Secondary Logo

Journal Logo

Original Research

Physical Activity and Urinary Incontinence Among Healthy, Older Women

Danforth, Kim N. ScD; Shah, Aparna Diwan MD; Townsend, Mary K.; Lifford, Karen L. MD; Curhan, Gary C. MD, ScD; Resnick, Neil M. MD; Grodstein, Francine ScD

Author Information
doi: 10.1097/01.AOG.0000255973.92450.24
  • Free

Several national organizations recommend that adults engage in at least 30 minutes of moderate-intensity physical activity 5 or more days per week.1,2 In 2003, less than half of adult women complied with this recommendation; in particular, among women aged 65 years and older, only 32% reported following these guidelines.3

Urinary incontinence (UI) sometimes occurs subsequent to strenuous, high-intensity physical activity, and concern about incontinence has been cited as one reason women do not exercise.4–7 However, very little is known about the association between low- or moderate-intensity physical activity and development of incontinence. Indeed, because regular physical activity may strengthen pelvic floor musculature, it is possible that moderate physical activity could decrease the risk of developing UI, especially stress UI.

We used data from a large prospective study of older women to examine the association between low- and moderate-intensity physical activity and the incidence of urinary incontinence.


The Nurses' Health Study began in 1976 when 121,701 female registered nurses in 14 U.S. states completed a self-administered, mailed questionnaire. At enrollment, study participants were 30–55 years of age. Follow-up questionnaires are mailed biennially to update information on lifestyle factors and health status. Information on urinary incontinence was requested in 2000 and again in 2002; in 2002, we also requested detailed information about the type of incontinence using a supplementary questionnaire that was mailed to women with incident incontinence. Through 2002, we have maintained 90% follow-up of participants. The study was approved by the Institutional Review Board of Brigham and Women's Hospital.

For these analyses of incident incontinence, we excluded women who were missing information on incontinence (n=31,355; incontinence questions were only included on the first of five questionnaire mailing cycles) or who reported prevalent incontinence in 2000 (n=51,401), as well as women missing information on physical activity (n=323). Women in the study population were similar to the entire cohort in key risk factors for incontinence, including mean age (66 versus 67 years, respectively), body mass index (BMI) (49% versus 46%, respectively, with BMI of 25 or more), and cigarette smoking status (11% versus 10% current smokers, respectively).

Because we were concerned about the possibility of major confounding by health status in these analyses, we made several additional exclusions to help minimize such confounding. We excluded 552 women who reported a history of major neurologic diseases (eg, stroke), because they may be particularly strong confounding variables. Additionally, we excluded 1,802 women who reported functional limitations (defined as significant limitations in climbing one flight of stairs, walking one block, bathing, or dressing), other difficulties walking, or ever having lived in a nursing home because these circumstances are highly associated with UI and would also severely limit activity.

In 2000, study participants were asked this question: “During the last 12 months, how often have you leaked or lost control of your urine?” Response categories were the following: never, less than once a month, once a month, 2–3 times a month, about once a week, and almost every day. Women who reported that they lost urine were then asked “When you lose your urine, how much usually leaks?” The response categories were these: a few drops, enough to wet underwear, enough to wet outer clothing, and enough to wet the floor. Both questions were repeated on the 2002 questionnaire.

Women who reported never leaking or leaking a few drops less than once a month in 2000 were considered at risk of incident incontinence. Among these women, cases of incident frequent incontinence were defined as those reporting incontinence at least weekly in 2002. Self-reported frequency and quantity of UI have been shown to be highly reproducible in this population,8 and in other studies, self-reported UI has been shown to be highly valid compared with clinical diagnosis.9

