|Home | About | Journals | Submit | Contact Us | Français|
To examine racial differences in urinary incontinence (UI) prevalence, frequency, quantity, type, and risk factors in a population-based sample of community-dwelling black and white women.
Women ages 35–64 were sampled from telephone records from three southeast Michigan counties. Women self-identifying as black or white race completed a telephone interview that assessed demographics, health history, lifestyle factors, and UI experience. Statistical analysis included descriptive statistics, factor analysis, and multivariable logistic regression to determine adjusted odds of UI. Estimates were weighted to reflect probability and nonresponse characteristics of the sample and to increase generalizability of the findings.
Interviews were completed by 1,922 black and 892 white women (response rate=69%). The overall prevalence of UI was 26.5%. By race, UI prevalence was 14.6% for black women and 33.1% for white women (p<0.001). Among incontinent women, there was no difference by race in the frequency of UI; however, black women reported more urine loss per episode (p<0.05). A larger proportion of white women with incontinence (39.2%) reported symptoms of pure stress incontinence compared to black women (25.0%), whereas a larger proportion of black women (23.8%) reported symptoms of pure urge incontinence compared to white women (11.0%). Risk factors for UI were generally similar for white and black women.
In this population-based study we observed racial differences in prevalence, quantity, and type of UI. Frequency of and risk factors for UI were generally similar for white and black women.
Urinary incontinence (UI) affects up to half of women, with prevalence estimates varying widely depending on the definition used and population studied.1–6 Prevalence estimates of UI have been based predominantly on older, white populations with limited information on the prevalence among other age or racial/ethnic groups.
A number of studies based on surveys 2, 7–13 and clinical evaluation 14, 15 suggest differences between black and white women in the prevalence and type of UI.15, 16 These studies are limited for several reasons including inadequate representation of black women, secondary analysis not specifically designed to examine racial differences in UI, or inclusion of select populations such as health maintenance organization members.
We conducted a population-based cross-sectional study to examine racial differences in prevalence, frequency, quantity, and type of UI in black and white community-dwelling women ages 35–64 years. In addition, we explored demographic, health history, lifestyle, and obstetric/gynecologic factors associated with continence status. The aim was to determine if a population-based epidemiologic study would confirm racial differences in UI. If confirmed, physiological or biological explanations for theses difference could be sought to advance our understanding of the pathophysiology, risk factors, and appropriate therapies for this condition.
Establishing the Prevalence of Incontinence (EPI) Study was conducted at the University of Michigan. The study population included community-dwelling black and white women ages 35–64 years from three southeast Michigan counties. Because it is hypothesized that black women have a lower prevalence of UI, our sample size was determined to ensure adequate representation of incontinent black and white women. Here we report the results of the survey component of the EPI Study.
Data collection commenced in the summer of 2002 and was completed in the fall of 2004. The sample was derived from 12,541 telephone numbers purchased from a commercial survey sampling firm. The name and address for the directory listing was appended to each record, which allowed us to control the likely racial distribution of the sample by referring to census information about the racial composition of geographic areas in southeastern Michigan. The sample was bought in several batches (while tracking rates for eligibility and consent) to meet the goal of interviewing approximately 1,500 black women and 1,000 white women. The power analysis was based on the number of continent and incontinent women of each racial group needed to analyze continence mechanism structure and function in the second (clinical) phase of the study. The research protocol was approved by the Institutional Review Board of the University of Michigan Medical School.
Prior to telephone contact, informational brochures and letters of introduction were mailed to the addresses associated with the purchased telephone numbers. Shortly thereafter, trained female professional telephone interviewers from the Survey Research Center (SRC) of the University of Michigan Institute for Social Research called the households to screen for eligible participants (Figure 1). Women who were between the ages of 35 and 64, who self-identified as black or white race, and who had not been pregnant in the past 12 months were eligible for study inclusion.
Eligible women were read a statement about the voluntary and confidential nature of their data, and were asked for verbal consent to continue with the interview. The interview assessed respondents’ demographic characteristics, health history, lifestyle factors, obstetric and gynecologic history, and UI experience. Women who reported losing urine at least 12 times in the past 12 months were asked additional incontinence questions regarding their urine loss, including frequency, amount, and instigating situations and activities. These questions were based on widely used measures of incontinence.17, 18 Assessment of depressive symptoms was based on the Center for Epidemiologic Studies - Depression (CES-D) questionnaire.19
In quantifying frequency of urine loss, women were asked “Typically, how many times a month do you lose urine?” In assessing amount of urine loss, women were asked “Typically, when you lose urine do you: (1) leak a few drops; (2) wet underwear or pad; (3) soak your outer clothes; or (4) drip onto the floor?” Type of incontinence was determined by ten questions that assessed how often women lost urine during certain activities or situations. The four possible responses included: often, sometimes, rarely, or never. We performed a principal components analysis to empirically derive two components indicative of type of incontinence (Table 1). We classified women as having stress symptoms if they reported “often” losing urine to at least one question that factored with the “stress” component, or as having urge symptoms if they reported “often” losing urine for at least one question that factored with the “urge” component. Women who reported “often” to one or more questions from only one component were classified as having pure stress or urge symptoms, whereas women who reported “often” to questions from both components were classified as having mixed incontinence. Women who did not reach a threshold of “often” on any question were classified as having incontinence “below threshold.” With regards to depression, women who answered affirmatively to any of the questions of having felt “depressed,” “sad,” or “lonely” in the prior week were categorized as having depressive symptoms.
