Fecal incontinence is commonly defined as the involuntary loss of solid or liquid feces or mucus⁶ and definitions can include consistency of stool and frequency of occurrence. Data on national rates of FI and anal incontinence are currently limited by variability in definitions, lack of inclusion of a wide range of age groups, and under-representation of minorities.

The Bowel Health questionnaire includes questions from the Fecal Incontinence Severity Index⁸, which asks about the frequency of accidental bowel leakage during the last month separately for gas, mucus, liquid, and solid stool. Frequency is assessed as 2 or more times a day, once a day, 2 or more times a week, once a week, 1 to 3 times a month, or never. For this study, FI was defined as any involuntary loss of mucus, liquid, or solid stool during the last 30 days; this definition of FI does not include gas. The Bowel Health questionnaire also asked survey participants how often they usually have bowel movements. For data analysis, their responses were merged into 3 ranges: <3/week, 3-21/week, and >21/week.

The validated Bristol Stool Scale⁹ was used to determine the participant’s usual stool consistency. This scale consists of 7 descriptions of stool characteristics and includes pictures of each stool type to aid participants in classifying their stools (http://www.cdc.gov/nchs/data/nhanes/nhanes_05_06/MECCAPI_handcards_0506.pdf) The Bristol Stool Scale may be used to classify individual bowel movements¹⁰ or to describe typical stool consistency¹¹. For the NHANES survey, participants were asked, “What is your usual or most common stool type?” Judgments of usual stool consistency have been shown to correlate r=0.57 with objective measures of whole gut transit time¹². For this study, we pooled stool type ratings 1 and 2 (hard and lumpy), ratings 3-5 (normal consistencies), and ratings 6 and 7 (mushy and watery).

Age was categorized as 20 to 29 years, 30 to 39 years, 40 to 54 years, 55 to 69 years, and 70 years and older. Based on self-reported information, participants were classified into Mexican American and other Hispanic, non-Hispanic white, non-Hispanic black, and other race/ethnicity groups. Marital status was classified as widowed, separated, or divorced; married or cohabitating with a partner; and never married. Education was defined as having less than a high school education, a high school diploma (including GED), or additional education beyond high school. Income was represented by the Poverty Income Ratio (PIR), which varies by family size and composition (www.census.gov/hhes/www/poverty/definitions.html#ratio). For this study, participants were classified as at or above the poverty threshold versus below the threshold. The body mass index (BMI) was calculated as measured weight (kg)/ height (m²); overweight was defined by a BMI of 25.0-29.9, and obesity was defined by a BMI ≥30. Vigorous physical activity was defined as activity done for at least 10 minutes in the last 30 days causing heavy sweating or large increases in breathing or heart rate; moderate physical activity was defined as activity done for at least 10 minutes causing only light sweating or a slight to moderate increase in breathing or heart rate. Pregnancy status was assessed by participant self-report and/or a urine pregnancy test.

Diabetes mellitus was defined as a positive response to any of three questions: (a) “Have you ever been told by a doctor or other health professional that you have diabetes or sugar diabetes?” (b) “Are you now taking insulin?” (c) “Are you now taking diabetic pills to lower your blood sugar?” Chronic illnesses were ascertained by 10 separate questions of the form: “Have you ever been told by a doctor that you have [Arthritis/ Congestive heart failure/ Coronary heart disease/ Angina or angina pectoris/ Heart attack/ Stroke/ Emphysema/ Chronic bronchitis/ Any liver condition/ Cancer or malignancy]?” The cumulative number of positive responses to these 10 questions was divided into three categories: 0, 1, and 2 or more. Self-described general health status was defined by the question “Would you say that in general your health is excellent, very good, good, fair, or poor?” Responses to this question were aggregated into two ranges: excellent, very good, or good health status versus fair or poor health.

Urinary incontinence (UI) was assessed using the Incontinence Severity Index^13,14, which consists of two questions: “How often do you have urinary leakage? Would you say never, less than once a month, a few times a month, a few times a week, or every day and/or night?” (responses coded 1-5) and “How much urine do you lose each time? Would you say drops, small splashes, or more?” (responses coded 1-3). This questionnaire is scored by multiplying responses to the two questions, and scores of 3 or greater are classified as at least moderately severe UI^{14, 15}. In this study, urinary incontinence was defined as a score of at least 3.

Data for adults 20 years of age and older from the NHANES 2005-06 were used in the analysis. SAS Survey Procedures (Research Triangle Institute, Research Triangle Park, North Carolina) was used to calculate all prevalence estimates and 95% confidence intervals of the estimates, incorporating the design effect, appropriate sample weights, stratification, and clustering of the complex NHANES sample design. The examination sample weights adjust for unequal probabilities of selection and non-response, and are post-stratified to population control totals for each sampling subdomain. Prevalence estimates (percents) for the national population and 95% confidence intervals are reported separately for women, men, and both women and men combined. Estimates with relative standard errors greater than 30% are identified as statistically unreliable.

An estimated 8.3% (CI 7.1, 9.5) of non-institutionalized adults in the United States report FI at least once during the last 30 days; this estimate corresponds to 18 million people as of April 1 to July 1, 2006. FI was found to be equally common in women (8.9%; CI 7.2, 10.5) and men (7.7%; CI 6.0, 9.4; p=0.31).

Liquid stool incontinence was the most common type of FI as shown in Table 2 and was reported equally often by women and men (p=0.69). Solid stool incontinence was more likely to be reported by women than by men (p=0.038). Incontinence of more than one consistency was reported by 26.1% (CI 21.1, 31.0) of participants with FI.

