Reducing caffeine intake is a common recommendation for women with urge urinary incontinence (UUI).1,2 This recommendation is based on evidence that caffeine promotes diuresis3,4 and increases detrusor pressure,5 which, together, might increase the likelihood of urgency-related involuntary urine loss. Yet, the available data on caffeine restriction and UI, derived from several small short-term trials, are equivocal. For example, a randomized crossover study6 and randomized trial,7 each including less than 100 adults with lower urinary tract symptoms, found no statistically significant improvement in UI with caffeine reduction over a period of 14 to 30 days. In addition, in a 3-week randomized trial of 224 women with UI,8 there was no significant difference in episodes of UI among women who received counseling about multiple behavioral UI management strategies, including caffeine intake, compared with those who received no intervention, although interpretation of these results is difficult because caffeine intake was one of several simultaneous interventions.
Clearly, there is still much to learn about the effect of caffeine on UI among women with incontinence, and the current literature has focused on the acute rather than any potential longer-term effects of caffeine intake. Therefore, we estimated the association between long-term caffeine intake and risk of subsequent UI progression over 2 years among 21,564 women with moderate UI enrolled in the Nurses' Health Study and Nurses' Health Study II.
MATERIALS AND METHODS
The Nurses' Health Study began in 1976 when 121,700 female nurses, aged 30–55 years, returned a mailed questionnaire about their health and lifestyle. In 1989, the Nurses' Health Study II was initiated when 116,430 female nurses aged 25–42 years returned a similar questionnaire. Participants provided informed consent by returning their questionnaires. In both cohorts, updated information has been collected using biennial questionnaires. During each questionnaire cycle, full-length questionnaires are sent in initial mailings, followed by abbreviated questionnaires to maximize participation. Questions about UI were included on the full-length 2002 and 2004 questionnaires in Nurses' Health Study and 2003 and 2005 questionnaires in Nurses' Health Study II. The Institutional Review Board of Brigham and Women's Hospital approved these studies.
For these analyses, we defined baseline as 2002 (Nurses' Health Study) and 2003 (Nurses' Health Study II) because both UI and caffeine intake were assessed on those questionnaires. After excluding deaths before baseline (n=15,152 for Nurses' Health Study; n=1,053 for Nurses' Health Study II) and nonresponse to the full-length questionnaire (n=24,374 for Nurses' Health Study; n=37,583 for Nurses' Health Study II) or the UI items at baseline (n=556 for Nurses' Health Study; n=144 for Nurses' Health Study II), there were 81,619 Nurses' Health Study and 77,650 Nurses' Health Study II participants who provided baseline information on UI frequency. Because our focus was progression of prevalent incontinence, we excluded women who reported no UI or insignificant UI (n=42,350 Nurses' Health Study and 46,683 Nurses' Health Study II participants with no UI or UI less than once per month) and women who already had frequent UI (n=24,315 Nurses' Health Study and 16,252 Nurses' Health Study II participants with UI once per week or more). Analyses utilized caffeine data from semiquantitative food frequency questionnaires administered at baseline and, in additional analyses, 4 years before baseline. Thus, we excluded 1,134 Nurses' Health Study and 1,432 Nurses' Health Study II participants who did not provide caffeine intake data on the food frequency questionnaires. In addition, we excluded women with major neurologic conditions (stroke, multiple sclerosis, Parkinson disease, or amyotrophic lateral sclerosis) or functional limitations (defined as difficulty climbing a flight of stairs, walking one block, bathing, or dressing), which might cause UI, as well as women missing data on body mass index (calculated as weight (kg)/[height (m)]2) and parity, which are important UI risk factors (n=1,596 for Nurses' Health Study; n=1,312 for Nurses' Health Study II). Finally, after excluding women missing UI data on the follow-up questionnaire (n=1,214 for Nurses' Health Study; n=1,417 for Nurses' Health Study II), there were 11,010 Nurses' Health Study and 10,554 Nurses' Health Study II participants available for analysis.
