152

Recurrent Urinary Tract Infections in Postmenopausal Women

Raul Raz,1 Yoshua Gennesin,1 Joseph Wasser,2 Zmira Stoler,2 Sophia Rosenfeld,1 Elimelech Rottensterich,2 and Walter E. Stamm3
From the 1Infectious Diseases Unit, Haemek Medical Center, and Technion Faculty of Medicine, and 2Zvulum Clinic, Haifa, Israel; and 3Department of Medicine, University of Washington, Seattle, Washington, USA

To evaluate factors associated with recurrent urinary tract infection (UTI) in postmenopausal women, we conducted a case-control study comparing 149 postmenopausal women referred to an infectious diseases outpatient clinic who had a history of recurrent UTI (case patients) with 53 age-matched women without a history of UTI (control patients). Each woman completed a questionnaire providing demographic data, history and clinical characteristics of prior infections, and information regarding risk factors for UTI. In addition, each patient underwent a gynecologic evaluation, renal ultrasound and urine ow studies, and blood group and secretor status testing. Three urologic factorsnamely, incontinence (41% of case patients vs. 9.0% of control patients; P ! .001), presence of a cystocele (19% vs. 0%; P ! .001), and postvoiding residual urine (28% vs. 2.0%; P = .00008)were all strongly associated with recurrent UTI. Multivariate analysis showed that urinary incontinence (odds ratio [OR], 5.79; 95% condence interval [CI], 2.0516.42; P = .0009 ), a history of UTI before menopause (OR, 4.85; 95% CI, 1.713.84; P = .003), and nonsecretor status (OR, 2.9; 95% CI, 1.286.25; P = .005) were most strongly associated with recurrent UTI in postmenopausal women. Prospective studies are needed to conrm these observations and to develop approaches for prevention.

Downloaded from http://cid.oxfordjournals.org/ by guest on January 3, 2013

Urinary tract infection (UTI) remains one of the most common bacterial infections seen in adult women of all ages [1,2]. On a global basis, an estimated 150 million UTIs occur annually, and in the United States, UTIs account for 16 billion dollars in direct health care costs [2]. Recent prospective studies found that the incidence of symptomatic UTIs among young women initiating contraception was as high as 0.50.7 case per person per year [3]. The prevalence of UTIs among women is thought to increase even further with age, with bacteriuria occurring in 10%15% of women aged 6570 years and 15%20% of women aged 180 years [46]. Despite the high prevalence of bacteriuria and the apparent high incidence of symptomatic UTIs among healthy postmenopausal women, the factors predisposing such women to UTI have been little explored as compared with those for either premenopausal women or older institutionalized women. For younger women, sexual intercourse, diaphragm and/or spermicide exposure, a history of UTI, nonsecretor status, and

recent antimicrobial exposure have been identied as important host characteristics associated with UTI [3, 711]. For older institutionalized women, urinary catheterization and functional status are the most important risk factors associated with UTI [12]. Many of the risk factors associated with UTI in both of these population groups were initially identied by studying women with recurrent UTIs, whose host characteristics associated with UTI may be more easily demonstrated than those of women with sporadic UTIs. We thus undertook a casecontrol study to assess factors associated with recurrent UTIs in healthy postmenopausal women.

Material and Methods

Study population and subject recruitment. Women with recurrent symptomatic UTI were referred from family practitioners to the Infectious Diseases Outpatient Clinic at Haemek Medical Center, Haifa, Israel. Approximately 2000 women with recurrent urinary infections visit the clinic annually. During the period of study, consecutive postmenopausal women with recurrent UTI (dened as 3 microbiologically documented episodes of symptomatic UTI managed in the clinic during the last year or 2 episodes during the last 6 months) were offered enrollment in the study. All of the enrolled women (case patients) were generally healthy, and all were ambulatory and did not have indwelling catheters. An age-matched group of women who visited the gastroenterology and metabolic units of the outpatient clinics and gave no history of UTI since the onset of menopause were invited to participate in the study as control patients. To minimize the number of control patients undergoing procedures that were of no benet to them, 1 control patient was selected for every 3 case patients

