Prevalence and Determinants of Vaginal Flora Alterations in Postmenopausal Women

Urinary tract infections (UTIs) affect >7 million women per year in the United States and cost the health care industry >$1 billion annually [1]. The incidence of UTI increases markedly in older women, but the pathogenic mechanisms and epidemiology of UTI in noninstitutionalized older women have been little studied. Although it has long been recognized that the colonization of the vaginal introitus by uropathogens is a critical event preceding UTI, the mechanisms that facilitate this disturbance of the vaginal ecology remain elusive, especially among essentially healthy, community-dwelling, postmenopausal women. We have recently demonstrated that the propensity for vaginal colonization with Escherichia coli the organism that most commonly causes UTI, is increased in premenopausal women who do not have lactobacillus-dominant vaginal flora, specifically H₂O₂-producing lactobacilli [2]. This issue has not been assessed in postmenopausal women, yet it is particularly relevant, because this population of women is at high risk for E. coli vaginal colonization and UTI and is less likely to be responsive to preventive measures such as antimicrobial prophylaxis [3, 4]. For this reason, we conducted a study aimed at characterizing the vaginal microbiology of community-dwelling postmenopausal women with and without recent UTI

Subjects and methodsThe study population consisted of women who were enrolled in a larger population-based, case-control, interview study of epidemiological risk factors associated with UTI in community-dwelling postmenopausal women and who, in addition, volunteered to make a clinic visit [5]. The study was conducted at Group Health Cooperative, a mixed-model nonprofit health maintenance organization in Washington state. Women aged 55–75 years who had been members of Group Health for at least 1 year and had an ICD-9 diagnosis of acute UTI and a urine culture positive for ⩾10⁵ cfu/mL of a uropathogen within the previous month were eligible to be case patients. Each month, computerized laboratory files were reviewed to identify possible cases. The medical records of potential cases were reviewed to confirm the presence of symptoms compatible with a diagnosis of acute cystitis, including dysuria, frequency, or urgency for ⩽2 weeks. Women were excluded if their index culture was obtained during treatment for a previously diagnosed UTI or if they had had asymptomatic bacteriuria within the past year. Control patients were randomly selected age-matched enrollees of Group Health without an ICD-9 diagnosis of acute UTI or a urine culture with ⩾10⁵ cfu/mL of a uropathogen in the preceding month

Potentially eligible case and control patients were mailed a letter describing the study and then contacted by telephone after permission was obtained from their primary care physicians. After a description of study procedures, verbal informed consent was obtained from those subjects willing to participate. Women were excluded if they had a severe medical disability that might interfere with voiding, dementia, severe psychiatric disease, known anatomic abnormalities of the urinary tract, a urinary drainage appliance in place, end-stage renal disease, or active cancer or if they lived in a nursing home. Eligible and willing participants then underwent a detailed phone interview regarding their UTI history and covariates of interest. After the interview, all subjects were invited to attend a voluntary clinic visit for additional studies. These visits were conducted after resolution of the cases’ index UTI; thus, all women were completely asymptomatic and had not taken antibiotics for at least 4 weeks. During the clinic visit, we collected urine specimens and swabs of the vaginal introitus for culture. Women who made these visits constituted the sample for our analysis. Our main outcomes of interest were the rates and determinants of vaginal colonization by E. coli and lactobacilli. All procedures used in this study were approved by the Human Subjects Committees of the University of Washington and Group Health

A culture of the vaginal introitus was obtained with a cotton swab and transported to the laboratory in Amies transport medium (Becton Dickinson). Semiquantitative cultures for aerobic gram-negative rods, cocci, and lactobacilli were done using standard laboratory methods [6, 7]. Results were categorized semiquantitatively as 0 (no growth), 1+ (colonies in the first streak zone), 2+ (1–50 colonies in the second streak zone), 3+ (confluence in the second streak zone or 1–10 colonies in the third streak zone), and 4+ (>10 colonies in the third streak zone)

We first characterized the participants by their UTI status and then evaluated factors that were independently associated with our main outcomes of interest—vaginal colonization by E. coli and lactobacilli. Data were analyzed using SPSS (version 10.0.5; SPSS). Risk factors that were potentially associated with E. coli and lactobacillus vaginal colonization were examined by using odds ratios and 95% confidence intervals (CIs), according to standard methods [8]. All variables that were univariately associated (P⩽.2) with the vaginal organism of interest were included in the multivariate model

