Both vaginal colonization by Candida species and vulvovaginitis associated with Candida occur more frequently among human immunodeficiency virus (HIV)–infected women, especially those with moderate to severe immune compromise.1–7 The Centers for Disease Control and Prevention (CDC) included vulvovaginal candidiasis among category B conditions, those for which clinical course or management is complicated by HIV infection.8 The CDC revised classification system indicates that vulvovaginal candidiasis associated with HIV is especially persistent, frequent, or poorly responsive to therapy. Similarly, the World Health Organization classifies vulvovaginal candidiasis that is chronic or poorly responsive to therapy as being associated with intermediate (moderate) HIV disease.9
Most studies of vulvovaginal candidiasis and HIV to date have been cross-sectional,1–5,7 have enrolled relatively few HIV-infected women, lacked appropriate matched controls, or have been restricted to specialized subpopulations such as women who are pregnant1 or have symptoms of vulvovaginitis.6 To redress this deficiency, we have used data from a large prospective study of women with or at risk for HIV infection to investigate the relative occurrence of vulvovaginal candidiasis in HIV-infected versus uninfected women. We also examined whether the vaginal yeast infections that occur among these two groups of women differ qualitatively.
MATERIALS AND METHODS
The HIV Epidemiology Research (HER) Study enrolled 1310 women aged 15–55 years (871 HIV infected and 439 at-risk HIV uninfected) from four US communities (the Bronx, Baltimore, Providence, and Detroit) between April 1993 and January 1995; details of this design and enrollment procedures have been reported.10 Institutional review boards approved the study protocol at each site and at the CDC; all participants gave informed consent. Each site contributed approximately one fourth of the participants. HIV-infected women without acquired immunodeficiency syndrome (AIDS)–defining conditions and HIV-uninfected women were selected so that about half reported risk related to injection drug use, and the remainder had increased risk due to sexual behavior but not injection drug use. Semiannual research visits included core interviews on medical history, obstetric and gynecologic history, substance use, and sexual behavior, as well as a physical examination with a gynecologic component. Candida was identified by culture (Sabouraud dextrose agar incubated at 30C). In addition, budding yeast and fungal hyphae were identified on Gram stains of vaginal secretions, which were read in a single laboratory (Wayne State University). This measure of yeast overgrowth showed less variability and better concordance with culture results than wet mounts, which were read by clinicians at each site.
We constructed three definitions of vulvovaginal candidiasis using the following data: 1) the presence of yeast in the vagina as measured by vaginal culture for Candida and a Gram stain of vaginal flora and 2) data on the presence of three signs on examination noted by a study clinician: vulvovaginal edema, erythema, or abnormal discharge. The three definitions were as follows: 1) vulvovaginal candidiasis 1, yeast on culture and/or Gram stain plus at least one sign; 2) vulvovaginal candidiasis 2, yeast on culture and/or Gram stain plus two or more signs; and 3) vulvovaginal candidiasis 3, budding yeast or hyphae on Gram stain plus at least one sign. At any given visit, a woman could meet one or more definitions of vulvovaginal candidiasis. These definitions represent vulvovaginal candidiasis of differing severity. Women with vulvovaginal candidiasis 2 are required to have more signs than those meeting only the definition of vulvovaginal candidiasis 1. As a group, women with vulvovaginal candidiasis 3 likely have a higher quantity of yeast present (as evidenced by the ability to detect yeast on Gram stain) than women meeting only the definition of vulvovaginal candidiasis 1 or vulvovaginal candidiasis 2 because not all of the latter necessarily have yeast detected on Gram stain.