Supplementary questionnaires were mailed to collect information on type of incontinence. Because of the large number of cases, we chose not to collect supplementary data on UI type from a random sample of approximately one fourth of the cases. Thus, information on UI type was requested from 1,913 women and was returned by 84%. The supplementary questionnaire asked about the circumstances of urine loss based on validated surveys for assessing incontinence type.10–12 Symptoms of stress UI were those associated with increased intra-abdominal pressure (ie, coughing, sneezing, laughing, lifting, brisk walking, or exercise), whereas symptoms of urge UI were those related to a sudden need to urinate. The dominant symptom was used to classify the type of incontinence, but if symptoms of stress and urge urinary incontinence were reported equally, type was classified as mixed. Women who did not receive or return the supplementary questionnaire or who reported symptoms other than stress, urge, or mixed incontinence were excluded from analyses of incontinence type. Cases that were excluded from the incontinence type analyses (n=789) were similar to those included (n=1,566) in mean age (68 versus 66 years, respectively), body mass index (56% versus 58% with BMI of 25 or more, respectively), and cigarette smoking (12% versus 10% current smokers, respectively). Thus, there is little possibility for any meaningful bias due to the incomplete information. In all analyses, noncases were defined as women who reported never leaking urine or leaking a few drops less than once a month again in 2002.

Questions on physical activity were first included in 1986 when participants were asked to report the average amount of time they spent per week on leisure-time activities, using 10 categories ranging from zero minutes to 11 or more hours per week. Participants were asked about the following activities: walking or hiking outdoors (including walking while playing golf); jogging (more than 10 minutes per mile); running (10 minutes or less per mile); bicycling (including stationary bike); swimming laps; tennis; calisthenics, aerobics, aerobic dance, or rowing machine; and squash or racquetball. Physical activity was reassessed in 1988, 1992, and every 2 years thereafter. Beginning in 1992, we also included further questions on lower intensity activities (yoga, stretching, toning) and other activities (eg, lawn mowing). We categorized physical activity as metabolic equivalent task hours by multiplying the duration of physical activity by the standard metabolic equivalent task score for that activity.13 One metabolic equivalent task hour is equal to the amount of energy required to sit quietly for an hour.

Because we believed that regular long-term physical activity was most likely to affect risk of developing incontinence, in the primary analyses we averaged the metabolic equivalent task hour values over questionnaire reports beginning in 1986. In addition, because we were concerned that very recent levels of activity might be influenced by early incontinence symptoms or more general health issues, we imposed a 2-year lag period between the last report of physical activity and the baseline of our incontinence analyses. Thus, we considered activity reports from 1986 through 1998 and did not include reports from 2000 in our calculation of long-term average physical activity. This cumulative average, combined with the 2-year lag, has several advantages: by averaging reports over many questionnaires, it provides a more accurate representation of physical activity patterns, and it also helps to limit confounding that may result from recent changes in activity due to frailty or other health issues in these older women.

In validation studies among a similar cohort of nurses, participants' responses to questions on activity one year apart were reasonably correlated (r=0.59), given the expected true changes that might occur over a 1-year period. Moreover, physical activity recalled for the previous year correlated strongly with past-week recalls of physical activity (r=0.79) and with physical activity logged in diaries during the year (r=0.62).14

Multivariable logistic regression models were used to calculate adjusted odds ratios and 95% confidence intervals. A separate model was used for each of the four incontinence case definitions (all cases, stress, urge, mixed), compared with noncases; in a given analysis, women who did not meet the case or non-case definition were excluded (eg, urge UI cases were excluded from analyses of stress UI). Physical activity was categorized in quintiles. All multivariable models included the following covariates: age (continuous), race or ethnicity (Caucasian, African-American, Hispanic-American, Asian-American, and other race or ethnicity), BMI (less than 22, 22–24, 25–29, 30 or more), parity (nulliparous, 1–2 births, 3 or more births), smoking (never, former, current), and postmenopausal hormone use (premenopausal or missing, never used hormone therapy, formerly used hormone therapy, currently use hormone therapy). Additional control for diabetes and use of medications that may affect continence (furosemide, thiazides, angiotensin-converting enzyme inhibitors, calcium channel blockers) did not impact effect estimates and therefore were not included in the final models. Tests of dose-response trends (P for trend) for physical activity were conducted by including metabolic equivalent task hours in the multivariable logistic regression model as a continuous variable.