To measure caffeine intake, we used validated food frequency questionnaires included on the Nurses' Health Study questionnaires in 1998 and 2002 and included on the Nurses' Health Study II questionnaires in 1999 and 2003. Participants were asked about longer-term caffeine intake; specifically, the food frequency questionnaire inquired how often, on average, during the previous year they consumed various items, including coffee with caffeine (one cup), tea with caffeine (one cup), caffeinated soda (one glass, bottle, or can), and chocolate (eg, bar or packet). There were nine response options ranging from none or less than one per month to six or more per day. Using U.S. Department of Agriculture food composition data supplemented with other sources,9–12 estimated caffeine contents were 137 mg per cup of coffee, 47 mg per cup of tea, 46 mg per can or bottle of cola beverage, and 7 mg per serving of chocolate. We calculated total caffeine intake by summing the caffeine content for specific items multiplied by weights proportional to the frequency of use of each item.
The reproducibility and validity of the food frequency questionnaires have been reported previously.13 In a Nurses' Health Study validation study, there was a strong linear association between intakes assessed with the food frequency questionnaire and intakes reported on four 1-week diet records completed over a 1-year period. For example, Pearson correlation coefficients were 0.78 for coffee, 0.93 for tea, and 0.84 for cola drinks.14
Questions about UI were included on the 2002 and 2004 questionnaires in the Nurses' Health Study and the 2003 and 2005 questionnaires in the Nurses' Health Study II. Participants were asked, “During the last 12 months, how often have you leaked or lost control of your urine?” Response options were never, less than once per month, two to three times per month, approximately once per week, and almost every day.15 A reliability study among 200 participants demonstrated high reproducibility of response to this question.16
Urinary incontinence progression case group individuals were women whose incontinence frequency increased from one to three times per month at baseline (ie, 2002 in Nurses' Health Study and 2003 in Nurses' Health Study II) to at least once per week 2 years later. Noncase group individuals were women whose UI frequency stayed the same or decreased over the 2-year period.
Urinary incontinence type was assessed on the 2004 Nurses' Health Study and 2005 Nurses' Health Study II questionnaires. Women were asked, “When you lose urine, what is the usual cause?” Urge urinary incontinence was defined as leaking usually caused by a sudden and urgent need to urinate. Stress UI was defined as leaking usually caused by coughing, sneezing, laughing, or performing physical activity. Type of UI was classified as mixed when stress and UUI symptoms were equally common.
We examined both baseline level of caffeine intake and previous change in caffeine intake in relation to risk of UI progression. For the analyses of level of caffeine intake, we categorized baseline intake into categories approximately corresponding to the amount of caffeine in one, two, three, or four or more cups of coffee (0–149, 150–299, 300–449, and 450–965 mg/d). For the analyses of change in caffeine intake, we first categorized caffeine intake at baseline and 4 years previously into five finer categories (0–149, 150–299, 300–449, 450–599, and 600–965 mg/d). We then defined caffeine intake as “stable” when the intake category remained the same at both time points. Women were considered to have increased their caffeine intake if they moved into a higher category, or to have decreased their intake if they moved into a lower category.
In separate analyses within each cohort, logistic regression models were used to calculate multivariable-adjusted odds ratios (ORs) and their 95% confidence intervals (CIs) for UI progression according to caffeine intake, both overall and by UI type. Model covariates, which reflected participants' status as of the baseline questionnaire, were UI frequency (once per month or two to three times per month) and potential UI risk factors identified from the literature, including age, parity, body mass index, cigarette smoking, race, and total fluid intake. Total fluid intake included milk, juice, tea, coffee, soda, punch, alcohol, and water. Analyses of change in caffeine intake were additionally adjusted for initial level of caffeine intake. Further adjustment for other potential confounding factors, such as physical activity (metabolic equivalent hours per week), menopausal status, postmenopausal hormone use, diabetes, and diuretic use did not change the results and thus were not included in the final multivariable logistic regression models.
