Constipation is among the most common gastrointestinal disorders. It is so prevalent, in fact, that it has been considered endemic in the elderly population. In the United States alone, more than 3 million prescriptions are written for cathartics yearly and over $800 million is spent for over-the-counter (OTC) laxatives. It is clear that constipation represents a major public health problem.
Despite its significant impact, the etiology of constipation remains largely unknown. The variety of symptoms and risk factors associated with constipation suggest that its etiology is likely to be multifactorial. Although epidemiologic studies cannot establish etiologic relationships, consistent epidemiologic distributions may suggest potential causative risk factors. The more uniform the epidemiologic pattern, the more likely an environmental agent(s) may be contributing to its etiology. Elucidation of the epidemiology of constipation, therefore, is helpful both in suggesting potential etiologic risk factors and in identifying populations that are at highest risk of developing this condition. Once high-risk populations are identified, they can be targeted for treatment or possibly even for interventions that might prevent the development of this often debilitating condition.
A description of the epidemiology of constipation is clearly dependent on how the disorder is defined. Attempts to provide an objective definition of constipation date back to the 1960s. In a commonly cited study, Connell and colleagues surveyed the bowel frequency of factory workers in England and found that more than 99% had bowel frequencies ranging between three per day and three per week. These data provided an easily measurable and reproducible definition that did not rely on an individual’s subjective impressions of their bowel function. As a result, constipation began to be defined as less than three bowel movements per week, and this definition continued as the predominant definition for the past 30 years. Unfortunately, infrequent bowel movements only capture a small percentage of individuals with constipation. A population-based survey of young adults done by Sandier and Drossman indicated that patients consider other symptoms to also represent constipation. Of those who completed the survey, 52% believed that constipation meant straining to pass stool, whereas 44% believed that constipation was present if the stools were hard. Other definitions included the inability to defecate when desired (34%) and abdominal discomfort associated with defecation. Infrequent bowel movements were considered constipation by only 32% of survey respondents. Thus, there appears to be a divergence between physicians’ and patients’ perspectives as to what constitutes constipation.
These differences, as well as the recognition that there are additional components to constipation, are the main reasons for the development of the Rome criteria for defining constipation. The initial intent of the Rome criteria for functional bowel diseases was to provide a consistent method of identifying individuals to facilitate enrollment of comparable patients into clinical trials. However, these criteria have increasingly been applied to clinical practice to better identify patients with constipation. The original Rome criteria have been updated, and now the Rome II criteria for functional constipation comprise two or more of the following abnormalities, which must be present at least 3 months during the previous year: less than three bowel movements per week, hard or lumpy stool, straining, a sensation of incomplete evacuation, a feeling of anorectal obstruction, or the need for manual maneuvers (digital disimpaction). Additionally, criteria for irritable bowel syndrome (IBS) must not be present; specifically, significant abdominal pain or discomfort must not be a primary complaint.
Even when using the Rome II criteria to define constipation in epidemiologic studies, many patients with this condition may not be captured. Defining constipation as hard stools, straining with defecation, or even infrequent defecation may be inadequate because individuals often complain of being constipated even though they do not have any of the established symptoms. Many individuals feel constipated based solely on the perception that their own expectations for bowel habits are not being met. This often manifests as a complaint that an individual is not having a daily bowel movement. Regardless of whether one meets the accepted definition of constipation, they are likely to seek treatment either by use of OTC products or by visiting a physician and requesting recommendations for therapy. Since constipation is a symptom-based disorder, patients’ perceptions that they are constipated make them constipated whether or not they demonstrate objective criteria to support the diagnosis.
Although this definition may seem to rest on a discussion of semantics, it is important when studying the epidemiology of constipation. The most reliable data sources for elucidation of the epidemiology of various diseases are population-based data. These sources, particularly the large databases, typically rely on International Classification of Diseases (ICD-9) diagnosis codes that ultimately are based on physician coding. If a physician’s perception is that constipation is present only when defecation frequency is less than three per week, the observed prevalence rates may underestimate the true prevalence of this condition. Even when epidemiologic studies utilize the Rome criteria to define constipation, they may underestimate the true prevalence of constipation. This possibility is supported by the findings of a recent systematic review of constipation where the prevalence of self-reported constipation was consistently higher than that defined by either the Rome I or Rome II criteria. On the other hand, relying solely on individuals’ perceptions that their bowel habits are not normal may lead to an overestimation of the true prevalence of constipation.