Because signs attributed to vulvovaginal candidiasis could be due to other conditions, we evaluated the role of three coexisting genital tract conditions in our risk models: bacterial vaginosis, trichomoniasis, and genital ulcers. Bacterial vaginosis was identified through Gram stains of vaginal specimens using a scoring system based on the relative frequencies of Lactobacillus sp, Gardnerella vaginalis, Bacteroides sp, and Mobiluncus sp morphotypes.11 Trichomoniasis was initially defined as the presence of trichomonads by wet mount (visits one to three); culture (in modified Diamond medium) was introduced for all women at visit four. Vulvar, vaginal, and cervical ulcers were visually identified by a study clinician on examination. Evaluations for gonorrhea and chlamydia were discontinued after visit five because of their very low occurrence, and these conditions were excluded from the analysis. Women with symptomatic vaginal or cervical infections at the time of the study visit were either treated or referred for treatment.
Human immunodeficiency virus status was ascertained through enzyme-linked immunosorbent assay and Western blot. We used three CD4 + T lymphocyte groups to categorize the level of immune system compromise: less than 200 cells per microliter, severe; 200–499, moderate; and 500 or more, least immunocompromised. When testing followed phlebotomy by more than 2 days, groups were determined by CD4 percentage instead of count, with cut points of 14% and 29%.8 Human immunodeficiency virus load was determined via a third-generation, branched deoxyribonucleic acid signal amplification assay (Chiron Corp., Emeryville, CA) with a lower quantification limit of 50 copies per milliliter. Mindful of the possible links between medical therapy and vulvovaginal candidiasis, we classified systemic and vaginal drugs using a published drug index12 with a view to possible class-level associations with vulvovaginal candidiasis. We focused on three classes of drugs: antiretrovirals, systemic antifungals, and systemic antibacterials. Antiretroviral regimens were classified as highly active antiretroviral combination therapy, other antiretroviral therapy, or no antiretrovirals; highly active antiretroviral combination therapy comprised regimens of multiple drugs with clinical evidence of effective and sustained viral suppression. Regimen classifications were adapted from US Department of Health and Human Services guidelines for preferred or alternative combinations,13 with unclassed regimens included if they were believed to have a high likelihood of exhibiting the same properties. Antiretroviral regimens not considered as highly effective (usually one- or two-drug regimens) were classified as antiretroviral therapy. Systemic antifungals included fluconazole, ketoconazole, and itraconazole. Insofar as our sample included women with Candida infection at other mucosal sites, systemic antifungals may not have been taken to treat vaginal Candida infection. We also included data on use of topical antifungals, as reported by the woman. Systemic antibacterials (eg, amoxicillin) included broad-spectrum antibacterials plus some narrow-spectrum antimicrobials with activity against common genital tract infections but excluded most antiparasitics and antimycobacterials. In addition to these drug classes, we considered antimycobacterials (eg, isoniazid), antiprotozoals (eg, atovaquone), and antivirals (eg, acyclovir). Additionally, we used data on diabetes therapies to identify women with that disease.
Pregnancy was ascertained through a urine test administered at the study site. Menses was considered recent if the last reported date was within 14 days of the gynecologic examination. Other factors considered included number of vaginal yeast infections in the previous 6 months (self-report); presence of a cervix; delivery in the past 8 weeks; number of pregnancies and live births; douching in the past 48 hours; and use of hormonal contraceptives, spermicides, diaphragms, or intrauterine devices.
We also accounted for potential links between vulvovaginal candidiasis prevalence and several sexual and drug-related factors, including trading sex for drugs or money; number of male partners; frequency of vaginal sex; consistency of condom use in vaginal sex; detection of sperm on Gram stain; number of female sex partners; frequency of sex with women; and use of tobacco, crack, and intravenous drugs.
Cumulative incidence of vulvovaginal candidiasis was estimated over the course of this study using the Kaplan-Meier product-limit survival estimator.14 Logistic models provided estimates of overall prevalence via methods that compensated for within-subject correlation across multiple observations.15 These prevalence estimates and odds ratios (ORs) averaged the characteristics of study participants over the course of the study. Inferences were based on standard errors estimated by robust methods using an m-dependent working correlation structure, which assumes that the correlation depends only on how far apart two visits were in time.15 Computations were performed in SAS 7 (SAS Institute Inc., Cary, NC).
