Overview

Fourteen epidemiological studies of UI prevalence in community settings have reported prevalence rates in a combined sample of adult women and men (Table 1). Five studies were conducted in the United States,^{8, 30, 39, 54, 120} two studies in Sweden,^{37, 114} three studies in the United Kingdom,^{36, 38, 43} and one study each in China,³³ the Netherlands,¹¹¹ Spain,³⁵ Japan,⁵⁸ and Australia.³³ The majority of these population-based surveys sampled middle-aged and older adult populations. Three studies included younger adults ages 18 years and over.^{36, 37, 54} There are methodological and reporting differences among the studies. Some studies used mailed questionnaires, whereas others involved in-home interviews. Most studies reported on the actual prevalence of UI in the sample; however, one study extrapolated the prevalence estimate using census statistics for the entire United States population.^{8, 9} There is limited information available on the incidence, progression, and remission of UI in the adult population combining men and women.

Table 1

Prevalence of UI in community dwelling adults by age and race.

Prevalence, severity, and impact of UI

Prevalence estimates for the adult population are fairly consistent, with most estimating between 16 to 22 percent.^{8, 9, 30, 35, 37, 38, 120} However, estimates varied according to age and gender. Adults ages 60 years and over have the highest rates of UI. Prevalence rates differ substantially by gender, with women having higher rates across all age groups. One large study of adults ages 40 and over reported a female to male ratio at 2:6.³⁸

Incidence of UI

The incident rate for the adult population varies by age, gender, and race. In a sample of 17,421 men and women ages 40 years and over, the 1-year incidence was 6 percent,³⁸ whereas in a study of 2,087 adults over age 70, the 2-year incidence was 20 percent, with 21 percent for urge UI and 20 percent for stress UI.³³ The largest survey involving 58,658 American men and women 65 years and over estimated a 2-year incidence rate of UI at 37 percent,³⁹ with rates being significantly higher in women than men, 44 and 28 percent, respectively. In this study, the incidence of UI measured as any urine loss experienced in the past 6 months increased with age beginning at 32 percent in adults 65–69 years and increasing to 54 percent in those over age 95. The impact of UI was greatest in the oldest age groups, with those rating a big impact varying from 14 percent of those 65–69 years to 27 percent for those 90–94, and 38 percent in those over 95 years. This study also found that non-Hispanics (38 percent) are more likely to be incontinent than Hispanics (31 percent), and American Indians and Whites had higher rates of UI than Blacks and Asians.³⁹ In contrast, Hispanics rated a slightly greater UI impact, e.g., a big problem (25 percent) compared to non-Hispanics (17 percent). A higher proportion of American Indians (30 percent) rated UI as having a big impact compared to Blacks (20 percent), Whites (17 percent), and Asians (15 percent).³⁹

Overview

We identified 117 epidemiological studies of UI prevalence in community dwelling women^{7, 8, 30, 35}–¹⁴⁸ that have been published over the past 17 years, with both broad and narrow age spans of women within specific countries or regions of countries. The majority of the studies have been conducted in the United States (40 studies),^{8, 30, 39, 44, 45, 47, 50, 54}–^{56, 59, 66, 67, 88}–^{90, 93, 102, 105, 110, 113, 118, 120, 123, 124, 126, 128, 130, 134, 136}–^{141, 144, 146}–¹⁴⁹ UK (10 studies),^{36, 38, 43, 52, 70, 77, 82, 103, 112, 132} or Northern Europe, including 13 studies from Sweden,^{37, 48, 53, 57, 61, 69, 71, 99, 114, 115, 119, 129, 131} four studies from Norway,^{49, 62, 80, 98} six studies from Denmark,^{51, 64, 73, 78, 96, 133} and one study from Finland. ¹⁰⁶

Prevalence rates regarding these variations can be attributed to differences in study populations, survey methodology (including sampling, definition, and measurement of UI), and reporting methods (Table 1). Although most studies of women are concentrated in middle-age and older women, other studies incorporate a broader age span with samples beginning at ages 15, 18, to 20 years and over. Study designs vary from probability-based methods of sampling of either large populations or random samples drawn from general practices, insurance plan enrollees; cross-sectional analyses of prospective cohorts of women who are participants in longitudinal studies or clinical trials in which UI is not a primary aim, cross-sectional studies of clinical populations or national panels, to case-control studies examining the effect of surgical interventions such as hysterectomy. Survey methodology involves mailed questionnaires, in-home interviews by trained interviewers, computer-assisted telephone interviews, or clinical evaluations. The definition and measurement of UI vary widely; over 20 definitions have been used (Table 1). Pooled prevalence of ever having UI was increased from 21 percent in 19–44 years old (17 studies) to 34 percent in 45–64 years old (45 studies), and to 39 percent among elderly women (11 studies) (Table 2). The differences across studies can create artifacts, as seen in Table 2, where the mean pooled prevalence for “ever” is less than the monthly rate. Younger females reported UI in the past year less frequently (9.61 percent, two studies), the prevalence increased to 35 percent among women 45–64 (11 studies) and to 41 percent among those older than 65 years old (13 studies). Elderly women reported UI in the past month (four studies) more often (56.7 percent).

Table 2

Pooled prevalence of UI in women by time of occurrence (random effects model, statistical test for between studies heterogeneity significant for all estimates).

Prevalence, severity, and impact of UI

were reported using standardized scales or instruments. Following the method used by the 3^rd International Consultation on Incontinence in summarizing prevalence estimates in women,⁵⁷⁹ the authors used the most inclusive definition of UI (“monthly UI,” “weekly UI,” and “daily UI”). Severity was measured by volume of urine lost, by categorical rating scale of the individual's perception, or by seeking professional help. The amount of urine lost was typically categorized as “drops” to “wets outer clothing” or “runs down the legs/floor.” In some studies, a higher percentage of women tended to lose drops^{45, 121, 122} in comparison to larger volumes, whereas in other studies, women were more likely to have damp underwear and/or wet underwear or clothing than losing drops.¹²⁸ The severity of the leakage has been shown to vary by the frequency of leakage, with infrequent leakage more likely to be associated with drops, whereas daily leakage with a greater number having to change undergarments or outergarments.⁴⁵ There are conflicting findings about whether younger or older populations rate their UI as more severe.^{38, 130} In one study, older women tended to rate their UI as more severe than much younger women.³⁸ In a few studies, severity was measured by the perception of how severe a problem UI is to how much daily life is restricted.^{37, 55, 190, 580}

Differences in reporting prevalence rates in UI types make comparisons challenging across studies, as some studies report frequencies within the incontinent group and others report the rates within the overall population (Table 3). Overall prevalence of monthly UI was the highest in elderly females (25.3 percent, 95 percent CI 14.1; 36.5) (two studies)^{111, 133} and in women 45–64 years old (20.5 percent, 95 percent CI 18.3; 22.8) (9 studies).^{43, 45, 46, 62, 116, 124, 132, 133, 148} Weekly incontinence was experienced more often by women older than 65 years (16.9 percent, 95 percent CI 15.1; 18.8)^{42, 94, 97, 148} and elderly women over 80 (29.9 percent, 95 percent CI 25.4; 34.4).⁴² Few studies reported daily UI, 5 percent of younger women (19–44 years old)⁹⁶ and 17 percent of women older than 65 years⁷⁹ reported having daily leakage. Severe UI, defined as wet clothes or severe enough to seek treatment, was experienced by 9 percent of women over 65 (8.96 percent, 95 percent CI 7.51; 6),^{59, 76, 77, 86, 128, 144} and 10 percent of middle-aged women (95 percent CI 9.37;11) (seven studies).^{40, 51, 60, 77, 86, 89, 131} Less than 2 percent of women younger than 45 years old experienced severe UI.⁸⁶

Table 3

Pooled prevalence of UI in women by frequency and severity (random effects model, statistical test for between studies heterogeneity significant for all estimates).

Racial/ethnic differences

The majority of epidemiological studies have been conducted in White women in North America, Europe, Australia, and New Zealand. Although several studies have been conducted in Asian countries, including Turkey,^{58, 60, 79, 95, 116, 121} because of methodological differences, it is difficult to directly compare those to studies in which varying racial/ethnic groups have been sampled. There have been a limited number of studies from Middle Eastern countries, with only one study from an African country.

Eight population-based studies in the United States have reported on racial/ethnic differences in the prevalence of UI in women (Table 1). The majority of these studies document a higher prevalence of UI (including all types) in White women as compared to Black, Hispanic, and Asian women.

Type of UI

In general, stress and mixed UI were the most common types in studies that have compared stress, urge, and mixed UI (16 out of 28 studies) (Table 4). Although surveys have found differences in the frequencies of the different types of UI by age and race, these differences are inconsistent across studies. Some surveys found higher rates of stress UI in young and middle-aged women,^{57, 74, 80, 104, 108} whereas others have found higher rates in older women.^{37, 85, 133, 145, 581} Similar inconsistencies are noted with urge UI, with some studies reporting higher rates in older women^{37, 106, 145} and others reporting it is more prevalent in middle-aged women^{108, 133} (Table 5). The prevalence of total UI increase in age categories from 19 percent in women 19–44 years (18 studies)^{57, 64, 68, 69, 75, 80, 85, 86, 92, 96, 98}–^{100, 119, 130, 133, 145, 148} to 29 percent in those older than 45 years. Stress incontinence was the most prevalent type in women 19–44 years old (12.8 percent, 95 percent CI 8.3; 17.4) (15 studies),^{37, 57, 64, 80, 85, 91, 95, 98, 104, 108, 109, 130, 133, 145, 124} and in those 45–64 years of age (21.5 percent, 95 percent CI 18.9; 24.1) (36 studies).^{37, 38, 44, 51, 57, 61, 70, 73}–^{75, 77, 78, 80, 84, 85, 87, 89, 95, 104, 108, 116, 117, 122, 124, 127, 129, 131, 135}–^{137, 139, 141, 145, 148, 149} However, mixed UI was the most prevalent type of incontinence in older women, 16.8 percent of women older than 65 (95 percent CI 13.7; 19.9) (19 studies)^{30, 35, 59, 66, 71, 72, 74, 80, 85, 90, 104, 106, 108, 110, 121, 128, 142, 143, 145} and 16 percent of elderly women (95percent CI 7.3; 24.4) reported mixed UI (seven studies).^{71, 74, 80, 84, 106, 124, 133} Prevalence of urge incontinence gradually increased from 5 percent in younger women (4.9 percent, 95 percent CI 3.7; 6.1)^{37, 57, 64, 80, 85, 91, 95, 98, 101, 104, 108, 130, 133, 145} to 10 percent in women 45–64 (10.2 percent, 95 percent CI 8.9; 11.5) (32 studies),^{37, 44, 51, 57, 61, 70, 73}–^{75, 77, 78, 80, 84, 85, 87, 89, 95, 102, 104, 108, 116, 117, 124, 126, 127, 129, 131, 137, 139, 145, 148, 149} and to 12 percent in women older than 65 years (12.2 percent, 95 percent CI 9.9; 14.5) (28 studies).^{30, 35, 37, 56, 59, 61, 63, 66, 71, 72, 74, 80, 84, 85, 90, 104, 106, 108, 110, 118, 121, 123, 125, 128, 133, 142, 143, 145}

Table 4

Prevalence of UI by incontinence type, age, and race in community dwelling women.

Table 5

Pooled prevalence of UI in women by type of incontinence (random effects model, statistical test for heterogeneity between studies significant for all estimates).

A small number of population-based surveys conducted in the United States have compared racial/ethnic differences in UI types.^{89, 118, 137, 149} In general, White women have higher rates of stress UI than Black, Hispanic, or Asian women,^{89, 149} although one study did find Hispanic women had a higher rate of stress UI than White, Black, and Asian women.¹³⁷ Black women are more likely to have higher rates of urge UI and mixed UI as compared to Whites and Hispanics.^{137, 149}

Summary

Evidence from large observational studies from different countries (level of evidence IIB) suggested that UI is a prevalent condition among women of all ages with overall prevalence close to the estimations of the National Health and Nutrition Examination Survey (38 percent).⁹ Prevalence increased in women older than 45 years of age with further small increases in elderly women. Stress UI is the most prevalent type if UI in women 45–64 years old, while elderly women experienced mixed UI more frequently. Incidence of UI (level of evidence IIA-B) gradually increases with age. Differences in definitions of UI contributed substantially to variability in the results from individual studies.

Type of UI

The prevalence of UI, defined in various ways (ever, current, any, greater than two times/week, or leakage within past 4 weeks, 2, 6, or 12 months), is estimated between 3 percent in men 30 years and over in a study conducted in the United Kingdom⁵² to 37 percent in men 39–91 years in a Norwegian study.⁵⁸⁶ In a large population-based study conducted in five countries (Canada, Germany, Italy, Sweden, and the United Kingdom), the prevalence rate for men ages 18 and over was 5 percent.¹⁴⁵

Studies in American men reported UI as involuntary leakage of urine during the last year, last month, or ever (Appendix Table F3). The prevalence estimations varied substantially depending on the definitions, with the higher prevalence of UI during the last year in males 19–44 years old (18.4 percent, 95 percent CI 4.5; 32.2) and 45–64 years old (24.6 percent 95 percent CI 19.92; 29.35) compared to UI during the last month (Table 9). Older males reported UI during the last month more frequently, from 29.2 percent (95 percent CI 24.4; 34.0) among those older than 65 years of age to 42.4 percent (95 percent CI 32.8; 52.0) in elderly males. Two percent of American men 45–64 years old ever experienced stress UI (95 percent CI 2.0; 2.0) and one percent reported stress UI during the last month. Urge UI was the more prevalent during the last year (6.7 percent, 95 percent CI 6.7; 6.7) of men 45–64 years old. Men over 65 years of age reported having urge UI during the last month (10.6 percent, 95 percent CI 10.6; 10.6).

Table 9

Pooled prevalence of UI in American males by definition and age categories (random effects model).

Severity of UI

Fewer studies provided estimates for severity of UI in American men (Appendix Table F4).^{54, 164, 181, 590, 591} A survey of 922 males older than 20 years who were recruited in the primary care clinic⁵⁴ reported that wet underwear less than once per month was experienced by 9 percent, monthly by 5 percent, weekly by 7 percent, and daily by 14 percent of the responders. A community-based cross-sectional survey of 778 men older than 40 years⁵⁹⁰ reported that 10.8 percent of the responders had wet underclothing during the last year. Among males 41–60 years old from primary care clinics in a Veterans Affairs facility, 4.4 to 4.8 percent experienced daily UI. ¹⁸¹ The prevalence of daily UI increased to 8.4 to 8.9 percent among those older than 60 years of age. Pooled analysis (Table 10) estimated that daily UI was experienced by 4.8 percent of males 45–64 years (95 percent CI 4.8; 4.8), 8.3 percent men over 65 years old (95 percent CI 7.0; 9.6), and 9.3 percent elderly men (95 percent CI 4.5; 14.1). Severe UI that required a change of underwear was reported by 2 percent of those 45–64 years old and 4 percent of elderly men (95 percent CI 3.9; 4.1).

Table 10

Pooled prevalence of UI in American males by severity and age categories (random effects model).

Racial/ethnic differences

The majority of studies have been conducted in White male populations (Table 1). Three studies from the United States provided data on prevalence rates in racial/ethnic groups using different survey methodology, including methods for estimating prevalence.^{8, 39, 181} In one large population-based survey using a weighted prevalence estimate, non-Hispanic Black men were found to have a higher rate of UI (21 percent) compared to non-Hispanic White men (16 percent) and Mexican-American men (14 percent).⁸ In the other study, non-Hispanics (38 percent) were more likely than Hispanics (31 percent) to have UI.³⁹ This latter study also found that American Indians and Whites had higher rates of UI than Asians and Blacks, respectively. A sample of male veterans receiving care in primary care clinics found similar rates of incontinence between Whites (32 percent) and Blacks (33 percent).¹⁸¹ In a cross-national comparison of UI prevalence rates, South Korean men had the lowest rates (4 percent) followed by men in France (7 percent), the United Kingdom (14 percent), and the Netherlands (16 percent).⁵⁸⁵

Progression and remission of UI in community dwelling men

There is limited evidence on the progression and remission of UI in men. Evidence indicates that when men became incontinent, they developed urge or other types of UI; those with urge UI alone either stayed as urge UI or developed mixed UI.⁴⁷ In one study over a 10-year period, 3 percent of men without either urgency or urgency with incontinence at baseline developed urge UI. There was a slight nonsignificant decline in men with urge UI at baseline to have it at the 10-year followup (5 percent vs. 4 percent, respectively).⁵⁸²

Overview

In contrast to community settings, there have been fewer epidemiological studies on the prevalence, incidence, progression, and remission of UI in LTC settings. Differences in study populations, survey methodology, and definitions and measurement of UI affect the variability in prevalence estimates. Also, the timing of the survey, e.g., at admission or at a later date, also adds to the variability in estimates. More recent studies have relied on the bladder continence item on the Minimum Data Set (MDS) to operationalize UI. Two studies report data from the same cohort but derive slightly different estimates.^{28, 166} There are scarce data on the severity and impact of UI.

Prevalence of UI in the LTC population

The majority of studies sample both men and women ages 65 years and over residing in LTC facilities (Table 11). Although most studies provide prevalence estimates by gender, some provided only a prevalence estimate that included both men and women. For the combined group, prevalence estimates using varying definitions (e.g., any daytime urinary incontinence, at least two episodes of urine loss in the past 2 weeks, urine loss at least twice a month, or medical record or staff report of UI) ranges from 30–77 percent.^{2, 28}–³¹ Measuring UI with the MDS, prevalence estimates varied from 30 percent in a large sample of 29,645 adults ages 20–109 years³¹ to 77 percent in a smaller sample of 380 adults 65 years and over.³⁰ In a recent population-based study involving 95,911 older nursing home residents from eight southeastern states, the prevalence rate at admission was 65 percent.³²

Table 11

Prevalence of UI and incontinence types in LTC populations by age, gender, and race.