Demographic characteristics of the study sample, self-reported UI frequency, and quantity of urine loss in the last 12 months were described overall and within racial groups, and differences by race were assessed using chi-square tests for categorical variables and F tests for continuous variables. Variables (demographic characteristics, health history, lifestyle factors, and obstetric/gynecologic history) found to be statistically significant in bivariate analyses with continence status were included in a multivariable logistic regression model to explore their independent associations with continence status by estimating adjusted odds ratios (OR) and 95% confidence intervals (CI). Weights were constructed based on age, race, and geographic location to adjust for the oversampling and for survey nonresponse. P-values less than 0.05 were considered statistically significant. All analyses were performed with Stata release 9.2, which permits the incorporation of weights and makes appropriate adjustments to the standard errors for complex survey sample designs.
Of the 3,692 households in which an eligible woman resided, interviews were completed by 1,922 black and 892 white women with a response rate of 69% (Figure 1). Characteristics of the study population, overall and by race, are shown in Table 2.
The overall prevalence of UI was 26.5% (95% CI: 24.3–28.7) based on self-reported number of episodes of urine loss in the past 12 months. The prevalence of UI was 14.6% (95% CI: 13.1–16.3) for black women and 33.1% (95% CI: 29.9–36.5) for white women (p<0.001). Overall, the mean age at onset of UI symptoms was 42.2 years (95% CI: 41.3–43.0); age at onset did not differ significantly by race. Among incontinent women, the mean number of urine leakage episodes per month was 14.9 (95% CI: 13.3–16.5). Mean episodes of urine leakage did not differ by race, with black women experiencing 14.8 episodes (95% CI: 12.9–16.7) and white women experiencing 14.9 episodes (95% CI: 13.0–16.9) per month. Among incontinent women, there was no significant difference by race in terms of the frequency of urinary incontinence (Table 3). However, a significant difference was observed in quantity of urine loss per episode (p<0.05). Half (50.1%) of white incontinent women reported losing a few drops of urine, compared to 37.0% of black women. Half of black incontinent women (50.6%) reported losing urine to the point of wetting their underwear or pad, compared to 37.7% of white incontinent women.
Types of UI are shown in Table 4. A significantly greater proportion of white women with UI (39.2%) reported symptoms of pure stress incontinence compared to black women (25.0%), whereas a greater proportion of black women (23.8%) reported symptoms of pure urge incontinence compared to white women (11.0%). There were no significant differences in self-reported mixed incontinence or “below threshold” incontinence symptoms by race
To identify factors for inclusion in multivariable logistic regression analyses, we conducted bivariate analyses of continence status with demographic characteristics, health history, lifestyle factors, and obstetric/gynecologic history (data available upon request). Demographic characteristics found to be significantly associated with continence status included age (p<0.001), race (p<0.001), and currently working for pay (p<0.05). Health history characteristics significantly associated with continence status included diabetes (p<0.001), mobility impairment (p<0.001), constipation (p<0.001), urinary tract infection (p<0.05), chronic lung disease (p<0.01), obesity (p<0.001), and depressive symptoms (p<0.001). The only lifestyle factor found to associated with continence status was participating in exercise involving bouncing at least once per week (p<0.01). Obstetric/gynecologic characteristics associated with continence status included a history of at least one vaginal delivery (p<0.05), surgery for prolapse or UI (p<0.001), hysterectomy (p<0.001), current estrogen use (p<0.05), and menopause status (p<0.001). Characteristics that were not significantly associated with continence status (p>0.05) and, thus, were not included in multivariable analyses, included: years of education completed, household income, marital status, lifting at least 30 pounds more than once per week, current cigarette smoking, and drinking over eight glasses of fluid per day.