Received 22 March 1999; revised 15 September 1999; electronically published 6 January 2000. This work was presented in part at the 37th Interscience Conference on Antimicrobial Agents and Chemotherapy held on 28 September to 1 October 1997 in Toronto, Canada Grant support: This work was supported by the National Institutes of Health (grants DK 40045 and DK 53369). Reprints or correspondence: Dr. Walter E. Stamm, Division of Allergy and Infectious Diseases, University of Washington School of Medicine, 1959 NE Pacic Street (356523), Seattle, WA 98195 (wes@u.washington.edu).
Clinical Infectious Diseases 2000; 30:1526 2000 by the Infectious Diseases Society of America. All rights reserved. 1058-4838/2000/3001-0023$03.00

CID 2000;30 (January)

Recurrent UTIs in Postmenopausal Women

153

enrolled. All participants provided informed consent, and the study was approved by the National Human Subjects Review Board, Jerusalem. Study procedures. Evaluation of both case and control patients was undertaken when the women were asymptomatic and abacteriuric. All of the case and control patients completed a nurseadministered questionnaire inquiring about their general medical history, marital status, number of pregnancies and deliveries, underlying diseases, and previous surgical procedures. They were also questioned about their history of UTIs, including age at rst UTI, factors potentially related to its appearance, and urinary incontinence. Urinary incontinence was dened as follows: urinary incontinence, involuntary loss of urine that is a social or hygienic problem and is objectively demonstrable; stress incontinence, dysfunction of the bladder outlet that leads to urinary leakage when intraabdominal pressure is raised above urethral resistance during exertion, such as coughing, bending, or lifting heavy objects; and urge incontinence, the urge to void but an inability to inhibit leakage long enough to reach the toilet. Cystitis and pyelonephritis were dened utilizing clinical criteria, specically dysuria; urgency and frequency (cystitis); and fever, chills, and/or loin pain (pyelonephritis). After completion of the questionnaire, the patients underwent renal ultrasonography with a postvoiding step to evaluate the volume of residual urine (Diasonic, Santa Clara, CA). The evaluation was conducted within several minutes of voiding and was performed by a technician who had no knowledge of the case or control status of the patient. The postvoiding renal volume was calculated automatically by the ultrasound equipment. The residual urine volume was dened as follows: mild, persistence of 50 mL of urine after micturition; moderate, persistence of a residual urine volume of 50100 mL; and severe, persistence of a residual urine volume 1100 mL. Urinary ow was measured by use of a uroowmeter; and women were classied as having a normal ow (015 mL/s) or as having mild (1015 mL/s), moderate (510 mL/ s), and severe (!5 mL/s) degrees of obstruction based on their peak urine ow rate. A gynecologic evaluation was performed by 1 gynecologist (J.W.) to determine the presence of uterine, bladder, or rectal prolapse. Prolapse was dened as a downward or forward displacement of 1 of the pelvic organs from its normal location. These displacements have usually been graded on a scale of 03; the grades rise with increasing severity of the prolapse, with 0 referring to no prolapse and 3 referring to complete prolapse. All forms of female genital prolapse are described with reference to the vaginal canal. Rectal prolapse refers to protrusion of the rectum into the posterior vaginal lumen. The gynecologist was not aware of the patients case or control status. For blood typing and secretor status testing, specimens of unstimulated whole saliva and blood were collected. Secretor status was determined by a hemagglutination inhibition assay [13]. Saliva was rst centrifuged at 4000 g for 15 minutes, and the supernatant was then boiled for 10 minutes and incubated at room temperature with 3 antisera (antisera to A, B, and H antigens). After 15 minutes, an equal volume of a suspension of 3% human erythrocytes in PBS (140 mM sodium chloride, 10 mM potassium phosphate, pH 7.4) was added. After 10 minutes, the presence or absence of hemag-

glutination was determined; saliva from secretors inhibited agglutination, whereas saliva from nonsecretors did not. Statistical analysis. SPSS software (SPSS, Chicago) was used to determine signicant differences between variables and to perform univariate and multivariate analyses for risk factors. In the univariate analysis, the x2 test for categorical variables and the Students t test for numerical variables were used. Logistic regression analysis was carried out to identify variables that were independently associated with recurrent UTI. The variables in the selected model were chosen by backward elimination. Variables included in the model were a history of UTI, urogenital interventions, secretor status, urinary ow, and age.