The prevalence of UI in female residents in LTC settings is estimated at between 60 and 78 percent.²⁸–^{30, 32, 164} In one study using data from the National Nursing Home Survey, the prevalence in women was estimated to be 74–85 percent.² However, in this same study when UI was identified from the medical record, the prevalence rate was <1 percent (1,366 per 100,000). The prevalence of UI in men ranges from 23–72 percent.²⁸–^{30, 32, 164}

Prevalence rates increase with advancing age in both men and women.^{2, 29, 30} There is limited data available on racial/ethnic differences in UI prevalence. Those studies available indicate conflicting findings. In one large study involving nursing home residents from eight states, there was a higher prevalence in Blacks (71 percent) compared to Whites (64 percent) at admission, but after admission the prevalence rate was more similar (78 percent vs. 74 percent, respectively.³²

Incidence of UI in LTC settings

Minimal data are available on the incidence of UI in LTC settings An early study involving 430 nursing home residents reported an incidence of 27 percent 2 months after admission and 19 percent at 1 year.¹⁶⁶ This is consistent with a later study that reported an incidence rate of 20 percent at 1 year.³³ The incidence of daytime UI between 2 months and 1 year after admission was higher in males (46 percent) than in females (16 percent).¹⁶⁶

Progression and remission of UI in LTC settings

Few studies have examined the progression and remission of UI in LTC settings in a way that can be easily interpreted. Although there are no large changes documented, the evidence available suggests that UI is a dynamic condition that does change over time, including improvement. One large study involving nursing home residents from eight states that examined UI at admission and within a 4-year post-admission period reported a 0 percent progression rate.³² In another study, the remission rate at 1 year was reported to be 10 percent.³³

Age

Age tends to be linearly associated with UI, with prevalence and incidence rates increasing steadily with advancing age.^{7, 41, 53, 55, 56, 62, 65, 80, 81, 83, 85, 99, 105, 133, 138, 143, 184} However, in older women an increase in incidence of weekly¹²³ UI was not significant when women 75–79 years were compared to those 70–74 years of age (Appendix Table F11). The large Canadian Study of Health and Aging cohort reported no significant increase in annual incidence of UI in women 75–84 or older than 85 years compared to women younger than 75 years of age.¹⁸⁶

Prevalent UI significantly increased with age in 20^{7, 55, 66, 67, 73, 74, 81, 84, 92, 96, 97, 105, 118, 124, 133, 138, 149, 185}–¹⁸⁷ of 28 studies that reported the association.^{7, 55, 62, 66, 67, 73, 74, 81, 84, 92, 93, 95}–^{97, 105, 118, 124, 132, 133, 138, 140, 144, 149, 185}–¹⁸⁹ Only one population-based cross-sectional study of 5,701 female participants in the Health and Retirement Study showed a decrease of 3 percent per 1 additional year in prevalence of moderate UI with 15 or fewer days with incontinence episodes in the last month.¹⁸⁹ Adjusted odds of prevalent stress UI increased by 110 percent per 5 year incremental increase in age in participants of the Heart & Estrogen/Progestin Replacement Study⁶⁶ with nonsignificant dose response association in the British cohort of 12,570 female respondents older than 40 years.¹³² The largest increase in odds of stress UI (239 percent) was reported among women over 80 years compared to those younger than 60 years¹³³ in the European population-based study of 5,795 elderly community dwelling females.¹³³

Prevalence of total weekly UI significantly increased by 106 percent per additional 1 year^{96, 149} and daily UI by 103 per additional 5 years of age.¹⁴⁰ Adjusted odds of occasional monthly UI was 1.2 and odds of severe UI 1.3 times in women 40–44 years compared to those younger than 40 years old.¹³⁸ Women 45–54 years old experienced UI 2.1 times more often and those older than 55 years of age 3.1 time more often than those 15–34 years old.⁷ Odds of reporting any UI was 3.7 times greater in women 50–59 than those 41–49 years old.⁸⁴ Women older than 70 years experienced UI 1.49⁸¹ to 2.9 times⁸⁴ more often than women under 50. The dose response association with severity of UI was shown in the Nurses' Health Study with odds ratio 1.45 for occasional, 1.56 for occasional severe, 1.67 for frequent and severe, and 1.81 for weekly severe UI in women older than 70 compared to those under 50.⁸⁴ Women 75–84 years old had 1.47¹⁸⁶ to 6.7⁹² times higher odds of UI compared to women under 75 years and those 15–24 years old, respectively.

Adjusted prevalence rates of weekly urge UI increased by 1.2⁶⁶ per 5 additional years of age and by 1.8 times ¹⁸⁷ per 10 additional years of age in a dose response model. Monthly urge UI was 1.8 time more frequent¹³³ among women 60–80 years old than for those under 60. Elderly women older than 80 years reported monthly UI four times more often than women younger than 60.¹³³ One community base Japanese survey of 968 women older than 40 years reported lower odds of urge UI among those older than 60.⁷⁴

Maternal age at the time of either the first or last delivery has been examined as a risk factor for subsequent UI; however, the effect age has on the development of UI appears to lessen by age 50.⁶³⁰ Several studies have found that age 35 years or older at first delivery increased the risk of UI or frequent UI when compared to younger women.^{105, 197} Similarly, one study found a lower risk in women who were 25 years or younger at their first delivery than their older counterparts (23 percent vs. 28 percent, respectively).⁶³⁰ Another study found that being age 30 or more at the second delivery doubled the odds of becoming incontinent.¹⁹⁷ Two studies reported that the strongest associations with UI were for women who were under the age of 22 years at their first delivery.^{105, 135} In one study examining maternal age at subsequent deliveries, women older than 30 and 40 years at a second vaginal delivery had lower rates of UI and urge UI.⁶⁴ Other studies have not been able to demonstrate an independent effect of age on UI in the immediate postpartum period.¹⁹⁸ Age at last delivery appears to be less of a risk factor.⁶³⁰

Gender

Female gender has been consistently documented as a strong predictor of UI in bivariate analyses,^{30, 36, 39, 47, 61, 74, 92, 97, 107, 114} with an age-adjusted odds of UI of 3.1.⁹² One in six women after age 30 have UI, compared to 1 in 20 men.⁵⁸⁷ Women have higher rates of UI than men in all age groups.^{81, 86} In early adult years, women have a two to four times higher rate of UI than men; however, in advanced years (e.g., ages 75 or 85 and over), even though women continue to have higher rates, the prevalence rates are more similar.⁸¹ In a study of 54,920 adults ≥30 years, the UI rate in women was 2.5 per 100 versus 1.4 per 100 in men.⁸⁶ Other studies find that UI prevalence is 6.8 times higher in women.⁷⁵

Race/Ethnicity

There is growing evidence that race/ethnicity is a predictor of UI, UI severity, and UI type in nonchildbearing and childbearing women (Appendix Table F12). In earlier studies, White women tended to have higher rates of UI than other ethnic/racial groups (Blacks, Asian Americans, including Japanese and Chinese, and Hispanics).^{30, 59, 89, 93, 105, 120, 138} In more recent studies, being non-White was associated with lower odds of having UI and severe UI,¹⁴⁴ with the exception of being Hispanic.¹³⁸ A higher UI incident rate in Whites was reported in two studies (OR 3.1, 95 percent CI 2; 4.8) (Figure 3).^{123, 149} African American women had higher odds of incident urge but lower odds of incident stress UI. Hispanic women developed weekly UI more often compared to Caucasian (OR 2.96, 95 percent CI 1.06; 8.18).

Figure

Figure 3. Association between incident UI in women of different races compared to Caucasian women (results from two studies) .

Prevalence of mild, moderate, and severe UI was lower among African American women than White women in all six studies that examined the association (Figure 4).^{93, 105, 137, 138, 149, 189} Prevalence of moderate UI was lower among Hispanic women compared to Caucasian women in four of five studies (Figure 5).^{93, 105, 138, 149, 189} However, evidence was not consistent. Prevalence of mild or severe UI was either less or the same in Hispanic women than in White women. Asian women had lower odds of moderate or severe UI (Appendix Figure F1) in two studies.^{105, 138}

Figure

Figure 4. Prevalent UI in African American women compared to Caucasian women.

Figure

Figure 5. Prevalent UI in Hispanic women compared to Caucasian women.

Race/ethnicity was associated with different types of UI, including stress, urge, or mixed UI (Appendix Table F12). After multiple adjustments, the risk of stress UI was higher in White women than in Black and Asian American women.^{118, 137} In one study White women had a 2.8 times higher risk of stress UI than Black women.⁶⁶ In contrast, the risk of urge UI was similar in Black and Asian American women compared to White women.⁵⁹ In other studies, White women had three times the risk of Black women with respect to urge UI.¹¹⁸

Obesity

Obesity has been consistently established as a strong predictor of UI, especially severe UI (Appendix Table F13). Incident monthly UI was 110 percent higher and daily 112 percent higher per one unit increase in BMI.¹⁴⁹ Obese females developed monthly UI 1.7 times more often (95 percent CI 1.2; 2.5) than women with normal weight.¹⁰³

Prevalent urge UI showed a positive dose response association 1.1 times per one unit increase in BMI,⁵¹ (OR 1.08, 95 percent CI 1.1; 1.1), 1.4 times in women with BMI 25–30 than <25 kg/m²,¹³³ and 1.8 times (OR 1.8 95 percent CI 1.4; 2.4) in those with BMI >30 kg/m2.¹³³ Asian overweight women experienced urge UI three times more often than those with normal weight (OR 3.4, 95 percent CI 1.2; 9.2).¹⁸⁷

The largest increase in stress UI was shown in obese women with uterovaginal prolapse compared to normal weight females without prolapse (OR 33.0, 95 percent CI 12.5; 87.1).⁹⁵ In contrast, underweight women after surgery for UI reported stress UI more often than females with greater BMI.⁶³¹ Severe UI was more prevalent in overweight women (BMI 25–29kg/m² OR 1.5, 95 percent CI 1.4; 1.6) and obese women (BMI >30 OR 3.1, 95 percent CI 2.9; 3.3).¹³⁸

The effect of BMI differed depending on definitions of UI (Figure 6). Odds of ever having UI were higher in underweight and overweight women.^{116, 191} Adjusted odds of UI in the past month were higher in overweight and obese females.^{83, 92, 133, 143} Weekly UI was more prevalent in obese but not overweight women with a dose response increase per one unit of BMI.^{66, 90, 92, 96, 138, 189} With the definition of UI in the past year (Figure 7), the majority of the studies showed a significant increase in UI corresponding to greater BMI.^{7, 51, 59, 68, 72, 144}

Figure

Figure 6. Association between prevalent UI and BMI in women depending on time of having UI.

Figure

Figure 7. Association between prevalent UI and BMI in women who reported involuntary urine loss in the past year.

The association between BMI and UI varied depending on frequency of leakage episodes (Figure 8) The majority of the studies reported a significant raise in daily UI corresponding to an increase in BMI per one unit (OR 1.44, 95 percent CI 1.3; 1.6), per five units (OR 1.6, 95 percent CI 1.4; 1.7), or in overweight women (OR 1.65, 95 percent CI 1.3; 2.1).^{55, 67, 70, 149, 189} Incremental increase per one unit in BMI was not associated with higher adjusted rates of at least monthly UI; however, obese women had greater odds of UI compared to women with normal weight.^{67, 104, 121, 124, 138, 149, 190} Underweight and obese women had increased odds of having at least weekly UI.^{67, 92, 138, 185} Daily and weekly stress UI were significantly associated with an increase in BMI in all studies (Figure 9).^{51, 66, 70, 118, 132, 133, 135, 149, 187} The association with UI at least monthly or in the past year was less consistent.

Figure

Figure 8. Association between severity of prevalent UI and BMI in women.

Figure

Figure 9. Association between severity of prevalent stress UI and BMI in women.

Studies in older women have found that a higher BMI (OR 1.3 per 5kg/m², CI 1.1; 1.6) was associated with stress UI but not urge UI.¹¹⁸ Other studies found that increasing BMI was associated with urge UI in women ages 40–60 years, whereas one study did not find an association.⁶⁶

Some studies have incorporated alternative anthropomorphic measures such as waist circumference, visceral fat area measured by computed tomography scan of the abdomen, and sagittal diameter^{66, 118, 149} with mixed outcomes (Appendix Table F14). One study found that waist to hip ratio was predictive of stress UI.⁶⁶ Other studies found increasing waist circumference is weakly associated with prevalent mixed UI (OR 1.0, 95 percent CI 1.0; 1.1 per 1cm increase in waist circumference), stress UI (OR 1.0, 95 percent CI 1.0; 1.1), and total monthly UI (OR 1.1, 95 percent CI 1.0; 1.1).¹⁴⁹

Evidence on maternal weight and maternal weight gain during pregnancy is conflicting. In one study, women with higher BMIs (>25kg/m²) and higher maternal BMIs (during pregnancy) increased their odds by 168 percent in developing UI at 3 months postpartum as compared to women with lower BMIs.¹⁹⁷ In another study, higher predelivery BMI (per 5kg/m²) increased the odds of UI at 6 months postpartum by 20 percent.²⁰³ In other studies, maternal weight gain was not associated with UI¹⁹⁹ or the development of stress UI 10 years after the index delivery.¹⁹⁶

Genetic influence

Evidence from four studies suggests a genetic influence on UI.⁶³²–⁶³⁵ In one study, daughters of mothers with stress UI had a higher prevalence of stress UI than age-matched controls, 71.4 percent versus 40.3 percent, respectively.⁶³² Furthermore, UI symptoms appeared 7 years earlier in “incontinent families” than in those without a history of incontinence.⁶³² In a case-control study examining relatives of women with urodynamically diagnosed UI, first-degree relatives had a UI prevalence of 20.3 percent versus 7.8 percent in relatives of the control group.⁶³⁵ Both mothers and sisters had significantly higher prevalence rates of stress UI than controls. Although daughters had a UI prevalence rate twice as high as controls, the difference was not significant. ⁶³⁵ In a population-based study, daughters of mothers with any type of UI had a 1.3 times higher risk of being incontinent, (95 percent CI 1.2; 1.4), a 1.5 times higher risk (95 percent CI 1.3; 1.8) of stress UI and 1.8 times higher risk (95 percent CI 0.8; 3.9) of urge UI.⁶³³ Daughters were also at higher risk of having severe incontinence if the mother had severe incontinence (1.9; 95 percent CI 1.3; 3.0). ⁶³³ Younger sisters of female siblings with UI, stress UI, or mixed UI had increased relative risks, respectively of 1.6 (95 percent CI 1.3; 1.9), 1.8 (95 percent CI 1.3; 2.3), and 1.7 (95 percent CI 1.1; 2.8).⁶³³ In another population-based study involving 548 monozygotic and 620 dizygotic twin pairs, the heritability for urge UI was 42 percent (95 percent CI 16; 63) among women ages 48–64 years and 49 (95 percent CI 29; 65) in those ages 70–94 years.⁶³⁴ Mixed UI also had a substantial genetic component; however, the role of genetic factors was less clear.⁶³⁴

Physical activity

There is limited data on the role of physical activity as a risk factor for UI (Appendix Table F14). Intensive physical activity in women 50–79 years of age in the Nurses' Health Study was associated with a significant reduction in incident UI (OR 0.9, 95 percent CI 0.8; 1.0).¹⁵⁶ Physically active women developed stress UI less frequently (OR 0.71, 95 percent CI 0.56; 0.91).¹⁵⁶ The same study reported a significant negative association between walking and incident urge UI (OR 0.7, 95 percent CI 0.5; 0.9).¹⁵⁶

In a large study of women ≥20 years, after adjusting for age, number of children, coughing, and wheezing/dyspnea, increasing levels of low physical activity had weak and negative associations with UI.¹⁹² No significant effects were noted for high intensity physical activity with respect to stress, urge, and mixed UI. Similarly, a large study of women ≥40 years did not find an effect for those with different levels of physical activity or participating in vigorous activities.¹⁰³ In a study examining the impact of physical activity on urine leakage in 665 primiparous women, high impact exercise before pregnancy was significantly associated with UI 1 year after childbirth, increasing the odds of UI by 40 percent.⁶³⁶ In two surveys, physical exercise in the past year was not associated with UI.^{55, 83}

Education and occupational status

Studies on occupational factors and UI risk are rare. In a study involving 1,581 Taiwanese women ≥20 years, having an occupation that involved weight lifting or other labor was not significantly associated with stress UI.⁹⁵ Women without private health insurance had higher adjusted odds of urinary symptoms (OR 2.6, 95 percent CI 1.4; 5).⁸⁴

Surprisingly, higher education was associated with increased odds of prevalent UI in the majority of the studies.^{67, 70, 72, 83, 93, 149, 189, 191} College graduates 42–52 years old experienced UI 1.31 times more often compared to women without degrees;⁹³ university graduates 50–64 years old reported UI 2 times more often (OR 2.0; 95 percent CI 1.8; 2.3)¹⁹¹ than females with lower education. One study found that higher education was an independent risk factor for UI for all ethnic groups except Chinese Americans but that educational level was not associated with moderate/severe UI.⁹³ However, in contrast to these studies, one study of postpartum incontinence reported that having less than a college education doubled the risk of UI.²⁰³

Marital status and social support

Married women had the same prevalence of UI as women who were widowed, divorced, or never married.⁸⁴ A large Study of Women's Health Across the Nation¹⁴⁹ suggested no association between level of social support and incident monthly UI. However, the same cohort reported higher prevalence rates of UI among women within the lower quartile of social support.¹⁴⁹ Psychological stress was associated with greater urinary symptoms including UI (OR 2.7, 95 percent CI 1.1; 7)⁸⁴ but not with UI in the past month.⁸³ Low social activity was not associated with risk of stress, urge, or mixed UI in women ≥70 years.¹⁰⁶ The interpretation of whether decreased social activity and poor social support precipitates UI or is a result of UI is unclear.

Dietary factors

Smoking

Smoking as a risk factor for UI has been assessed in several studies.^{55, 72, 83, 93, 103, 118, 138, 192} (Appendix Table F16). Six studies^{92, 93, 138, 186, 189, 192} of 12^{67, 72, 83, 92, 93, 138, 143, 186, 189, 191, 192} found a significant positive association between smoking and UI. One study suggests that smoking status per se is not associated with UI, but rather examining the dose-response relationship between numbers of cigarettes smoked, either as a former or current smoker, may be more relevant.¹⁹² In this study both former and current smoking was associated with UI, but only for those who smoked more than 20 cigarettes per day, raising the odds of UI by 2.7 and 3.0, respectively.¹⁹² However, severe incontinence was weakly associated with smoking, regardless of number of cigarettes. Interestingly, adjusted odds of current smokers in the same study¹⁹² had higher odds of mixed and total but significantly lower odds of stress UI. Neither smoking status nor a dose response between numbers of cigarettes smoked per day was found to be a risk factor in a survey involving 2,767 Italian women.⁷² Although a survey of 1,262 women in the United Kingdom found that the number of cigarettes was not associated with UI, being a former smoker did increase the odds by 30 percent.⁸³ In a large study of 83,355 American women ages 37–54 years, former smoking was not associated with UI, whereas current smoking was significantly associated with frequent and severe UI.¹³⁸ Current smoking increases the odds of moderate/severe UI by 20–55 percent.^{93, 138} Smoking during pregnancy independently increased the odds of UI by 290 percent (odds ratio 2.9, 95 percent CI 1.4; 3).¹⁹⁸

Functional status

Parity

We identified 24 observational studies that reported odds ratios of UI in association with parity (Table 18).^{55, 59, 64, 68, 70, 72, 83, 92, 93, 95, 96, 105, 119, 124, 133, 135, 138, 149, 185, 191, 193}–¹⁹⁶ Incident UI was not associated with parity.^{149, 193} A positive significant association between prevalent UI and parity was reported in 13 studies,^{68, 72, 83, 93, 96, 105, 119, 124, 138, 185, 191, 193, 195} while six studies did not find a significant increase in prevalent UI in relation to parity (Figure 10).^{59, 64, 92, 133, 149, 194} The number of births did not show a dose response association with prevalent UI (odds ratio per one additional birth 1.0, 95 percent CI 1.0; 1.1) but did with moderate severe UI (odds ratio per one additional birth 1.1, 95 percent CI 1.0; 1.1) and severe UI (odds ratio per one additional birth 1.1, 95 percent CI 1.0; 1.1). All six studies that measured prevalent stress UI reported a significant positive association with parity (Figure 11).^{72, 95, 133, 149, 195, 196} Prevalent stress UI and severe stress UI did not show a significant dose response association with the number of births. Only one study¹⁹⁵ of four ^{72, 133, 195, 196} found an increase in urge UI corresponding to parity.

Table 18

Association between parity and UI in females.

Figure

Figure 10. Odds ratios of UI in females with different numbers of births.

Figure

Figure 11. Odds ratios of stress UI in females with different numbers of births.