Under the HER Study design, we were able to ascertain vulvovaginal candidiasis status every 6 months. For each pair of visits separated by 6 months, vulvovaginal candidiasis at the second visit was designated as persistence if it was preceded by vulvovaginal candidiasis, escalation if preceded by subclinical colonization, and acquisition if preceded by noncolonization. We estimated the association of vulvovaginal candidiasis at the current visit with vulvovaginal candidiasis status at the preceding visit by including previous vulvovaginal candidiasis status as an additional predictor in the multivariable model.
We evaluated each of the candidate predictors for vulvovaginal candidiasis individually. These initial models adjusted only for HIV status, study site, risk cohort (injection drug use, heterosexual sex), and visit number. Predictors were retained as candidates for subsequent multivariable models if they demonstrated at least a moderate association with vulvovaginal candidiasis (P < .2).
Multivariable models were then built using variables identified in this process plus several preselected variables thought to be related to clinical course and management of vulvovaginal candidiasis. Seven preselected predictors—use of antibacterial drugs, pregnancy, diabetes, self-reported vaginal yeast infections, bacterial vaginosis, trichomoniasis, and genital ulcers—were thus included in the final model by design. Additional predictors were retained if they demonstrated statistical evidence of an association with vulvovaginal candidiasis. After selection of a multivariable model we checked for effect modification of HIV status on the other predictors. Then, we explored associations of vulvovaginal candidiasis with risk factors in HIV-infected women, looking at two characteristics of the stage of HIV disease (CD4 group and viral load) and two HIV-related medical therapies (antiretrovirals and systemic antifungals).
Finally, we examined the risk of more severe vaginal yeast infections among women with vulvovaginal candidiasis 1. This was accomplished using multivariable models that estimated the risk of vulvovaginal candidiasis 2 or vulvovaginal candidiasis 3 among women with vulvovaginal candidiasis 1.
In this analysis, we limited our attention to the first ten visits from 1277 subjects (33 women were excluded because they seroconverted within that period or because we did not have complete information on the risk factors under consideration). The analysis covered 8648 visits spanning April 1993 to February 1999. Among 856 HIV-infected participants, we considered 5705 visits, and among 421 HIV-uninfected participants, 2943 visits. Approximately 75% of all women made at least five visits (mean 6.9). When considering acquisition, escalation, and persistence, we restricted our attention to pairs of consecutive visits.
For the 1277 women in our analysis (Table 1), mean age over all study visits was 37 years, with no difference by HIV status. (More complete demographic information on the cohort at enrollment is included in Smith et al.10) Human immunodeficiency virus–uninfected women were more likely to report postsecondary education and recent menstrual periods and to be pregnant. No significant differences were seen for bacterial vaginosis, trichomoniasis, or diabetes. Human immunodeficiency virus–infected women were more likely to have genital ulcers on examination and to report recent vaginal yeast infections and current antibacterial use. Human immunodeficiency virus–infected women showed moderate immune compromise (as indicated by CD4 count) at almost half their visits; at the remaining visits, women were evenly divided between the highest and lowest CD4 groups. Human immunodeficiency virus–infected women reported highly active antiretroviral combination therapy use at 8% and other antiretroviral use at 35%. After 1995, however, highly active antiretroviral combination therapy use was reported at 19% of visits, whereas other antiretroviral regimens were reported at 34%. Human immunodeficiency virus–infected women reported use of systemic antifungals at 10% of visits.
Cumulative incidence of vulvovaginal candidiasis for each of the three definitions was higher among HIV-infected women than among HIV-uninfected women (Figure 1). After 4.5 years of follow-up the cumulative incidence of vulvovaginal candidiasis 1 was 64.5% among HIV-infected women who did not have vulvovaginal candidiasis 1 at the baseline visit. The 4.5-year cumulative incidences among HIV-infected women of vulvovaginal candidiasis 2 and vulvovaginal candidiasis 3 were 28% and 38.1%, respectively. The cumulative incidence of each condition at 4.5 years was roughly ten to 20 percentage points lower among HIV-uninfected women.