ResultsOf the 899 case and 911 control subjects who participated in interviews, 463 women (269 case and 194 control subjects) made a clinic visit and were eligible for our analyses. Women who agreed to make a clinic visit were similar to women who did not make a visit, with regard to mean age, race, income levels, history of UTI, presence of diabetes mellitus, and sexual activity. Case women who made a visit were more likely than case women who did not make a visit to have used hormone replacement therapy (HRT) during the preceding year (57% vs. 45%, respectively) and to have postgraduate education (64% vs. 55%, respectively). The subgroup of case and control women included in the study were comparable to each other with regard to mean age, ethnicity, education and income level, parity, prior genitourinary surgery, and occurrence of a UTI at age <18 years (table 1). The proportions of women with a diagnosis of diabetes mellitus, use of HRT, or being sexually active during the previous year were also similar between case and control subjects. Case women were more likely to report incontinence, defined as accidental leakage of urine, ⩾1 time per week on average, compared with control subjects

The prevalence of vaginal colonization with lactobacilli, E. coli enterococci, other gram-negative and -positive bacteria, and yeast are shown in figure 1. The overall prevalence of lactobacilli in any quantity was similar in case and control subjects (62% and 61%, respectively). Heavy growth of lactobacilli, defined as >10 colonies in the third streak when plated on agar, was also similar in case and control subjects (48% and 45%, respectively). The prevalence of vaginal E. coli was 39% overall and was significantly increased in case subjects (44%) versus control subjects (32%) (P=.01). Enterococci were likewise present in 39% of women and were significantly increased in case subjects (44%) versus control subjects (32%) (P<.01). The rates of colonization with the other organisms were similar in both groups of women

The prevalence of vaginal colonization with E. coli enterococci, and other gram-negative organisms were compared between women with heavy growth of lactobacilli and those with light or no growth of lactobacilli. Women with heavy growth of lactobacilli were less likely to have vaginal colonization with E. coli than were women with light or no growth of lactobacilli (34% vs. 45%; P=.025). Vaginal colonization with enterococci and gram-negative rods other than E. coli was not associated with the presence or quantity of vaginal lactobacilli

The use of HRT was defined as the self-reported use of any hormone medication other than birth control pills, including estrogen pills, capsules, vaginal creams, or patches or progestin pills. The vast majority (93%) of the women taking HRT were using systemic estrogen. Approximately one-half of these women were also taking progesterone. Eighty-one percent of women using systemic or topical HRT had vaginal colonization with lactobacilli, compared with 41% of women not using HRT (P<.001; figure 2). Heavy growth of lactobacilli was demonstrated in 65% of women using systemic HRT, 54% of women using topical HRT, and 25% of women using no HRT (P<.001; figure 2). There was no association of HRT use with E. coli vaginal colonization (38% of women using HRT and 40% of women not using HRT had vaginal colonization with E. coli)

Exposures that were not significantly associated with the prevalence of vaginal lactobacilli or E. coli included age, incontinence, frequency of sexual activity, and lifetime history of UTIs (data not shown). Having a diagnosis of diabetes mellitus also was not significantly associated with the presence of lactobacilli. However, there was an increased prevalence of E. coli colonization in women with insulin-treated diabetes (62%) versus women without diabetes or with diabetes not treated with insulin (38%; P=.05)

In a multivariable model, we found that the factors independently associated with E. coli vaginal colonization included case-control status (or history of recent UTI) and the absence or low levels of vaginal lactobacilli (table 2). There was also a statistical trend for a 3-fold increased likelihood of E. coli vaginal colonization among women with insulin-treated diabetes mellitus (table 2). The only variable independently associated with vaginal lactobacilli colonization was the use of HRT during the past year, with a 5–6-fold increased likelihood of having vaginal colonization with lactobacilli among women who used HRT, compared with women who did not (P<.001). This was true for colonization with any lactobacilli as well as for heavy amounts of lactobacilli

DiscussionThe vaginal ecology plays an important role in the pathogenesis of UTI in women. For this reason, the vaginal microbial flora have been extensively characterized in young, sexually active women, who are at high risk for UTI. However, little is known about the vaginal flora of community-dwelling postmenopausal women. Our study describes the prevalence of vaginal lactobacilli, as well as potential uropathogens, in an essentially healthy, asymptomatic sample of postmenopausal women. In addition, we have evaluated the association between recent UTI and other common exposures possibly associated with UTI in this age group, such as incontinence, diabetes, and the use of HRT, with vaginal microbial flora

Our data demonstrate that community-dwelling postmenopausal women have a much lower prevalence of vaginal lactobacilli than healthy premenopausal women. Eschenbach et al. [9] demonstrated that 96% of healthy premenopausal women had H₂O₂-producing lactobacilli. Only 62% of women in our study had any vaginal lactobacilli, and fewer than half had heavy growth of lactobacilli. In a study by Hillier et al. [10], 49% of postmenopausal women who were not using estrogen had any lactobacilli, and only 38% had H₂O₂-producing strains of lactobacilli. This is similar to the 41% lactobacillus colonization rate in the non–HRT-using group of women in our study. We did not characterize the lactobacilli as to H₂O₂ production, so we cannot comment on this issue