The role of parity is complex and changes as a woman ages. One study reported that age modified the association between parity and UI with a significant association in young and middle aged women and attenuation of the association in older females.¹⁹⁵ Particularly in studies of perimenopausal and postmenopausal an association has not been found between parity and UI or with UI type,^{118, 128} suggesting that aging tends to diminish this effect. In a study involving separate multivariate analyses in three age cohorts (ages 18–23, 45–50, and 70–75 years), the effect of increasing parity declined with age.⁶⁸ In other studies involving both cross-sectional and longitudinal analyses, the role of parity changes after 1 year. Odds ratios associated with parity are higher in women <60 years than in women ≥60 years.¹⁰⁵ Certain risk factors, such as vaginal delivery, are strong predictors of UI in younger women, but with changes associated with aging and menopause, this effect seems to disappear. In premenopausal women, the number of vaginal childbirths was strongly associated with UI.⁹⁶ In other studies, the relative risk of having stress UI 10 years after childbirth was not associated with the number of vaginal deliveries.¹⁹⁶

Obstetric and fetal factors

Mode of Delivery

The majority of the studies report that spontaneous vaginal deliveries are more likely to be associated with UI,^{7, 197, 199} stress UI and severe stress UI,¹³⁵ and subsequent stress UI surgery⁶³¹ compared to Cesarean delivery (labored and unlabored). Cesarean section was associated with lower odds of UI in seven studies,^{119, 197, 199}–²⁰³ by 80 percent²⁰¹ to 41 percent²⁰⁰ and did not show a significant association in five studies^{135, 141, 196, 204, 205} (Table 19). Two studies reported higher odds of UI when women after Cesarean section were compared with nulliparous females (Figure 12).^{7, 206} Vaginal delivery compared to Cesarean increased the risk of total UI with little evidence of association with stress and urge UI. Adjusted odds of UI was higher after forceps delivery by 150 percent¹⁹⁹ to 187 percent¹⁹⁸ compared to vaginal delivery (Figure 13), by 310 percent to 430 percent¹⁹⁹ compared to Cesarean section, and by 430 percent compared to nulliparous women.²⁰⁶ Vacuum delivery was associated with random changes in UI in the majority of the studies (Table 19). One study found that previous pregnancy was a risk factor for severe stress UI in women who reached the age of 50, although the mode of delivery had less effect.¹³⁵ Breech delivery was not associated with an increased risk of UI or any UI type.⁹⁸

Table 19

Odds ratio of UI according to delivery.

Figure

Figure 12. Odds ratio of UI after vaginal delivery compared to Cesarean section.

Figure

Figure 13. Odds ratio of UI after forceps delivery compared to vaginal or no forceps.

Few prospective studies reported protective effects of Cesarean section on UI compared to vaginal delivery.^{119, 197, 202, 205}

Menstrual cycle and menopause

Because of the presence of estrogen and progesterone receptors in the lower urinary tract, female UI is assumed to be associated with a woman's hormonal status and the fluctuations in it (Appendix Table. F19). One study 2,158 premenopausal women in Denmark examining the role of hormonal variation found that self-reported UI the day before completing a survey questionnaire was strongly associated with a recent decrease in bleeding duration (OR 2.2, 95 percent CI 1.3; 3.6).⁹⁶

Because of the increased prevalence of UI in the perimenopausal years, menopause has been assumed to be a key risk factor in UI. One study found that perimenopausal status was independently associated with UI, increasing the odds by 127 percent (OR 1.3, 95 percent CI 1.1; 1.5).⁹³ The large Women's Health in the Lund Area study found an increase in adjusted odds of UI by 144 percent among premenopausal women (OR 1.4, 95 percent CI 1.1; 1.8) and by 1.5 percent in depressed premenopausal women (OR 1.5, 95 percent CI 1.1; 2.0).¹⁹¹ Studies examining menopause assessed by self report have not found it to be an independent risk factor for stress UI.^{70, 95} Although menopause was not found to be significantly associated with UI, the number of years past menopause did increase the odds of UI by 15 percent, suggesting the aging process may have a greater role in the development of UI than hormonal status.⁹⁰

Gynecological or abdominal surgery

There is mixed evidence related to prior gynecological surgery as a risk factor (Appendix Table F20). Five^{55, 83, 108, 124, 138} studies of eight^{55, 62, 72, 83, 108, 124, 138, 144, 191} that examined this association reported significantly higher adjusted rates of total UI among women after hysterectomy. The increase was 160 percent (OR 1.6, 95 percent CI 1.1; 2.1) for UI in the past month,¹⁰⁸ 130 percent for at least monthly UI (OR 1.3, 95 percent CI 1.1; 1.6),¹²⁴ 140 percent for daily UI (OR 1.4, 95 percent CI 1.1, 1.6),⁵⁵ and 160 percent for severe UI (OR 1.6; 95 percent 1.5; 1.7).¹³⁸ In contrast, rates of stress UI were either the same after hysterectomy in four studies^{87, 108, 118, 135} or were less in the Women's Health Australia study (OR 0.8, 95percent CI 0.7; 0.9).⁶⁸ Women after hysterectomy also had the same prevalence of urge UI in three studies.^{70, 74, 108} Only one European cross-sectional study reported higher adjusted odds of bothersome urge UI among women after hysterectomy (OR 2.6, 95 percent CI 1.4; 4.4).⁸⁷ Depressed women after hysterectomy had UI 1.3 times more often (OR 1.3, 95 percent CI 1.0;1.7).¹⁹¹ Diabetics after hysterectomy also had increased adjusted rates of weekly UI (OR 2.3, 95 percent CI 1.0; 5.2).¹⁴⁰

Evidence on the relationship between UI and other gynecological conditions and procedures is conflicting (Appendix Table F21). Women with previous gynecological surgery had stress UI twice as often (OR 2, 95 percent CI 1.1; 3.7).⁹⁵ Women with prolapse (OR 4.11, 95 percent CI 2.15; 7.86)⁹² and after prolapse surgery had increased odds of UI.^{62, 68} In another study, pelvic organ prolapse surgery was not significantly associated with UI.¹³⁵ Women who had both a hysterectomy and prolapse repair were 1.8 to 2.3 times more likely to have UI compared to women without these surgeries.⁶⁸

Other surgeries on the uterus, excluding hysterectomies, led to conflicting findings, depending on the measurement of UI, when UI was confirmed by objective measures, they did raise the odds of having UI by 2.2 times, but when self report was used as the UI measure, there was no association.⁶² In another study of premenopausal women, having had abdominal or gynecological surgery was associated with a 170 percent higher risk when compared to women without this surgery;⁹⁶ and in another study involving both premenopausal and postmenopausal women, prior gynecological surgery doubled the odds of having UI.⁹⁵

Pelvic floor muscle contraction strength

Two studies examined the role of pelvic floor muscle contraction or exercises on either the prevalence or incidence of UI. In one cross-sectional study involving 507 women who completed a clinical evaluation, poor ability to contract pelvic floor muscles was strongly associated with UI (adjusted odds ratio of 3.5 for objectively confirmed according to the ICS definition UI and 4.5 for self-reported UI) depending on the definition of UI.⁶² In a study of primiparous women, inability to interrupt urine flow doubled the odds of being incontinent.⁶³⁶

Other factors

One clinical study found that abnormal findings on a gynecological examination were significantly associated with having UI in women ages 50–74.⁶² Vaginal symptoms (dryness, discharge, itching, dyspareunia) in postmenopausal women are significantly associated with any UI and severe UI.¹⁴⁴ In this same study, atrophic vaginitis was significantly associated with any UI but not severe UI. Vaginal colonization with E. coli was not independently associated with any or severe UI.¹⁴⁴

In a study of 665 primiparous women, perceived discomfort in the lower abdomen increased the odds of UI by 3.6 times.⁶³⁶ One large survey of middle-aged American women found that fibroids were associated with prevalent UI but not incident UI.¹⁴⁹

Childhood voiding dysfunction

There is limited evidence available on the role of childhood voiding dysfunctions as a risk factor for UI in adulthood. Evidence available suggests that childhood nocturnal enuresis is associated with the development of UI in adulthood, particularly urge UI. In one study, the odds of having UI were increased by 2.4 times among premenopausal women.⁹⁶ In two other studies, childhood nocturnal enuresis increased the odds of urge UI by 2.7 times.^{70, 637} but was not associated with stress UI.⁷⁰ It was also associated with increasing the risk of severe UI almost 3-fold.(Kuh, 1999, #1578). Childhood daytime incontinence has also shown to be associated with adult urge UI (OR 2.6, 95 percent CI 1.1; 5.9).⁶³⁷

Lower urinary tract symptoms

Few epidemiological studies have examined the association of lower urinary tract symptoms (urgency, frequency, nocturia, dysuria, difficulty with bladder emptying) as independent risk factors for either prevalent or incident UI. Findings are inconsistent across studies (Appendix Table F22). Urinary frequency was independently associated with UI in women ≥60 years,⁴² whereas it was not associated with UI risk in women ages 20–84 years.¹³⁹ Urgency increased the odds of UI by 9.3 times in women ages 60–84 years, but was not found to be associated in those ≥85 years.⁴² However, in another study with women ages 20–84 years, urgency did not predict UI risk.¹³⁹ Nocturia increased the odds of UI in women ≥85 years but not in those ages 60–84.⁴² Stinging or burning urine was significantly associated with UI in three age cohorts (ages 18–23, 45–50, and 70–75 years).⁶⁸

In a study of postpartum women, frequency of urination increased the odds of having UI one year after delivery.¹⁹⁸

Urinary tract infections

There is inconsistent data on the role of urinary tract infections as a risk factor for UI (Appendix Table 23). Women with urinary tract infections had higher rates of UI in 11 studies^{66, 72, 74, 90, 92, 97, 108, 116, 121, 127, 144} of 15^{66, 70, 72, 74, 90, 92, 96, 97, 106, 108, 116, 121, 127, 144, 187} that examined the association. Women with recurrent urinary tract infection had the highest increase in UI by 230 percent for weekly UI (OR 2.3, 95 percent CI 1.3; 3.9)¹²⁷ and for monthly UI (OR 2.3, 95 percent CI 1.6; 3.1),¹⁰⁸ 220 percent for UI in the past year (OR 2.2, 95 percent CI 1.4; 3.4),⁷² and by 470 percent for ever having UI (OR 4.7, 95 percent CI 4.7; 8.9).¹¹⁶

Several studies reported that menopausal status can influence the association between urinary tract infections and UI. One study in premenopausal women,⁹⁶ one study in postmenopausal women,¹⁰⁶ and one study that included both premenopausal and postmenopausal women did not find a significant increase in odds of UI among women with urinary tract infection.¹³⁹ In contrast, one survey in postmenopausal women found that the lifetime number of urinary tract infections (six or more) increased the risk of UI by 1.9 times for any UI and 2.0 times for severe UI.¹⁴⁴ Two studies found urinary tract infection increased the odds of UI by 4.8 times in women who were perimenopausal and by 3.4 in women who were menopausal.^{90, 152}

Urinary tract infections have been associated with UI type; in one survey two or more urinary tract infections in the past year doubled the odds of having urge UI.⁶⁶ In another study, cystitis was significantly associated with stress UI at baseline and 1 year later, increasing the odds of stress UI by 50 to 90 percent.¹³²

Postvoid residual bladder volume

One study found that postvoid residual bladder volume was not associated with UI in postmenopausal women.¹²⁸

Bladder or urinary surgery

Several studies have examined whether prior bladder or urinary surgery is a risk factor for UI. Some studies have not found significant associations,¹⁴⁴ whereas another study found that UI surgery doubled the odds of having UI in perimenopausal women.¹³⁵

Arthritis and musculoskeletal disorders

Evidence from three studies,^{7, 86, 118} suggested that women with arthritis had higher rates of UI (Figure 14). Two studies found that arthritis was associated with increased odds of UI by 80–88 percent.^{7, 86} Arthritis was also an independent risk factor for UI type. In a study involving women ages 70–79 years, arthritis was significantly associated with both urge and stress UI.¹¹⁸

Figure

Figure 14. Association between arthritis and UI in women.

Joint pain was significantly associated with stress UI in a 1-year followup study, increasing the odds of UI by 40 percent.¹³² A history of hip fracture increased the odds of UI by 38 percent.⁸¹ Osteoporosis was associated with UI in one study,⁷ but the association can be confound by age.

Diabetes

There is growing evidence to suggest that diabetes mellitus increases the odds of having UI (Appendix Table F24). Two studies^{134, 149} of three^{134, 149, 186} reported that women with diabetes develop UI more often. Incidence of weekly UI was higher by 147 percent (OR 1.5; 95 percent CI 1.2; 1.9) in women with duration of diabetes more than 10 years.¹³⁴ The same Nurses' Health Study cohort showed that incidence of severe UI was increased by 175 percent (OR 1.8, 95 percent CI 1.3, 2.3) and very severe by 126 percent (OR 2.6; 95 percent CI 1.4; 5.0).¹³⁴ The Study of Women's Health Across the Nation found 302 percent increase in developing monthly UI in women with diabetes independent on other risk factors (OR 3.0, 95 percent CI 1.1; 8.1).¹⁴⁹

Prevalence of stress UI was greater in women with diabetes in three studies^{95, 133, 149} of six^{66, 95, 108, 133, 135, 149} that examined the association (Figure 15). Pooled odds ratio of prevalent stress UI was not significant (OR 1.4, 95 percent 0.9; 2.1). The result of meta-analysis was sensitive to one study, The Heart & Estrogen/Progestin Replacement Study Research Group, randomized trial of 2,763 women taking combination hormone therapy to prevent coronary heart disease, that did not find a significant association between diabetes and stress UI.⁶⁶ However, the majority of the studies reported a significant increase in adjusted odds of total UI among women with diabetes (Figure 16)^{55, 66, 72, 86, 93, 97, 108, 116, 124, 127, 128, 133, 134, 138, 149, 186} Pooled analysis of 16 studies resulted in an odds ratio of 1.4 (95 percent CI 1.2; 1.5) of having prevalent UI in women with diabetes. Four of five studies found a significant increase in adjusted odds of urge UI among women with diabetes (Figure 17).^{66, 108, 118, 133, 149} Pooled analysis of five studies estimated that diabetic women had urge UI 1.7 times more frequently than nondiabetics (95 percent CI 1.2; 2.2).

Figure

Figure 15. Association between prevalent stress UI and diabetes in females.

Figure

Figure 16. Association between prevalent UI and diabetes in women.

Figure

Figure 17. Association between prevalent urge UI and diabetes in women.

Diabetic complications such as macroalbuminuria, retinopathy, and/or peripheral neuropathy significantly increased the odds of having UI in two studies.^{128, 140} Diabetic neuropathy was associated with an adjusted odds ratio of 2.4 and macroalbuminuria increasing the odds by 3.8.¹⁴⁰ However, in another study, a BMI adjustment decreased the strength of these associations.¹²⁸ This same study also found that neither diabetes treatment (diet, pill, or insulin) nor duration of treatment was associated with UI after adjustment for BMI. Blood glucose control as measured by HbA_1c was also not associated with UI.¹²⁸Women who have insulin-dependent diabetes were found to have a 3.5 times higher risk of urge UI when compared to women without diabetes.¹²⁸

Stroke

One prospective cohort, the Canadian Study of Health and Aging, found that elderly community dwelling women had higher risk of developing UI after stroke (OR 1.6, 95 percent CI 1.1; 2.2).²⁰⁷ Prevalence of UI was significantly higher among women after stroke in five^{55, 86, 97, 133, 186}of six^{55, 67, 86, 97, 133, 186} studies that examined this association (Figure 18) Pooled analysis of six studies estimated an increase by 167 percent of UI in women with history of stroke (OR 1.7 95 percent CI 1.4; 2.1). Severe UI was more prevalent in women with history of stroke in one study (OR 1.9, 95 percent CI 1.4; 2.8).⁶⁷ Paraplegia was also associated with UI (OR 1.6, 95 percent CI 1.1; 2.6).¹³³

Figure

Figure 18. Association between stroke and prevalent UI in women.

Neurological disorders

There is conflicting evidence on neurological diseases and prevalent UI (Appendix Table F25). Women with any neurological diseases had greater rates of ever having UI (OR 3.8, 95 percent CI 1.7; 8.6).¹¹⁶ Parkinson's disease was associated with increased odds of having UI in one study (OR 2.3, 95 percent CI 1.1; 4.5)⁸¹ while two other studies^{97, 133} did not find a significant association.

Pulmonary disorders

Asthma and chronic obstructive pulmonary disease (COPD) have been studied as potential risk factors for UI in several studies with conflicting findings (Appendix Table F26). Although most studies find COPD to predict UI risk,^{81, 124, 133} in a large study involving 29,520 women,⁸⁶ COPD was not associated with UI. In a smaller survey involving 1,531 women ≥65 years, COPD increased the odds of UI by 53 percent.⁸¹ In other studies, COPD was associated with the type of UI. In women ages 55–75 years, the odds of having stress UI were increased five-fold, although it was not associated with urge UI risk.¹¹⁸ Asthma was not associated with UI risk in one study,⁸⁶ whereas in another study it increased the odds of stress UI by 50 percent at baseline but was not significantly associated 1 year later.¹³² Several studies indicated that frequent or prolonged coughing increases the odds of UI by 33 to 60 percent^{114, 638}

Comorbidity and poor health

Some studies have examined the role of comorbidity in relation to UI (Appendix Table F26). Elderly community dwelling women with kidney problems (OR 1.7, 95 percent CI 1.2; 2.3) and foot diseases (OR1.4, 95 percent CI 1.0; 1.8) were at risk of developing UI.¹⁸⁶ Women with two or more comorbid diseases had higher adjusted rates of UI (OR 5.9, 95 percent CI 3.7; 9.6).⁹² Increased comorbidity index was associated with higher adjusted odds of UI in three studies.^{121, 124, 189}

In studies involving community dwelling females, poor self-rated health increased the odds of having daily UI by 60 percent⁵⁵ and was weakly associated with stress UI.¹¹⁸ In another study, poor health was not associated with stress or urge UI in post-menopausal women; however, it did increase the odds of having mixed UI by 143 percent (OR 1.4, 95 percent CI 1.1; 1.8).⁶⁶ In a large survey of women ages 40–55 years, poor health was associated with increased odds of urge UI but not mixed or stress UI.¹⁴⁹ Women with poor or fair health experienced UI 2.6¹⁸⁷ to 2.9⁸⁴ times more often.

Cardiovascular disorders

There is limited evidence on the role of cardiovascular disorders such as heart problems and hypertension on UI (Appendix Table F26). Hypertension was strongly associated with UI in two studies,^{83, 127} whereas two studies did not find a significant association with any UI⁶³⁹ or with stress UI.⁹⁵ In large surveys, heart problems,⁶³⁹ including congestive heart failure,^{55, 97} were not associated with UI.

