The needs and problems with classification of constipation subtypes
Because constipation embraces a heterogeneous group of disorders and aetiologies, sub-classification is important to minimise this heterogeneity and to facilitate epidemiological, aetiological, pathophysiological and therapeutic enquiry. Within the pathogenetic pathway (from the cellular level, via local neuromuscular dysfunction, through dysfunctional integrated mechanical responses, and ultimately to symptoms) it helps to pinpoint possible ‘levels’ and types of disease biomarkers and, thereby subclassify homogenous groups. If an aberrant physiological measurement is specific for a pathophysiological process, it may guide us to a possible therapeutic target. Of more immediate clinical importance, identification of such markers should dictate a specific investigative algorithm and, if the measure is predictive of outcome, it should dictate-specific therapy. Finally, classification should assist in assigning patient eligibility criteria and valid outcome measures, both vital for controlled clinical trials. We will examine critically the evidence supporting symptom- and measurement-based criteria to classify constipation subtypes.
Symptom-based classification of constipation subtypes
Definitions of constipation. Constipation is a common and heterogeneous disorder, which is characterised by difficult and/or, infrequent stool passage.16,17 However, a number of incompatible definitions and classifications have been applied in the literature, and a disconnect between patient and physician perceptions of what should constitute constipation have emerged.1,18–22
The American College of Gastroenterology (ACG) Chronic Constipation Task Force members have produced evidence-based recommendations on the management of chronic constipation.18 They also recommended that physicians use a simple broad definition of chronic constipation: ‘unsatisfactory defecation characterized by infrequent stools, difficult stool passage or both at least for previous 3 months. Difficult stool passage includes straining, a sense of difficulty passing stool, incomplete evacuation, hard/lumpy stools, prolonged time to stool, or need for manual manoeuvres to pass stool’. This recommendation was also based on expert opinion. They criticised the previous Rome criteria arguing these are probably too detailed for use in the clinical setting.
After publication of the ACG task force recommendations, the Rome III criteria for functional constipation were developed, again based on expert consensus.1 Functional constipation was defined solely by symptomatic features, by the presence of two or more of the following six symptoms: hard stools, straining, sensation of incomplete evacuation, sensation of anorectal blockage, the use of manual manoeuvres during evacuation, and infrequent bowel movements (<3 per week). Notably, the Rome III criteria recognised subgroups of functional constipation based on both symptom and physiological criteria. Thus, the Rome III criteria for dyssynergic defaecation was based on patients meeting the diagnostic criteria for functional constipation, and having at least two objective findings (positive balloon expulsion test or imaging, inappropriate contraction of the pelvic floor muscles or inadequate propulsive forces).23 The definition indicated that experts considered symptoms alone inadequate to identify an evacuation disorder in clinical practice.
Also of note is that the Rome III criteria do not require colonic transit studies (scintigraphic or radio-opaque markers) for a diagnosis of functional constipation, although an objective measure of slow transit could arguably increase the specificity of symptom-based criteria for diagnosis.24–27 Further, the Rome III criteria recommended that constipation-predominant irritable bowel syndrome (IBS-C) should be differentiated from functional constipation, and suggest this can be performed by applying symptom-based criteria for IBS rather than objective testing.1,18,23 However, this approach has not been formally validated. In the surgical literature, much has been written about this topic, and often IBS is considered likely in the setting of chronic symptoms of constipation and normal transit.28–30 The differentiation here again is largely based on opinion, and the importance of differentiating IBS in terms of optimising management is not established. Indeed, drugs that are efficacious in chronic (functional) constipation such as tegaserod or lubiprostone have been very similarly efficacious in IBS-C.31–34 Data such as these led the ACG task force members to suggest differentiating these entities is artificial.18