The mean age of the study population in 2000 was 65.9 years and ranged from 54 to 79 years. On average, walking made up almost half of total physical activity among our participants (46% of metabolic equivalent task hours per week). Biking and other low-impact aerobic exercise (eg, aerobic dance, rowing machines) each accounted for about 10% of total activity. Running was rare in our population, making up less than 1% of total activity, on average. As expected, BMI was progressively lower with increasing physical activity (Table 1). Current cigarette smoking was somewhat more common in the women with lower levels of physical activity. The distribution of other potential confounding factors was fairly similar across the physical activity groups.

Table 1
Table 1:
Characteristics of the Nurses' Health Study Participants at Baseline

There were 2,355 cases of incident urinary incontinence in our population. After adjusting for potential confounding factors, increasing levels of total physical activity were associated with decreasing incidence of UI (P for trend <.01, Table 2). Women with higher levels of total physical activity had a significant 15–20% lower risk of developing UI compared with the lowest quintile (for quintile 4: OR 0.85, 95% CI 0.75–0.98; for quintile 5: OR 0.81, 95% CI 0.71–0.93).

Table 2
Table 2:
Physical Activity and Risk of Incident Urinary Incontinence

In addition, when we separately examined walking, the most common activity in this cohort, we also found a strong trend of decreasing risk of developing UI with increasing levels of walking (P for trend <.01, Table 2). Women in the top quintile of walking had a 26% lower risk compared with women in the lowest quintile (95% CI 0.63–0.88).

Among the women that further provided information on incontinence-type symptoms and could be clearly classified by their UI type, we also explored the association of physical activity with stress (n=791 cases), urge (n=335 cases), and mixed (n=440 cases) urinary incontinence (Table 3). We found that higher levels of physical activity were significantly associated with decreasing stress UI (total activity: P for trend =.01; walking: P for trend =.01). For example, women in the top quintile of total physical activity were approximately 30% less likely to develop stress UI compared with women in the bottom quintile (OR 0.71, 95% CI 0.56–0.91), with similar, borderline significant results for walking (OR 0.76, 95% CI 0.57–1.02). We found no significant relation between physical activity and urge or mixed UI, although there were substantially fewer cases of urge and mixed UI than stress UI. Thus, there was less power to detect associations (Table 3).

Table 3
Table 3:
Physical Activity and Risk of Incident Stress, Urge, and Mixed Urinary Incontinence

In additional analyses, results were generally similar when we specifically examined severe incontinence, defined as at least weekly leaking of enough urine to wet the underwear. Although confidence intervals were somewhat wider than in analyses of frequent incontinence, the odds ratios for severe incontinence were 0.80 (95% CI 0.65–1.00), comparing extreme quintiles of total physical activity, and 0.63 (95% CI 0.48–0.82), comparing extreme quintiles of walking. For both total activity and walking, we also found suggestions of dose-response trends of decreasing risk of severe UI with increasing activity (total activity: P for trend =.04, walking: P for trend =.05).

In secondary analyses, we further explored how physical activity at different time periods during our follow-up was associated with urinary incontinence, including more recent activity (cumulative average from 1994–1998) and most recent (only at baseline in 2000). Findings remained robust in all these analyses, and were entirely consistent with those presented above (data not shown). We also conducted several analyses to help consider potential residual confounding by body mass index. We separately analyzed the relation of activity to UI within strata of BMI (less than 25, 25–29, 30 or more); the associations between physical activity and UI remained similar regardless of BMI category. In particular, even among thin women (BMI less than 25), there was still an inverse relation between activity level and risk of UI. Moreover, adjustment for weight change in addition to BMI did not change our results either. Finally, adjusting for BMI continuously, instead of categorically, did not alter results.


