vaginal yeast colonization in a nonpregnant population of young, sexually active women over a 12-month period; 2) to identify demographic, behavioral, and historical risk factors associated with vaginal yeast colonization; and 3) to explore the correlation between yeast colonization and clinical outcomes related to perceived vulvovaginal candidiasis (symptoms and antifungal medication use).
MATERIALS AND METHODS
Asymptomatic, 18- to 30-year-old, nonpregnant, sexually active women were recruited from 3 locations in the greater Pittsburgh area over a 3-year period (1998–2000): the Allegheny County Health Department Sexually Transmitted Diseases Clinic, the University of Pittsburgh Student Health Center, and the Family Health Council Clinic of Aliquippa. Women were excluded if they were pregnant, were currently experiencing genital tract symptoms or bleeding, were currently using systemic antimicrobials, had used any intravaginal products (douche, antifungal products, spermicides) in the preceding 24 hours, or had plans to move from the area during the study period. All women provided written informed consent before enrollment into a Magee-Womens Hospital Institutional Review Board–approved protocol.
All women were interviewed using a structured questionnaire that included detailed questions regarding demographic factors, personal/sexual habits, lower genital tract symptoms, and vaginitis diagnoses/treatments received over the past 4 months. At enrollment, one sterile swab was inserted into the vagina and a sample of vaginal fluid was obtained. The swab was then placed in Amies transport media (MML Diagnostics Packaging, Troutdale, OR). The swab was stored at room temperature and taken to the laboratory within 24 hours of collection. Validation studies were done as quality control measures for the vaginal swabs to demonstrate that the number of yeast colonies did not decrease or increase at room temperature for 24 hours. After enrollment and swab collection, patients were instructed to return at 4-month intervals for a total of 4 visits during 1 year of follow-up. At each follow-up visit the same questionnaire was completed, and a vaginal swab was obtained and processed in the same manner. A total of 1,248 women were recruited into the cohort. Greater than 80% (4,012/4,992) of the scheduled visits for these women were accomplished. A smaller subset of 709 women completed all 4 visits.
At all 4 visits, a swab containing vaginal fluid from the participant was first used to inoculate 5% sheep blood agar plates (Prepared Media Laboratories, Tualatin, OR) and was then placed in a selective broth media suspension containing colistin and nalidixic acid (Prepared Media Laboratories). A validation study was performed to show that the selective broth media used was appropriate for detection of vaginal yeasts. A 106 suspension of Candida albicans American Type Culture Collection (ATCC) 10231 was made using the National Committee of Clinical Laboratory Standards guidelines. Dilutions of 10−1, 10−2, 10−3, 10−4, and 10−5 were made using the 106 suspension. A volume of 0.1 mL for each dilution was inoculated into a selective broth media tube. These were then subcultured at 24, 48, and 72 hours onto Sabouraud agar plates (Gibson Laboratories Inc, Lexington, KY) to determine the length of time yeast is maintained in the broth and at what dilution yeast is viable in the selective broth media. The blood agar plates were incubated at 37°C in 5–7% CO2 for 24 hours. After 72 hours of incubation, the selective broth suspension was subcultured onto Sabouraud Dextrose agar plates (Gibson Laboratories Inc), which were incubated for an additional 48 hours. Both the blood agar and the Sabouraud agar plates were examined for yeast growth. Heavy growth was defined as yeast growth on the original blood agar plates, and light growth was defined as yeast growth only after application of broth enrichment techniques. All yeast isolates were then identified on the basis of germ tube formation. The non-Albicans species were further identified using API 20C AUX (bioMerieux sa, Etoile, France). Additional assessments of vaginal flora included detection of lactobacilli and group B streptococci (GBS) by culture and evaluation of the vaginal flora by Gram-stained vaginal smear, as previously described.13
Vaginal yeast colonization was defined as detection of yeast from the agar plate or broth medium. All statistical analyses were performed using Stata 8.0 statistical software (Stata Corporation, College Station, TX), and all statistical tests were evaluated at the 0.05 level of significance. Generalized estimating equations were used to identify factors among the cohort of 1,248 women that were associated with vaginal yeast colonization. The generalized estimating equation is a regression method that is able to model the marginal expectation of repeated, correlated outcomes with both time-dependent and time-independent covariates. An exchangeable working correlation matrix was specified, and modified sandwich estimates of the variance were calculated. Statistical inference was based on the generalized Wald test statistic.14 Multivariable generalized estimating equation models were developed using forward stepwise regression. Variables were retained in the model if the Wald test statistic had a P value of .05 or less. The population-averaged odds ratios derived from both the univariable and multivariable generalized estimating equation models are presented along with the 95% confidence intervals and P values.