Results
Comparability of Case and Control Patients

Downloaded from http://cid.oxfordjournals.org/ by guest on January 3, 2013

One hundred forty-nine postmenopausal women with recurrent UTI (case patients) and 53 postmenopausal women without UTI (control patients) completed the study and were included in the nal study population (tables 1 and 2). The mean age ( SD) of the patients was 65.7 7.2 years and that of the control patients was 66.6 6.6 years. The mean number ( SD) of previous childbirths per patient was also similar in the 2 groups: 2.5 1.2 for case patients and 2.3 1.4 for control patients. Marital status and the systemic use of estrogen were proportionally similar in the 2 groups. When asked to identify factors possibly associated with their recurrent UTIs, most case patients (78%) did not relate any specic factor to the onset of recurrent UTI. However, 27 case patients (18%) believed that their UTIs started after a gynecologic surgical intervention. Only 3 case patients (2%) related their UTIs to sexual activity.
Factors Associated with Recurrent UTIs

Underlying medical and surgical conditions. A history of UTI before menopause or in childhood was more frequent in case patients than in control patients (table 1). Thus, only 6 (11%) of 53 control patients had had at least 1 episode of UTI before menopause as compared with 44 (30%) of the case patients (OR, 3.14; 95% CI, 1.328.36; P = .008). No differences were seen in the prevalence of other underlying diseases in the 2 groups, including diabetes mellitus (table 1). More case patients than control patients had undergone urogenital surgery (27% vs. 13%, respectively; OR, 2.41; 95% CI, 1.015.78; P = .04; table 2). The reasons for surgery were total hysterectomy due to myoma or uterine carcinoma in 33 case patients and 5 control patients and prolapse of the urinary bladder in 7 case patients and 2 control patients. Urologic factors. Incontinence was present in 61 (41%) of 153 case patients as compared with 5 (9%) of 53 control patients (OR, 6.65; 95% CI, 2.517.7; P ! .001; table 2). None of the control patients had a cystocele observed during gynecologic evaluation in contrast to 19% of the case patients (P ! .001).

154

Raz et al.

CID 2000;30 (January)

Table 1. Univariate analysis of demographic and historical factors associated with recurrent UTI in postmenopausal women.
Case patients (n = 149) 65.7 2.5 115 10 24 41 108 30 10 44 (28) (72) (20) (7) (30) 7.2 1.2 Control patients (n = 53) 66.6 2.3 41 3 9 8 45 9 0 6 (15) (85) (17) (11) 6.6 1.4

Factor Age, y No. of children Marital status Married Divorced Widowed Current estrogen use Oral and/or vaginal None Diabetes mellitus UTI in childhood UTI before menopause NOTE. infection.

OR (95% CI) ( 4.28 to 0.17) ( 0.61 to 0.15) 0.9 (0.421.96)

P .07 .24 .8

2.14 (0.895.68) 1.23 (0.542.8) Undened 3.14 (1.328.36)

.07 .62 .06 .008

Data are no. (%) of subjects or mean

SD. UTI, urinary tract

Of the case patients, 9 (6%) had a mild cystocele, 18 (12%) had a moderate cystocele, and 2 (1%) had severe prolapse of the bladder. Residual urinary volume was considerably more common in case patients than in control patients. Thus, only 1 control patient (2%) had a measured residual urine volume as compared with 41 (28%) of the case patients (P = .0008 ). Most case patients had mild or moderate residual urine volumes; only 6 case patients had large residual urine volumes. Urinary ow was reduced in 67 (45%) of the case patients and only 12 (23%) of the control patients (OR, 2.79; 95% CI, 1.316.29; P = .004). Most of the case patients had only mild or moderate degrees of obstruction to urine ow (table 2). Blood group and secretor status. There were no differences between the case patients and control patients in the distribution of blood groups or Rh factor. However, 76 (51%) of the case patients were nonsecretors versus 16 (30.2%) of the control patients (OR, 2.41; 95% CI, 1.234.7; P = .008; table 3). Logistic regression analysis. To better dene those factors independently associated with recurrent UTI in the case patients, we carried out logistic regression analysis (table 4). Even though women with either a cystocele or a residual urine volume after voiding had a very high positive predictive value for recurrent UTIs, they were excluded from the logistic regression analysis since these conditions were essentially absent in the control patients. The factors independently associated with recurrent UTIs in the multivariate analysis were urinary incontinence (OR, 5.79; 95% CI, 2.0516.42; P = .009 ), a history of UTI (OR, 4.85; 95% CI, 1.713.84; P = .003), and nonsecretor status (OR, 2.9; 95% CI, 1.286.25; P = .005).