In our large-scale, prospective investigation, we found that moderate-intensity physical activity, including walking, was associated with an approximately 20–25% reduced risk of developing UI in older women. Existing epidemiologic data on physical activity and incontinence consist predominantly of small-scale cross-sectional studies and focus largely upon very high-impact activities15–17 in specific populations (eg, Olympic athletes17). In contrast, in these participants of the Nurses' Health Study, most activity consisted of walking and other low-impact activities (eg, bicycling). Limited data from cross-sectional studies of lower-impact activity generally support our findings. The Norwegian EPINCONT study,16 a large cross-sectional study, found a significant inverse association between low-impact physical activity and incontinence (OR 0.8, 95% CI 0.7–0.9 for 3 or more hours per week compared with less than 1 hour). This relation was specific to stress UI (OR 0.8, 95% CI 0.7–0.9 for stress UI; OR 1.0, 95% CI 0.8–1.3 for urge UI), consistent with our results. When examining severe incontinence, risk of urge UI was also reduced (OR 0.4, 95% CI 0.3–0.7). However, these results for severe incontinence may be particularly difficult to interpret. Because the EPINCONT study is cross-sectional, severe incontinence may be most likely to result in reduced activity, rather than reduced activity leading to less incontinence. In our prospective study, results incorporating severity of incontinence were generally similar to our primary results. Finally, in a prospective study, subjects with type 2 diabetes were assigned an intensive lifestyle intervention, which included both physical activity and dietary modification. This regimen was associated with significantly reduced odds of stress UI among female participants, although it is somewhat difficult to disentangle the impact of activity compared with diet modification.4

A likely mechanism explaining possible benefits of physical activity on development of incontinence, especially stress UI, is via strengthening of pelvic floor muscles. Evidence suggests that activity-induced increases in intra-abdominal pressure may result in simultaneous contraction (and strengthening) of the pelvic floor musculature.18 Although some benefits of activity could also be incurred from weight loss that may accompany physical activity, our results appeared to be independent of body mass index or weight change.

Several limitations should be considered. Classification of urinary incontinence frequency and type was based on self-report. However, we have demonstrated high reliability of self-reported urinary incontinence frequency in this population.8 Moreover, Diokno et al9 found 83% agreement between self-reported incontinence and clinical diagnosis. In addition, Sandvik et al12 found good specificity (92% for stress UI, 95% for urge UI), although poor sensitivity (59% for stress UI, 62% for urge UI) of self-reported incontinence-type symptoms compared with clinical diagnoses. However, in a prospective cohort study, in which misclassification of cases is unlikely to be related to the exposure, high specificity is most critical for validly identifying relative risk relations.19

In interpreting our results, we were concerned that initial minor symptoms of UI or other related health problems may lead women to reduce activity, falsely yielding an inverse relation between activity and UI. However, this should not be a major issue in our prospective study, where activity levels were considered only among mainly healthy women with no incontinence. Furthermore, by averaging activity over a long period, we minimized the effect of recent or transient changes in exercise patterns in response to subclinical or minor incontinence symptoms. In addition, we imposed a 2-year lag period between last report of physical activity and the baseline of our incontinence study (ie, in primary analyses, we considered activity only through 1998 and incident incontinence from 2000 to 2002). This should also have helped to minimize any influence of minor leaking at baseline on categorization of physical activity.

An important concern in this observational study is the possibility for confounding. In particular, conditions affecting physical mobility or functioning might both decrease exercise and increase incontinence. We had extensive data on a large variety of health variables, and we excluded women from our study population who reported a variety of functional limitations or neurologic conditions. However, it is not possible to eliminate confounding in observational studies.

Finally, our study population is a select cohort of largely Caucasian health professionals, and results may not be generalizable to all women. Nonetheless, estimates of urinary incontinence prevalence among these women were previously found to be similar to those of similarly aged, Caucasian, community-dwelling women.8 Furthermore, our prior results for other incontinence risk factors (eg, parity, diabetes) were comparable to those found in studies of more general populations,8,20 suggesting that relations observed in our study population are applicable to broader populations of older Caucasian women. Differences in urinary incontinence prevalence across racial and ethnic categories have been documented, and it is possible that incontinence risk factors may also vary by race and ethnicity. Thus, our findings may not be applicable to non-Caucasian women.

In conclusion, our study found that low-impact physical activity is associated with a reduced risk of UI. Our results appeared somewhat stronger for stress UI than urge UI. Although findings will need to be confirmed in future studies, our results suggest that women who avoid exercise due to concern about becoming incontinent might be reassured that low-impact activity does not appear to increase the risk of developing incontinence.