Shown in Table 5 are odds ratios for UI adjusted for variables found to be significant in bivariate analyses (demographic characteristics, health history, lifestyle factors, and obstetric/gynecologic history). Compared to women ages 35–44, the odds of UI were significantly higher among women ages 45–54 (OR=1.7; 95% CI: 1.2–2.4) and among women ages 55–64 (OR=1.8; 95% CI: 1.1–2.8). Black women had approximately one-third the odds of UI (OR=0.3, 95% CI: 0.2–0.3) as compared to white women. Several health history characteristics were significantly associated with approximately two-fold higher odds of UI, including mobility impairment (OR=1.9; 95% CI: 1.2–2.8), constipation (OR=2.1; 95% CI: 1.4–3.1), obesity (OR=2.3; 95% CI: 1.8–3.0), and depressive symptoms (OR=1.5, 95% CI: 1.1–1.9). With respect to obstetric/gynecologic history, women who had at least one vaginal delivery (OR=1.5; 95% CI: 1.1–2.0) and those who had undergone hysterectomy (OR=1.5; 95% CI: 1.1–2.2) had significantly higher odds of urinary incontinence.
When the adjusted odds ratios were examined within the racial groups, the risk factors were found to be generally similar for black and white women. This conclusion was supported by testing the significance of adding interaction terms for each predictor by race to the overall model. The F statistic for the main effects model was statistically significant (p<0.001), whereas the F statistic for the block of interaction terms was not significant (p=0.25).
In this large, population-based cross-sectional study specifically designed to assess differences in UI among black and white women, we found that black women had approximately one-third the odds of UI as compared to white women after adjusting for multiple risk factors. Strengths of this study include in-depth assessment of self-reported UI experience and comprehensive examination of associations of continence status with health history, lifestyle factors, and obstetric/gynecologic history among black and white incontinent women.
Several health history factors were significantly associated with nearly 2-fold increased odds of UI in both black and white women, including mobility impairment, constipation, obesity, and depressive symptoms. These risk factors are consistent with other population-based studies that evaluated racial differences.6, 20 Despite adequate representation of both racial groups, there was little evidence to support the idea that risk factors for UI differed between black and white women. This is in contrast to earlier work by Bump et al showing different risk factors for black versus white women; however, his study acknowledged that its small sample size limited its ability to make conclusions about racial differences.14 Within the age span we examined, older age appeared to have a greater impact on UI among white as compared to black women.
A factor analysis guided our assessment of incontinence symptom types. Similar to prior reports, we found different types of UI reported by each racial group.6, 14, 20, 21 White women reported stress incontinence symptoms more commonly than black women, whereas twice as many black women reported symptoms of urge incontinence as compared to white women. In a large cohort study from a health maintenance organization, Thom found similar results for stress urinary incontinence symptoms, but in their population, the difference in urge UI by race was relatively small and not statistically significant.20 Almost half of incontinent women have either mixed or “below threshold” symptoms.
The overall prevalence of incontinence for the survey sample was 26.5% based on self-reported number of episodes of urine loss in the past 12 months. A significant difference in prevalence of incontinence was observed between racial groups, 14.6% of black women versus 33.1% of white women. Studies that defined UI as at least monthly urine loss have reported similar prevalence UI rates for white women (23–34.5%) versus black women (14–25.2%).2, 8–10 However, different prevalence rates have been reported in other population-based studies with differing definitions of UI and population studied.4, 12, 13
Half of black incontinent women reported losing urine to the point of wetting their underwear or pad while only a third of white incontinent women complained of this quantity of urine loss. It is likely that a larger amount of urine loss is associated with the different types of urine loss reported by race, with urge incontinent episodes having greater volume versus stress incontinent leakage. Other reports have not noted this volume difference, possibly because the quantity or severity of incontinence is reported using a Sandvik scale that combines frequency and volume.20 While a useful research and clinical tool, an increase in volume and a decrease in frequency alone may not be detected using this method. In contrast to the differences in quantity lost, there was no significant difference by race in terms of the frequency of UI. This finding is consistent with a study in 1999, where Fultz et al also reported that the frequency in urine loss was similar between black and white women.8
Several important factors must be considered when interpreting the findings from this study. Women were recruited from Southeastern Michigan, thus limiting the generalizability of our study to women of European or African descent in other parts of the United States. Moreover, we did not recruit Latinas or Asian American women, although these groups are a rapidly growing part of the U.S. population. Our results may only be generalizable to women between the ages of 35–64, and may not be applicable to younger or older populations. As a cross-sectional study, associations do not imply causality; longitudinal studies will be needed to further evaluate these predictive factors.
At a population level, epidemiologic studies can help us understand burden of disease and associated risk factors. The definitions of incontinence and frequency and quantity of urine loss may differ from patients seeking treatment. Since any individual woman – black or white –can suffer from UI, including stress, urge or mixed symptoms, race should not be used to direct differential evaluation and treatment for incontinent women.
In this population-based study we observed racial differences in incontinence prevalence, type and quantity of urine loss. With adequate representation of these racial groups, there is little evidence to support the idea that the profile of risk factors for UI differed between white and black women. Additional studies of EPI Study participants will focus on exploring these biological and non-biological differences, including urodynamic function and anatomical measures of physiological and biological variability.
We gratefully acknowledge research support from the National Institutes of Child Health and Human Development Grant R01 HD 041123.
Sponsored by NICHD R01 HD 041123
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.