Nevertheless, reliable criteria are necessary to define constipation in order to examine its epidemiology in an organized manner. There are a number of methods to define constipation, and each definition has its own strengths and limitations. There is not a single best method to define constipation. For purposes of interpretation and application of the findings of epidemiologic studies of constipation, however, it is important to know the definition upon which the results were based. This will enable practitioners’ to apply the results to their own patient population. For this discussion, the epidemiologic patterns of constipation will be discussed in the context of the specific definition upon which they were based.
Clinical Presentation and Natural History
Constipation may be associated with a number of different diseases or conditions. The known etiologies of constipation include mechanical obstruction, metabolic disturbances, neurologic disorders, and medication side effects. A large proportion of patients with constipation do not have a known cause and suffer from idiopathic constipation.
The symptoms of constipation may start at any time, although they typically begin later in life. Constipation usually commences insidiously without any obvious inciting event. Early in its course, infrequent or difficult defecation may represent the only symptom. As constipation progresses in severity, patients usually develop bloating and crampy abdominal discomfort that may be worse after meals. The pain is often constant, located in the lower abdomen and is not generally relieved with defecation, which distinguishes chronic constipation from irritable bowel syndrome. The pain is also generally less severe than that seen in patients with IBS. Those who have suffered with constipation for many years may additionally note fatigue, malaise, anorexia, or other constitutional symptoms. The specific constipation symptoms vary according to the type of constipation. Slow-transit constipation, for example, is often associated with infrequent defecation, bloating, fatigue, and malaise. By contrast, disordered defecation commonly presents with hard stools, straining, rectal pressure, and feelings of incomplete evacuation.
There has been little if any study of the natural history of constipation. Consequently, the course of constipation remains unclear. It would appear that constipation represents a slowly progressive disorder that rarely if ever resolves. This is partly supported by Talley and colleagues, who estimated the stability of the symptoms of constipation among residents of Olmstead County, Minnesota. They compared the results of two surveys completed 15 months apart, finding that 89% of the population surveyed had no change in their symptoms of constipation during the intervening months. Furthermore, the typical clinical presentation of a patient with severe constipation is of one who has been self-medicating for years with OTC laxatives, often requiring increasing doses to achieve consistent relief.
Despite its apparent progressive course, however, constipation rarely leads to severe morbidity. Hospitalization for constipation is uncommon and mortality from constipation is quite rare. Potential complications of chronic constipation include fecal impaction, fecal incontinence, sigmoid volvulus, and stercoral ulcerations of the sigmoid colon or rectum.
It is difficult to estimate the incidence of this common disorder because of the widespread availability of OTC therapies. The overwhelming majority of patients self-medicate when they initially develop symptoms of constipation, making it difficult to capture these individuals at the time they initially develop constipation. Since incidence defines the number of new cases per specified time period, it is essential to determine the frequency and time frame of new cases. Two studies have provided estimates of the incidence of constipation. Talley and colleagues observed onset rates of 40/1000 person-years when resurveying white residents of Olmsted County a median of 15 months after an initial survey of the same population. The corresponding constipation symptom disappearance rate was 309/1000 person-years. A study by Everhart et al showed that over a 10-year period between the first National Health and Nutrition Examination Surrey (NHANES I) and National Health and Nutrition Evaluation Survey Epidemiologic Follow-up Study (NHEFS), there was a 27.3% increase in the number of patients self-reporting constipation.
A third study examined the incidence of constipation among nursing-home patients. Although this investigation may not provide insight into the incidence of constipation in the general population, it does provide interesting comparative data because nursing home patients represent a high-risk population. In this retrospective database study of nursing home residents, the incidence of constipation after admission to a nursing-home facility was estimated to be 7% in the first 3 months. This correlates to an incidence rate of 280/1000 person years, sevenfold higher than that seen in ambulatory Olmstead county residents. While the population studied was biased toward a high-risk population, such as the elderly, immobile, and disabled, the investigators captured data on 21,012 Medicare and Medicaid beneficiaries to obtain this estimate of incidence.