Gastrointestinal diseases and procedures

Other surgical procedures

One study found that varicose veins and hemorrhoids were independently associated with UI at 1-year post-partum in primiparous women by 50 percent.⁶⁴⁰ Previous surgery for UI was associated with increased odds of weekly UI (OR 1.4, 95 percent CI 1.2; 1.8)⁸³ and stress UI (OR 1.3, 95 percent CI 1.3; 4).¹³⁵ In one study, urological surgeries were not associated with prevalent UI in women.⁷²

Incident UI

Only one study reported incident UI in women with diabetes treated with pharmacological agents and found that insulin administration but not oral medications were association with a 350 percent increase in developing of UI (OR 3.5, 95 percent CI 1.6; 7.9).¹²³

Diuretics

The evidence on the role of diuretics on UI risk is conflicting. Earlier studies that examined bivariate associations indicated that diuretics were associated with UI in older adults.⁶⁴¹ However, large studies involving multivariate analyses did not find that diuretics including nonthiazide diuretics increased the odds of having UI.^{55, 86} Diuretics is a strong predictor of UI type, particularly urge UI.¹³³ One study found that diuretics significantly increased the odds for stress and urge UI in women ages 40–60 years by two to four times.⁷⁸

Estrogen

Several epidemiological studies have examined estrogen therapy as a risk factor for UI (Appendix Table F29).^{55, 59, 105, 118, 144, 191} The risk of UI is elevated among women taking postmenopausal hormones (oral and transdermal estrogen with and without progestin) as compared to those who have never taken them.¹⁵³ Women taking transdermal estrogen with and without progestin (RR 1.7, 95 percent CI 1.4; 2.06 and RR 1.5, 95 percent CI 1.2; 1.8, respectively) had a slightly higher risk of UI than those who took the oral forms (RR 1.5, 95 percent CI 1.4; 1.6 and RR 1.3, 95 percent CI 1.2; 1.4). This same study also found there was little risk after cessation of hormones and a decreasing risk of incontinence with increasing time since last hormone use.¹⁰⁵ Ten years after stopping hormone use, the risk was identical in women who had and had not taken hormone therapy. Two studies found that oral hormone replacement therapy increased the odds of having UI by 1.9 times.^{55, 59} Vaginal estrogen cream use was significantly associated with any UI in women ages 55–75 years, although this same association was not found with severe UI.¹⁴⁴ Systemic hormone replacement therapy is strongly associated with urge and stress UI. In a study of 1,584 women ages 70–79 years, current oral estrogen use increased the odds of urge UI by 70 percent and stress UI by 98 percent, respectively.¹¹⁸ Hormone use for menstrual disorders was also an independent risk factor for UI.⁹⁶ Former use of oral contraceptives increases the odds of UI by 18–20 percent, although current use was not associated.¹³⁸

Psychotropic medications

A few studies have examined whether antidepressants in general or specific types of antidepressants are a risk factor for UI. In a large survey, antidepressant use was associated with increased odds of having UI by 75 percent.⁸⁶ In a study involving 6,642 women, the use of selective serotonin reuptake inhibitors (SSRIs) or serotonin norephinephrine reuptake inhibitors (SNRIs) was not associated with UI in multivariate models that included all women, depressed women only, and nondepressed women who reported depressive symptoms.¹⁹¹ However, in a subanalysis of depressed women on SSRI/SNRIs, there was a significant association between drug use and UI.¹⁹¹ Tranquilizers have been shown to increase the risk of UI by 65 percent.⁸⁶ The contribution of psychotropic medications, independent from baseline disease on UI, has not been examined in prospective observational studies. The results from RCTs showed that serotonin-noradrenaline reuptake inhibitor, duloxetine improved stress UI,⁶⁴² suggesting that depression rather than the antidepressant could be considered as a risk factor for UI.

Other medications

In a large survey of 29,520 women, pain, narcotics, asthma, hypertensive, and heart medications were not associated with an increased risk of UI.⁸⁶ In this same survey, antibiotics were significantly associated with UI, increasing the odds by 64 percent, laxatives by 67 percent, and hypnotics by 52 percent. However, another study did not find an association between sleeping medications and UI risk.¹⁰⁶ Polypharmacy (e.g., three drugs or more) was not associated with UI risk in women ages 70 years and over.¹⁰⁶

Summary

Limited evidence, Level IIA from prospective cohort studies suggested that increased BMI, diabetes, comorbidities, cognitive decline, and hormone therapy were associated with developing UI in community dwelling females. Prevalence of UI was higher in aging and depressed women, after stroke, vaginal trauma, and in women with physical dependency. The strength of the association depends on definitions of UI by time, type, and severity. Differences in assessment of associated factors may also contribute to the effects on UI. Hormone status of women modified the effects of other associated factors. Comparisons across the studies were difficult to make due to methodological heterogeneity in adjustment and statistical models.

Association between risk factors and UI in LTC settings

The prevalence of UI increased with the length of stay in nursing homes from 39 percent at 2 weeks to 44 percent at 1 year after admission.²⁸ It is not clear if this is an effect of prolonged exposure or a difference in case mix (Table 24). The majority of residents with cognitive impairment experienced UI (72 to 84 percent).¹ The proportion of incontinent patients increased responding to severity of impairment, from 60 percent in mild to 93 percent in severely demented.³⁴ Physical dependency was associated with a higher prevalence of UI, from 26 percent in independent residents to 81 percent immobile.³⁴ The prevalence of UI among dependent residents was more than 70 percent in seven countries and varied from 87 percent in Iceland to 72 percent in the United States.¹ The prevalence of UI among residents with diabetes ranged from 55 percent²⁹ to 65 percent.³⁴ Almost all (93 percent) residents with FI also experienced UI.²⁹ One study of 9,013 patients with multiple sclerosis reported 21 percent UI and 9 percent of frequent UI.⁶⁵⁴ The estimations were not consistent; however, the prevalence of UI among patients with urinary tract infection in Italy was 63 percent in one study¹ and 81 percent in another.²⁹ Few studies examined adjusted odds ratios of UI among residents in LTC independent of other confounding factors. Aging was associated with increased odds of UI by 3 percent per year to 24 percent per 5 years of age (Table 25).^{4, 165} In contrast with higher incidence of UI in males,¹⁶⁶ prevalence of UI was lower in males than females in two studies^{29, 165} of three ^{29, 165, 655} that examined this association. Race, BMI, diabetes, arthritis, and cardiovascular diseases did not show significant association with UI, but the evidence was limited to one state survey of nursing homes.^{4, 165}

Table 24

Prevalence of UI among residents in LTC with risk factors.

Table 25

Association between UI and risk factors (adjusted odds ratios) in LTC settings.

Stroke increased the adjusted odds of UI by 20 to 40 percent.^{4, 165} Physical dependency was a strong and independent risk factor for UI in several studies (Table 26).^{4, 29, 164, 165} Impairment of ADLs was associated with three to four times larger odds of UI.^{4, 164} Residents with dependency in toilet use experienced UI six times more often.¹⁶⁵ The odds of UI were seven times higher among wheelchair users and bedridden residents.²⁹ Mental impairment was associated with 192 to 361 percent higher prevalence of UI.²⁹

Table 26

Association between dependency, mental, gastrointestinal diseases, and UI (adjusted odds ratios) in LTC settings.

FI was associated with ten²⁹ to 20¹⁶⁵ times larger odds of UI. The conditions that assume intensive bowel control and frequent checking of wet condition of residents, including tube feeding and diarrhea, were associated with lower UI. In conclusion, consistent evidence suggests that restrictions in ADL, physical dependency, and cognitive impairment are risk factors for UI in LTC. Factors that induce UI also seem to affect FI.

Age

In the past decade, a large body of evidence has suggested an association between FI and patient age. Age was a strong predictor of FI and combined UI and FI in the majority of the studies (79 percent or 22/28) (Table 27). The association between age and FI was examined among women only.^{187, 208}–²¹³ Adjusted odds of AI increased by 87 percent for every additional 10 years of age in Asian women and by 36 percent in White women.¹⁸⁷ The frequent comorbidity of FI and UI may be attributable to the common cause of both disorders, such as pelvic floor sphincter injuries.^{4, 58, 162, 186} Women over age 42 reported combined UI and FI 16 times more often than younger females after adjustment for other risk factors.¹⁸⁶ Japanese adults over age 75 had 319 percent higher odds of combined UI and FI compared to those 65–74 years old.⁵⁸ The association between age and FI was not linear.⁶⁵⁷ However, one study reported a 30 percent increase in the adjusted odds of FI corresponding to a 10 year increment in age in women.²²⁵ The threshold age for women was related to menopausal age; no significant increase in FI was shown when women older than 30 or 35 years were compared to women younger than 29 years of age.²³⁰ The results became significant when 30–49 year old women were compared to those 50–69 years old (211 percent increase in FI) or those 70–90 years old (222 percent increase in FI).¹⁶³ No studies reported adjusted odds ratios of FI in men, with an inconsistent increase in the crude odds of FI after 35,⁶⁵⁸ 44,⁶⁵⁹ or 65 years of age.²²¹ Studies with women and men analyzed together reported a 2 percent increase in FI in adults older than 40 years²²² and a 271 percent increase in FI after age 75⁵⁸ after adjustment for other risk factors.

Table 27

Association between age and FI.

Age not only increases the risk of FI, it likely modifies the effects of other risk factors, including birth related anal trauma. Maternal age older than 25 years was associated with greater odds of FI²¹⁴ and persistent FI.²¹⁵ Women 20–24 years old at the time of delivery did not experience a larger risk of FI compared to younger pregnant women.²¹⁸ Less expected was the fact that pregnant women below age 20 experienced FI more often than women over 30.²¹⁸

The interaction between age and other risk factors requires future research. Pooled analysis indicated that studies with older adults reported higher crude rates of FI, but the estimation had low validity due to significant statistical heterogeneity between studies.⁵¹⁸Age, but not study quality, contributed to differences in rates of FI.⁵¹⁸ The authors used different age categories which made pooling more questionable. Pooled analysis of patients' individual data should give better information of the curve-linear association and meaningful cut-points for men and women independent of other risk factors.

Gender

This line of research shows mixed results (Table 28). Traditionally, FI is believed to be more prevalent in women.^{6, 622} Two recent studies confirmed this finding by reporting a greater prevalence of FI in women.^{7, 508} One study reported greater prevalence of combined incontinence (FI + UI) in community-living elderly women than men.⁶⁶⁰ However, three different groups of researchers found no differences in the prevalence of FI in women and men.^{4, 216, 217} Four other studies showed that male gender is an independent risk factor for FI^{3, 58, 222, 224} and combined incontinence.⁵⁸ One study indicated that men ages 75–79 or 85–89 are more likely to have FI, combined incontinence, and less severe FI than women in the same group.¹¹⁴ The majority of these studies involved a large sample (more than 4,000) elderly people (>65 years) and adjusted for at least one risk factor. However, a definite association between gender and FI cannot be drawn given varying study designs, sampling and statistical methods, FI definitions, and varying control for potential risk factors. Valid synthesis of different studies would be possible when relative risk of incident incontinence, adjusted for age and gender-specific confounding factors, was examined in prospective cohorts.

Table 28

Association between gender and FI (women as a reference gender).

Race

Limited evidence from one study suggested that Asian women had a higher prevalence of birth related FI compared to other ethnic groups after adjustment for mode of delivery, maternal age, perineal trauma, BMI, and total number of births (Appendix Table F34).²¹⁵ The prevalence of combined UI and FI was two times higher in non-White than White residents of nursing homes in Wisconsin, independent of other risk factors identified in the Health Care Financing Administration Minimum Data Set.⁴ Younger Taiwanese women living in the community (mean age of 43 years) had a prevalence of FI of 2.5 percent,²¹³ a rate similar or less than that reported in non-Asian women of a similar age group.^{218, 232} The association between race and severity of FI and quality of life related to FI in men and women from the community and in LTC facilities requires future investigation.

Diet, BMI, and diabetes

The highest prevalence of FI (44 percent) was reported among women with morbid obesity undergoing laparoscopic weight loss surgery (Appendix Table F35). The prevalence of FI among patients with diabetes mellitus varied from 1 percent in women^{163, 211} to 13 percent in a study of both genders;⁶²⁶ 3 percent of men and women with diabetes experienced incontinence of solid feces and 13 percent solid or liquid feces.⁶²⁶

Adults with decreased serum levels of vitamin B12 (300pg/ml or less) had a significantly higher prevalence of combined UI and FI (Table 29).⁶⁶¹ No other dietary intakes have been shown to increase the risk of FI. Increased BMI and the presence of diabetes in residents of nursing homes were not associated with higher rates of combined UI and FI.^{3, 4} Higher BMI in women did not increase the risk of FI^{208, 210}–²¹³ but was associated with higher prevalence of flatus⁶⁶² and liquid feces incontinence.¹⁸⁵ One study of 8,774 women reported a 70 percent increase in risk of persistent postnatal FI in obese women (Appendix Table F36).²¹⁸

Table 29

Association between FI and diet, BMI, and diabetes.

Inconsistent evidence suggested that men and women with diabetes had FI more frequently than nondiabetics with rates higher by 70 percent²¹⁶ to 300 percent⁵⁸ in older adults and by 207 percent in younger adults²¹⁷ after adjustment for other risk factors.

Physical activity

Physically active adults had a 70 percent lower risk of fecal urgency after adjustment for other risk factors (Appendix Table F37).¹²⁵ Distance runners had a FI prevalence of 12 percent in one study with no formal control comparison.⁶⁶³ The association between physical activity and incidence of FI and quality of life related to FI in adults in community and LTC requires future investigation.

Smoking and pulmonary diseases

The association between smoking and FI is inconsistent in direction and effect size (Table 30).^{125, 185, 186, 208, 218, 226, 227} White women with obstructive pulmonary diseases had a 208 percent higher rate of FI after adjustment for possible confounding factors.¹⁸⁷ Women with chronic bronchitis had liquid incontinence 3.5 times more often in a bivariate analysis.¹⁸⁵ Older men and women with protracted coughing experienced increased risk of combined UI and FI by 65 percent.¹¹⁴

Table 30

Association between FI, smoking, and pulmonary diseases.

Muscle-skeletal disorders

The crude prevalence of FI was 71 percent higher among older men with arthritis with attenuation of the association in multivariate analysis (Table 31).¹⁸⁶ Foot and ankle problems in elderly adults were associated with a two-fold increase in prevalence but not incidence of FI.¹⁸⁶ Arthritis in older men and women increased the risk of liquid FI by 80 percent in bivariate analysis of 1,352 adults over 75 years old.¹⁵⁸

Table 31

Association between FI and muscle-skeletal diseases.

Neoplasm

Older patients with rectal cancer had the highest prevalence of FI (55 percent),⁶⁶⁴ but other studies showed lower prevalence of FI among patients with rectal cancer^{665, 666} (Appendix Table F38). Patients after proctocolectomy with J-pouch anastomosis reported 25 percent prevalence of FI with lower rates after W-pouch anastomosis.⁶⁶⁷

Comorbidities

High rates of medical comorbidities in women were associated with a 2.6 times higher adjusted risk of FI (Table 32).¹⁶³ Incremental increases in the Charlson Comorbidity Scale scores in women were associated with 76 percent higher adjusted odds of FI.²¹⁹ The presence of kidney diseases in older adults was associated with a significant risk of incident FI by 48 percent in women and 94 percent in men.¹⁸⁶ Men suffering from transient ischemic attack had a three times higher prevalence of FI.²¹⁹ Women with arterial hypertension reported FI 2.4 times more often compared to normotensive females after adjustment for other risk factors.²¹³ Heart disease in nursing home residents did not increase the risk of FI.^{3, 4}

Table 32

Association between FI and comorbidities.

Stroke and other neurological disorders

The prevalence of FI was 30 percent²²¹ to 34 percent²²⁰ after acute stroke (Appendix Table F39). The absolute risk of FI was less than 10 percent several months after stroke in most studies.²²⁰–^{222, 668} Stroke was associated with a significant increase in adjusted odds of FI in the majority of the studies (Table 33). Combined UI and FI was five times more frequent in community dwelling adults after stoke⁵⁸ but not in residents of nursing homes.⁴ Incident FI occurred twice as frequently in women but not in men after a stroke.¹⁸⁶ Adjusted prevalence of FI was three²¹⁶ to five⁵⁸ times higher in stroke survivors compared to adults without a stroke. Incremental 10-year increase in age at the time of a stroke was associated with a 45 percent higher probability of FI 6 months after the stroke.²²⁰ Diabetics experienced FI 2.2 times more often after stroke compared to non diabetics after adjustment for confounding factors.²²⁰ Severity of stroke was positively associated with FI.^{220, 221} Functional dependency and need for assistance with toilet access 3 months after stroke in men and women was associated with a 349 percent increase in FI. Residents of nursing homes after stroke suffered from FI 1.2–1.3 times more often.⁴ Adjustment for other risk factors of FI attenuated the strength of association, but a higher odds ratio was still highly significant (4.9, 95 percent CI 14; 16).⁵⁸

Table 33

Association between FI and stroke.

Several studies examined the prevalence of FI in patients with neurological diseases (Appendix Table F40). The prevalence of FI was 10 percent in men and 5 percent in women with Parkinson's Disease.⁶⁶⁹ Two studies of patients with multiple sclerosis reported that 21 percent⁵²⁴ to 51 percent⁶⁷⁰ experienced FI. Among patients with spinal cord injury, subjects with tetraplegia had the highest prevalence of FI (14 percent)⁶⁷¹ compared to 4 percent in cases with paraplegia, and 2 percent in patients with incomplete injury.⁶⁷¹

Functional impairments and dependency status in association to FI in residents of nursing homes

were examined in one study using Wisconsin annual nursing home surveys (Table 34).^{3, 4} Tube feeding was associated with an 8–9 percent increase in the odds of FI,³ dependence in daily activities with 6–7.3 higher odds of FI, and dependency in eating with four times higher odds of FI after adjustment for all risk factors in the model.³ Nursing home institutionalization was examined in one study of 430 residents.¹⁶⁶ Increased length of stay from 2 weeks to 1 year was associated with a seven times higher prevalence of FI.¹⁶⁶ Residents with pressure ulcers experienced FI 2.3 to 2.6 times more often; this increase was independent of other FI risk factors. (Appendix Table F41)³

Table 34

Association between FI and functional status of residents in nursing homes.

Functional status and mental health in association with FI in community dwelling adults

Poor general health and dementia increased the risk of combined UI and FI in adults after adjustment for confounding factors (Table 35).⁵⁸ Crude rates of combined incontinence were significantly higher in older adults with memory problems, edema in legs, and slow healing wounds.¹¹⁴ The odds ratios of FI were significantly higher by 188 percent in women and by 218 percent in men with poor general health,²¹⁹ by 400 percent in adults with functional limitations,²²² and by 80 percent in adults with decreased mobility and severe/mild physical activity impairment.²²³ Major depression in women was associated with 273 percent higher rates of FI independent of other risk factors.¹⁶³ Geriatric depression in men was associated with a 283 percent higher risk of FI.²¹⁹ Adults with dementia had FI twice as often in two studies after adjustment for other risk factors of FI.^{223, 224} Impaired cognitive status in men was associated with a five-fold increase in the adjusted odds ratio of FI.¹⁸⁶ Older men and women with Mini-Mental State Examination (MMSE) score <15 had a 250 percent increase in the adjusted rates of FI.²²³

Table 35

Association between FI with functional status and mental health in adults.

Chronic gastrointestinal conditions

Constipation, chronic diarrhea, and fecal impaction increased the adjusted relative odds of FI, but not combined incontinence, in residents of nursing homes (Appendix Table F42).^{3, 4} Constipation increased the risk of FI by 209–211 percent in women (Table 36).^{187, 212} Women with irritable bowel syndrome experienced FI 2.7 to 6.3 times more often after adjustment for other risk factors.^{187, 211, 212} Crude rates of combined incontinence were significantly higher in men and women with chronic gastrointestinal disorders.¹¹⁴ The adjusted rates of FI were 240 percent²²⁵ to 450 percent²¹⁹ higher in women with diarrhea, 190 percent higher in irritable bowel syndrome patients, 250 percent higher in females with anal fistula, and 140 percent higher after cholecystectomy.²²⁵ Adults with incomplete bowel evacuation had four times higher odds of FI.²¹⁷ Frequent diarrhea and water stools were associated with higher rates of liquid FI (Appendix Table F43).²²⁴ One study reported increased crude rates of severe FI in association with chronic gastrointestinal problems in adults.¹⁵⁸

Table 36

Association between FI and gastrointestinal conditions in adults.

