During their lifetime, up to 50% of women report having had at least one urinary tract infection,1,2 and approximately 50% of women experience urinary incontinence.3 These two common conditions are associated in multiple, large, population-based studies,4–10 suggesting that women with a history of urinary tract infection are more likely to have incontinence and vice versa. However such studies were not adequately designed to clearly define the temporal relationship between these two frequently co-occurring conditions.
Medical textbooks frequently cite incontinence as a symptom of urinary tract infection,11 presumably due to inflammation and irritability of the mucosa and smooth muscle of the bladder wall.5,12,13 On the other hand, structural problems such as cystocele,14 postsurgical changes after hysterectomy,15,16 use of estrogen replacement,17 or disruptions in the vaginal flora stemming from chronic perineal dampness18 may result in incontinence, which may, in turn, be associated with UTI. The objective of this study was to evaluate the rate of self-reported urine loss, comparing women who experienced an incident microbiologically confirmed UTI with those who did not and also to assess a temporal relationship between these two conditions.
MATERIALS AND METHODS
We analyzed data from daily diaries completed by 1,017 postmenopausal women during a 2-year, prospective, observational study, the methods of which have been described previously.19–22 Briefly, between 1999 and 2001 women between the ages of 55 years and 75 years were randomly selected from enrollment files from Group Health Cooperative, a large health maintenance organization in the Pacific Northwest. These women had been continuously enrolled at Group Health Cooperative for the previous year, had no menstrual cycle in the preceding 12 months, resided in Pierce, King, or Snohomish counties in Washington State, lived in the community, and had no debilitating illness or active malignancy. The Group Health Cooperative diabetes registry was used to increase the proportion of women with diabetes in the study sample to approximately 20%, because diabetes was a primary exposure of interest. Participants were asked to make baseline and annual visits during 2 years of follow-up, and to maintain a daily study diary. Participants who completed the baseline visit and at least 1 month of diaries were included in this analysis. The study was approved by the human subjects committees of the University of Washington and the Group Health Cooperative Center for Health Studies, and all participants provided informed consent before enrollment.
We ascertained information on urinary incontinence from two sources, annual questionnaires and daily diaries. At entry, participants completed a questionnaire that assessed urinary health along with demographic characteristics, general health, hormone use, and physical function. Regarding urinary incontinence, participants were questioned about accidental leakage in the past year, the frequency of leakage in the past month, and the usual amount of urine lost. As described previously,22 we used these variables to approximate an established measure of incontinence severity, the Sandvik index, which categorizes incontinence into: amount (1=few drops/damp underwear, 2=more [moderately to completely wet underwear/wet clothes]) multiplied by frequency (1=less than monthly, 2=monthly, 3=weekly, 4=daily) to equal slight (1–2), moderate, (3–4) or severe (6–8).21,23
Using check boxes, participants were asked to record information in the diary each day about incontinence and other urinary symptoms, including pain, burning, frequency, and urgency in a monthly diary throughout the 2-year follow-up. The check box pertaining to urinary incontinence read “Urine loss enough to wet underwear.”
Our assessment of incident symptomatic UTI has been described previously.20 Briefly, participants were instructed to collect a urine sample at the onset of symptoms, and send the sample to the study site. Urine specimens were collected using a Dipslide (Unipath, Ogdensburg, NY) home testing kit, supplied to the participants at enrollment along with instructions. We also conducted a systematic surveillance of the health plan’s computerized clinical records. All participants were also interviewed over the telephone to verify the presence of urinary symptoms. We defined the presence of a UTI on the basis of a midstream, clean-catch, urine specimen yielding 105 or more colony forming units (CFU/mL) of a uropathogen plus at least two of the following urinary symptoms: dysuria, urgency, or frequency.
Only culture-confirmed, symptomatic UTIs were included in the analysis. Urinary symptoms recorded in the diary were incorporated into the UTI onset definition to approximate onset date as accurately as possible. “Burning or painful urination” recorded in the diary up to 7 days before the date of positive urine culture was used to define the infection onset date. If burning was not reported in the diary during this window, infection onset date was defined as the date of urine culture.
Study time was defined as calendar days on study, beginning with the first diary entry after the baseline visit and ending with the last entry. Among women who did not experience a UTI during the study, we calculated the monthly rate of urine loss as reported in the diaries. Among women who did experience a UTI during the study, we approximated the monthly rate of urine loss not temporally related to infection, ie, the basal rate. To do this we excluded diary entries 14 days preinfection and 14 days postinfection for each UTI and then calculated the basal monthly rate. This 28-day time period, rather than a narrower time period, was chosen to capture all of the potential symptomatic days surrounding an acute UTI. Rates were stratified by diabetes, body mass index (BMI), age, hormone use, and history of hysterectomy. Rates were compared using either the Wilcoxon rank sum test to account for the nonnormality of the data, as verified by the Shapiro-Wilk test, or the linear test for trend. To assess confounding, we performed linear regression, modeling the log-transformed rate of urine loss as the dependent variable and any incident UTI as the independent variable. We defined confounding as a substantial change in the independent variable after individual adjustment for diabetes status, age, body mass index, hormone use, and history of hysterectomy. We accounted for the oversampling of women with diabetes in the original study design by applying the prevalence of diabetes in the subset of randomly selected women in our cohort before supplementation with women from the diabetes registry to the diabetes-specific rates in our sample.