Most participants with FI reported that it occurred one to three times per month (Table 2). However, 2.8% (CI 1.9, 3.7) of women and 2.6% (CI 1.7, 3.4) of men had FI at least once a week. A frequency of one or more FI episodes per day was reported by 0.9% (CI 0.6, 1.2) or 90 people out of a thousand.

The NHANES provided an opportunity to assess how prevalence estimates would be affected by including accidental loss of gas in the definition, which would correspond to the definition of anal incontinence¹⁶. Leakage of gas at least once in the last month was reported by 51.0% (CI: 45.7, 56.3) of women and 46.2% (CI: 42.4, 49.9) of men, and daily leakage of gas was reported by 20.5% (CI: 17.6, 23.4) of women and 24.0% (CI: 20.7, 27.3) of men.

As shown in Figure 1, there was a linear progression in the prevalence of FI with age for both women and men (p<0.01 for both). For women and men combined, prevalence increased from 2.6% (CI 1.5, 3.8) at 20-29 years of age up to 15.3% (CI 11.5, 19.0) in participants aged 70 and over. However, FI was not significantly associated with race/ethnicity, marital status, education, or family income for women or men after adjusting for age (data not shown).

Table 3 lists candidate risk factors that were significantly associated with FI in either women or men when tested individually; associations are expressed as age-adjusted odds ratios and 95% confidence intervals. This table also shows the final multivariate models for women and men. Supplemental tables S1 and S2 (see journal website) provide estimates of the prevalence of FI by all risk factors considered in this study.

Participants who reported that their bowel movements were usually mushy or watery were substantially more likely to report FI compared to participants with normal stool consistency (Figure 2) even after multivariate adjustment for other risk factors. For women, the findings for stool frequency paralleled those for stool consistency: having more than 21 bowel movements per week was significant in the final model. However, after multivariate adjustment this relationship did not hold for men.

The inability to engage in physical activity was significantly associated with FI in women but not men (Table 3). In the bivariate analyses, engaging in vigorous physical activity appeared to be protective in women, but this association did not survive multivariate adjustment. Similarly, in bivariate analyses obesity (BMI ≥30) was associated with a higher prevalence of FI compared to normal or low weight (BMI <25) in women, but this association was no longer significant after multivariate adjustment.

Women with chronic illnesses were at increased risk of FI, but this association was not significant for men. However, in men (but not women) self-reported poor health was significantly associated with FI.

In women, the prevalence of FI increased with increasing numbers of vaginal deliveries from 5.9% (CI 3.2, 8.6) in women with no vaginal deliveries to 15.1% (CI 11.9, 18.3) in women with 4 or more vaginal deliveries (p<0.0001; Supplemental Table S2). However, number of vaginal deliveries was no longer significantly associated with FI after adjustment for age (p=0.09) and other risk factors (p=0.57).

This study has limitations. First, prevalence estimates were limited to non-institutionalized individuals, and because the prevalence of FI is higher in nursing homes than in the community, the overall prevalence of FI is likely underestimated, especially among older participants. Secondly, the volume of stool lost, in addition to the frequency and type of stool lost, may influence the impact of FI on quality of life¹⁸, and volume of incontinence was not assessed. Finally, some possible risk factors of FI such as hemorrhoids and rectal prolapse were not assessed.

FI is frequently reported to be more common in the elderly^{19, 20}, and this is often attributed to the confounding of age with poorer health status and with other risk factors such as mobility restriction. This survey confirms a strong association between FI and age, with the prevalence of FI increasing from 2.6% at ages 20-30 years up to 15.3% in people aged 70 or older. However, this effect of aging does not appear to be explained by poorer health in older participants: multivariate regression analysis showed age to be a strong predictor of FI after adjusting for the number of chronic illnesses, overall health status, and physical activity level. The mechanisms that could explain this effect of age on the prevalence of FI are unknown.

Women were found in some prior studies^{21, 22} to have a higher prevalence of FI than men, and the speculation has been that sex differences occur because obstetrical injuries are a major risk factor for FI. However, in this survey of the whole adult lifespan, there was no significant difference in the prevalence of FI between women and men. In univariate analysis the number of vaginal deliveries was associated with increased prevalence of FI in women, but after adjustment for age and other risk factors, number of vaginal deliveries was not found to be a significant independent predictor of FI. Other studies which included a broad range of ages^23-25 have also failed to find a sex difference in FI prevalence or an effect of obstetrical injury²⁶. This suggests that obstetrical injuries, while important because they are potentially preventable, are not the most common causes of FI in women. In addition, even though there have been some reports of differential rates of obstetric injuries by racial/ethnic group, no differences in rates of FI by race were noted here.

An aim of this study was to identify modifiable risk factors which can be incorporated into prevention or treatment strategies. Our data confirm previous reports of an association between FI and diarrhea^{27, 28} and suggest that treating diarrhea may be an effective method for treating or preventing FI. There are studies showing that the treatment of diarrhea-associated FI with loperamide²⁹ or fiber supplements³⁰ is effective at least short-term, although more research to identify long term effective treatments is needed.

Previous authors have reported an association between poor overall health and FI^{31, 32}. The number of chronic illnesses and self-described health status are correlated, and in multivariate regression analyses which adjust for this correlation, the number of chronic illnesses was found to be significantly associated with FI in women, while in men it was self-reported poor or fair health status that was a significant predictor of FI.

Urinary incontinence was significantly associated with FI, as others have shown^33,34. The association is believed to be related to the common innervation of both the external anal and external urethral sphincters by sacral nerves, the interactions between afferents from the rectum and bladder at the level of the spinal cord³⁵, and the exposure of the rectum and bladder to the same traumas because both are located close together in the pelvis.

01