Table: Prevalence of constipation by gender
|Study||Population source||Year||Constipation definition||Overall prevalence (%)||Males (%)||Females (%)|
|Talley||Olmsted county||1991||Strain and hard Or <3/wk||17.4||13.9||20.8|
|Rome 1 FC||19.2||18.3||20.1|
|Rome 1 DD||11.0||5.2||16.5|
|Drossman||Householder||1993||Rome 1 FC||3.6||2.4||4.8|
|Rome 1 DD||13.8||11.5||16.0|
|Stewart||U.S. EPOC||1997||Rome II FC||14.7||12.0||16.0|
DD, Disordered defecation; EPOC, Epidemiology of Constipation (study); FC, Functional constipation; NHANES, National Health and Nutrition Examination Survey; NHIS, National Health Interview Survey.
The prevalence of constipation ranges from 3.4% to 27.2% depending on the definition of constipation utilized (Table: Prevalence of constipation by gender). This wide range is the result of both differing case definitions of constipation and the effects of varied ascertainment methods. A recent systematic review of all population-based epidemiologic studies of constipation in North America identified 10 different studies demonstrating an average preva lence rate of 14.8%. In these studies, differing case definitions of constipation were employed, including self-reported constipation and answers to questions defining Rome I or Rome II criteria for constipation. Interestingly, one of the studies in this systematic review compared the three different definitions of constipation among the same individuals, finding that self-report led to the highest prevalence of constipation (27.2%), whereas Rome I (14.9%) and Rome II criteria (16.7%) provided similar prevalence rates. However, the overall average prevalence of constipation among the 10 studies was quite similar to that identified by Rome I or Rome II.
Although population-based surveys tend to be the most reliable, case ascertainment is frequently based on ICD-9 coding, which provides the opportunity for significant variability in case definition. A number of regional studies have been performed that benefit from the ability to more precisely define constipation. These studies even allow analysis of the epidemiology of the different subsets of constipation such as normal-transit (functional) and slow-transit constipation or disordered defecation constipation.
Talley et al performed two mail surveys of white adults between the ages of 30 and 64 residing in Olmsted County, observing prevalence rates ranging from 11% for cases defined by Rome I outlet obstruction to 19% for cases defined by Rome I functional constipation. In the same manner, a meta-analysis of 30 regional epidemiologic studies revealed a broad range of prevalence rates, again depending on the specific definition of constipation employed. Prevalence rates ranged from 1.4% for infrequent defecation (less than three bowel movements weekly) to 16.9% for straining.
Based on an analysis of all the published population-based data, the overall prevalence of constipation is approximately 15% corresponding to 42 million people in the United States alone.
The relationship between age and constipation prevalence has been evaluated in numerous studies. Unfortunately, most of these studies have divided age groups differently, with some being much more aggressive in subdividing elderly subjects by age. In general, however, constipation demonstrates a progressive increase in prevalence with increasing age. Harari et al and Johanson et al, observed a trend toward increased constipation with increasing age in the National Health Interview Survey (NHIS) data, as did Sandier et al in the NHANES I data. This same age distribution is seen in other large databases. Physician visits data such as the National Ambulatory Medical Care Survey (NAMCS) and the National Disease and Therapeutic Index (NDTI) as well as hospitalization data sources such as the National Hospital Discharge Survey (NHDS) likewise demonstrate a clear and progressive increase in constipation associated with increasing age.
The majority of epidemiologic studies that have examined the prevalence of constipation by gender report a higher prevalence of constipation in females (Table: Prevalence of constipation by gender), with female-to-male ratios ranging from 1.01:1 to 3.77:1. This was true across a range of case criteria, although the higher ratios were typically observed in the studies that used self-reported constipation (average 2.65) rather than Rome criteria (average 1.75).
The prevalence of constipation is higher among non-Caucasian populations with nonwhite-to-white ratios ranging from 1.13 to 2.89. Self-report again generated the highest ratio, and Rome II criteria the smallest. The prevalence of constipation among different racial or ethnic groups is more difficult to identify. Nonwhite racial groups are not typically broken down any further for analysis because of small numbers of nonwhite participants among population-based North American studies. Moreover, it is difficult to compare prevalence rates of constipation among different countries to examine the influence of race or ethnicity since the definitions of constipation can vary significantly.