To estimate whether there was an acute increase in urine loss after a UTI, we compared the rate of urine loss within the 3-day period after UTI onset to the basal rate of urine loss among women who experienced a UTI, limited to the first UTI per woman. The Wilcoxon signed rank test was used to test the difference between these two rates. All analyses were performed with SAS 9.1 (SAS Institute, Cary, NC).
Among the 1,017 women enrolled in the study, 913 (90%) completed at least one monthly diary during follow-up, and 50% completed a daily diary over the entire 2 years of diaries. Women who completed diaries were comparable to the complete cohort in terms of demographics, overall health, and frequency of urinary incontinence. In our sample, the majority of women were white and either married or living as married, with good physical function scores. At baseline, two thirds of women reported an incontinence episode within the previous year. Of these women, approximately two thirds reported urine loss at least once per week or urine loss sufficient to dampen underwear (Table 1). During the study, loss of urine was reported in the diary by 547 women (60%), among whom the average rate of reporting this symptom was 4.7 times per month. At least once during follow-up, 44% of women reported an urgent need to urinate (2.5 reports per month), 30% reported an increase in urinary frequency (1.4 reports per month), and 17% reported burning or painful urination (0.5 reports per month).
The mean rate of urine loss was 2.83 episodes per woman-month as reported in the diaries in the entire cohort. After excluding the 14 days before and 14 days after each UTI, this rate decreased to 2.81 episodes of urine loss per month. Both of these rates were equivalent to approximately 9.4% of diary days with report of urine loss. The prevalence of diabetes in the randomly selected subset of our cohort was 6.3%, before oversampling from the diabetes registry. To assess the effect of oversampling women with diabetes in our study, we applied the 6.3% prevalence to the mean monthly rates of urine loss stratified by diabetes. This resulted in a slight change in the overall monthly rate of urine loss for all study days from 2.83 to 2.60 (8.8% of days diaries were completed) and for the basal rate from 2.81 to 2.57 (8.6% of days diaries were completed).
The basal monthly rate of urine loss was lower among women who did not experience a UTI (2.64 times per month, or 8.8% of study days) compared with the rate of women who did (4.60 times per month, or 15.3% of study days) (Wilcoxon P=.04) (Fig. 1 and Table 2). These rates did not substantially change after adjustment for potential confounding variables, including age, BMI, diabetes status, and hormone use. During the 3-day period immediately after the first UTI (n=78), urine loss was reported between 21% and 24% of days (Fig. 1). Women without diabetes compared with women with diabetes had lower rates of urine loss (2.49 and 4.21 episodes per women per month, respectively, Wilcoxon P=.03) (Table 2). Women who reported oral or vaginal estrogen use within the past year at baseline had higher rates of urine loss compared with those who did not (2.97 and 2.61 episodes per women per month, respectively, Wilcoxon P=.001) (Table 2). The rate of urine loss increased with increasing BMI (linear test for trend, P<.001), but not with age (linear test for trend, P>.20) (Table 2). While the exclusion of the 14-day time period before and after UTI onset did not change these comparisons, we did observe a decrease in rates among women with UTI when these days were omitted (4.84 to 4.64 episodes per month, respectively; P<.001) (Table 2).
To evaluate the presence of an acute increase in the rate of urine loss immediately after onset of a UTI, we compared a woman’s daily rate of urine loss during the 3-day period after a UTI onset (0.23 reports per day) to her basal rate (0.15 reports per day) among women who had a UTI, limited to the first UTI if a woman experienced multiple infections (n=78). Although the rate of urine loss in the 3 days after UTI onset was 1.5 times higher than the basal rate, this difference was not statistically significant (P=.26).
This large, prospective study of community-dwelling postmenopausal women demonstrates, by use of daily diary data, the temporal relationship between urinary incontinence and symptomatic, microbiologically confirmed UTI. We found that women who developed a UTI during the course of the study had a higher basal rate of urinary incontinence during infection-free intervals, excluding the 14-day time period before and after the incident UTI compared with women who did not. Diary documentation of urine loss demonstrated a short, 3-day increase in incontinence immediately after the onset of urinary tract infection. Although this acute increase in urine loss was 1.5 times higher than the basal rate, this difference did not reach statistical significance. The lack of statistical association may be due to insufficient power to detect a difference caused by the somewhat small sample size (78 women who experienced UTIs). To have 80% power to detect a difference between these two rates if one truly did exist, we would have needed 174 women with a UTI available for this analysis.