Subgroups of chronic constipation. In the absence of structural or metabolic aetiologies, chronic (or functional) constipation is generally considered to represent a disorder of colonic transit, abnormal evacuation or abnormal rectal perception based on physiological measures including colonic transit testing and anorectal manometry with balloon expulsion.1,18–20,24 Three broad (and overlapping) categories of chronic constipation are currently described: (i) normal transit constipation, (ii) slow transit constipation (STC) and (iii) EDs.18,24,25 Nyam et al.25 observed that among 1009 constipated patients who were studied at a referral centre, 59% had normal colonic transit and normal pelvic floor function, 25% had pelvic floor dysfunction, 13% had STC and 3% had STC and pelvic floor dysfunction. By contrast, Koch et al.30 determined that only 8% of 190 patients had normal colonic transit and normal pelvic floor function, whereas 59% had ‘disordered defaecation’; 27% had STC and 6% had combined disordered defaecation and STC. Why such a disparity in findings? This is almost certainly test-dependent; the Nyam et al.25 study assessed evacuatory function by anorectal manometry and showed a low yield for an ED (poor sensitivity, good specificity for dyssynergic defaecation only?), whereas the Koch et al.30 study used defaecography to evaluate evacuatory function, which images both structural and ‘functional’ obstructive ‘causes’, and demonstrated a diagnostic yield for ED twice the magnitude of the Nyam study (good sensitivity, poor specificity?). Other studies have likewise reported varying distributions of subgroups of constipation, but these studies were from referral centres and not the general population. Thus, we do not know the real distribution of subgroups based on physiologic testing in chronic constipation, as most patients do not seek heath care.
Empiric evidence for constipation subgroups. One approach that can provide empiric evidence that true subgroups of constipation exist applies the mathematical techniques of factor and/or cluster analysis, to determine whether certain symptoms and/or subjects group together more than expected by chance, suggesting the existence of true syndromes. Talley et al.35 reported that distinct clusters of gastrointestinal symptoms existed from four national populations (United State, Sweden, Australia and Germany). They showed that all four nations studied had subgroups with IBS type symptoms, and separately constipation and diarrhoea (in three of four countries). However, two studies specifically have applied factor or cluster analysis to determine if subgroups of constipation can be identified in the outpatient setting, but no such cluster could be established.36,37
Physiological testing (gold standard) vs symptoms in chronic constipation. As it is important to identify those who have slow colonic transit and those who have an ED, as this can change management, a key question is whether clinical features point clinicians in the right direction or act as surrogates for physiological testing which is not available everywhere. A number of investigators have attempted to discriminate the subgroups of constipation by applying symptom profiles, scoring system, demographics and even psychological characteristics,29,30,36–39 but arguably this has been all unsatisfactory.
Glia et al.29 studied 134 patients with chronic constipation and showed that patients with slow transit (vs normal transit) more often reported less than three bowel movements per week (84%vs 46%), constipation since childhood (58%vs 22%) and laxative dependence (87%vs 44%). Further, pelvic floor dysfunction was associated with a higher prevalence of backache (55%vs 33%) and a lower prevalence of normal stool frequency (19%vs 35%), compared with normal pelvic floor function. They concluded that symptoms may be good predictors of slower transit times but poorer predictors of pelvic floor function. Mertz et al.36 evaluated the correlation of scores derived from the factor analysis and the results of physiological tests. Symptoms of slow transit (infrequent defaecation and no urge) discriminated slow transit constipation, but there were no symptoms which identified pelvic floor physiological measures.
Grotz et al.38 studied 184 constipated patients, and observed that a feeling of incomplete evacuation was more commonly reported in patients with pelvic floor dysfunction compared to patients with normal transit constipation. In addition, a more regular defaecation pattern, taking a position other than sitting to defaecate, and feelings of incomplete evacuation were more commonly reported in constipated patients with STC than in constipated patients with normal transit. In the multivariate analysis, however, they could not identify any colonic symptoms as discrimators.
In the Koch et al.30 study of 190 patients with chronic constipation, they observed that straining was the most commonly reported symptom in patients with constipation across all physiological subgroups (82–94%). Although they found that infrequent bowel movements and bloating were associated with STC, while feelings of incomplete evacuation, a sense of anal blockage and self digitation were associated with disordered defaecation, there were no sufficiently sensitive and specific symptoms to distinguish among the subgroups.