recurrent UTI differ by age and functional status. At least 3 groups of women with recurrent UTI should probably be distinguished: premenopausal women, healthy postmenopausal women (the group studied here), and elderly institutionalized women. Scholes et al. [14] reported that the major risk factors for recurrent UTI in premenopausal women were sexual intercourse and spermicide exposure. In addition, a history of UTI, a history of UTIs in childhood, and a history of maternal UTIs may be associated with recurrent UTI in premenopausal women, which suggests a possible genetic predisposition to recurrent UTI in at least some women [14]. Urologic factors (such as incontinence, postvoiding residual urine volume, abnormal urinary ow, and cystocele) were infrequent and not associated with recurrent UTI in these younger women [15]. We undertook this study to ascertain risk factors for recurrent UTIs in healthy postmenopausal women who were not institutionalized and did not have a catheter. Our ndings indicate that most of the risk factors associated with recurrent UTIs in these women differ from those found in premenopausal women. In particular, mechanical and/or physiological factors that affect bladder emptying were strongly associated with recurrent UTIs in the women we studied but have generally not been associated with recurrent UTI in premenopausal women. A signicant postvoiding residual urine volume was present in more than one-quarter of our case patients and in only 1 control patient and therefore had a very high positive predictive value for recurrent UTI. Similarly, 18% of the case patients in our study had low-grade cystoceles, whereas none of the control patients had cystoceles (positive predictive value for recurrent UTI, 100%). Urinary incontinence (stress, urge, or both) was also observed more frequently in our case patients than in our control patients (41% vs. 9%, respectively) and was strongly associated with recurrent UTI in the multivariate analysis. Incontinence is very common in older women and has been estimated in populationbased studies to occur in 17%46% of community-dwelling women aged 160 years [16]. Further studies are needed to conrm the association of incontinence with recurrent UTI and to ascertain how incontinence predisposes women to recurrent infection. Urinary incontinence increases in frequency with age and in the presence of other associated risk factors, including obesity, medical illnesses, smoking, and alcohol consumption.
Table 2. Univariate analysis of urologic factors associated with recurrent UTI in postmenopausal women.
Case patients (n = 149) 41 67 40 61 29 (23) (45) (27) (41) (19) Control patients (n = 53) 1 12 7 5 0 (2) (23) (13) (9)

Downloaded from http://cid.oxfordjournals.org/ by guest on January 3, 2013

Factor Residual urine volume Reduced urine ow Urogynecologic surgery Incontinence Any cystocele

OR (95% CI) Undened 2.79 (1.316.29) 2.41 (1.015.78) 6.65 (2.517.7) Undened

P .00008 .004 .04 !.001 !.001

Discussion Recurrent UTI remains an important public health problem in women of all ages. The data reported here and in other recent studies of recurrent UTI suggest that the factors underlying

NOTE. Data are no (%) of subjects with the indicated factor. UTI, urinary tract infection.

CID 2000;30 (January)

Recurrent UTIs in Postmenopausal Women

155

Table 3. Univariate analysis of blood group and secretor status in postmenopausal women with recurrent UTI and in control patients.
Case patients (n = 149) 39 65 38 5 17 130 105 76 (26.5) (44.2) (25.5) (3.4) (11.6) (88.4) (71) (51) Control patients (n = 53) 19 20 11 3 9 44 32 16 (35.8) (37.7) (20.8) (5.7) (17) (83) (60.4) (30.2)

Factor Blood group O A B AB Rh factor Positive Negative P blood group Nonsecretor NOTE.

OR (95% CI) 1 0.63 (0.281.42) 0.59 (0.231.53) 1.23 (0.717.1) 1.56 (0.653.76) 1.6 (0.833.08) 2.41 (1.234.7)

P .48

.31

.16 .008

Data are no. (%) of patients. UTI, urinary tract infection.

