Defining constipation to estimate its prevalence in the community: results from a national survey

Constipation is a common condition in the community which represents a significant burden for both individuals and health care systems [1]. For the individual, constipation is associated with pain and symptoms which negatively impact quality of life [2]. From the health care system perspective, considerable costs are associated with the diagnosis and treatment of constipation [3‐7]. Since constipation is such a burden, it is important to know its prevalence but estimating the prevalence of constipation in the community can be challenging.

Three large systematic reviews including 50 epidemiological studies of community-dwelling adult populations have shown that the prevalence varies widely, with estimates for constipation ranging from 2 to 35% [8‐10]. This wide range may be in part due to differences in populations, because of various factors such as age groups, culture, diet and environment, but it may also be due to differences in the way constipation was defined in each study [8]. Although the Rome criteria have been developed for use as a standard definition of chronic constipation, most epidemiological studies have used a variety of other definitions of constipation. The various definitions used have included self-reported constipation or questions based on one or some of the symptoms as detailed in the Rome criteria.

The Rome criteria comprise a set of clinical symptoms which are internationally recognised as the gold standard in the diagnosis of chronic constipation [11, 12]. These criteria were first developed in 1994 (Rome I) and subsequently revised in 2000 (Rome II), 2006 (Rome III) and 2016 (Rome IV) [13]. Whilst useful in clinical research and pharmaceutical trials, the Rome criteria have been found cumbersome to use in clinical practice [13] and have not always been used in epidemiological research [14]. The majority of epidemiological studies have not investigated chronic constipation and have instead used simple definitions of constipation such as self-reported constipation over a defined time period, or specific symptoms, such as bowel motion frequency [8, 15].

It is not known to what extent the definition of constipation impacts prevalence estimates and whether any of these simple measures are valid alternatives to the Rome criteria for estimating the prevalence of chronic constipation. The aims of this study were to explore the impact that different definitions used to identify individuals with constipation have on population prevalence estimates and to compare the performance of simple definitions with the Rome III criteria.

This study compared the prevalence of constipation in the study population using five simple definitions with the prevalence estimated using the Rome III criteria. The sensitivity, specificity and predictive values for each of the simple definitions were determined using the Rome III criteria as the gold standard for chronic constipation and modified Rome III criteria for sub-chronic constipation.

This study found that the definition of constipation used to estimate prevalence has considerable impact on the estimates of prevalence obtained. We found six-fold differences in the prevalence of constipation estimated using different definitions. Using the Rome III criteria in a large nationally representative sample of community-dwelling adults, we estimate that one in four adults is chronically constipated. None of the simple definitions tested in this study performed well enough in terms of sensitivity, specificity, PPV and NPV to be considered as suitable proxies for definitions of chronic or sub-chronic constipation based on the Rome III criteria.

Using the Rome III criteria, we estimated the prevalence of chronic constipation in the Australian community-dwelling adult population to be 24.0%. Previous studies have estimated the prevalence of chronic constipation among Australian adults to range from 2.8 to 30.7% [16, 17, 20‐23]. Different definitions used in these studies may be the most important factor contributing to the wide range, however, differences in sample populations, sample sizes and data collection methods may also be relevant. Our survey was conducted online with a large nationally representative sample, whereas all previous Australian studies were mail surveys focussed on participants from certain geographical regions within Australia which were unlikely to be representative of the national population.

Although it is frequently difficult to achieve both high sensitivity and high specificity when testing against a gold standard [18], none of the commonly used simple definitions tested in our study adequately identified individuals with constipation compared with the Rome III criteria for either chronic or sub-chronic constipation. To some extent this might have been expected since we compared very simple definitions with a more complex definition as the gold standard but use of sensitivity, specificity, positive and negative predictive values is the appropriate method to quantify the accuracy of alternative diagnostic tests against a definitive gold standard diagnostic test [18]. We observed considerable variation in prevalence estimates with the different constipation definitions which indicates that differences in the way constipation is defined may explain the wide variation in the prevalence of constipation reported in the literature [8‐10]. It also illustrates the importance of using a suitable definition to identify individuals with constipation when conducting prevalence studies. The diversity of estimates of constipation prevalence in the Australian community further illustrates the issue of different results obtained in different studies with different definitions of constipation.