Initial models for the frequency of vulvovaginal candidiasis indicated a significant association with HIV status. Over the course of the study, the average prevalence of vulvovaginal candidiasis 1 was 16.1% among HIV-infected women and 9.0% among HIV-uninfected women (P < .01) (Table 2). The average prevalences of vulvovaginal candidiasis 2 and vulvovaginal candidiasis 3 were lower and differed significantly by HIV status (P < .01 for each) (Table 2).
The risk of vulvovaginal candidiasis 1 at a given visit varied by vaginal Candida status at the previous consecutive visit. Vulvovaginal candidiasis 1 was least frequent among women who did not have Candida colonization at the preceding visit (acquisition), higher among those with preceding subclinical Candida colonization (escalation), and most frequent among women with preceding vulvovaginal candidiasis 1 (persistence) (Table 2). These differences in frequency were significant even when corrected for multiple comparisons (P < .01). Furthermore, HIV infection was associated with increased prevalence of vulvovaginal candidiasis 1 at a given visit, regardless of the woman's status at the preceding visit: HIV-infected women had 1.74 times the odds for acquisition, escalation, or persistence relative to HIV-uninfected women (95% confidence interval [CI] 1.45, 2.09). Results for vulvovaginal candidiasis 2 and vulvovaginal candidiasis 3 were similar (vulvovaginal candidiasis 2: OR 1.60; 95% CI 1.15, 2.23; vulvovaginal candidiasis 3: OR 1.58; 95% CI 1.24, 2.00).
Similar results were obtained in a multivariable model that adjusted for other risk factors for vulvovaginal candidiasis. Persistent cases occurred more frequently than escalated cases, which in turn occurred more frequently than acquired cases, and the HIV-associated risk seen after stratifying on the woman's vulvovaginal candidiasis status at the previous visit remained after adjusting for other factors (data not shown).
Regardless of the definition used, vulvovaginal candidiasis was elevated among HIV-infected women. The HIV-associated ORs for vulvovaginal candidiasis 1, vulvovaginal candidiasis 2, and vulvovaginal candidiasis 3 were 1.88 (95% CI 1.58, 2.25), 1.68 (1.23, 2.28), and 1.63 (1.29, 2.06), respectively, after adjustment for antibacterial use, age, education, genital tract infections, recent menses, diabetes, and pregnancy. Only bacterial vaginosis appeared to act as an HIV effect modifier.
In a model that considered HIV-infected women separately, those with a CD4 count below 200 cells per microliter showed higher prevalence of vulvovaginal candidiasis (all definitions) (Table 3). Among women with CD4 counts of 200–499 cells per microliter, only vulvovaginal candidiasis 1 was significantly more prevalent (Table 3). Use of highly active antiretroviral combination therapy or less active antiretroviral regimens was not associated with prevalence of vulvovaginal candidiasis. Reported use of systemic antifungal agents was associated with decreased prevalence of vulvovaginal candidiasis 1.
When HIV viral load was added to the models, the ORs obtained for CD4 of less than 200 and 200–499 were reduced and remained significant for vulvovaginal candidiasis 1 only. Viral loads of greater than 10,000 and 1000–10,000 were significantly associated with increased risk of vulvovaginal candidiasis 1 (OR 1.66; 95% CI 1.23, 2.24 and OR 1.49; 95% CI 1.12, 1.99, respectively), vulvovaginal candidiasis 2 (OR 3.68; 95% CI 1.96, 6.92 and OR 2.78; 95% CI 1.50, 5.16, respectively), and vulvovaginal candidiasis 3 (OR 2.11; 95% CI 1.36, 3.28 and OR 1.82; 95% CI 1.18, 2.79, respectively).