A recent study by Brown et al. [17] showed that obesity and a history of hysterectomy had a substantial impact on the prevalence of urinary incontinence; in this study, hysterectomy was performed primarily for leiomyomas (65%70% of case patients) and infrequently for prolapse (10%) or incontinence (4%). In our study, 27% of the case patients had undergone gynecologic surgery (33 of 40 were hysterectomies) as compared with only 13% of the control patients. Most of these procedures were performed for leiomyomas and not for incontinence or prolapse. Thus, recurrent episodes of UTI in these women had their onset after urogenital surgery. Hysterectomy may cause damage to the urethral and bladder supportive structures or to the pelvic plexus, leading to subsequent incontinence later in life [18]. In addition, other gynecologic procedures that alter the normal anatomy of the pelvis are also likely to increase the risk of recurrent UTIs. However, in our study, there was an insufcient number of such procedures, given our sample size, to adequately assess them as risk factors for recurrent UTIs. Inherited factors may also play a role in predisposing postmenopausal women to recurrent UTIs. Thus case patients in our study reported a history of childhood and premenopausal UTIs signicantly more frequently than did the control patients, which suggests a lifelong factor or factors that predispose them to urinary infection. A history of premenopausal UTIs remained strongly correlated with recurrent UTI in postmenopausal women in the multivariate analysis. Nonsecretor status was also an inherited characteristic strongly associated with recurrent UTI in the multivariate regression analysis. It is interesting that the previously reported association of nonsecretor status with recurrent UTI has been most evident in populations of women aged 135 years. The mean ages of the populations described by Sheinfeld et al. [9] and Kinane et al. [10], in whom nonsecretor status was associated with recurrent UTI, were 38 and 55 years, respectively, and both populations included many postmenopausal women. The mechanism through which nonsecretor status predisposes to recurrent UTI is probably the presence on vaginal and uroepithelial cells of 2 unique nonsecretor-associated glycolipids that serve as binding sites for specic Escherichia coli

adhesions [19, 20]. It is possible that expression of these nonsecretor-associated glycolipids varies with age or hormonal status. Alternatively, nonsecretor status may become more important in older women when other risk factors such as sexual activity and spermicide exposure recede. Finally, although not directly evaluated in this study, the reduced levels of estrogenic hormones present after menopause appear to contribute to the occurrence of recurrent UTI in postmenopausal women. In a prior double-blind, placebocontrolled intervention study, we demonstrated that topically applied intravaginal estrogen markedly reduced the incidence of recurrent UTI [12]; concomitantly, E. coli vaginal colonization was reduced by one-half, lactobacillus colonization was reestablished in most women, and the vaginal pH fell signicantly. In contrast, the effects of oral estrogen on the risk of UTI are less clear and need further study. Prior investigators have reported both increased and decreased risks of UTI associated with the use of estrogenic hormones [21, 22]. In our current study, estrogen use was relatively infrequent in both case and control patients and was not related to recurrent UTI. Given the small number of estrogen users, however, the study lacked power to appropriately address this issue. We undertook this case-control study to identify factors potentially associated with recurrent UTI, so that such factors could be subsequently further studied in more denitive prospective studies. Several limitations of our study deserve discussion. Our case patients were referred by family practitioners to a special consultation clinic for evaluation of recurrent UTI. Although this is standard practice for such practitioners, we cannot be certain whether the referred patients studied in our clinic are representative of all women with recurrent UTI. Similarly, our age-matched control patients were recruited and enrolled in other specialty clinics at the hospital and are probably not representative of all women in the community who do not have recurrent UTI. Nevertheless, the case and control patients were quite comparable in terms of demographic factors and by virtue of all being outpatients recruited from specialty clinics at the same medical center. Finally, we were unable to carefully assess the relationship between sexual intercourse and recurrent UTI in the study because of the reticence of our patients to discuss sexual practices. However, only 2% of women with recurrent UTI thought that coitus was related to UTI, and it is thus unlikely that sexual intercourse was a major predisposing factor in women in the study. Further study of this issue, however, is warranted. De-

Downloaded from http://cid.oxfordjournals.org/ by guest on January 3, 2013

Table 4. Multivariate analysis of factors predisposing postmenopausal women to UTI.

Factor Urinary incontinence History of UTI Nonsecretor NOTE. OR (95% CI) 5.79 (2.0516.42) 4.85 (1.713.84) 2.9 (1.286.25) P .0009 .003 .005

UTI, urinary tract infection.

156

Raz et al.

CID 2000;30 (January)

spite these limitations, our study identied strong associations between recurrent UTI and several plausible variables of interest that should now be further studied prospectively. In conclusion, our data support the view that several factors importantly inuence the incidence of recurrent UTI among postmenopausal women. These factors include a history of UTIs, nonsecretor status, and possibly other inherited predispositions, as well as urodynamic factors, especially incontinence, residual urine volume, and presence of a cystocele. These major factors predisposing to recurrent UTI in postmenopausal women differ importantly from the factors that predominate in premenopausal women with recurrent UTI. However, these factors have been identied primarily in case-control or intervention studies, and a prospective cohort study of postmenopausal women to determine more precisely the incidence of and risk factors for recurrent UTI in this group would be of great interest. Proposed interventions to reduce the incidence of recurrent UTI among postmenopausal women must take the multifactorial nature of this condition into account.
Acknowledgments We thank Frances Nachmani for her assistance in typing the manuscript, Dr. H. Ben-Arie for performing the secretor status studies, and Dr. Kvitko for determining the blood groups.