After calculating cohort-specific ORs, we combined the log ORs, weighted by the inverse of their variances, using a random effects model for meta-analysis. Between-study heterogeneity was tested by the Q statistic.17
We conducted several secondary analyses to further estimate the association between caffeine and incontinence progression. First, in addition to controlling for initial frequency of UI, to estimate whether the association between caffeine and UI progression was influenced by UI frequency at baseline, we conducted separate analyses among women with UI once per month at baseline and UI two to three times per month at baseline. In addition, to minimize the possibility of misclassification of UI progression, we repeated the analyses after defining UI progression more conservatively as a change from UI one to three times per month to daily UI; noncase group individuals remained women whose UI frequency stayed the same or decreased and women with more moderate progression were excluded. To estimate the association between baseline level of caffeine intake and UI progression among those with the most stable caffeine intake over time, in whom the likelihood of misclassification may be lowest, we repeated the analyses after excluding 3,071 participants who decreased their caffeine intake by more than one category during the 4 years before baseline. Finally, if any negative effect of caffeine might be limited to women more predisposed to UI progression, then we repeated the analyses among the oldest women with mental health or physical function limitations. Mental health limitations were defined as a score less than 84 (the median value among women without UI) on the Short-Form 36 Health Status Survey mental health subscale18 or self-reported regular use of antidepressant medication or diagnosis of depression. Limited physical function was defined as a score less than 90 (the median value among women without UI) on the Short-Form 36 Health Status Survey physical function subscale.18
For all analyses, two-tailed P<.05 was considered statistically significant. Data were analyzed using SAS 9.2.
At baseline, Nurses' Health Study participants were aged 56–81 years and Nurses' Health Study II participants were aged 39–56. In general, characteristics of women were fairly similar across categories of caffeine intake, although women who consumed more caffeine had higher mean daily fluid intake and were more likely to have ever smoked cigarettes (Table 1).
There was little change in the proportion of women with UI progression across categories of baseline level of caffeine intake (data not shown in tables). For example, among Nurses' Health Study and Nurses' Health Study II participants with UI one to three times per month at baseline, the percentages reporting an increase in UI frequency to at least weekly after 2 years of follow-up were 22%, 22%, 24%, and 21% in women consuming 0–149, 150–299, 300–449, and 450 mg or more of caffeine per day, respectively.
After adjusting for potential confounding factors, we found no association between baseline level of caffeine intake and subsequent odds of UI progression over 2 years (Table 2). For example, among women in both cohorts, the OR for UI progression was 0.87 (95% CI 0.70–1.08) comparing the highest compared with the lowest category of caffeine intake. Results were comparable when women with UI once per month at baseline were analyzed separately from women with UI two to three times per month (data not shown). In addition, when we repeated the analyses defining UI progression as an increase in UI frequency to daily rather than at least weekly (data not shown in table), results in the combined cohorts were similar to those reported here (OR 0.85, 95% CI 0.59–1.21 comparing 450 or more compared with 0–149 mg of caffeine per day).
We also conducted analyses of specific types of UI progression. Comparing those in the highest compared with lowest caffeine intake category, there was no indication of significant increased odds of UUI progression with higher caffeine intake (combined cohorts: OR 0.84, 95% CI 0.57–1.25; Table 2). Higher caffeine intake also was not associated with odds of stress or mixed UI progression (data not shown in table). For example, ORs comparing the highest compared with lowest caffeine intake categories were 0.93 (95% CI 0.69–1.25) for women with stress UI and 0.68 (95% CI 0.45–1.04) for women with mixed UI.
In secondary analyses, we restricted the study population to women with stable caffeine intake over the previous 4 years; results were similar to those reported herein (data not shown). In addition, results were comparable to those in the main analyses when we restricted the study population to older women in the Nurses' Health Study with depressive symptoms or limited physical function.
We were also interested in assessing the relation between change in caffeine intake and subsequent risk of UI progression. In unadjusted analyses, the percentage of women with UI progression was similar among women with stable caffeine intake (22%) and those who increased (20%) or decreased (22%) their caffeine intake during the 4 years before baseline. After adjusting for potential confounding factors, odds of subsequent UI progression were similar among women who increased (combined cohorts: OR 1.08, 95% CI 0.95–1.22) or decreased their caffeine intake (OR 1.16, 95% CI 1.02–1.31) compared with women with stable caffeine intake (Table 3). When we separately examined women with UI once per month compared with two to three times per month at baseline, results were similar to those in the primary analyses (data not shown in table). For example, comparing women who increased their caffeine intake with those whose intake remained stable, ORs were 1.01 (95% CI 0.78–1.31) for women with UI once per month and 1.10 (95% CI 0.89–1.37) for women with UI two to three times per month. In addition, results in analyses using a stricter definition of UI progression (ie, an increase to daily UI) were consistent with those reported herein (data not shown).