In addition to the HIV-specific items listed above, the model evaluated the independent effect of risk factors not specific to HIV (Table 3). Bacterial vaginosis and recent menses were independently associated with reduced odds for having vulvovaginal candidiasis 1, 2, or 3. Odds of vulvovaginal candidiasis 1 and 2 decreased with age and education and increased with trichomoniasis. Genital ulcers, self-reported recent vaginal yeast infections, and pregnancy were each associated with increased odds of vulvovaginal candidiasis 1, 2, or 3, and diabetes was associated with increased risk of vulvovaginal candidiasis 1 and vulvovaginal candidiasis 3. No other predictor listed in Materials and Methods demonstrated an independent association with vulvovaginal candidiasis—not race, antibacterial drugs, contraceptive practices (including use of hormonal contraceptives), sexual behaviors, or nontherapeutic drug use. Neither reported current use of topical antifungals nor reported use of antifungal cream after a prior episode (data not shown) was significantly associated with current vulvovaginal candidiasis.
Finally, we examined whether vaginal yeast infections are more severe among HIV-infected women. Among women who had yeast on either culture or Gram stain and had at least one clinician-reported sign (ie, among women with vulvovaginal candidiasis 1), the risk of more symptomatic vulvovaginitis (the presence of two or more signs or vulvovaginal candidiasis 2) was not elevated among HIV-infected women (adjusted OR 0.97, P = .86). Similarly, among women with vulvovaginal candidiasis 1, the risk of extensive yeast overgrowth (vulvovaginal candidiasis 3) was essentially equal in HIV-infected and HIV-uninfected women (adjusted OR = 0.96, P = .77).
Vulvovaginal candidiasis incidence was higher in HIV-infected women than in HIV-uninfected women for all three definitions we used, regardless of vulvovaginal candidiasis status 6 months earlier. However, among women with vaginal infections, the risk of more severe yeast vulvovaginitis did not differ by HIV status. Additionally, among HIV-infected women, vulvovaginal candidiasis prevalence was related to the severity of HIV disease: Both lower CD4 count and higher HIV viral load were associated with increased prevalence. Within a CD4 count stratum, the prevalence of vulvovaginal candidiasis among women using highly active antiretroviral combination therapy did not differ from that among women on no antiretroviral therapy. This implies that increases in CD4 count among women taking highly active antiretroviral combination therapy may be associated with decreases in risk for vulvovaginal candidiasis, as women take on the lower risk associated with higher CD4 counts. Conversely, if a woman's CD4 count does not increase on highly active antiretroviral combination therapy, her vulvovaginal candidiasis risk will remain unchanged. Reported use of systemic antifungals, but not topical antifungals, was associated with a decreased risk of vulvovaginal candidiasis 1.
Our analysis used three definitions of vulvovaginal candidiasis based on Candida culture, the presence of yeast on Gram stain, and signs of vulvovaginitis. Using an alternative definition of vulvovaginal candidiasis, a cross-sectional analysis of baseline HER Study data found no significant difference in the prevalence of vulvovaginal candidiasis between HIV-infected and HIV-uninfected women.16 In that analysis, vulvovaginal candidiasis was defined as a positive Candida culture plus abnormal discharge and vaginal erythema or edema. Using that more restrictive definition, Candida vulvovaginitis was found in only 3.1% of HIV-infected women and 1.8% of HIV-uninfected women. Two other cross-sectional studies reported a positive association of vulvovaginal candidiasis with HIV and HIV-associated immune compromise, using two other definitions of vulvovaginal candidiasis: 1) positive Candida culture plus either clinician-reported abnormal white discharge or at least one self-reported symptom2 and 2) presence of budding yeast or hyphae on potassium hydroxide mounts of vaginal specimens.3
The increased risk of vulvovaginal candidiasis in women with low CD4 counts has been attributed to defective cell-mediated immunity, as seen in immunosuppressed patients who have increased risk for chronic mucocutaneous candidiasis and recurrent Candida vulvovaginitis.17,18 Other researchers have reported slightly larger association between antibacterial use and vulvovaginal candidiasis prevalence than reported here, but the associations did not achieve statistical significance in these studies either.1,3 Their findings and ours are consistent with a beneficial response to therapy among HIV-infected women and no significant risk for vulvovaginal candidiasis in women taking antibacterial agents.