References 1. Schappert SM. National ambulatory medical care survey: 1992 summary. Advanced data from vital and health statistics, no 253. Hyattsville, MD: National Center for Health Statistics, 1994, DHHS publication:941250. 2. Ronald AR. Editorial comment: sexually transmitted diseases and urinary tract infections. Curr Opin Infect Dis 1995; 8:13. 3. 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; 335:46874. 4. Nygaard IE, Johnson JM. Urinary tract infection in elderly women. Am Fam Physician 1996; 53:17582. 5. Nicolle LE. Urinary tract infection in the elderly. J Antimicrob Chemother 1994; 33(Suppl A):99109.

6. Boscia JA, Kobasa WD, Knight RA, et al. Epidemiology of bacteriuria in an elderly ambulatory population. Am J Med 1986; 80:20814. 7. Strom BL, Collins M, West SL, Kreisberg J, Weller S. Sexual activity, contraceptive use and other risk factors for symptomatic and asymptomatic bacteriuria: a case-control study. Ann Intern Med 1987; 107:81623. 8. Foxman B, Fredrichs RR. Epidemiology of urinary tract infection. I. Diaphragm use and sexual intercourse. Am J Public Health 1985; 75:130813. 9. Sheinfeld J, Schaeffer AJ, Cordon-Cardo C, Rogatko A, Fair WR. Association of the Lewis blood-group phenotype with recurrent urinary tract infections in women. N Engl J Med 1989; 320:7737. 10. Kinane DF, Blackwell CC, Brettle RP, Weir DM, Winstanley FP, Elton RA. ABO blood group, secretor status and susceptibility to recurrent urinary tract infection in women. BMJ 1982; 285:79. 11. Smith HS, Hughes JP, Hooton TM, et al. Antecedent antimicrobial use increases the risk of uncomplicated cystitis in young women. Clin Infect Dis 1997; 25:638. 12. Raz R, Stamm WE. A controlled trial of intravaginal estriol in postmenopausal women with recurrent urinary tract infections. N Engl J Med 1993; 329:7536. 13. Eriksson W, Partanen K, Frank RD, et al. ABH secretion polymorphism in Icelanders, Aland Icelanders, Finns, Finnish Lapps, Komi and Greenland Eskimos: a review and new data. Ann Hum Biol 1986; 3:27385. 14. Scholes D, Hooton TM, Roberts PL, et al. Risk factors for recurrent UTI in young women [abstract]. Clin Infect Dis 1997; 25:440. 15. Hooton TM, Stapleton AE, Roberts PL, et al. Perineal anatomy and urine voiding characteristics in young women with and without recurrent urinary tract infections. Clin Infect Dis 2000, in press. 16. Versi E. Incontinence in the climacteric. Clin Obstet Gynecol 1990; 33:3928. 17. Brown JS, Seeley DG, Fong J, et al. Urinary incontinence in older women: who is at risk? Obstet Gynecol 1996; 87:71521. 18. Taylor T, Smith A, Fulton M. Effects of hysterectomy on bowel and bladder function. Int J Colorectal Dis 1990; 5:22831. 19. Lomberg HB, Cedergren H, Lefer B, Nilson AS, Svanberg-Eden C. Inuence of blood group on the availability of receptors for attachment of uropathogenic E. coli. Infect Immun 1986; 51:91926. 20. Stapleton A, Hooton TM, Fennell C, Roberts DL, Stamm W. Effect of secretor status on vaginal and rectal colonization with mbriated Escherichia coli in women with and without recurrent urinary tract infection. J Infect Dis 1995; 171:71720. 21. Orlander JD, Jick SS, Dean AD, Jick H. Urinary tract infections and estrogen use in older women. J Am Geriatr Soc 1992; 40:81720. 22. Kirkengen AL, Andersen P, Gjersoe E, Johannessen GR, Johnson N, Bodd E. Oestriol in the prophylactic treatment of recurrent urinary tract infections in postmenopausal women. Scand J Prim Health Care 1992; 10: 13942.

Downloaded from http://cid.oxfordjournals.org/ by guest on January 3, 2013