When considering the prevalence of individual symptoms, the most common symptoms were straining, hard stools and incomplete evacuation which have similarly been reported as the three most common core symptoms of constipation in other studies [24]. Only 42 to 46% of participants who were regarded as constipated (sub-chronic or chronic) by the Rome III criteria reported experiencing fewer than 3 bowel motions per week. Also, less than 20% of all participants reported fewer than 3 bowel motions per week yet almost 60% had self-reported constipation in the last 12 months. Similar findings have been reported in Japan where 28% of participants in an online survey considered themselves to be constipated, but only 8% reported a bowel motion frequency of fewer than 3 per week [25]. In our study, simple definitions for identifying constipation which used bowel motion frequency performed poorly against the Rome III criteria, demonstrating that bowel motion frequency should not be used to identify constipation, unless used as one of the symptoms of the Rome criteria.

Other research has found that estimates of prevalence using self-reported constipation are greater than prevalence figures for chronic constipation derived from Rome criteria in the same population [26‐29]. We found that self-reported constipation over 2 weeks approximated the Rome III prevalence estimate for chronic constipation. However, importantly both the sensitivity and PPV of the simple definition, “self-reported constipation in the past 2 weeks”, were low, indicating that this simple definition did not identify the same individuals with constipation as identified by the Rome III criteria. One factor to consider is that self-reported constipation is any constipation experienced during the defined time period whereas unmodified Rome criteria provide an indication of chronic constipation only. A further consideration is that with self-reported constipation, constipation is self-defined and consequently different individuals may have different perceptions of constipation [8, 30, 31]. It could be argued that this is the true definition of constipation which should be used in clinical practice but in prevalence studies of any constipation, our results suggest that defining constipation as “self-reported constipation for the last 3 months” might be considered as an alternative to other constipation definitions since it compared more favourably to the gold standards for both chronic and sub-chronic constipation than the other definitions in terms of sensitivity, specificity, PPV and NPV.

Our study showed that the period of time used in the simple definitions affected the estimated prevalence. Looking at the three definitions using self-reported constipation, as the time period increased the estimated prevalence also increased, as might be expected in any estimate of period prevalence [32]. Also, the time periods which have been used with self-reported constipation often differ to the periods specified in the Rome criteria. Many international epidemiological studies have used self-reported constipation over 12 months as the definition of any constipation but our results indicate that this may over-estimate the prevalence of constipation.

A major strength of our study is that it was conducted in a large population-based sample which was nationally representative in terms of age, gender and location of the Australian community-dwelling adult population. One limitation of our study was that although we asked if participants had been medically diagnosed with irritable bowel syndrome (IBS), we did not include questions regarding the Rome III diagnostic criteria for IBS. Consequently, our estimated prevalence of 24.0% using the Rome III criteria may include both functional constipation and constipation due to IBS. This may be one reason why our prevalence estimates for chronic and sub-chronic constipation were much higher than the most recent study in Australia where it was estimated that almost 4% of the population may experience constipation-predominant IBS [16]. Furthermore, it should be noted that, although our survey was conducted prior to publication of Rome IV criteria, the Rome IV criteria for functional constipation are essentially the same as Rome III criteria. [14]. A further limitation is the possibility that those with constipation were more likely to complete the questionnaire resulting in some selection bias which may in turn over-estimate the prevalence.

This study highlights the importance of careful selection of the definition since it has a major bearing on the estimated prevalence of constipation. All of the simple definitions included in this study are commonly used in research, yet none provided a valid alternative to the Rome III criteria which are considered to be the gold standard for diagnosis of chronic constipation, or to the modified Rome III criteria for diagnosis of sub-chronic constipation.