The observational nature of our study limited some aspects of the analysis. Because we had incomplete data on the occurrence and treatment of symptomatic vulvovaginitis between visits, we were unable to include such episodes in our analysis. The study did not systematically collect data on response to treatment for vulvovaginal candidiasis. Although vulvovaginal candidiasis prevalence was not significantly decreased in either HIV-infected or HIV-uninfected women who reported topical antifungal use, topical therapy for vulvovaginal candidiasis is available over the counter and could have been used for conditions that were not really vulvovaginal candidiasis. It was not possible for us to analyze precisely the effectiveness of topical therapy for vulvovaginal candidiasis.
Despite these limitations, several strengths of this large, prospective design permitted a substantially improved understanding of the association of HIV status with vulvovaginal candidiasis. These include the following: a large number of HIV-infected and HIV-uninfected women; follow-up over several years; the fact that the HIV-infected and HIV-uninfected women were from the same communities and were matched at enrollment by HIV risk behavior; the fact that the study visits were not symptom driven; and the use of several definitions of vulvovaginal candidiasis, which allowed us to examine severity as well as incidence and prevalence. In addition, the detailed data collection at each study visit allowed us to control for both HIV-related covariates (CD4 count, viral load, and the use of antiretroviral and systemic antifungal agents) and covariates not related to HIV (such as coexistence of bacterial vaginosis, trichomoniasis, or genital ulcers; age; diabetes; and pregnancy).
Many practitioners have interpreted the inclusion of vulvovaginal candidiasis as a category B condition to mean that a patient with unknown HIV risk and recurrent vaginitis symptoms may have a sentinel event for HIV. Our results, however, do not support the use of vulvovaginal candidiasis status as a reliable sentinel for HIV. First, vulvovaginal candidiasis is common in HIV-uninfected women. Second, although it is more common in untreated HIV-infected women than in HIV-uninfected women, the increase is not substantial among women with high CD4 counts or low viral load. Although women with very low CD4 and high viral load have higher rates of vulvovaginal candidiasis, they are likely to manifest other, more classically AIDS-related sentinel conditions as well. Finally, among women with vulvovaginal candidiasis, the condition does not appear to differ substantively in severity between HIV-infected and HIV-uninfected women.
In conclusion, the results of this analysis will be useful clinically in helping to anticipate those HIV-infected women most likely to experience vulvovaginal candidiasis. These include women with low CD4 counts or high HIV viral loads. Our findings also imply that vulvovaginal candidiasis risk may be modulated by antiretroviral therapy; a decrease in vulvovaginal candidiasis risk can be expected for women whose immune function improves on highly active antiretroviral combination therapy but not for those without substantial CD4 increases. We also found high rates for vulvovaginal candidiasis among women with vulvovaginal candidiasis (or subclinical Candida colonization) within the past 6 months. Our data on treatment of women with vulvovaginal candidiasis are not definitive. Nonetheless, a high persistence rate was seen among HIV-infected women, even those who report intercurrent use of topical antifungals or current systemic antifungal use. This high persistence (or reinfection) rate implies that careful follow-up of HIV-infected women after treatment is warranted.
1. Burns DN, Tuomala R, Chang B, Hershow R, Minkoff H, Rodriguez E, et al. Vaginal colonization or infection with Candida albicans in human immunodeficiency virus-infected women during pregnancy and during the postpartum period. Women and Infants Transmission Study Group. Clin Infect Dis 1997;24:201–10.
2. Duerr A, Sierra MF, Feldman J, Clarke LM, Ehrlich I, DeHovitz J. Immune compromise and the prevalence of Candida vulvovaginitis in human immunodeficiency virus-infected women. Obstet Gynecol 1997;90:252–6.