Surgical procedures

The majority of women (97 percent) after surgery for rectal prolapse experienced FI (Appendix Table F44).⁶⁷² Hemorrhoid surgery in women was associated with a 270 percent increase in the odds of FI.²¹¹ Patients with inflammatory bowel disease after ileal pouch-anal anastomosis and pouchitis had a 43 percent prevalence of occasional and 11 percent of frequent FI.⁶⁷³ Perianal surgery of the skin outside of the anal canal was associated with a five times higher adjusted odds ratio of FI.²¹⁷ The odds of FI were not significantly different among women after anorectal surgery after adjustment for parity and delivery.²²⁵ Isolated flatus incontinence at least once a week occurred 1.5 times more often after previous lower abdominal or urological surgery.²²⁷ The association was not consistent across different surgeries and studies. Gynecological surgery increased the adjusted odds ratio of AI by 180 percent (Appendix Table F45).²¹³

Drug administration

Crude rates of combined UI and FI were higher after diuretic and laxative use (Appendix Table F46).⁶⁶¹ One large study of 6,099 adults over 65 years old reported a significant increase in the adjusted odds of FI after anticonvulsants, antidepressants, anti Parkinson medications, antipsychotic, narcotics, and hypnotics.²¹⁶ Hormone replacement therapy in morbidly obese women before laparoscopic weight loss surgery²⁰⁸ and contraceptive use²²⁵ did not show a significant association with FI.

Prostate diseases

The prevalence of FI was less than 10 percent in all studies among men with prostate diseases (Appendix Table F47).^{180, 674, 675} Prostate diseases, including prostate cancer, were associated with an increased adjusted odds ratio of FI (Table 37).^{180, 219} Both baseline prostate diseases and treatments have been associated with an increased risk of FI. Baseline FI before perineal prostatectomy for localized prostate cancer was associated with 600 percent higher adjusted rates of FI and 400 percent higher rates of severe FI after surgery.⁶⁷⁶ Patients with positive surgical margins experienced severe FI seven times more often after surgery.⁶⁷⁶ Radical prostatectomy was associated with a significant increase in the adjusted odds ratio of FI by 530 percent compared to healthy controls.¹⁸⁰ External beam radiation was associated with higher rates of bowel symptoms than radical prostatectomy independent of other confounding factors.^{677, 678}

Table 37

Association between FI and prostate diseases.

Anal trauma

Consistent evidence suggests that anal trauma was associated with significant risk of FI independent of other risk factors (Table 38). The adjusted odds of FI were 4.4 times higher in women with anal sphincter tear.²³² Women with anal injury not related to childbirth had 240 percent increases in FI.²²⁵ The adjusted odds ratio of bothersome FI was 16 times higher in women with anal sphincter defect.⁶⁷⁹ A combination of anal injury and pelvic organ prolapse increased the odds of bothersome FI by seven times. Women with anal injury, pelvic organ prolapse, and UI had an adjusted odds ratio of 55 to experience bothersome FI.⁶⁷⁹ Perianal injury increased risk of FI by 262 percent in men and women in one study.²¹⁷ The independent contribution of anal injuries in males and females by age and race categories requires future investigation. Abortions were not associated with FI in one study (Appendix Table F48).²²⁷ Menopause did not increase the adjusted odds of FI^{210, 213} but was associated with a higher prevalence of FI in another study²²⁶ (Appendix Table F49).

Table 38

Association between FI and anal trauma in females.

FI in association with UI

The prevalence of FI was 25 percent²¹⁰ to 28 percent⁶⁸⁰ in women with stress UI and 41 percent in women with overactive bladder.²¹⁰ UI was associated with 11–12 times higher odds of FI in residents of nursing homes in one study (Appendix Table F50).³ Several studies reported an increase in the crude odds of FI in patients with UI^{187, 208, 210}–^{213, 681} but only one study²¹¹ found a significant (1.9 times) increase in the adjusted odds ratio of FI in 881 women with UI (Table 39). Urinary symptoms other than UI in men and women were associated with a significant increase in crude odds of combined UI and FI in one study.¹¹⁴ UI in men and women was associated with 2.9²²³ to eight times greater odds of FI.²²² The adjusted odds of FI in women with UI were two to six times higher compared to continent females.^{162, 163, 219, 226} The presence of stage II pelvic organ prolapse and/or urinary incontinence compared to stage 0 or I pelvic organ prolapse and no UI increased the adjusted odds of bothersome FI by 4.9 times.⁶⁷⁹

Table 39

Association between FI and UI.

Prolapse

Most studies reported a prevalence of FI of more than 10 percent in women with genital prolapse (Appendix Table F51). The adjusted odds of FI were 1.9 times in patients with rectal prolapse²¹¹ and 3.2²¹³ to five times in women with utero-vaginal prolapse (Appendix Table F52). One study reported a significant increase in the crude odds of flatus, liquid, and solid FI in women with genital prolapse.¹⁸⁵ More than a third of men and women with rectal prolapse reported FI.⁶⁸² Burch colposuspension was associated with greater adjusted odds of flatus and liquid FI compared to healthy controls 14 years after surgery.⁶⁸³ Women after Burch colposuspension also experienced more severe flatus and liquid incontinence with greater impact on their quality of life (Appendix Table F53).⁶⁸³

FI related to pregnancy and birth

Consistent evidence suggested a significant dose response increase in the absolute risk of FI after vaginal trauma. The prevalence of FI was 7 percent at 3 months and 5 percent at 12 months postpartum (Appendix Table F54).⁶⁸⁴ Women with sphincter tears experienced higher rates of FI, from 15 percent among nullipara to 30 percent after more than two deliveries.⁶⁸⁵ The highest prevalence (40 percent) was reported in nullipara with fourth degree sphincter tear after index delivery.⁶⁸⁶ Less than 1 percent of women experienced FI after vacuum delivery.²¹¹ The prevalence was 2 percent²¹¹ to 5 percent⁶⁸⁷ after forceps delivery and 13 percent²¹¹ after episiotomy or vaginal delivery.²¹⁰ The prevalence of combined UI and FI was 6 percent at 10 months⁶⁸⁸ and 2 percent at 24 months postpartum.⁶⁸⁴ Pregnant women⁶⁵⁷ and women after episiotomy⁶²⁸ reported 13 percent FI. The prevalence of FI was lower by 1 percent in nulliparous and 3 percent after spontaneous vaginal delivery.¹⁶³ The prevalence of FI was lowest (0.3 percent) after Cesarean sections.¹⁶³ The prevalence of flatus was 4 percent in pregnant women with one previous delivery,²²⁷ 6 percent at 3 months after delivery,⁶⁸⁴ 8 percent after instrumental delivery,⁶⁸⁹ and 25 percent after previous multiple pregnancies⁶⁸⁹ and among multiparous women.⁶⁵⁸ One small study reported the highest prevalence of flatus 30 years postpartum in women with anal disruption (59 to 76 percent) and episiotomy (70 percent).⁶⁹⁰ The prevalence of liquid incontinence was less than 6 percent^{227, 658} and solid incontinence less than 5 percent^{227, 658, 684, 688} in pregnant women and after different modes of delivery. Women with obstetric sphincter tears experienced higher rates of incident FI and severe FI (Appendix Table F55). More than 5 percent of women with complete third-degree tears after the first pregnancy and more than two deliveries after that experienced severe FI.⁶⁸⁵ Almost 7 percent of women 12 months after instrumental delivery and/or a high birth weight infant reported solid and liquid FI.⁶⁵⁸ Complete FI with liquid and solid incontinence was reported by 12 percent of women 13 years after an obstetrical anal sphincter tear.⁶⁹¹ More than 2 percent of young women 3 months postpartum complained about weekly solid or liquid incontinence.²²⁸

The relative effects of different risk factors were less consistent across the studies and depended on the definitions of FI and adjustment for confounding factors (Table 40). The adjusted odds of FI were significantly increased after delivery of heavy babies,²⁰² among women with narrow subpubic arch angle,⁶⁹² and with FI during pregnancy,²⁰⁹ after increased number of births,^{215, 226} and a high degree of perineal injury.^{187, 209, 228}–²³¹ No independent significant risk factors were reported for solid FI.^{185, 693, 694} Caesarean section was associated with a reduced risk of frequent FI²¹⁴ in contrast with forceps delivery followed by frequent and persistent FI^{214, 215, 218} (Table 41). The number of births²¹⁵ and anal injury^{218, 227} were associated with higher adjusted odds of severe FI.

Table 40

Association between FI and pregnancy related risk factors.

Table 41

Association between risk of FI and severity of FI and factors related to pregnancy and birth.

Fetal characteristics were examined in association to FI in 11 studies (Appendix Table F56).^{202, 205, 218, 226}–^{228, 232, 662, 689, 692, 695} Fetal weight was associated with random changes in AI^{205, 228, 232, 692} and FI,^{662, 689} but with a significant increase in the adjusted odds of FI by 240 percent in one study of 3,887 women.²⁰² Head circumference of the baby did not show a significant relationship with FI^{205, 692} or fecal urgency.²⁰⁵

Women with FI during pregnancy experienced higher rates of FI after delivery (Table 42).^{209, 662} Women with narrow subpubic arch angle experienced FI almost nine times more often independent of other risk factors.⁶⁹² The number of births associated with FI was examined in 18 studies,^{163, 185, 186, 208, 212, 218, 227, 229, 230, 657, 658, 695, 696} but only five found a significant increase in odds of FI independent on other risk factors.^{209, 213, 215, 226, 232} Women giving birth had FI three times more often compared to childless females.²³² Four studies reported dose response associations between FI and the number of births.^{209, 213, 215, 226} Adjusted odds of persistent FI were 3.2 times greater among women after four or more deliveries.²¹⁵

Table 42

Association between baseline FI, number of pregnancies and FI.

The association between FI and delivery management was examined in several studies (Appendix Table F57). Mediolateral episiotomy after the first pregnancy was associated with lower odds by 83 percent of FI in one study²²⁹ of seven.^{202, 227}–^{229, 695, 697} Prolonged pushing time was associated with 22 percent higher odds of solid FI.²¹⁸ Cesarean surgery was shown to reduce the risk of FI by 42 percent and frequent FI by 64 percent in one study²¹⁴ of 14.^{163, 185, 202, 205, 208, 214, 215, 218, 225, 226, 228, 692, 695, 698} Forceps delivery was associated with increased adjusted odds of FI in one multicenter study²¹⁵ with random results in others.^{202, 214, 215, 218, 226}–^{228, 692, 699}–⁷⁰¹ An inconsistent increase in FI was shown after instrumental delivery in one study.²³⁰ The operative delivery in women with stage II pelvic organ prolapse and/or UI was associated with a 4.5 times greater adjusted odds of bothersome FI.⁶⁷⁹ Vacuum extraction did not show a significant association with postpartum FI.^{214, 215, 218, 227, 694, 701} Vaginal delivery was not shown as a significant risk factor for FI.^{7, 163, 185, 210, 212, 213, 225, 229, 231, 695}

Birth related perianal trauma was associated with a large increase in adjusted rates of FI in eight of 24 studies (Table 43)^{187, 209, 218, 228}–²³² Women with sphincter tears had 230 percent²⁰⁹ to 280 percent^{231, 232} higher rates of FI. The degree of perineal damage showed dose response association with FI.^{229, 230} Women with stage IV perineal damage experienced FI more often compared to damage of grades I–II.^{228, 229} Full thickness of the internal anal sphincter defect was associated with 510 percent higher rates of FI after adjustment for other risk factors.²³⁰

Table 43

Association between FI and birth related anal trauma.

Summary

In conclusion, aging and dependency in daily activities were strongly associated with the development of FI in LTC facilities. Neurological disorders, perineal trauma, and rectal diseases, including neoplasm, are among the independent risk factors of FI in community dwelling adults. Women are at risk of FI related to pregnancy and birth. The strength of the association varied depending on the definitions of FI. Primary prevention of diseases associated with increased risk of FI could reduce the incidence and severity of incontinence in adults. Strategies to reduce the risk of FI related to pregnancy and birth are not well documented and require future research.

Implementation of evidence-based guidelines

The implementation of guidance-driven continence care in nursing homes^{238, 708} did not improve the incontinence status of the residents (Appendix Table F58). The protocol included staff education and direct patient care implemented by advanced practice gerontological nurses in a consecutive cohort of newly admitted residents in three licensed proprietary nursing homes. A quasi-experimental study⁷⁰⁸ evaluated UI using the Incontinence Monitoring Schedule with frequent observations for 3 days before and after the continence protocol was implemented (Appendix Table F58). The implementation resulted in a nonsignificant tendency to reduce the proportion of incontinence episodes and total prevalence of UI from 71 percent in patients after regular care to 64 percent in the intervention group (Appendix Table F59).⁷⁰⁸ A multicenter uncontrolled intervention of a computerized quality management model in nursing homes, based on guidelines for medical assessment and individualized prompted voiding protocols of UI, was associated with a nonsignificant reduction in wet episodes among total checks by nurses from 31 to 17 percent.²³⁸

Conservative management programs

Conservative management programs in nursing homes^{709, 710} resulted in inconsistent improvement in UI in female residents (Appendix Table F60). Behavioral therapy included hourly wet checks, prompted and assisted toileting, positive reinforcement, and bladder training implemented by trained nursing research assistants. Nursing research assistants reported an improvement in UI in females after active intervention three times more often compared to usual care (26 percent vs. 8 percent, RR 3.1, 95 percent CI 1.3; 7.4) (Appendix Table F61)⁷¹⁰ However, the changes in urinary wet episodes were not statistically significant (Appendix Table F62).⁷⁰⁹ Conservative management programs reduced the progression of UI in the majority of RCTs that included male and female residents of nursing homes (Appendix Table F63).²³³–²³⁷ The effect size was small and differed depending on the definition of the outcomes. Integrated incontinence care that included fluid prompting, prompted toileting, and regular wet checks in combination with exercises reduced the frequency of UI by 1.6 percent (95 percent CI -2.5;-0.8) compared to regular care (Appendix Table F64).²³⁵ Prompted voiding treatment increased the proportion of appropriate toileting to 60 percent compared to 17 percent after regular care (mean difference 1.7 percent, 95 percent CI 1.3; 2.2.²³³ Individualized habit training for UI with verbal encouragement administered by nursing staff²³⁷ significantly decreased the frequency of incontinence episodes by 19 percent in elderly residents without neurological diseases but not in memory-impaired residents.²³⁶ Individualized prompted voiding and intensive endurance and pelvic floor muscle training significantly reduced the proportion of wet checks by nurses among total checks to 25 percent compared to 50 percent after usual care.²³⁴ One RCT²³⁹ examined the effects of rehabilitation based on the Functional Independence Measure scale on risk of UI in 34 patients with acute hemispheric stroke (Appendix Table F65). The number cured at discharge was four times larger after active rehabilitation compared to a conventional rehabilitation program (RR 4.1, 95 percent CI 1.5; 11.2) (Appendix Table F66).

One RCT of 32 women examined the effects of oral estrogen (0.63 mg) combined with progesterone (2.5 mg) and regular toileting assistance (prompted voiding) by trained research and did not report significance on rates of positive cough test or appropriate toileting rate.⁷¹¹

Level of evidence

The evidence from nine RCTs²³³–^{237, 708}–⁷¹¹ and one nonrandomized trial²³⁸ was analyzed in relation to a variety of clinical outcomes in residents of nursing homes. One trial reported preplanned intention to treat analysis; four justified sample size;^{234, 237, 709, 710} allocation concealment was unclear in all open label trials. Baseline characteristics of the participants did not differ and confirmed adequacy of randomization in five trials.²³⁴–^{236, 708, 711} Adherence of nursing staff to the protocol was not consistently analyzed, thus preventing valid comparisons of the treatment effects between the studies.

Summary

In conclusion, in contrast to numerous studies of the management of UI, only a few RCTs aimed at slowing progression in LTC settings. Individualized management programs that included prompted voiding, exercise, and positive feedback modestly improved the severity of symptoms. Active rehabilitation with self management of urinary symptoms resulted in continence in the majority of the patients after stroke. Fully powered RCTs would better estimate the effectiveness of individualized evidence-based conservative management programs. Cluster trials with randomization of nursing units, adequate allocation concealment, and preplanned intention to treat analysis are needed to provide valid results. The effectiveness of clinical interventions should be examined in subgroups of residents by cognitive and physical functioning, gender, and ethnicity. The effects of conservative management programs on the quality of life in these populations have not been investigated in RCTs. Staff participation in developing combined management protocols and adherence to protocols should be continuously measured as quality of care indicators in LTC facilities.

Implementation of evidence-based guidelines

(Appendix Table F67). National evidence-based guidelines were implemented in gynecology units in four district general hospitals across Scotland. Women were advised to follow the recommended appropriate protocol for 1 year. This intervention did not improve self-reported severity of UI and quality of life in women at 1 year of followup (Appendix Table F68).⁷¹²

Conservative management programs

Urinary continence service delivered by specially trained nurses included advice on diet and fluids, bladder training, and pelvic floor muscle awareness²⁴⁰ in a large, well designed RCT of 3,746 incontinent men and women living in private households (Appendix Table F69). The proportion of cured patients was 1.5 times higher after the continence service (RR 1.5, 95 percent CI 1.3; 1.7) (Appendix Table F70). More patients reported improvement in UI 6 months after active treatment (RR 1.2, 95 percent CI 1.1; 1.3). The continence service also reduced progression of urgency to very strong or overwhelming by 30 percent (RR 0.7, 95 percent CI 0.7; 0.8). The complex rehabilitation program by nurse continence advisors and consulting urogynecologists including bladder training, gradual increase in fluid intake, pelvic floor muscle training, and transvaginal electrical stimulation resulted in urinary continence in 50 percent of women but with no statistically significant differences compared to usual care (Appendix Table F71).⁵¹⁰ Small significant improvement compared to usual care was detected in two quality of life measures, the Life Urogenital Distress Inventory (2.9 percent) and in the Short Form Urogenital Distress Inventory (16 percent) (Appendix Table F72). Community-based interventions, including education, bladder training, managing the urge to urinate, and pelvic floor muscle training, resulted in continence in 35 percent of women vs. 30 percent after usual care (RR 1.7, 95 percent CI 1; 2.9).²⁴¹ The majority of women reported more than 50 percent improvement in incontinence (RR 1.6, 95 percent CI 1.1; 2.2).²⁴¹ A continence management program implemented by nurse continence advisers with physician expertise included lifestyle modification sessions for 446 incontinent subjects.²⁴² Daily incontinence events were significantly reduced after active treatment with no changes in pad use (Appendix Table F73).²⁴² Educational programs about bladder health and recorded incontinence episodes in a voiding diary, in addition to listening to an audiotape daily, were examined in one RCT.⁷¹³ The combination of education and cognitive interventions significantly reduced the number of incontinent episodes (RR 1.7, 95 percent CI 1.1; 2.7) but did not improve perceived incontinence measured using the Urinary Incontinence and Frequency Comfort Questionnaire (Appendix Table F74).

Dietary and other lifestyle interventions in females

(Appendix Table F75). A standard low calorie liquid diet, increased physical activity to 60 minutes per day, training by a nutritionist, and exercise supervised by a physiologist or behavioral therapist improved stress UI in 92 percent, urge UI in 70 percent, and overall UI in 60 percent of participating women (RR of improvement in overall UI 3.5, 95 percent CI 1.3; 9.1) (Appendix Table F76).⁷¹⁴ Self-selected diet with low fat and low cholesterol foods and 25g of soy protein (Appendix Table F75) did not reduce the risk of stress and urge UI in women compared to a diet without soy (Appendix Table F77).⁷¹⁵ Increasing fluid intake by 500cc did not change the frequency of UI.⁷¹⁶ Restriction in caffeine intake in combination with increasing decaffeinated fluids significantly reduced the number of UI episodes by 87 percent but did not change urine loss in the pad weight test (Appendix Table F77).⁷¹⁷ A significant decrease in daytime episodes of involuntary urine loss after caffeine restriction was reported in another RCT.⁷¹⁸ A positive insignificant tendency to reduce the risk of stress UI was observed after the lifestyle interventions associated with the Diabetes Prevention Program RCT (RR 0.9, 95 percent CI 0.7; 1) (Appendix Table F78).²⁴⁵

In adults, bladder training, fluid intake modification, and restricted caffeine intake to <100mg/day significantly decreased the number of voids by 6 percent, and daily urgency episodes by 21 percent, with random differences in daily leakage episodes compared to bladder training alone (Appendix Table F79).⁷¹⁹

Attributable to conservative management program events

We analyzed the number of avoided/excessive events per 1,000 treated from RCTs that reported significant effects of conservative combined interventions on UI (Table 44). Compared to usual care, continence service resulted in 90 additional cases of resolved UI (continent) 1,000 treated community dwelling adults (95 percent CI 49; 136).²⁴⁰ A specially designed rehabilitation program resulted in 722 additional cases of resolved UI (continent) per 1,000 treated adults with stroke (95 percent CI 121; 235).²³⁹ With continence service, UI was improved in an additional 100 subjects per 1,000 subjects treated (95 percent CI 56; 146), quality of life improved in 90 subjects (95 percent CI 52; 129), and satisfaction with present urinary symptoms in 110 (95 percent CI 66; 157).²⁴⁰ A computerized quality management model for medical assessment and individualized prompted voiding would avoid progression of UI in 140 per 1,000 treated residents of nursing homes (95 percent CI 13; 216).²³⁸

Table 44

Comparative effectiveness of combined conservative management interventions on UI (significant relative benefit of continence shown from individual studies).