1. Centers for Disease Control and Prevention. Physical activity for everyone: recommendations. Available at: Retrieved December 20, 2006.
2. U.S. Department of Health and Human Services. Physical activity and health: a report of the surgeon general. Atlanta (GA): U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion; 1996.
3. Division of Nutrition and Physical Activity National Center for Chronic Disease Prevention and Health Promotion. U.S. physical activity statistics: state demographic data comparison. Available at: Retrieved December 20, 2006.
4. Brown JS, Wing R, Barrett-Connor E, Nyberg LM, Kusek JW, Orchard TJ, et al. Lifestyle intervention is associated with lower prevalence of urinary incontinence: the Diabetes Prevention Program. Diabetes Care 2006;29:385–90.
5. Thyssen HH, Clevin L, Olesen S, Lose G. Urinary incontinence in elite female athletes and dancers. Int Urogynecol J Pelvic Floor Dysfunct 2002;13:15–7.
6. Nygaard I, Girts T, Fultz NH, Kinchen K, Pohl G, Sternfeld B. Is urinary incontinence a barrier to exercise in women? Obstet Gynecol 2005;106:307–14.
7. Jones WK. Understanding barriers to physical activity is a first step in removing them. Am J Prev Med 2003;25 suppl:2–4.
8. Grodstein F, Fretts R, Lifford K, Resnick N, Curhan G. Association of age, race, and obstetric history with urinary symptoms among women in the Nurses' Health Study. Am J Obstet Gynecol 2003;189:428–34.
9. Diokno AC, Brown MB, Brock BM, Herzog AR, Normolle DP. Clinical and cystometric characteristics of continent and incontinent noninstitutionalized elderly. J Urol 1988;140:567–71.
10. Hannestad YS, Rortveit G, Sandvik H, Hunskaar S. A community-based epidemiological survey of female urinary incontinence: the Norwegian EPINCONT study. Epidemiology of Incontinence in the County of Nord-Trondelag. J Clin Epidemiol 2000;53:1150–7.
11. Lemack GE, Zimmern PE. Predictability of urodynamic findings based on the Urogenital Distress Inventory-6 questionnaire. Urology 1999;54:461–6.
12. Sandvik H, Hunskaar S, Vanvik A, Bratt H, Seim A, Hermstad R. Diagnostic classification of female urinary incontinence: an epidemiological survey corrected for validity. J Clin Epidemiol 1995;48:339–43.
13. Ainsworth BE, Haskell WL, Leon AS, Jacobs DR Montoye HJ, Sallis JF, et al. Compendium of physical activities: classification of energy costs of human physical activities. Med Sci Sports Exerc 1993;25:71–80.
14. Wolf AM, Hunter DJ, Colditz GA, Manson JE, Stampfer MJ, Corsano KA, et al. Reproducibility and validity of a self-administered physical activity questionnaire. Int J Epidemiol 1994;23:991–9.
15. Eliasson K, Nordlander I, Mattsson E, Larson B, Hammarstrom M. Prevalence of urinary leakage in nulliparous women with respect to physical activity and micturition habits. Int Urogynecol J Pelvic Floor Dysfunct 2004;15:149–53.
16. Hannestad YS, Rortveit G, Daltveit AK, Hunskaar S. Are smoking and other lifestyle factors associated with female urinary incontinence? The Norwegian EPINCONT Study. BJOG 2003;110:247–54.
17. Jiang K, Novi JM, Darnell S, Arya LA. Exercise and urinary incontinence in women. Obstet Gynecol Surv 2004;59:717–21.
18. Bo K. Urinary incontinence, pelvic floor dysfunction, exercise and sport. Sports Med 2004;34:451–64.
19. Rothman KJ, Greenland S. Modern epidemiology. 2nd ed. Philadelphia (PA): Lippincott-Raven; 1998.
20. Lifford KL, Curhan GC, Hu FB, Barbieri RL, Grodstein F. Type 2 diabetes mellitus and risk of developing urinary incontinence. J Am Geriatr Soc 2005;53:1851–7.
© 2007 by The American College of Obstetricians and Gynecologists. Published by Wolters Kluwer Health, Inc. All rights reserved.