When we looked specifically at women with UUI progression (Table 3), there was no indication of an association between increased (combined cohorts: OR 1.10, 95% CI 0.85–1.43) or decreased caffeine intake (combined cohorts: OR 1.15, 95% CI 0.90–1.45) and UI progression. In addition, there was no association between increased or decreased caffeine intake and risk of stress or mixed UI progression (data not shown).
Overall, we observed no association between longer-term caffeine intake, defined as average intake during the previous year, or previous change in caffeine intake and subsequent odds of UI progression, including UUI progression, over 2 years among women with moderate UI. Although these data do not address acute effects of caffeine on continence mechanisms, they indicate that longer-term caffeine levels, or long-term changes in caffeine intake, are not related to UI progression. Compared with other large U.S. studies,19,20 caffeine intake among the women in our cohorts was comparable or slightly lower (lower intake in our study might be expected because all women had moderate UI at baseline), and thus our findings on levels of caffeine intake are not likely to vary from other populations.
Our results are consistent with data from two small randomized trials.6,7 For example, Swithinbank et al6 observed no change in UI over 2 weeks after an intervention to replace caffeinated with decaffeinated beverages in 69 women (mean age 55 years) with stress incontinence or idiopathic detrusor overactivity.
Several mechanisms have been hypothesized to link caffeine intake with exacerbation of UI, such as increased diuresis3,4 and increased detrusor pressure during bladder filling,5 which could promote urgency, and possibly UUI, particularly in women with underlying detrusor overactivity. It is unknown how long-term compared with acute caffeine intake may influence these mechanisms. Nonetheless, the results of our study and other epidemiologic studies of acute caffeine reduction do not support the hypothesized mechanisms.
Several limitations of our study should be considered when interpreting the results. First, caffeine intake and UI data were self-reported rather than objectively measured, and thus absolute caffeine intake and prevalence of UI progression may be underestimated or overestimated. However, validation of the main contributors to caffeine intake among these women (coffee, tea, and soda) indicated that self-reports are reasonably accurate in ranking individuals according to intake level and thus are not likely a major source of bias.14 Regarding UI, previous studies have demonstrated high reliability21 and validity22 of self-reported UI; moreover, there is high specificity of self-reported UI type compared with clinical diagnoses,23 which is key to valid risk factor estimation. Second, we did not have information on caffeine consumption patterns (eg, consuming large quantities at once compared with moderate quantities throughout the day) and thus could not examine this aspect of caffeine intake in relation to UI progression. Third, in a large epidemiologic study, measurement of acute caffeine intake is not feasible, and our food frequency questionnaire requests information about longer-term food and beverage consumption (ie, average intake over the previous year). Nonetheless, understanding the relation between longer-term caffeine habits and UI progression is important, and information on the role of caffeine in worsening UI is largely based on a few small studies. In addition, we did not assess UI treatment among women in current analysis, and thus it is unclear how treatment may have affected our UI progression estimates. However, data from previous studies in these cohorts indicate that a minority of women seek treatment.24,25 Furthermore, it is unlikely that treatment is related to caffeine intake, suggesting lack of treatment data would not bias our results. In addition, we did not collect data on urinary symptoms other than UI, such as urgency, daytime frequency, and nocturia, and thus cannot make any conclusions about their potential associations with long-term caffeine intake. Finally, more than 95% of our study participants are white; therefore, our findings may not be generalizable to nonwhite women, in whom UI progression rates may be different from those in white women.26
In conclusion, if confirmed in other studies, then these results suggest that women with moderate UI should not be concerned that their regular caffeine consumption, defined here as intake over 1 year, will influence their risk of development of more frequent UI. In addition, these data do not support an association between long-term increases or decreases in caffeine intake and risk of UI progression over 2 years.