Primary prevention of female UI related to pregnancy and birth

was examined in eight large RCTs with more than 100 women²⁴⁶–²⁵³ (two studies were published twice with different duration of followup^{246, 251}) and one smaller trial²⁵⁴ (Appendix Table F80). Conservative advice about self-administered pelvic floor muscle training at 5, 7, and 9 months after delivery supplemented with bladder training did not change the risk of stress UI and tended to decrease the risk of severe UI at 12 months postpartum (Appendix Table F81).^{251, 252} No improvement was seen in women with severe (>1 week) UI at baseline.²⁵² Self-rated severity was significantly reduced by 87 percent (95 percent CI -93.0; -81.1) compared to usual postpartum care (Appendix Table F82). Pelvic floor muscle training that occurred within 48 hours of delivery reduced the risk of self-reported UI 3 months postpartum (RR 0.8, 95 percent CI 0.7; 1.0),²⁴⁶ but the effect was attenuated at 12 months²⁴⁸ (Appendix Table F83). Pelvic floor muscle training that started 1 month before delivery reduced the risk of postpartum stress incontinence (RR 0.6, 95 percent CI 0.4; 0.9).²⁴⁷ Standardized instruction in pelvic floor muscle training started before delivery did not decrease the risk of stress UI compared to regular care in a small RCT without justified sample size.²⁵⁴ In addition to visits and phone calls by community midwives, reminders, feedback to self measure squeeze pressure, and voiding diary, pelvic floor muscle training did not reduce the risk of incontinence 3 months postpartum in an RCT of 1,800 participants.²⁵³ The combination of pelvic floor muscle training with biofeedback and intravaginal electrostimulation resulted in a continence rate of 19 percent vs. 2 percent after usual care (RR of continence 11.0, 95 percent CI 1.5; 82.8) with no significant improvement in urodynamic outcomes (Appendix Table F84).²⁴⁹ Self-administered perineal massage daily from the 34^th or 35^th week of pregnancy until delivery did not improve the continence rate in a large sample of pregnant women with (n = 493) and without (n = 1,034) a previous vaginal birth.²⁵⁰

We estimated that assessment of UI by nurses with conservative advice on pelvic floor muscle training supplemented with bladder training could avoid 121 cases of UI (95 percent CI 7; 152) and 59 cases of severe incontinence (95 percent CI 5; 99) among 1,000 treated (Table 45).²⁵¹

Table 45

Comparative effectiveness of combined conservative management intervention on female UI related to pregnancy and birth (primary prevention).

Pelvic floor muscle training for secondary prevention of UI in community dwelling females

Complex behavioral modification programs, including pelvic floor muscle training, bladder training, and individualized testing of knowledge, adherence, and skills, were implemented in 359 postmenopausal, continent women 55 years and older (Appendix Table F85).^{243, 244} Behavioral modification did not cure UI at 1 year of followup, although the improvement rate in urinary symptoms was 36 percent higher after intervention vs. standard care (RR 1.4, 95 percent CI 1.1; 1.7).²⁴³

The effects of behavioral interventions on UI in females were examined in 21 RCTs.²⁶⁷–²⁸⁷ The effects of pelvic floor muscle training on stress UI in women were examined in 18 RCTs (Appendix Table F86).²⁸⁸–^{290, 292, 299, 316, 720}–⁷³¹ The rates of cure of UI, improvement or progression of UI, continuous severity measures, and urodynamic outcomes after active treatments were compared to usual care or active controls.

Clinical interventions for primary prevention of UI in community dwelling males

Outcome - Continence

The behavioral interventions on UI in males with prostate diseases were examined in 12 RCTs (Appendix Table F90).²⁵⁵–²⁶⁶ Pelvic floor rehabilitation programs resulted in continence rates from 77 percent²⁵⁵ to 99 percent²⁵⁷ (Appendix Table F91). The highest continence rate was reported in a large well designed RCT of early pelvic floor rehabilitation in patients who had radical retropubic prostatectomy for clinical stage T1 or T2 prostate cancer.²⁵⁷ The majority of patients (99 percent) reported continence after the intervention that included verbal explanations, palpation, and Kegel exercises with a small significant relative benefit compared to usual care (RR 1.1, 95 percent CI 1.1; 1.2).²⁵⁷ The relative effect in the same RCT was larger when continence status was measured with a scale specific for UI (RR 1.3, 95 percent CI 1.2; 1.5).²⁵⁷ Continence rates in the control groups were more than 60 percent across all RCTs with no statistically significant differences compared to active treatments (Figure 29).

Figure

Figure 29. Relative benefit of urinary continence after active treatment compared to usual care in males with urological diseases (efficacy RCTs, primary prevention). PFMT pelvic floor muscle training; PFMT + BFB pelvic floor muscle (more...)

Behavioral clinical interventions for secondary prevention of urinary continence in adults

The effects of behavioral interventions in males and females were examined in four RCTs (Appendix Table F93).^{43, 732}–⁷³⁴ One large RCT of pelvic floor muscle training and bladder retraining supervised by non-specialist nurses in 561 adults 35 years and older with regular UI reported a significant increase in cure (RR 15.5, 95 percent CI 2.1; 112.5) and improvement of UI (RR 12.6, 95 percent CI 6; 26.2) (Appendix Table F94).⁴³

Surrogate outcomes

Pooled estimates of behavioral interventions on urinary continence in adults

The differences in populations and treatments, including difference in training regimes and frequency, made a pooled analysis questionable. We compared the relative benefit of urinary continence after behavioral interventions to usual care (Table 46). Pelvic floor muscle training, biofeedback assisted pelvic floor muscle training, and a combination of pelvic floor muscle training with bladder training showed an inconsistent relative benefit across studies. Pooled relative benefits of pelvic floor muscle training in females (RR 7.1, 95 percent CI 2.8; 18)^{280, 288, 289, 299} and pelvic floor muscle training combined with biofeedback (RR 11.2, 95 percent CI 2.2; 56.4)^{288, 289} were sensitive to one small RCT²⁸⁸ with 2 month followup (Figure 30). Pelvic floor muscle training combined with bladder training increased continence by 175 percent compared to usual care (pooled RR 1.8, 95 percent CI 1.1; 2.9).^{241, 243, 280, 290}

Table 46

Relative benefit of urinary continence after behavioral interventions in adults compared to regular care (random effects models, secondary prevention).

Figure

Figure 30. Relative benefit of urinary continence after active treatment compared to usual care in females (efficacy RCTs, secondary prevention). PFMT pelvic floor muscle training; BT bladder training; PFMT+BT pelvic floor muscle (more...)

The majority of RCTs demonstrated significant improvement in UI after pelvic floor muscle training compared to regular care. An improvement was shown eight times more often after pelvic floor muscle training (RR 8.1, 95 percent CI 2.3; 28.4), twice as frequently after biofeedback assisted pelvic floor muscle training (RR 1.9, 95 percent CI 1.0; 3.5), and 2.5 times more often after pelvic floor muscle training and bladder training (RR 2.5, 95 percent CI 1.4; 4.6) (Figure 31).

Figure

Figure 31. Relative benefit of improvement in UI after active treatment compared to usual care (pooled results RCTs, random effects model). PFMT pelvic floor muscle training; PFMT + BFB pelvic floor muscle training with biofeedback; (more...)

Attributable benefits of behavioral interventions

Few RCTs showed significant relative benefit of behavioral interventions on urinary continence (Table 47). The number of avoided stress UI cases per 1,000 varied from 170²⁴⁹ to 749²⁸⁸ when treated with behavioral interventions. The relative benefits on total UI was smaller.^{43, 257} Ignoring different definitions of improvements and possible attenuation of the improvement over time, weight reduction would result in improved stress UI in 990 adults per 1,000 treated (Table 48).⁷¹⁴

Table 47

Comparative effectiveness of behavioral interventions on urinary continence in adults from the community (significant results from individual RCTs).

Table 48

Comparative effectiveness of behavioral interventions to improve UI in adults (significant results from individual RCTs).

Intensive lifestyle changes would avoid 54 cases of stress UI per 1,000 treated (Table 49).²⁴⁵ Pelvic floor muscle training and strategies to improve adherence among pregnant women would avoid 390 cases of UI per 1,000 treated (Table 50).²⁴⁶

Table 49

Comparative effectiveness of behavioral interventions on UI in adults.

Table 50

Comparative effectiveness of behavioral interventions on severity of UI in adults.

Level of evidence

The evidence from RCTs (level I) was analyzed to compare the effects of behavioral interventions on UI.

Summary

In conclusion, behavioral interventions resulted in improvement in UI across RCTs, but the estimation of the overall effect was difficult due to heterogeneity between studies. The long-term continence outcomes among subjects that experienced improvement while participating in RCTs are restricted to 1 year of followup. Several large well-designed RCTs reported a significant benefit of behavioral interventions on cure of incontinence with greater effects on stress UI in females and very limited evidence on urge incontinence and male incontinence. Weight reduction in combination with physical activity showed promising protection against incident UI. Preventive complex pelvic floor rehabilitation programs in specific populations, including males undergoing prostate surgery and pregnant women, demonstrated relative benefit on several measures of urinary continence. The effects on quality of life and objective instrumental outcomes are inconsistent and small. Active interventions provided better outcomes compared to usual care with no relative benefit compared to each other.

Patient outcomes

Surrogate outcomes

Events attributable to stimulation therapy

We estimated that functional magnetic stimulation would avoid 390 cases of urge UI per 1,000 treated women (Table 51). Transvaginal electric stimulation with individually adjusted intensity would result in improvement in UI in 180 women per 1,000 treated.³⁰² Home-managed electrical stimulation with vaginal/anal stimulators would avoid 50 cases of UI per 1,000 treated women (Table 52).³⁰⁴

Table 51

Comparative effectiveness of electrical stimulation or neuromodulation on UI in adults.

Table 52

Comparative effectiveness of electrical stimulation on any UI in adults.

Sacral nerve stimulation would avoid 385³⁰⁶ to 460³⁰⁵ cases of urge UI per 1,000 treated adults. Electrical stimulation with surgical first stage lead placement would improve UI in more than 400 women per 1,000 treated compared to placebo.³⁰³

Summary

In conclusion, evidence from several small RCTs (<100 subjects), three large RCTs, and one prospective cohort suggested that electrical and magnetic stimulation of pelvic floor and sacral nerve roots neuromodulation can improve predominantly urge UI in adults but the curative effects are not consistent.

Patient outcomes

Outcomes

UI

The risk of complete urinary retention was lower after intraurethral collagen submucosal injection compared to surgical procedures, including needle bladder neck suspensions, Burch colposuspensions, and intravaginal slings (RR 0.1, 95 percent CI 0.0; 0.9). However, no RCTs showed a significant relative benefit of bulking agents compared to placebo⁷⁵⁰ or other treatments.^{307, 309, 310, 749} The incidence of urgency was higher after the synthetic bulking agent compared to bovine collagen (RR 2.1, 95 percent CI 1.3; 3.3).⁷⁴⁹ Bulking agents did not result in consistent improvement in urodynamic outcomes, pad weight test results, or quality of life measures (Appendix Table F118). We estimated that 119 patients would avoid complete urinary retention after collagen injection compared to surgical interventions (Table 53). In conclusion, no evidence suggests that bulking agents improve efficacy compared to placebo or effectiveness compared to other treatments in females.

Table 53

Comparative effectiveness of bulking agents on UI in females.

Transurethral radiofrequency energy collagen micro-remodeling was examined in one RCT of 110 women with stress UI and showed no significant improvement in a quality of life questionnaire in women with stress UI and bladder outlet hypermobility compared to placebo injections.⁷⁵¹

Patient outcome - UI

The effects of different obstetric strategies to reduce the risk of UI were examined in seven large RCTs³⁸⁶–^{389, 391, 752, 753} and one smaller RCT³⁸² with justified sample size that evaluated surgical techniques for the primary repair of obstetric anal sphincter lacerations on UI (Appendix Table F119). Planned Cesarean delivery compared to vaginal birth reduced the risk of stress UI by 40 percent (RR 0.6, 95 percent CI 0.4; 0.9) at 3 months but not 2 years postpartum (Appendix Table F120).³⁸⁹ Caesarean section was more protective against UI compared to vaginal delivery 3 months postpartum in the large RCT with no intention to treat and a 4.3 percent rate of cross treatments decided by the attending physician.⁷⁵² Caesarean section reduced the risk of any UI by 70 percent (RR 0.3, 95 percent CI 0.1; 0.7) in primiparous women and in the subgroup of primiparous women with a history of UI (RR 0.3, 95 percent CI 0.1; 0.9).⁷⁵² The risk of stress UI was reduced by 70 percent in primiparous women (RR 0.3, 95 percent CI 0.2; 0.5), with the same reduction in females with and without previous UI. The protective effect was not seen in multiparous women.

Restrictive vs. liberal policies for episiotomy^{386, 753} did not reduce the risk of UI 1.5–3 years postpartum. Mediolateral episiotomy compared to no episiotomy³⁸⁷ did not decrease the risk of stress or urge UI 3 months after delivery. The same proportion of women requiring assisted vaginal delivery reported stress and urge UI after delivery with vacuum extractor or forceps delivery 5 years postpartum.³⁹¹ End-to-end repair compared to overlapping technique did not result in a lower risk of UI in 51 women with complete third or fourth degree anal sphincter laceration who underwent primary repair at the time of vaginal delivery.³⁸² Episiotomy did not affect urodynamic outcomes (Appendix Table F121).^{386, 387}

Patient outcome - UI

Three powered RCTs^{320, 754, 755} and two smaller RCTs with no justification for sample size^{319, 756} examined the effects of hysterectomy on the incidence and progression of UI (Appendix Table F122). Total abdominal hysterectomy was compared to subtotal supracervical hysterectomy in women with benign diseases of the uterus (Appendix Table 123). No differences in total, urge, or stress UI were detected between the two treatments. Intrafascial total abdominal hysterectomy significantly reduced the risk of urge UI compared to the extrafascial approach.³¹⁹

Surrogate outcome - Objective measures of severity of UI

Two RCTs evaluated urodynamic outcomes and self-reported severity of UI and did not show differences between subtotal and total hysterectomy (Appendix Table F124).^{754, 756}

Patient outcomes

Surrogate outcomes

Patient outcomes

Continence

Forty RCTs reported continence rates after different surgical procedures (Table 54).^{319, 345}–^{351, 353, 355}–^{362, 486, 487, 764, 767}–^{770, 772}–^{774, 777, 778 321, 343, 344, 491, 757, 766, 771, 775, 776} The definitions varied from subjective cure with no symptoms or complaints, and need for pad use to objective cure during urodynamic evaluation and stress test. The continence rates demonstrated a substantial variability after the same procedures from 3.1 percent⁷⁷⁰ to 85.1 percent³⁴⁶ after laparoscopic colposuspension and from 8.7 percent⁷⁷⁷ to 93.3 percent⁴⁹⁰ after laparoscopic Burch procedures. Burch retropubic urethropexy resulted in continence in more than 90 percent of women and provided the largest relative benefit compared to modified anterior colporrhaphy (RR 5.1, 95 percent CI 1.8; 14.1)³⁵³ and anterior colporrhaphy with Kelly plication (RR 1.4, 95 percent CI 1.1; 1.8).³⁴⁵ The relative benefit of continence after Burch urethropexy was larger in another RCT that examined the same procedures (RR 2.3, 95 percent CI 1.3; 3.8).⁷⁷⁴ Burch colposuspension with abdominal hysterectomy compared to anterior colporrhaphy with vaginal hysterectomy resulted in higher continence rates (RR 1.6, 95 percent CI 1.1; 2.3) in women undergoing surgery for primary stress incontinence and a concurrent grade 2 or 3 cystocele.³⁶² The relative benefit of the same procedure differed depending on the definition of cure. For example, the same trial that showed a 60 percent relative benefit for urinary continence with the Burch procedure compared to anterior colporrhaphy reported a 300 percent relative benefit on cure defined as a negative cotton swab test (RR 3.2, 95 percent CI 1.6; 6.4).³⁶² The largest relative benefit of Burch retropubic suspension or pubovaginal sling was shown when compared with Oxybutynin 5mg three times/day (RR 41.3, 95percent CI 2.6; 645).⁷⁶⁹

Table 54

Effects of surgical interventions on urinary continence in females (secondary prevention) (sorted by rate of urinary continence after active treatment, from highest to lowest).

Surrogate outcomes

Possible predictors of better effect

The trials used stratified randomization by a surgeon⁷⁶¹ and by participating centers^{343, 771, 779} to estimate valid treatment effects independent of provider differences. Patient age, previous bladder neck surgery,⁷⁷¹ and concomitant rectocele repair for symptomatic rectocele⁷⁷⁹ were balanced between treatment groups with stratified randomization to avoid effect measure modification. One trial reported the incidence of UI after sarcopexy combined with a Burch colposuspension compared to colposacropexy without prophylactic colposuspension in subgroups with baseline maximum urethral closure pressure above and below 35cmH₂O.³²¹ No evidence suggested a significant effect modification by patient baseline conditions.

We analyzed the comparative effectiveness of different surgical procedures on urinary continenc e in females (Table 55). The laparoscopic Burch procedure with sutures did not provide significant benefit compared to staples.^{361, 777} The laparoscopic Burch procedure compared to open Burch surgery resulted in higher continence rates in one RCT, lower in another, and random differences in three RCTs. Sling procedures were compared to Burch colposuspension with significant relative benefit of continence in one RCT and random differences in four RCTs. The Burch procedure compared to the tension-free tape procedure did not result in higher cure rates in four RCTs. An increase in continence was observed after the Burch procedure compared to colporrhaphy in two RCTs. One RCT reported higher continence rates after Burch operation vs. retropubic urethropexy with random differences in two other RCTs. No significant differences in cure were observed in two RCTs that compared retropubic urethrocystopexy and pelvic floor repair. Tension-free tape procedures resulted in a very small relative benefit in continence compared to the sling procedure in five RCTs, while six RCTs reported random differences.

Table 55

Comparative effectiveness of surgical procedures on urinary continence in females (results from RCTs).

We estimated that different surgical procedures may result in a large number of avoided stress and total UI in women with no consistent benefit of one procedure over others (Tables 56–58).

Table 56

Comparative effectiveness of surgical interventions on urinary continence in females (secondary prevention).

Table 57

Comparative effectiveness of surgical interventions on improvement of urinary continence in adults.

Table 58

Comparative effectiveness of surgical interventions on urinary continence in adults.