A total of 1,248 women were enrolled. Approximately 60% of the cohort was Caucasian, greater than 80% were 18–25 years of age, 75% of the women attained more than a high school education, and the majority of women were of low gravidity (0–1).
Among a subset of women who completed all 4 visits (n = 709), different patterns of yeast colonization were identified. Seventy percent (496/709) demonstrated colonization at one point or another, most having intermittent colonization (470/496), and only 4% (26/709) demonstrating persistent colonization at all 4 visits. Thirty percent (213/709) of the women were never colonized by yeast during the year of follow-up.
Including both plate- and broth-positive women, the average point prevalence of yeast positivity at any one visit was 30%, with approximately half of the women being heavily colonized and half being lightly colonized. The vast majority (98%) of women colonized with yeast were colonized with Candida albicans. The other 2% were colonized by various Candida spp, including glabrata, parapsilosis, dubliensis, and famata.
Numerous demographic and historical variables appear to be associated with yeast colonization when evaluated individually. These include age greater than 26, cigarette smoking in the past 4 months, use of marijuana in the past 4 months, condom use, sexual activity both in the past 4 months and also in the past 5 days, 2 or more sexual partners in the previous 4 months, colonization with lactobacilli, and colonization with GBS. In addition, multiple symptoms also appear to be associated with vaginal yeast colonization, including dysuria, abnormal and/or increased discharge, pruritus, and burning. However, multivariable analysis revealed only 2 symptoms that are predictive of yeast colonization: vulvovaginal burning (odds ratio [OR] 1.4, 95% confidence interval [CI] 1.1–1.7) and pruritus (OR 1.7, 95% CI 1.4–2.0) (Table 1). Moreover, on multivariable analysis, 5 demographic/historical factors are independently predictive of yeast colonization (Table 2).
Evaluation was then done to investigate whether any correlation existed between vaginal yeast colonization and antifungal use. The analysis failed to demonstrate an association between antecedent yeast colonization (those positive at a previous visit) and subsequently reported antifungal use (OR 1.1, 95% CI 0.8–1.6). Women who reported antifungal use in the previous 4 months were neither more nor less likely to be colonized at the subsequent visit. In addition, there was no correlation between colonization category over the 12-month study period and antifungal use, with statistically equal numbers of women reporting use of antifungals in the colonized and noncolonized categories (23.5% versus 24%, respectively). Of note, 25% of women lacking vaginal yeast colonization at any point in the 12-month study reported use of vaginal antifungals. The lack of association was irrespective of colonization density (high versus low).
This investigation demonstrates that 70% of young, sexually active women will be vaginally colonized by yeast at some time during a period of 1 year. Risk factors for colonization include depomedroxyprogesterone use in the previous 4 months, marijuana use in the past 4 months, recent (within 5 days) sexual intercourse, and concurrent colonization with lactobacilli and GBS. Despite frequent colonization and a correlation between symptoms often associated with vulvovaginal yeast infection (pruritus and burning), we were unable to demonstrate an association between colonization and antifungal use.
Previous studies have focused on point prevalence of yeast colonization, with approximately 20% of women being colonized at any one point in time. The point prevalence in this study at each visit averaged 30%, a number slightly higher than, but in accordance with, previous published data.1 This slightly higher rate of yeast colonization may be due in part to the use of a selective broth enrichment technique that enhanced the sensitivity for the detection of light colonization. The fact that 70% of women are colonized with yeast and that colonization is more likely in women that have concurrent Lactobacillus supports the concept that Candida albicans is a member of the normal vaginal flora of many healthy asymptomatic women. This finding suggests that host factors play an important role in the genesis of patient symptoms caused by yeast, since a large percentage of colonized women have no symptoms.