1. Hashim H, Al Mousa R. Management of fluid intake in patients with overactive bladder. Curr Urol Rep 2009;10:428–33.
2. Wyman JF, Burgio KL, Newman DK. Practical aspects of lifestyle modifications and behavioural interventions in the treatment of overactive bladder and urgency urinary incontinence. Int J Clin Pract 2009;63:1177–91.
3. Lohsiriwat S, Hirunsai M, Chaiyaprasithi B. Effect of caffeine on bladder function in patients with overactive bladder symptoms. Urol Ann ;3:14–8.
4. Maughan RJ, Griffin J. Caffeine ingestion and fluid balance: a review. J Hum Nutr Diet 2003;16:411–20.
5. Creighton SM, Stanton SL. Caffeine: does it affect your bladder? Br J Urol 1990;66:613–4.
6. Swithinbank L, Hashim H, Abrams P. The effect of fluid intake on urinary symptoms in women. J Urol 2005;174:187–9.
7. Bryant CM, Dowell CJ, Fairbrother G. Caffeine reduction education to improve urinary symptoms. Br J Nurs 2002;11:560–5.
8. Kincade JE, Dougherty MC, Carlson JR, Hunter GS, Busby-Whitehead J. Randomized clinical trial of efficacy of self-monitoring techniques to treat urinary incontinence in women. Neurourol Urodyn 2007;26:507–11.
9. Grand AN, Bell LN. Caffeine content of fountain and private-label store brand carbonated beverages. J Am Diet Assoc 1997;97:179–82.
10. U.S. Department of Agriculture. Composition of foods—raw, processed, and prepared. Agricultural Handbook No 8 Series. Washington, DC: Department of Agricuture, Government Printing Office; 1993.
11. U.S. Department of Agriculture. USDA nutrient database for standard reference. Washington, DC: Department of Agriculture, Government Printing Office; 1996.
12. U.S. Department of Agriculture. USDA nutrient database for standard reference. Washington, DC: Department of Agriculture, Government Printing Office; 2001.
13. Willett WC, Sampson L, Stampfer MJ, Rosner B, Bain C, Witschi J, et al.. Reproducibility and validity of a semiquantitative food frequency questionnaire. Am J Epidemiol 1985;122:51–65.
14. Salvini S, Hunter DJ, Sampson L, Stampfer MJ, Colditz GA, Rosner B, et al.. Food-based validation of a dietary questionnaire: the effects of week-to-week variation in food consumption. Int J Epidemiol 1989;18:858–67.
15. Resnick NM, Beckett LA, Branch LG, Scherr PA, Wetle T. Short-term variability of self report of incontinence in older persons. J Am Geriatr Soc 1994;42:202–7.
16. 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.
17. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials 1986;7:177–88.
18. Ware JE, Snow KK, Kosinski M, Gandek B. SF-36 health survey: manual and interpretation guide. Boston: The Health Institute, New England Medical Center; 1993.
19. Choi HK, Curhan G. Coffee, tea, and caffeine consumption and serum uric acid level: the third national health and nutrition examination survey. Arthritis Rheum 2007;57:816–21.
20. Shen J, Johnson VM, Sullivan LM, Jacques PF, Magnani JW, Lubitz SA, et al.. Dietary factors and incident atrial fibrillation: the Framingham Heart Study. Am J Clin Nutr 2011;93:261–6.
21. 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.
22. 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.
23. 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.
24. Lifford KL, Townsend MK, Curhan GC, Resnick NM, Grodstein F. The epidemiology of urinary incontinence in older women: incidence, progression, and remission. J Am Geriatr Soc 2008;56:1191–8.
25. Townsend MK, Danforth KN, Lifford KL, Rosner B, Curhan GC, Resnick NM, et al.. Incidence and remission of urinary incontinence in middle-aged women. Am J Obstet Gynecol 2007;197:167e1–5.
26. Townsend MK, Curhan GC, Resnick NM, Grodstein F. Original research: rates of remission, improvement, and progression of urinary incontinence in Asian, Black, and White women. Am J Nurs 2011;111:26–33.