Before this study, multiple large population-based studies have demonstrated associations between urinary incontinence and UTI.4–10 Because these studies assessed incontinence and UTI at a single time point or, at most, annually, the temporal relationship between the two conditions has remained unclear. It is plausible that chronic urinary incontinence may predispose to UTIs due to its association with abnormal genitourinary anatomy or function and environmental factors affecting vaginal bacterial ecology. It is also clinically reasonable that UTIs might result in urge urinary incontinence. In the Heart and Estrogen/Progestin Replacement Study,5 UTIs were prospectively associated with urge incontinence, and there are data to suggest that UTIs lead to incontinence by stimulating bladder detrusor activity, inhibiting alpha adrenergic receptors in the urethra, and decreasing bladder sphincter pressure.12,13 By incorporating urine loss information collected in daily diaries during a study of incident UTI, we were able to illustrate this temporal relationship in detail.
Our population sample was supplemented with women with diabetes so that the relationship between diabetes and UTI could be explored. Because diabetes is also associated with urinary incontinence21,22 we accounted for the oversampling by weighting the rates and proportions reported here by the distribution of diabetes in the original random sample of Group Health Cooperative women. We did not find a large effect of the oversampling on the rates. Of note, the prevalence of diabetes at Group Health Cooperative is much lower than the general population of women of comparable age in Washington State. According to the National Diabetes Surveillance System, the prevalence of diabetes in Washington State among women aged 45–64 years and 65–75 years in 2000 was 7.1% and 14.5%, respectively.24 Thus, using the distribution of diabetes in Washington State would have resulted in an even smaller change after standardization.
Although the observational design of this research prevents definitive conclusions regarding causation, the prospective evaluation allows us to address temporal sequence, a key consideration in the assessment of a potential causal association. Other study strengths include a strong adherence to the diary protocol by participants, with 50% of women completing diaries throughout the 2-year study, as well as strong concordance between diary reports and questionnaire data.
Aside from these strengths, there are a number of limitations. Misclassification may have occurred regarding the timing of incontinence surrounding the onset of UTI, despite our attempts to define the onset of UTI as accurately as possible. Women were instructed to contact study personnel with the first symptoms of a UTI, and we used diary entries as well as interviews near to the time of UTI onset to improve accuracy in the recording of onset of symptoms of infection. The managed care setting in which all participants had prepaid health care access would be expected to have improved our ability to capture UTIs and also determine their onset. Although we did not find a large bias due to confounding, our reported association between urinary incontinence (UI) and UTI may be attributable to a mutual relationship with a third, related factor.
The high prevalence of hormone replacement therapy (HRT) use (62%) in our cohort may have affected our findings. In our study, the rate of UI was higher among women reporting hormone use. If hormone use increased the rate of UI, then the high prevalence of hormone use in our cohort might have resulted in higher rates of UI than what would be expected in a population with a lower prevalence of HRT. Of note, a previous analysis of the same data found that HRT was not associated with incident UTI, therefore it is unlikely that this variable was a confounder in this analysis.20
We chose to compare the 3-day post-UTI onset time period based upon the graphic depiction in Figure 1 and clinical knowledge regarding the expected duration of UTI symptoms. Thus, a small 1–3-day shift in UTI onset date could modify our conclusions. Misclassification of urine loss as reported in the diaries might have occurred, most likely due to failure to report symptoms by women who were less symptomatic, but real-time data collected by diaries should be more accurate than data collected by retrospective recall.25,26
Urinary incontinence and UTI are common, burdensome urinary problems in postmenopausal women. Our data suggest that chronic basal urinary incontinence is associated with incident UTIs. Also, the rate of urine loss may increase in the 3-day period after UTI compared with the basal rate; however, our study may have lacked statistical power to find an association. The results from this study provide a stepping stone for future research that can clarify the mechanistic aspects of the temporal relationships described here and more precisely control for confounding factors. This could offer physicians and patients with a clearer understanding of the causal relationships between urinary incontinence and infection and subsequent avenues for intervention.
1. Kunin CM. Urinary tract infections in females. Clin Infect Dis 1994;18:1–10.
2. Kunin CM. Urinary tract infections: detection, prevention, and management. 5th ed. Baltimore (MD): Williams & Wilkins; 1997.
3. Melville JL, Katon W, Delaney K, Newton K. Urinary incontinence in US women: a population-based study. Arch Intern Med 2005;165:537–42.
4. Brown JS, Grady D, Ouslander JG, Herzog AR, Varner RE, Posner SF. Prevalence of urinary incontinence and associated risk factors in postmenopausal women. Heart & Estrogen/Progestin Replacement Study (HERS) Research Group. Obstet Gynecol 1999;94:66–70.