Other investigators have evaluated epidemiological factors for vaginal yeast colonization with differing results. Cotch and colleagues3 conducted a longitudinal study of pregnant women beginning in midgestation (23–26 weeks). Among numerous demographic and behavioral factors evaluated, 3 factors (African-American race, unmarried status, and Lactobacillus spp colonization) were found to predict vaginal yeast colonization. Our data are consistent in that we found women were 90% more likely (OR 1.9, 95% CI 1.6–2.3) to be colonized with yeast if they also had Lactobacillus spp colonization. The HIV Epidemiology Research Study found that vaginal yeast colonization was more prevalent among HIV+ women (irrespective of CD4+ count), women reporting recent intravenous drug use, and those currently on insulin and/or oral hypoglycemic therapy.4 We are unable to evaluate diabetes or HIV serostatus as a predictive factor for yeast colonization in our cohort because of the low numbers of women with these conditions (6 and 1, respectively) in our study.
The role of hormonal contraceptives on rates of vaginal yeast colonization has been evaluated in multiple studies. Using a prospective study design, Eschenbach and colleagues5 evaluated the effect of oral contraceptive pills (OCPs) on vaginal flora by performing vaginal cultures on asymptomatic women before and 2 months after initiation of combined OCPs. There were no substantive changes in vaginal flora constituents, including no change in the percentage of women colonized by Candida albicans. The authors concluded that OCPs do not produce detectable changes in vaginal flora. A similar study published by the same group of investigators evaluated the effects of depomedroxyprogesterone on the vaginal flora. This investigation revealed a protective effect of depomedroxyprogesterone against vaginal yeast colonization, with a significant decrease in the percentage of women colonized by Candida albicans after treatment with cyclic depomedroxyprogesterone.6 A separate prospective evaluation of the effects of various contraceptives among 103 sexually active women demonstrated a significant decrease in prevalence of women colonized by Candida albicans from before initiation (16%) of OCPs to 4 weeks after use (5%).7 However, few effects on other constituents of the vaginal flora were observed in the OCP group, prompting the authors to conclude that OCPs do not have a marked affect on the vaginal flora. Our investigation failed to demonstrate any relation between OCP use and vaginal yeast colonization. However, recent depomedroxyprogesterone use was found to be predictive of vaginal yeast colonization (OR 1.4, 95% CI 1.1–1.7), in contrast to the study noted above.6 Differences in study design and size limitations may contribute to the different findings. The odds ratio of 1.4 in this study suggests a modest increase in risk. Future longitudinal investigations may supplement the understanding of the relationship between depomedroxyprogesterone and other hormonal methods of contraception and yeast colonization.
This investigation, using different approaches, failed to find a correlation between yeast colonization and antifungal use. The finding that colonization does not predict subsequent antifungal use and that equal percentages of women report use of vaginal antifungal preparations regardless of their colonization category (including 25% of women with no yeast colonization) suggests that a subjective diagnosis of vulvovaginal candidiasis may be inaccurate. This lack of correlation naturally raises questions about the availability and use of OTC antifungal medication for self-diagnosed vulvovaginitis.
Others have investigated self-diagnosis and self-treatment with antifungal preparations using different study designs with similar concerning findings. Linhares and colleagues10 compared culture-positive and -negative women with vaginitis symptoms and found no difference between the 2 groups in symptoms of pruritus, burning, and discharge, all of which are typically ascribed to vulvovaginal candidiasis. Ferris and investigators12 reported the results of a questionnaire study assessing women's ability to diagnose the correct genital infectious condition before and after reading “classic case scenarios.” They found that, after reading the scenarios, more than 30% of the women would mistakenly use an OTC antifungal preparation for other diagnoses (pelvic inflammatory disease, bacterial vaginosis, urinary tract infection, and trichomoniasis).
A more recent study by Ferris et al11 investigated self-diagnosis by using a unique study design. Patients were evaluated clinically by the investigators within 24 hours of a purchase of OTC antifungal products after self-diagnosis. The authors found that only 33% of these women had yeast vulvovaginitis, with the rest having other vaginitides, mixed infections (some with sexually transmitted diseases), or noninfectious causes of vaginal complaints. These studies taken together, in addition to our findings, suggest that vulvovaginal symptoms are poor predictors of actual diagnosis. Moreover, self-diagnosis may often lead to incorrect and/or unnecessary treatment.