The influence of socioeconomic status on the prevalence of constipation also appears to be constant among published studies. Although the specific breakdown by income groups was different across studies, subjects with lower incomes consistently demonstrated significantly higher rates of constipation than their wealthier counterparts. This effect was usually less dramatic in studies defining constipation using the Rome criteria.
An inverse correlation of years of education with prevalence of constipation has also been reported. There appears to be a trend toward increased prevalence with less education in the NHANES I data. A similar trend toward increasing self-reported constipation with less education was seen in the NHIS data set by Johanson. An association of constipation with lower education is less consistent in several other studies, particularly among those defining constipation using the Rome criteria. This finding, therefore, may simply represent a surrogate marker for socioeconomic status.
Constipation demonstrates a distinctive geographic distribution. Analysis of the prevalence of self-reported constipation by region in the United States reveals that constipation is more common in the South and Midwest. A more refined analysis of Medicare data examining the prevalence of constipation by individual states revealed that constipation was more common in rural states, northern or mountainous states, and poorer states. The latter finding is not surprising given the socioeconomic distribution of constipation observed in other epidemiologic studies. The uniformity of findings among the different studies serves to support the validity of the observed geographic distribution.
This unique geographic pattern of constipation seems to suggest the influence of three global environmental factors: rural living, colder temperatures, and lower socioeconomic status. How these factors influence the development of constipation remains speculative. However, these factors likely act through effects on the diet. Conceivably, poorer individuals living in colder climates may consume less fresh fruits and vegetables related to diminished availability or increased cost. It may be hypothesized, then, that the absence of fresh fruits and vegetables may play an important role in the development of constipation.
Patients Seeking Health Care
A number of studies have utilized health care databases to examine various aspects of the epidemiology of constipation. Although these databases have been used at times to provide estimates of the prevalence of constipation, they are not true population-based data sources. These data sources may be biased and underestimate the true prevalence of constipation because entrance is dependent on health care seeking. Since a large proportion of patients with self-reported constipation self-medicate rather than seek health care for their constipation, they would never be included in a health care database. Nevertheless, studies from individuals seeking health care are still helpful in examining the demographic distributions of constipation because there are not likely to be any systematic differences among individuals who self-medicate and those who seek medical attention for their symptoms of constipation. Referral population studies are beneficial in corroborating the demographic patterns of constipation observed in population-based prevalence data. For example, the identification of similar male-to-female ratios of constipation in physician visit data serves to substantiate the validity of similar findings in other data sets. Another benefit of studies of referral populations is their ability to evaluate large numbers (millions) of subjects and generate hypotheses regarding causality.
Utilizing data from the NHIS, the NHDS, and the NDTI, Johanson and colleagues found associations between constipation and increasing age, female gender, low income, and decreased education. Although these databases were not specifically population based, the demographic distributions identified by analysis of these databases were analogous to those observed in population-based studies, adding support to the validity of these distributions.
Patients with Concomitant Disease
Constipation occurs commonly among patients with other diseases. In many instances, these conditions are the actual cause of an individual’s constipation. For example, hypothyroidism is well known to cause constipation. In other cases the associations may be coincidence or may be the result of shared etiologic risk factors. To investigate this possibility, Johanson et al examined the Health Care Financing Administration (HCFA) database comprising 11 million Medicare beneficiaries to assess the association of constipation with other coexisting diseases. Not surprisingly, a number of recognized causes of constipation were found to be strongly associated with constipation including laxative abuse [odds ratio (OR) 18.8], Hirschsprung’s disease (OR 6.5), intestinal obstruction (OR 6.3), and hypothyroidism (OR 1.6). These findings served to corroborate the validity of the results.
Subsequent analysis revealed the largest group of conditions associated with constipation was neurologic and psychiatric disorders. A number of dramatic associations between constipation and neuropsychiatric and spinal diagnoses were observed, including herpes zoster (OR 5.1), depression (OR 6.5), multiple sclerosis (OR 3.9), Parkinson’s disease (OR 3.2), vertebral column fracture (OR 10.1), and sprains and strains of the sacroiliac (OR 7.7) region. These associations suggest a potential link between central nervous system (CNS) function and constipation. Of particular interest was the strong association between herpes zoster and constipation. The zoster virus resides in the posterior root ganglia and can damage the ganglionic or spinal neurons. This association hints at a possible viral contribution to the onset of constipation among some patients with idiopathic constipation.