Summary

In conclusion, a large body of evidence suggests that surgical procedures of stress UI in women result in continence and improvement in UI in more than 75 percent of treated women. Different surgical procedures have comparable effectiveness with no differences in incident urge incontinence. The absolute risk of de novo urgency and urge incontinence after surgery for stress UI requires an RCT, powered to estimate the balance between benefits and harms in continent women with prolapse and patients with stress incontinence. The comparative effectiveness of behavioral interventions and surgery are not known. Individual patient factors that may modify the effects of different procedures also require future research.

Effects of clinical interventions on UI in adults with adenocarcinoma of the rectum

One large RCT of 1,861 patients with histologically confirmed adenocarcinoma of the rectum compared the effects of surgery and adjuvant radiotherapy on UI (Appendix Table F136). The incidence of UI and incontinence that required pad use was the same among patients after total mesorectal excision and adjuvant radiotherapy before surgery (Appendix Table F137).

Effects of clinical interventions on UI in males with urological diseases

Eleven RCTs examined clinical interventions to prevent UI in patients with prostate cancer⁷⁸⁰–⁷⁹⁰ and three RCTs included patients with benign prostate diseases^{489, 791}–⁷⁹³ (Appendix Table F138).

Effects of hormone therapy on UI in community dwelling women

The effects of systemic oral hormone therapy on continence status in postmenopausal women were examined in 19 RCTs (Appendix Table F143).^{352, 365}–^{367, 369, 513, 794}–⁸⁰⁶ The effects of transdermal and intravaginal estrogen therapy to improve continence status in postmenopausal women were administered in 10 RCTs (Appendix Table F144).^{363, 364, 368, 370, 490, 807}–⁸¹¹

Summary

In conclusion, consistent evidence suggests that system hormone therapy increased the risk and progression of UI in postmenopausal females (Table 59). The relative harm varied from 42 to 106 percent for stress UI, 129 to 906 percent for any, and differed substantially for trials that measured urge UI.

Table 59

Comparative effectiveness of hormone therapy compared to placebo on risk of UI in females.

Few RCTs demonstrated relative benefits of local estrogen administration for treatment of stress UI by 64 percent³⁶³ and urge by 55 percent.³⁶⁴

The effective clinical interventions that resulted in long-term stress urinary continence in more than 75 percent of participating females (Table 60) included Burch colposuspension techniques and sling and tension-free tape procedures. Total urinary continence in more than 75 percent of women was observed in several RCTs after tension-free tape and sling procedures (Table 60). The effective clinical interventions that resulted in long-term urinary continence in males with prostatic diseases included pelvic floor muscle training, electrical stimulation with biofeedback, radical retropubic prostatectomy with bladder neck preservation, and implantation of an artificial urethral sphincter (Table 61). Overall, consistent evidence from RCTs suggests that pelvic floor muscle training augmented with biofeedback, Burch colposuspension, and tension-free tape and sling procedures are effective to reduce the absolute risk of stress UI in women (Tables 62 and 63).

Table 60

Clinical interventions that resulted in stress urinary continence in more than 75% of participating females.

Table 61

Clinical interventions that resulted in urinary continence in more than 75% of participating males.

Table 62

Clinical interventions that resulted in urinary continence at 6 or more months of followup (results from RCTs sorted by treatments and rates of continence after active treatments [from highest to lowest]).

Table 63

Surgical interventions that resulted in significant difference in urinary continence at 6 or more months of followup (results from RCTs).

Tension-free tape and sling procedures decreased the absolute risk of total UI. No clinical interventions demonstrated consistent reduction in absolute risk of urge UI.

The largest relative benefits with more than 300 excessive cases of resolved stress incontinence were reported in RCTs per 1,000 females treated with pelvic floor muscle training, Burch colposuspensions, and tension-free vaginal tape compared to usual care or other treatments (Table 64). Few RCTs reported large relative benefits on total urinary continence in females with local estrogen, Burch colposuspension with abdominal hysterectomy, and pubovaginal sling. Few small RCTs reported more than 300 cases of resolved urge incontinence in males and females after electrical or magnetic stimulation.

Table 64

Clinical interventions that resulted in a larger number of attributable events of urinary continence in adults from the community (number of additional continence cases >300 per 1,000 treated).

Effects of pharmacological agents on UI

Three systematic Cochrane reviews reported UI after drug administration. The reviews analyzed randomized trials of anticholinergic drugs on overactive bladder,⁸¹² of adrenergic drugs for urinary incontinence,⁸¹³ and of serotonin and noradrenaline reuptake inhibitors for stress UI in adults.⁶⁴² We estimated attributable events of cure, improvement, or progression from the RCTs published after 1990. Adrenergic drugs were better than placebo in two of five RCTs (Table 65). Continence rates after pelvic floor muscle exercise were comparable with rate after adrenergic drugs. A combination of phenylpropanolamine that was withdrawn from the market when it was linked to intracerebral hemorrhage with estrogen resulted in continence in 0–40 percent of women. No trials included males. Clenbuterol resulted in continence 2 to 4.6 times more often than placebo in two trials but was not more effective compared to pelvic floor muscle training.

Table 65

Effects of adrenergic drugs on self-reported urinary continence in females (results from RCTs).

The Cochrane review of 61 trials of adults with overactive bladder syndrome that compared anticholinergic drugs with placebo treatment or no treatment reported the combined outcome of cure or improvement in UI from eight RCTs (Table 66). The other 53 RCTs did not report patient outcomes. No trials were conducted in men. Administration of propiverine (two RCTs) or extended release tolteradine (two RCTs) for 2–12 weeks resulted in greater rates of cure or improvement compared to placebo. Pooled analysis of six RCTs of different drugs resulted in 139 percent higher rates of cure of improvement (RR 1.4, 95 percent CI 1.3; 1.5).⁸¹² The authors also reported significant reduction in daily leakage episodes (mean difference -0.5, 95 percent CI -0.7; -0.4) after drug administration.

Table 66

Effects of anticholinergic drugs compared to placebo on self-reported urinary continence or improvement in UI in adults with overactive bladder (results from RCTs).

The review of eight RCTs of duloxetine administered for 3–12 weeks in patients with predominantly stress UI concluded that the drug failed to show better curative effects than placebo (Table 67). However, improvement rates and quality of life scores were better after active duloxetine compared to placebo. The long-term effects of pharmacological agents on UI are not clear. The effects on surrogates reported in many trials may not reflect patient outcomes. The comparative effectiveness of drugs compared to other conservative interventions remains unclear.

Table 67

Effects of serotonin and noradrenaline reuptake inhibitor duloxetine on stress UI (results from RCTs).

One multicenter RCT of 695 males 40 years or older with overactive bladder examined the effects of alpha 1 adrenoreceptor blocker tamsulosin for treatment of benign prostatic hyperplasia.⁸¹⁴ Men were treated with 0.4mg of tamsulosin, 4mg of tolterodine ER, both tolterodine ER plus tamsulosin, or placebo for 12 weeks. Urge incontinence rates were the same after tamsulosin (RR 1.1, 95 percent 0.8; 1.6), tolterodine (RR 1.1, 95 percent CI 0.8; 1.7), or the combinations of both drugs (RR 1.1, 95 percent CI 0.8; 1.6) compared to placebo.

Baseline mechanisms of FI

comprise loss of structural and functional integrity of pudendal nerve activity, of pelvic floor muscles, of rectal compliance and of the anal sphincter and loss of rectal sensation.^{17, 20, 23, 24} Disposition of the rectum into the anus (rectal prolapse) or protrusion of the rectum through the vagina (rectocele) also can result in FI. Clinical types of FI include urge incontinence (discharge of feces despite active attempts to retain bowel contents), passive incontinence (involuntary discharge of stool or flatus without awareness), and fecal seepage (leakage of small amount of stool without awareness or staining of undergarments following an otherwise normal evacuation). The aim of effective clinical intervention is to restore the structure and function of these mechanisms, improve the strength and coordination between the pelvic floor and the anal sphincter muscles during voluntary squeeze and following rectal perception, enhance anorectal sensory perception and control, regulate pudendal nerve activity, and provide normal anatomical positioning of the rectum.

Conservative management of FI in nursing homes

Consistent benefits of combined nurse led interventions on the severity of incontinence were reported in three RCTs that examined the effects of conservative management of FI in nursing homes (Appendix Table F151).²³³–²³⁵ The largest improvement (10 percent increase in appropriate toileting) was achieved after prompted voiding treatment, including checks for incontinence, offering toileting assistance, prompted voiding, and social reinforcement of appropriate toileting (Appendix Table F152).²³³ The percentage of incontinent wet checks among total checks by nurses was reduced by 3 percent.²³³ Integrated incontinence care and frequent exercises showed a significant reduction of 6 percent in the proportion of wet episodes.²³⁵ Functional incidental training that included prompted voiding combined with individualized endurance and strength-training exercises reduced the number of wet stool checks among total checks by 1.2 percent.²³⁴

Pharmacological interventions on FI in LTC settings

One RCT examined the effects of laxative agents in elderly patients with FI associated with chronic fecal impaction (Appendix Table F151).⁸¹⁷ The number of FI episodes per patient day did not differ when an osmotic agent with a rectal stimulant and weekly tap-water enemas were compared to single osmotic laxative (Appendix Table F153).⁸¹⁷

Evidence-based conservative management of FI

included the assessment of patient history and rectal examination, patient education on regular toilet habits, pelvic floor and sphincter-strengthening exercises, and individualized doses of laxatives.⁵⁰⁹ Implementation of this program did not reduce the frequency of self-reported fecal episodes (Appendix Table F154) and did not improve quality of life (Appendix Table 155) in stroke patients with constipation and FI (Appendix Tables F156 and F157).

Summary

In conclusion, very few well-designed RCTs examined the clinical interventions on FI in nursing homes and LTC settings. A small improvement in severity of FI was reported after integrated care in nursing homes that included prompted toileting and exercise. The effects of conservative management of FI in subgroups by gender, race, and ethnicity remain unknown. Clinical interventions that can reduce the risk of incidence and progression of FI in LTC settings have not beet tested in RCTs.

Summary

The effects of behavioral changes supervised by health care providers that would provide long-term protection from postnatal FI require further investigation.

Patient outcomes

Surrogate outcome - Objectively measured mean maximum resting pressure

was increased by 5 percent and squeeze pressure by 3 percent compared to the control intervention (Appendix Table F164).³⁷²

Patient outcomes

Surrogate outcome - Subjectively measure severity of FI and quality of life with FI

Hospital-based training and biofeedback services combined with sphincter exercises and electromyelogram biofeedback devices significantly reduced the number of fecal accidents per week by 47 percent.³⁷⁶ The same intervention without the device reduced fecal episodes by 52 percent.³⁷⁶ In a small crossover RCT of 23 patients with idiopathic FI associated with abnormal perineal descent and frequent irritable bowel syndrome, pelvic floor muscle training with active sensory retraining reduced FI episodes by 37 percent compared to exercises with no instructions.³⁷³ One RCT reported a significant reduction in weekly pad use for FI by 78 percent after hospital biofeedback in combination with an electromyelogram device, by 61 percent after hospital-based computer-assisted sphincter pressure biofeedback, and by 89 percent after lifestyle changes and sphincter exercises.³⁷⁶

Behavioral interventions did not improve total FI scores in the majority of the trials.^{375, 376, 502, 818} Only one RCT reported a small significant improvement in St Mark FI scores by 9 percent.³⁷⁵ One RCT showed significant improvement in quality of life after pelvic floor muscle training and biofeedback compared to baseline levels by 4 percent in Pescatori scores of family and relationship and by 16 percent in Pescatori scores of travel restriction.³⁷⁴

Summary

In conclusion, four RCTs of behavioral treatments of FI reported small (<40 percent) improvement in severity and quality of life (Appendix Table F169).³⁷²–³⁷⁵ Limited evidence suggested a significant reduction in FI after complex behavioral interventions, including lifestyle changes and exercises augmented with biofeedback (Appendix Table F170).³⁷⁶

Patient outcomes

Surrogate outcomes

Summary

In conclusion, electrical stimulation did not improve FI in the majority of the RCTs. The significant relative improvement after sacral nerve stimulation in patients with severe baseline FI requires future confirmation in a large well designed RCT with long-term followup.

Botulinum toxin

Loperamide

Two RCTs examined the effects of loperamide in patients after restorative proctocolectomy for ulcerative colitis³⁷⁸ and in patients with chronic diarrhea and FI.⁸²³

Patient outcomes - Continence

All patients were continent after rectal administration of isosorbide-5-mononitrate⁴⁰⁴ and after glycerine trinitrate ointment⁴⁰⁵ (Appendix Table F177). Healing of anal fissures was observed four times more often after isosorbide-5-mononitrate compared to placebo (RR 4, 95 percent CI 1; 13).⁴⁰⁴ We estimated that rectal administration of isosorbide-5-mononitrate would result in 580 healed patients among 1,000 treated.

Surrogate outcome - Subjective and objective measures of severity of FI

Chemical sphincterotomy with oral nifedipine⁴⁰³ or rectal administration of isosorbide-5-mononitrate⁴⁰⁴ did not result in consistent improvement in the Wexner scores and anal manometry outcomes (Appendix Table F176).

Patient outcome - Improvement in FI

Topical 10 percent phenylephrine, alpha-adrenergic agonist increasing anal sphincter pressure, applied to the anus improved FI symptoms three times more often than placebo in patients with dysfunctional anal sphincter (RR 3, 95 percent CI 1; 13).⁸²⁴

Surrogate outcome - Subjective measures of severity of FI

Anal administration of phenylephrine significantly improved FI scores and FI symptom scores⁵⁰³ but not the Wexner scale scores⁸²⁴ (Appendix Table F176).

Patient outcome - FI

Oral sucralfate did not prevent FI in prostate cancer patients treated with radiotherapy at a 3 month followup.³⁷⁹

Patient outcome - Improvement in FI

Medical bowel confinement and regular diet did not improve FI after anorectal reconstructive surgery at 13 months of followup.⁸²⁵

Summary

Synthesizing the evidence from all available clinical trials, a few interventions³⁷⁸–³⁸⁰ resulted in fecal continence in more than 50 percent of patients but were not statistically better than placebo^{378, 379} or surgery³⁸⁰ (Appendix Tables F158 and F159s). Chemical sphincterotomy with isosorbide-5-mononitrate,⁴⁰⁴ glycerine trinitrate, and Botox resulted in 100 percent continence rate.⁴⁰⁵ Administration of phenylephrine to improve sphincter function in patients with FI after ileoanal pouch construction demonstrated significant efficacy compared to placebo.⁸²⁴ Reproducibility of the significant effects of chemical sphincterotomy using different agents should be confirmed in the large RCTs.

Patient outcome - FI

Defects of internal or external sphincter did not differ after two techniques (Appendix Table F181).³⁸³ The rates of self-reported fecal and flatus incontinence were the same after both interventions.³⁸² Only one RCT reported a significant reduction in fecal urgency at 6 and 12 months postpartum after end-to-end technique compared to overlapping repair.³⁸¹ We estimated that 250 per 1,000 treated women would not experience fecal urgency at 6 months and 283 at 12 months postpartum.

Surrogate outcome - Subjective and objective measures of FI

Self-reported symptoms of FI and anal manometry were comparable after overlap or end-to-end anal sphincter repair (Appendix Tables F182 and F183).⁸²⁸ FI scores and instrumental outcomes did not differ after two interventions (Appendix Table F184).^{383, 384}

Patient outcome - FI

Self-reported FI 18 months postpartum was the same after restrictive or liberal episiotomy.³⁸⁶ The same proportion of women after mediolateral episiotomy or vaginal birth experienced FI (Appendix Table F185).³⁸⁷

Cesarean vs. vaginal delivery resulted in the same rates of fecal and flatus incontinence at 3 months and 2 years postpartum.^{388, 389} However, only 43 percent of women randomized to vaginal delivery underwent Cesarean section.

Delayed or immediate pushing in the second stage of labor with epidural analgesia demonstrated the same rates of FI and sphincter defects in ultrasound at 3 months postpartum.³⁹⁰

Delivery with a vacuum extractor tended to increase the risk of anal (RR 3.5, 95 percent CI 1.0; 12.2) ³⁹³ or FI (RR 1.8) ³⁹¹ compared to forceps delivery with borderline significance of the differences (Appendix Table F185).

Surrogate outcome - Subjective and objective measures of FI

Self-reported FI scores and instrumental outcomes did not differ after all tested treatments.(Appendix Table F186)

Summary

No evidence suggested that Cesarean section or instrumental delivery can prevent FI postpartum compared to vaginal delivery.

Patient outcomes - FI

Self-reported FI and fecal urgency at 3 months and 1 year of followup did not differ among women who underwent ultrasound examinations compared to clinical examinations alone. Severe FI was reduced by 60 percent after ultrasound examination at 3 months (RR 0.4, 95 percent CI 0.2; 0.7) with no differences at 1 year of followup.

In conclusion, effective clinical interventions to reduce the risk of FI related to obstetric trauma are not well known and require future investigation.

Patient outcome - Continence

Both treatments resulted in continence rate of 40 percent (Appendix Table F188). Marlex mesh posterior rectopexy alone cured 56 percent and sigmoidectomy combined with a sutured posterior rectopexy cured 46 percent of 29 women with full-thickness rectal prolapse with no significant differences between treatments.⁴¹⁴

Patient outcome - Continence

Gluteus maximus transposition failed to cure 33 percent and total pelvic floor postanal repair with anterior levatoroplasty failed to cure 42 percent of women with post obstetric FI (Appendix Table F189).⁴²⁸ Total pelvic floor repair cured 75 percent, postanal repair cured 33 percent, and anterior levatoroplasty and sphincter plication cured 33 percent of 36 women 28–75 years with neuropathic FI with no statistically significant differences between groups.⁸²⁹ Adjuvant biofeedback following anal sphincter repair and anterior overlapping sphincter repair cured <6 percent of 38 females with persistent FI and external anal sphincter defect with no statistically significant differences between groups.⁸³⁰

Surrogate outcome - Subjective and objective measures of severity of FI

Self-reported FI scores, patient satisfaction, and quality of life were comparable after the tested interventions (Appendix Table F190).^{385, 428} Surgeries resulted in random differences in instrumental outcomes inconsistent in the direction and effect size (Appendix Table F191). Small changes in maximum resting pressure were observed after adjuvant internal anal sphincter plication in pelvic floor repair compared to pelvic floor repair alone in 33 women with severe neuropathic FI related to vaginal delivery.⁸³¹

Patient outcome - FI

FI did not differ at 12 months after total abdominal or subtotal abdominal hysterectomy in a well-designed RCT of 279 women with benign uterine diseases.⁷⁵⁴

Summary

Overall, limited evidence from small RCTs suggested that surgical procedures result in comparable risk of FI and quality of life. The effect size varied depending on the definition of FI and baseline cause and severity of FI. The effectiveness of the interventions in the subgroups of different age and severity categories require future investigation.

Patient outcomes

Surrogate outcome - Subjective measure of severity of FI

Self-reported stool leakage tended to be greater after radiotherapy compared to active surveillance (Appendix Table F193).⁷⁹⁰

Surrogate outcomes - Subjective measures of severity of FI

Transanal irrigation showed a small positive effect compared to conservative bowel management in 87 patients after spinal cord injury and symptoms of neurogenic bowel dysfunction (Appendix Table F194).⁸³² Neurogenic bowel dysfunction scores were reduced by 3 percent, St Mark's FI scores by 7 percent, and Cleveland Clinic constipation scores by 6 percent. The other incontinence outcomes, including the American Society of Colon and Rectal Surgeons FI scores, influence of FI on daily activities, and the number of episodes of clothes change, however, showed no differences (Appendix Table F194).

Patient outcomes

Surrogate outcome - Objective measure of severity of FI

Anorectal outcomes demonstrated inconsistent changes across four RCTs,^{397, 400}–⁴⁰² with significant differences in one.³⁹⁷ (Appendix Table F197).