There are some limitations to the present study. This was a prospective study. However, patients at each visit were asked to retrospectively report symptoms and medication use for the previous 4-month period. This may predispose the data to recall bias. However, women were not informed of their yeast colonization status, and the recall bias should have been similar with colonized and noncolonized women. We enrolled 1,248 women and analyzed them based on the number of total visits; however, only 709 women completed all 4 visits. This may have subjected the data to a nonquantifiable dropout bias. A further limitation of this study is that the majority of women were not examined when they had their symptoms and obtained treatment. It is therefore possible that women who had no yeast colonization detected at our previous visit could have had an intervening symptomatic vaginal yeast infection that was successfully treated, thereby eradicating the yeast and rendering it undetectable by our subsequent interval culture. Lastly, because our subjects were between the ages of 18 and 30 years and tended to be from the midlower socioeconomic class, these results may not be completely generalizable to the entire female population. Despite these limitations, the large numbers and length of follow-up in this study add to the importance of our findings.
In summary, the present study supports the concept that Candida albicans is part of the normal vaginal flora of asymptomatic sexually active young women manifesting mostly in an intermittent manner. The epidemiological findings over a 12-month period suggest that recent sexual activity and use of injectable contraceptives are associated with vaginal yeast colonization. Moreover, our findings contribute to the growing body of evidence that questions the practice of self-diagnosis and self-management of vulvovaginal symptoms.
1. Sobel JD. Epidemiology and pathogenesis of recurrent vulvovaginal candidiasis. Am J Obstet Gynecol 1985;152:924–35.
2. Sobel JD. Vaginitis. N Engl J Med 1997;337:1896–903.
3. Cotch MF, Hillier SL, Gibbs RS, Eschenbach DA. Epidemiology and outcomes associated with moderate to heavy Candida colonization during pregnancy. Vaginal Infections and Prematurity Study Group. Am J Obstet Gynecol 1998;178:374–80.
4. Schuman P, Sobel JD, Ohmit SE, Mayer KH, Carpenter CC, Rompalo A, et al. Mucosal candidal colonization and candidiasis in women with or at risk for human immunodeficiency virus infection. HIV Epidemiology Research Study (HERS) Group. Clin Infect Dis 1998;27:1161–7.
5. Eschenbach DA, Patton DL, Meier A, Thwin SS, Aura J, Stapleton A, et al. Effects of oral contraceptive pill use on vaginal flora and vaginal epithelium. Contraception 2000;62:107–12.
6. Miller L, Patton DL, Meier A, Thwin SS, Hooton TM, Eschenbach DA. Depomedroxy- progesterone-induced hypoestrogenism and changes in vaginal flora and epithelium. Obstet Gynecol 2000;96:431–9.
7. Gupta K, Hillier SL, Hooton TM, Roberts PL, Stamm WE. Effects of contraceptive method on the vaginal microbial flora: a prospective evaluation. J Infect Dis 2000;181:595–601.
8. Nyirjesy P, Weitz MV, Grody MH, Lorber B. Over-the-counter and alternative medicines in the treatment of chronic vaginal symptoms. Obstet Gynecol 1997;90:50–3.
9. Lipsky MS, Waters T. The “prescription-to-otc switch” movement: its effects on antifungal vaginitis preparations. Arch Fam Med 1999;8:297–300.
10. Linhares IM, Witkin SS, Miranda SD, Fonseca AM, Pinotti JA, Ledger WJ. Differentiation between women with vulvovaginal symptoms who are positive or negative for Candida species by culture. Infect Dis Obstet Gynecol 2001;9:221–25.
11. Ferris DG, Nyirjesy P, Sobel JD, Soper D, Pavletic A, Litaker MS. Over-the-counter antifungal drug misuse associated patient-diagnosed vulvovaginal candidiasis. Obstet Gynecol 2002;99:419–25.
12. Ferris DG, Dekle C, Litaker MS. Women's use of over-the-counter antifungal medications for gynecologic symptoms. J Fam Pract 1996;42:595–600.
13. Meyn LA, Moore DM, Hillier SL, Krohn MA. Association of sexual activity with colonization and vaginal acquisition of Group B streptococcus in non-pregnant women. Am J Epidemiol 2002;155:949–57.
14. Hardin JW, Hilbe JM. Generalized estimating equations. Boca Raton (FL): Chapman & Hall/CRC; 2003.