The prevalence of constipation among populations of patients with specific neurologic disease has also been studied. Hinds et al observed a constipation prevalence rate of 43%, among 280 outpatients with multiple sclerosis regardless of the severity of their disability. Han et al examined 72 spinal cord injury patients and found that 31% demonstrated severe constipation and 24% had difficulty evacuating their stool. Han et al further found that anal massage was used by 35%, abdominal massage by 29%, and manual digitation by 18% to assist bowel movements. De Looze et al also studied spinal cord injury patients. They found that 58% had constipation and required manual maneuvers or laxatives to facilitate successful defecation. Both quadriplegia, as compared to paraplegia, and anticholinergic medication usage significantly increased the risk of constipation, while the presence of intact rectal sensation did not reduce the risk. Looking at the association between constipation and neurologic disease from another perspective, a population-based study of a cohort of 51- to 75-year-old men in Hawaii found that less than one bowel movement per day predicted future onset of Parkinson’s disease with an OR of 2.7 at 24 years of follow-up, with an average onset at 12 years after the initial assessment.
Finally, a recent study of developmentally delayed individuals demonstrated a high prevalence of constipation among this population as well. Bohmer and colleagues studied the bowel habits and laxative use patterns of a random population of 215 individuals with IQs less than 50. The authors defined constipation as less than three bowel movements per week or the need to use laxatives more than three times per week. Even with this relatively strict definition of constipation, they found that 70% of these individuals were constipated. When compared to a control group, the constipated patients were more likely to have cerebral palsy, be nonambu-latory, use anticonvulsants, and have an IQ less than 35. This study further supports the possibility of a shared risk factor between individual neurologic diseases and constipation.
Constipation is also widely believed to be associated with diabetes mellitus. This association has been observed in studies by Talley’s group in Australia, Lithner’s group in Sweden, and Enck et al in Germany. However, the only population-based study specifically examining the prevalence of gastrointestinal symptoms among diabetics was performed by Locke’s group at the Mayo Clinic. They surveyed random samples of Olmsted County residents with type 1 and type 2 diabetes as well as two age- and gender-stratified control groups without diabetes. They did not observe any difference in the prevalence of constipation among patients with type 1 (12% vs. 14%) or type 2 (10% vs. 12%) diabetes, when compared with controls. There was a trend toward constipation and/or laxative use being more common among individuals with type 1 diabetes particularly among men (p < .15), but this difference was observed only in those individuals using calcium channel Mockers, which are known to cause constipation as a side effect of the medication.
Pappagallo surveyed 76 opioid-using pain clinic patients with chronic noncancer pain and found that 40% had less than three stools per week, despite 80% of them taking medications to prevent constipation. Leroi et al interviewed 344 Canadian gastrointestinal (GI) clinic patients with functional lower GI disorders, and found that 40% reported a history of sexual abuse, compared to 10% of those being seen for organic disorders. Of those who reported a history of sexual abuse, 88% reported constipation. A more recent study by Hobbis et al used a matched case-control design to find that in two sets of age- and gender-matched controls (Crohn’s disease patient controls and healthy controls) there were no differences in the rates of abuse between patients with functional bowel disorders and controls.
Thus, constipation appears to be strongly associated with a number of neurologic diseases, with opioid use, and possibly with a history of sexual abuse, although the association between constipation and sexual abuse remains to be confirmed.
The epidemiology of constipation demonstrates a consistent pattern with several key points. No true population-based incidence or natural history studies have been published to date. The prevalence of constipation ranges from 3% to 27%, averaging approximately 15%, the variance resulting from differing case definitions of constipation and the effects of varying ascertainment methods. Constipation increases progressively with age, and this increase is particularly marked after the age of 65 years. It is twice as common in females than males and is also more common in nonwhites than whites, although the distribution by race is less consistent than the distributions by age or gender. Constipation is also more common among those with lower socioeconomic status and education levels.
Constipation demonstrates a distinct geographic distribution, being more common in rural states, northern or mountainous states, and poorer states. The unique geographic pattern of constipation seems to suggest the influence of three global environmental factors: rural living, colder temperatures, and lower socioeconomic status. Finally, constipation is more common among patients with a wide variety of neurologic diseases, indicating the possibility of a shared environmental risk factor.