Patient outcome - FI

Incision, drainage, and fistulectomy with primary partial internal sphincterectomy compared to incision and drainage alone did not reduce the risk of flatus incontinence and soiling in patients with anorectal abscesses (Appendix Table F198).⁴¹⁵ Special fibrin sealant combined with intra-adhesive cefoxitin, surgical closure of primary opening, or both modifications resulted in fecal continence in all patients with chronic anal fistula.⁴¹⁶ Adjuvant radiotherapy before total excision in patients with rectal cancer consistently increased the risk of FI in three RCTs, by 60 percent during the day (RR 1.6, 95 percent CI 1.4; 1.9), by 90 percent during the night (RR 1.9, 95 percent CI 1.4; 2.6), by 70 percent in utilization of pads (RR 1.7, 95 percent CI 1.4; 2.1) compared to mesorectal excision,⁴²⁴ and by 220 percent (RR 2.2, 95 percent CI 1.2; 4.2) compared to the low anterior resection.⁴²⁴ The harm of adjuvant radiotherapy reported in an RCT of 171 patients who survived for a minimum of 5 years after treatment with an increased relative risk of incontinence of liquid stool (RR 2.1, 95 percent CI 1.3; 3.2), incontinence of solid stool (RR 4.1, 95 percent CI 1.2; 14.2), and utilization of pads (RR 2.2, 95 percent CI 1.4; 3.5) ⁴²⁵ was larger. We estimated that surgery without adjuvant radiation would avoid 240 cases of daytime FI per 1,000 treated,⁴²⁴ 310 cases of self-reported FI,⁴²⁴ 260 cases of loose stool, and 110 cases of loose solids.⁴²⁵ Adjuvant radiation also restricted social activities almost three time more often compared to surgery (RR 2.9, 95 percent CI 1.4; 5.8).⁴²⁵

Surrogate outcome - Subjective measures of severity of FI

Postoperative radiotherapy, in addition to intensive daily pelvic floor muscle training with intra-anal biofeedback, improved the modified Cleveland Incontinence Scores at 1 year by 5.6 percent (mean difference -0.6, 95 percent CI -1.0; -0.2) compared to behavioral treatments without radiation (Appendix Table F199).⁴²⁶

The clinical effects of glyceril trinitrate application were compared to lateral sphincterotomy with no significant difference in FI among patients with chronic anal fissure.⁴¹⁹–⁴²³ Quality of life was the same after both treatments.⁴²²

The trial that compared dermal island flap anoplasty with cutaneous advancement flap into the rectum to conventional open fistulotomy or seton insertion in 20 patients with fistula-in-ano reported comparable FI scores (Appendix Table F199).⁴¹⁸ Another RCT did not show difference in FI Wexner scores after internal sphincterotomy compared to the application of 0.25 percent nitroglycerin tid to perianal area in 90 patients with symptomatic chronic anal fissures (Appendix Table F199).⁴²²

A trial that compared dermal island flap anoplasty with cutaneous advancement flap into the rectum with conventional open fistulotomy or section in 20 patients with fistula-in-ano reported comparable FI scores.⁴¹⁸ Another RCT did not show differences in FI. Wexner scores after internal sphincterotomy compared the application of 0.25 percent nitroglycerin tid to the perineal area in 90 patients with symptomatic chronic anal fissures⁴²² (Appendix Table F199).

Patient outcome - FI

The risk of complete FI was reduced by 60 percent (RR 0.4, 95 percent CI 0.2; 0.7) after posterior sutured abdominal rectopexy combined with sigmoidectomy and end-to-end stapled anastomosis compared to polyglycolic acid mesh rectopexy without sigmoidectomy (Appendix Table F200).⁴⁰⁶

Surrogate outcome - Subjective and objective measures of severity of FI

Surgical procedures did not reduce the risk of FI, self-reported severity scores, or instrumental outcomes (Appendix Table F201).

Patient outcomes

FI

Implantation of artificial bowel sphincter reduced urge FI by 80 percent (RR 0.2, 95 percent CI 0.2; 0.3), pad utilization by 40 percent (RR 0.6, 95 percent CI 0.5; 0.7), and restrictions in physical activities by 60 percent (RR 0.4, 95 percent CI 0.3; 0.5) compared to baseline levels (Table 68 and Appendix Table F202).⁴²⁷

Table 68

Comparative effectiveness of surgical interventions in anal sphincter on FI in adults (events).

Progression of FI was reduced by 80 percent (RR 0.2, 95 percent CI 0.1; 0.9) after closed compared to open lateral sphincterotomy.⁸³³ Improvement in FI was increased by 40–50 percent after anterior levatorplasty with mobilization of external sphincter compared to anal (and vaginal in women) plug electrostimulation of the pelvic floor⁸³⁷ in patients with idiopathic severe refractory FI. Physical restrictions were reduced by 50 percent (RR 0.5, 95 percent CI 0.2;0.9).

Surrogate outcome - Subjective and objective measures of severity of FI

Artificial bowel sphincter reduced FI scores (0–120 for complete FI) from 106 to 48 (mean difference -1.7, 95 percent CI -1.4;-2.0) or by 2 percent compared to the baseline condition.⁴²⁷ Artificial bowel sphincters reduced the Cleveland Clinic Scoring and the American Medical Systems quality of life questionnaire specific to FI compared to usual supportive care.⁸³⁹(Appendix Table F203).

Guidance by endoanal ultrasound of injections of silicon biomaterial into inter-sphincteric space and internal anal sphincter improved the Wexner and visual analog scores and FI Quality of Life Index compared to injections without a guide (Appendix Table F204).⁸³⁸

All procedures resulted in small inconsistent changes in anal manometry outcomes (Appendix Table F204).

Surrogate outcome - Dynamic graciloplasty

One multicenter noncontrolled nonramdomized clinical trial (Dynamic Graciloplasty Therapy Study Group) reported significant improvements quality of life (Medical Outcomes Study Short Form 36 physical function and social functioning). The majority of nonstoma patients (56 to 62 percent) at 12–24 months had continent stools/total number of stools ≥50 percent. In the stoma patients 37.5 percent at 12 months to 43 percent at 24 months reported having continent stools.^{429, 430}

Summary

In conclusion, implanting an artificial bowel sphincter demonstrated some consistent effects to reduce the risk of FI. Adjuvant radiation in patients with prostate cancer resulted in a consistent significant increase in FI compared to surgery alone. The protective effects on FI of surgeries for hemorrhoids, rectal prolapse, rectal cancer, and anal fissures are not replicable across RCTs and require future confirmation in well-designed long-term RCTs.

Patient outcome - Continence

The majority of patients with cancer were continent after standardized resection with J pouch anastomosis or straight coloanal anastomosis across RCTs (Table 69). Rates of FI did not differ between treatment groups. Abdominal colectomy with hand sewing or double stapling the pouch to the anal canal provided the highest rates of continence in patients with ulcerative colitis with no differences between the two surgeries.⁴³¹ All patients were continent after duplicated or quadruplicated ileal reservoirs in restorative proctocolectomy with no differences between them.^{432, 433} Abdominal rectopexy with or without sigmoidectomy resulted in the same continence rates in patients with rectal prolapse.⁸⁵⁴

Table 69

Surgical interventions (diversion) on FI in adults (events).

Fecal continence was reported in 75 percent of 1,885 patients operated for chronic ulcerative colitis at 1 year and gradually decreased to 59 percent after 20 years of followup (Appendix Table F205).⁸⁵³ The proportion of patients continent after surgery increased during the last decades from 57 percent in 1981-1985 to 70 percent from 1996-2000 (Appendix Table F206).⁸⁵³

Surrogate outcome - Subjective and objective measures of severity of FI

Severity of FI differed between types of surgical anastomosis in patients with cancer (Appendix Table F207). Low anterior resection with straight anastomosis increased FI scores by 61 percent compared to J pouch anastomosis (mean difference in a scale from 0 to18 for complete FI, 1.2, 95 percent CI 0.8; 1.7).⁸⁴⁴ In contrast, another RCT showed a significant increase in the Wexner FI scores after J anastomosis (mean difference 1.3, 95 percent CI 0.5; 2.0),⁸⁴¹ The Fecal Incontinence Severity Index decreased after J anastomosis (mean difference -4.6, 95 percent CI -5.6; -3.5) and the FI quality of life scale improved (mean difference 0.7, 95 percent CI 0.1; 1.3).⁸⁵⁶ Severity of FI did not differ after the tested interventions in patients with ulcerative colitis or rectal prolapse.

Anal manometry outcomes demonstrated random changes after surgical interventions with inconsistent direction and effect size. Smaller RCTs reported greater differences (Appendix Table F208).

Summary

In conclusion, different types of anastomosis (J-pouch-anal or straight) combined with resection of cancer resulted in comparable fecal continence rates across the RCTs. Surgical interventions in patients with ulcerative colitis were comparable between active treatments for fecal continence in the majority of the patients during the first 10 years after surgery with an increased risk of FI during longer period of followup. The effects of treatments on quality of life were inconsistent in direction and effect size.

Clinical interventions resulted in comparable incidence and progression of FI in the majority of the RCTs. Significant relative benefits on fecal continence were reported in small trials of <100 subjects (Table 70). Definite harm of adjuvant radiotherapy compared to surgery alone was shown in RCTs of 597 patients⁴²⁴ and 171 patients⁴²⁵ with rectal cancer. Short-term benefits of pelvic floor muscle training supplemented with bladder training and supervised by nurses was reported in RCTs of 747 pregnant women.²⁵¹ Early detection of anal sphincter tears after vaginal delivery, followed by immediate surgical repair, significantly reduced the risk of FI 3 months and 1 year postpartum in RCTs of 752 pregnant women.³⁸⁵ The significant results of small RCTs need future confirmation in well-designed long-term trials.

Table 70

Comparative effectiveness of clinical interventions on FI in community dwelling adults (significant differences only shown).

Valid questionnaires to assess UI (Appendix Table F209)

The Symptom and Quality of Life Committee of the International Consultation on Incontinence (ICI) systematically reviewed available questionnaires and scales to diagnose UI and evaluate quality of life in patients with UI.⁸⁶⁶ The committee highly recommended the questionnaires with rigorous validity, reliability, and responsiveness in several clinical studies graded with the highest level of evidence (level A) (Appendix Table F210).⁸⁶⁶ The ICI Questionnaire combined symptoms of UI and impact on life for both genders.⁸⁶⁸ The Bristol Female Lower Urinary Tract Symptoms Questionnaire⁸⁶⁹ and the Stress and Urge Incontinence and Quality of Life Questionnaire (SUIQQ)⁸⁷⁰ are recommended to estimate the severity and impact on life of stress and urge UI in women (Appendix Table F211).⁸⁶⁶ The ICS male questionnaire measures voiding patterns, stress, urge UI, and impact on quality of life related to UI.⁸⁷¹ The overactive bladder questionnaire demonstrated validity and reliability to measure symptoms of overactive bladder in males and females.⁸⁷² Most of the questionnaires are gender specific and include UI and other symptoms of lower urinary tract problems in females (Urogenital Distress Inventory;⁸⁷³ Urogenital Distress Inventory-6;⁸⁷⁴ Incontinence Severity Index;⁸⁷⁵ Bristol Female Lower Urinary Tract Symptoms Questionnaire⁸⁷⁶ and in males ICS Male, Lower Urinary Tract Symptoms⁸⁷⁷ and Danish Prostatic Symptom Score, Lower Urinary Tract Symptoms⁸⁷⁸). The questionnaires on Quality of Life in Persons with UI⁸⁷⁹ and the Incontinence Classification System⁸⁸⁰ estimate perceived UI, social interactions, personal strain, impact on sexual life, and overall health in both genders. Several scales are recognized as valid tools to measure quality of life in incontinent women: King's Health Questionnaire;⁸⁸¹ Incontinence Impact Questionnaire;⁸⁸² Incontinence Impact Questionnaire-7; UI Severity Score (UISS);^{883, 884} and CONTILIFE-Urinary Incontinence-Specific Measure of Quality of Life.⁸⁸⁵ The committee did not find good evidence to recommend scales of quality of life in males. Standard recommendations developed by the Symptom and Quality of Life Committee of the ICI did not include sensitivity, specificity, and positive predictive likelihood of different questionnaires compared to objective physician diagnosis or instrumental methods. The assessment of validity based on significant correlation coefficients can be biased by the sample size of the studies and the distribution of the responses.

Diagnostic value of questionnaires and scales for UI

Several studies examined the diagnostic value of questionnaires and validated scales compared to multichannel video urodynamic testing,^{452, 480, 506, 886}–⁸⁸⁹ physician diagnosis based on medical history, urine analysis, and micturition charts and diaries and urodynamics,^{447, 890}–⁸⁹² detailed interviews with nurses,⁸⁹³ or other objective tests including the pad test,⁸⁹⁴ provocative cough stress test, and computerized urethra-cystometrogram (Table 71).⁸⁹⁵ One study tested a simple screening tool to detect common geriatric problems in community dwelling adults including UI.⁴⁷⁹ The scales included questions on the presence and severity of symptoms of UI and quality of life. The majority of studies examined females only; three studies included both males and females.^{479, 480, 892} An epidemiologic survey could identify only 56 percent of women who had urge UI and 66 percent of women who had stress UI.⁴⁴⁷ Survey questions for geriatric UI could identify only 60 percent of incontinent adults.⁴⁸⁰ The highest sensitivity of 98 percent was reported in a nonrandomized multicenter study of 531 patients with overactive bladder and urge UI detected with a simple, five-item questionnaire.⁸⁹²

Table 71

Diagnostic value of validated questionnaires and scales to detect persons at risk and patients with UI.

The sample size of the studies was not associated with the sensitivity of the scales to detect UI (Appendix Table F212). The specificity of the scales varied from 21 percent⁴⁸⁰ to 100 percent.⁸⁹⁵ The lowest specificity was demonstrated for a survey in elderly patients to detect urge UI.⁴⁸⁰ The Medical, Epidemiologic, and Social Aspects of Aging Urinary Incontinence Questionnaire did not provide any false positive results of stress UI.⁸⁹⁵ Such large variability in specificity of the scales could not be explained by differences in sample size (Appendix Table F212). The scales had comparable specificity for different types of UI; 95 percent for mixed UI,⁴⁷⁹ 96 percent for stress UI,⁴⁸⁰ and 96 percent for urge UI.⁴⁴⁷

Three scales suggested a greater likelihood of UI compared to urodynamic evaluations^{480, 895} and physician diagnosis^{447, 479} (Appendix Table F213). The same scale resulted in larger positive likelihood ratios for urge (positive LR 14) and lower for stress UI (positive LR 5.5).⁴⁴⁷ The majority of scales resulted in only a small (positive LR <4) probability of UI diagnosed with multichannel urodynamics,^{452, 480, 506, 887}–⁸⁸⁹ pad test,⁸⁹⁴ or physician diagnosis.^{890, 891} The moderate diagnostic value of the Urogenital Distress Inventory was reported in a systematic review with a pooled likelihood ratio of 2.2, 95 percent CI 1.5; 2.9.^{452, 888} Larger studies tended to report lower likelihood ratios, but the sample size could explain only small proportions of differences in ratios (Appendix Table F212). Across the studies, the consistent number needed to screen to detect one case of UI was two to three patients (Table 71).

Several studies estimated the validity of the questionnaires with correlation coefficients. One study of 384 older women reported a significant correlation (correlation coefficient ratio 0.2–0.7) between the Incontinence Impact Questionnaire and the Urogenital Distress Inventory with the clinical diagnosis of UI.⁸⁹⁶ The Incontinence Severity Index was correlated with the 48-hour “pad weighing” test (correlation coefficient 0.6-0.6).^{875, 897}

A large RCT reported a significant correlation between the Incontinence Quality of Life Questionnaire, Patient Global Impression of Improvement and Severity Questionnaire, and stress pad test results.⁵⁰⁷ Secondary analyses of three clinical drug trials reported a significant correlation between the Urgency Perception Scale and patient voiding diary variables and other patient assessments, including perception of bladder condition, perception of treatment benefit, the Medical Outcomes Study Short-Form 36 health survey, the King's Health Questionnaire, the OAB Questionnaire, and the Overall Treatment Effect scale.⁸⁹⁸ The significance of correlation coefficients depends on the sample size of the studies. For example, a large RCT reported a correlation coefficient ratio of 0.2–0.4 as significant.⁵⁰⁷

In conclusion, two questionnaires demonstrated good diagnostic value compared to multichannel urodynamics to detect UI in epidemiologic surveys.^{447, 895} Two screening questionnaires may accurately detect UI in older adults.^{479, 480} Several questionnaires of severity and quality of life of UI that demonstrated rigorous validity and reliability were recommended for clinical practice and research. The clinical and economical value of self-reported perceived UI needs future research.

Diagnostic value of clinical history compared to multichannel urodynamic for stress UI in community dwelling adults

The diagnostic value of clinical history to detect stress UI compared to a “gold standard” multichannel urodynamics was examined in 24 studies with inconsistent results.⁴³⁵–^{455, 899} Most of the studies included females; one evaluated stress UI in men after radical prostatectomy⁸⁹⁹ (Table 72). The authors used ICS criteria to define urodynamic UI, noted during filling cystometry as the involuntary leakage of urine during increased abdominal pressure, in the absence of a detrusor contraction.⁴⁸³ Clinical history assessments varied but were poorly reported. The common assessment included duration and severity of involuntary urine leakage, exacerbating factors, baseline diseases, self management, and impact on quality of life.

Table 72

Diagnostic value of clinical history compared to multichannel urodynamics - “gold standard” to detect persons at risk and patients with UI.

The sensitivity of the clinical history compared to multichannel urodynamics varied from 44 percent in the largest study⁴³⁷ to 100 percent in several studies in females^{435, 443, 446} and males.⁸⁹⁹ The larger study tended to report lower sensitivity; 20 percent of the variation in sensitivity was attributable to sample size (Appendix Figure F2).

Specificity also varied substantially from 13 percent in a study with elderly women⁴⁵⁵ to 95 percent in the smallest study of 46 patients with complaints of UI or voiding disorders.⁴³⁵ The largest study of 1,455 women reported specificity of 87 percent.⁴³⁷ Clinical history resulted in 50 percent false positive results in males.⁸⁹⁹ The variability in specificity could not be explained by the sample size of the studies; only 3 percent was attributable to the number of subjects (Appendix Table F212). Positive predictive values as the proportion of true positives among all positives also varied from 66 percent in a study of 66 females⁴⁴⁵ to 92 percent in a study of 566 patients.⁴⁵³ Larger studies tended to report greater positive predictive values of clinical history (Appendix Figure F3); 22 percent of the differences in positive predictive values were attributable to the sample size of the study. Only one study of 46 women showed that clinical history was associated with a definite increase in the likelihood of disease (positive likelihood ratio of 20).⁴³⁵ One study of 198 Japanese women reported a moderate increase in the likelihood of stress UI identified with clinical history compared to multichannel urodynamics (positive likelihood ratio of 7.7).⁴³⁶ The majority of the studies found only small increases (positive likelihood ratios >2) in the likelihood of stress UI identified with clinical history⁴⁴⁰–^{447, 899} or minimal (positive likelihood ratios >1).^{442, 448}–⁴⁵⁵ The differences in sample size could not explain the variability in positive likelihood ratios (Appendix Table F212). One systematic review reported a small positive likelihood ratio (2.09, 95 percent CI 1.83; 2.35) associated with the pooled sensitivity and specificity in females using studies published before 2003.⁸⁶⁷ The review reported highly significant heterogeneity between studies (p<0.0001) in sensitivity and specificity with no clear explanations for substantial variability in diagnostic values of clinical history to detect stress UI.