5. Brown JS, Vittinghoff E, Kanaya AM, Agarwal SK, Hulley S, Foxman B, et al. Urinary tract infections in postmenopausal women: effect of hormone therapy and risk factors. Obstet Gynecol 2001;98:1045–52.
6. Diokno AC, Brock BM, Herzog AR, Bromberg J. Medical correlates of urinary incontinence in the elderly. Urology 1990;36:129–38.
7. Foxman B, Somsel P, Tallman P, Gillespie B, Raz R, Colodner R, et al. Urinary tract infection among women aged 40 to 65: behavioral and sexual risk factors. J Clin Epidemiol 2001;54:710–8.
8. Hu KK, Boyko EJ, Scholes D, Normand E, Chen CL, Grafton J, et al. Risk factors for urinary tract infections in postmenopausal women. Arch Intern Med 2004;164:989–93.
9. Sherburn M, Guthrie JR, Dudley EC, O’Connell HE, Dennerstein L. Is incontinence associated with menopause? Obstet Gynecol 2001;98:628–33.
10. Parazzini F, Chiaffarino F, Lavezzari M, Giambanco V; VIVA Study Group. Risk factors for stress, urge or mixed urinary incontinence in Italy. BJOG 2003;110:927–33.
11. Kunin CM. Detection, prevention and management of urinary tract infections. 4th ed. Philadelphia (PA): Lea & Febiger; 1987.
12. Bhatia NN, Bergman A. Cystometry: unstable bladder and urinary tract infection. Br J Urol 1986;58:134–7.
13. Nergardh A, Boreus LO, Holme T. The inhibitory effect of coli-endotoxin on alpha-adrenergic receptor functions in the lower urinary tract. An in vitro study in cats. Scand J Urol Nephrol 1977;11:219–24.
14. Romanzi LJ, Chaikin DC, Blaivas JG. The effect of genital prolapse on voiding. J Urol 1999;161:581–6.
15. van der Vaart CH, van der Bom JG, de Leeuw JR, Roovers JP, Heintz AP. The contribution of hysterectomy to the occurrence of urge and stress urinary incontinence symptoms. BJOG 2002;109:149–54.
16. Brown JS, Sawaya G, Thom DH, Grady D. Hysterectomy and urinary incontinence: a systematic review. Lancet 2000;356:535–9.
17. Grady D, Brown JS, Vittinghoff E, Applegate W, Varner E, Snyder T; HERS Research Group. Postmenopausal hormones and incontinence: the Heart and Estrogen/Progestin Replacement Study. Obstet Gynecol 2001;97:116–20.
18. Pabich WL, Fihn SD, Stamm WE, Scholes D, Boyko EJ, Gupta K. Prevalence and determinants of vaginal flora alterations in postmenopausal women. J Infect Dis 2003;188:1054–8.
19. Boyko EJ, Fihn SD, Scholes D, Abraham L, Monsey B. Risk of urinary tract infection and asymptomatic bacteriuria among diabetic and nondiabetic postmenopausal women. Am J Epidemiol 2005;161:557–64.
20. Jackson SL, Boyko EJ, Scholes D, Abraham L, Gupta K, Fihn SD. Predictors of urinary tract infection after menopause: a prospective study. Am J Med 2004;117:903–11.
21. Jackson SL, Scholes D, Boyko EJ, Abraham L, Fihn SD. Urinary incontinence and diabetes in postmenopausal women. Diabetes Care 2005;28:1730–8.
22. Jackson SL, Scholes D, Boyko EJ, Abraham L, Fihn SD. Predictors of urinary incontinence in a prospective cohort of postmenopausal women. Obstet Gynecol 2006;108:855–62.
23. Sandvik H, Seim A, Vanvik A, Hunskaar S. A severity index for epidemiological surveys of female urinary incontinence: comparison with 48-hour pad-weighing tests. Neurourol Urodyn 2000;19:137–45.
24. Centers for Disease Control and Prevention. National Center for Chronic Disease Prevention and Health Promotion. Data & trends. National diabetes surveillance system. State-specific estimates of diagnosed diabetes among adults. Prevalence of diagnosed diabetes per 100 adult population, by age and state, United States, 1994–2005. Available at: http://www.cdc.gov/diabetes/statistics/prev/state/tPrevalenceTotal.htm
. Retrieved July 31, 2007.
25. Armstrong BK, White E, Saracci R. Principles of exposure measurement in epidemiology. In: Monographs in epidemiology and biostatistics. Vol 21. New York (NY): Oxford University Press; 1992.
© 2008 The American College of Obstetricians and Gynecologists
26. Hornsby PP, Wilcox AJ. Validity of questionnaire information on frequency of coitus. Am J Epidemiol 1989;130:94–9.