3. Greenblatt RM, Bacchetti P, Barkan S, Augenbraun M, Silver S, Delapenha R, et al. Lower genital tract infections among HIV-infected and high-risk uninfected women: Findings of the Women's Interagency HIV Study (WIHS). Sex Transm Dis 1999;26:143–251.
4. Imam N, Carpenter CC, Mayer KH, Fisher A, Stein M, Danforth SB. Hierarchical pattern of mucosal Candida infections in HIV-seropositive women. Am J Med 1990; 89:142–6.
5. Rhoads JL, Wright DC, Redfield RR, Burke DS. Chronic vaginal candidiasis in women with human immunodeficiency virus infection. JAMA 1987;257:3105–7.
6. Spinillo A, Michelone G, Cavana C, Colonna L, Capuzzo E, Nicola S. Clinical and microbiological characteristics of symptomatic vulvovaginal candidiasis in HIV-seropositive women. Genitourin Med 1994;70:268–72.
7. Williams AB, Andrews S, Tashima K, Mezger J, Yu C. Factors associated with vaginal yeast infections in HIV-positive women. J Assoc Nurses AIDS Care 1998;9:47–52.
8. Centers for Disease Control and Prevention. 1993 revised classification system for HIV infection and expanded surveillance case definition for AIDS among adolescents and adults. MMWR Morb Mortal Wkly Rep 1992;41(RR-17):1–19.
9. World Health Organization. Acquired immunodeficiency syndrome (AIDS): Interim proposal for a WHO staging system for HIV infection and disease. Wkly Epidemiol Rec 1990;65(29):221–4.
10. Smith DK, Warren DL, Vlahov D, Schuman P, Stein MD, Greenberg BL, et al. Design and baseline participant characteristics of the Human Immunodeficiency Virus Epidemiology Research (HER) Study: A prospective cohort study of human immunodeficiency virus infection in US women. Am J Epidemiol 1997;146:459–69.
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. United States Pharmacopeial Convention. USP DI. Vol I, drug information for the health care professional. Rockville, Maryland: USP, 1999.
13. Panel on Clinical Practices for Treatment of HIV Infection. Guidelines for the use of antiretroviral agents in HIV-infected adults and adolescents. May 5, 1999. Available at: http://hivatis.org/guidelines/AAMay599.pdf
. Accessed 1999 Jul 14.
14. Kaplan EL, Meier P. Nonparametric estimator for incomplete observations. J Am Stat Assoc 1958;53:457–81.
15. Diggle PJ, Liang K-Y, Zeger SL. Analysis of longitudinal data. Oxford, United Kingdom: Clarendon Press, 1994.
16. 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.
17. Witkin S. Immunologic factors influencing susceptibility to recurrent candidal vaginitis. Clin Obstet Gynecol 1991;34:662–8.
The HER Study group consists of Robert S. Klein, MD, Ellie Schoenbaum, MD, Julia Arnsten, MD, MPH, Robert D. Burk, MD, Chee Jen Chang, PhD, Penelope Demas, PhD, and Andrea Howard, MD, MSc, from Montefiore Medical Center and the Albert Einstein College of Medicine; Paula Schuman, MD, and Jack Sobel, MD, from the Wayne State University School of Medicine; Anne Rompalo, MD, David Vlahov, PhD, and David Celentano, PhD, from the Johns Hopkins University School of Medicine; Charles Carpenter, MD, and Kenneth Mayer, MD, from the Brown University School of Medicine; Ann Duerr, MD, Lytt I. Gardner, PhD, Scott Holmberg, MD, Denise Jamieson, MD, Jan Moore, PhD, Ruby Phelps, Dawn Smith, MD, and Dora Warren, PhD, from the Centers for Disease Control and Prevention; and Katherine Davenny, PhD, from the National Institute of Drug Abuse.
18. Sobel JD. Candidal vulvovaginitis. Clin Obstet Gynecol 1993;36:153–65.