We also found that E. coli vaginal colonization was unexpectedly common, occurring in 39% of the study women. This rate is much higher than that reported in healthy premenopausal women [2, 11, 12] and is similar to the 35%–45% rate of E. coli colonization demonstrated in asymptomatic premenopausal women with recurrent UTI [2]. Of importance, the rate of colonization with E. coli was inversely associated with the presence of lactobacilli, as has also been demonstrated in younger women [2, 13]. This suggests that the presence of vaginal lactobacilli may play a role in the prevention of vaginal colonization by uropathogens and subsequent UTI in community-dwelling postmenopausal women. The fact that the restoration of vaginal lactobacilli via topical estrogen in postmenopausal women with recurrent UTI was associated with a marked reduction in UTI incidence is also consistent with a protective role of vaginal lactobacilli [14]

In agreement with the results of previous studies of vaginal flora and UTI in premenopausal women, we found that postmenopausal women with a recent history of symptomatic UTI were at increased risk of E. coli vaginal colonization [15–17]. Of interest, sexual activity was not associated with E. coli vaginal colonization. This is in contrast to the results of studies in premenopausal women, for whom sexual activity and the use of a diaphragm with spermicide are 2 of the strongest determinants of E. coli vaginal colonization and acute UTI [12, 18]. However, the frequency of sexual activity has not been shown to be associated with UTI risk in older women [19, 20]. It is possible that the frequency of sexual intercourse in postmenopausal women is too low to significantly affect the vaginal flora and the subsequent risk of UTI or that other, more strongly associated factors obscure a smaller role of sexual activity in UTI risk in this population

Insulin-treated diabetes was associated with an almost 3-fold increased risk of E. coli vaginal colonization. The wide CIs and significance values consistent with a statistical trend in the multivariate model are likely due to the small numbers of women with this exposure. This is one of the first clinical studies to have demonstrated an increased prevalence of E. coli vaginal colonization among women with insulin-treated diabetes. An in vitro study has shown that type 1 fimbriated E. coli adhere in significantly higher numbers to uroepithelial cells from women with diabetes, compared with control women [21]. Recently, Boyko et al. [5] determined that postmenopausal women with diabetes that requires treatment with oral hypoglycemic agents or insulin were at increased risk of symptomatic UTI. Similarly, both insulin-dependent and non–insulin-dependent diabetes were shown to be associated with UTI in postmenopausal women in a study by Brown et al. [20]. Our findings suggest that an increased propensity for E. coli vaginal colonization among women with insulin-treated diabetes may be one of the mechanisms for the increased risk of UTI in this group. Diabetes not requiring insulin was not an independent risk factor for E. coli vaginal colonization or for UTI in this population [5]

Perhaps most interesting is the very strong association we demonstrated between the presence of vaginal lactobacilli and HRT use. The association was seen regardless of the type of estrogen or mode of delivery (systemic vs. topical). This finding is consistent with the results of Raz and Stamm [14], which showed repletion of lactobacilli in 61% of women with recurrent UTI who were treated with topical estrogen. Cauci et al. [22] also demonstrated markedly lower rates of vaginal lactobacilli, as assessed by Gram’s stain, in postmenopausal women not using HRT (44% had no lactobacilli), compared with postmenopausal women using HRT (7% had no lactobacilli). Given the current controversy over the possible benefits and risks associated with HRT, our finding of a beneficial effect of HRT on the vaginal microbial flora is particularly important and supports further study of the relationships among the use of HRT, vaginal microbial flora, and the risk of UTI in postmenopausal women

Our study has some potential limitations. The patients were not seen at the time of acute UTI; thus, there was a temporal lag of several weeks between the time of infection and the ascertainment of the vaginal microbiology. In the context of the current analyses, this does not pose a limitation; rather, the finding of high rates of E. coli colonization is all the more compelling because it is occurring among asymptomatic women. High rates of vaginal E. coli colonization are expected at the time of acute UTI but not necessarily during asymptomatic periods. In addition, because vaginal cultures were done at least 4 weeks after acute UTI, it is unlikely that antibiotics used to treat the UTI affected our findings. Recall bias is a concern with case-control studies, but this was minimized by interviewing subjects as soon as possible after the index UTI and before the results of the vaginal cultures were available. Last, our population was primarily white, and the exposures related to alterations in vaginal flora that we identified should be assessed in other ethnic groups

In conclusion, the present results provide important information regarding the vaginal ecology of community-dwelling postmenopausal women with and without a recent UTI. The relative depletion of lactobacilli, especially among women not using HRT, and the increased prevalence of E. coli in women without lactobacilli support a potential role for the restoration of vaginal lactobacilli via a lactobacillus probiotic or topical estrogen. In addition, the finding of increased E. coli colonization among insulin-using women with diabetes provides insight into the pathophysiology of the increased UTI risk seen in this population

• Received February 12, 2003.
• Accepted April 11, 2003.

• © 2003 by the Infectious Diseases Society of America