All of the above studies in outpatients had major limitations, and it is conceivable that symptoms do identify physiological subgroups reliably but this has been missed. In particular, sample sizes have been likely too small to identify robust subgroups, and the quality and breadth of symptom assessment (a key to success) has varied.
Two scoring system38,40 have been developed to try and discriminate subgroups of constipation based on symptoms. Although they created classification functions for each subgroup, the level of discrimination for these was low, and they concluded that symptom profiling could not effectively differentiate pathophysiologic subgroups.
In conclusion, chronic constipation has multiple pathophysiologies, and currently symptom-based definitions are still considered the gold standard for diagnosis. However, symptom-based criteria cannot effectively discriminate among subgroups of chronic constipation due to substantial symptom overlap among the physiological entities.
Measurement-based classification of constipation subtypes
Why haven’t objective physiological measures solved the problem? There are many reasons why the currently adopted measurement techniques are falling short of our expectations and needs: (i) The marked heterogeneity of constipation itself. The fact that the incidence of delayed transit varies from 36% to 80% and evacuation disorders from 16% to 75%41 suggests there are, as yet, unidentified functional markers of constipation, and a need for better standardisation of tests; (ii) Suboptimal specificity of tests in separating normality from abnormality (e.g. 20% and 16% of healthy controls demonstrate a dyssynergic pattern on manometry and an abnormal balloon expulsion test respectively);42 (iii) There is marked overlap among the current subtypes of constipation. Up to 75% of patients with a proven ED have delayed colonic transit;41–43 (iv) The current ‘anatomical division’, whereby the colon is responsible for transit and the anorectum is solely responsible for evacuation, is a gross oversimplification. It is clearly recognised that there is close co-ordination between colonic contractile activity and anal sphincter relaxation,44 that pan-colonic propulsive activity is necessary in the immediate predefaecatory and often through the phase of rectal evacuation,45 and such activity is capable of emptying the entire colon.46 Further, Karlbom et al.47 found that defaecographically defined rectal evacuation was less efficient in those with pure STC compared with those with an isolated evacuation disorder or combined slow transit with evacuation disorder. An examination of ‘simulated’ defaecation, particularly in the absence of priming colonic propulsive movements and in a laboratory environment is far from a daily ritual; (v) The umbrella category of evacuation disorder comprises anatomical as well as functional abnormalities. Even these anatomical ‘abnormalities’ are relatively ubiquitous in healthy controls. For example, a rectocoele is commonly present in postpartum females, and the normal range of evacuation from the healthy rectum varies widely from 13% to 100%48 and (vi) There has been little rigorous evaluation of the utility and predictive value of these physiological tests.41
What are the desirable attributes of a measurement tool? The attributes of a good measurement tool include proven: (i) test–retest reliability (and intra- and inter-observer reliability); (ii) validity for the task; (iii) high accuracy (i.e. sensitivity and specificity) in the case of a tool purporting to be a ‘diagnostic’ test and (iv) clinical utility (i.e. influence management and predict therapeutic outcome).49 A prerequisite to assess validity and diagnostic accuracy is a gold standard comparator. If it claims to be a biomarker, it should have a histopathological correlate or an independent functional ‘signature’. For example, the failed rectoanal inhibitory reflex compared with aganglionosis is specific and sensitive for Hirschprung’s disease. Unfortunately, comparing anorectal manometry with defaecography, for example, is doomed because both have major problems with specificity (see above). In light of these constraints and the intrinsic limitations of current tests, arguably the most relevant question is whether they can predict therapeutic outcome.
Clinical utility of colonic transit measurement in predicting outcome. A critical appraisal of currents tests of colonic transit can be found in Chapter 3 of this Supplement. There are no controlled data evaluating the clinical utility of colonic transit measurement. Two uncontrolled studies found that the demonstration of pan-colonic transit delay (inertia) predicted a significantly worse outcome in response to bran supplements and laxatives when compared with normal transit constipation or distal transit delay.50,51
A critical review of 32 studies reporting outcome from colectomy, noted widely varying satisfaction rates and suggested, although imperfect that physiological testing (transit measures and anorectal function studies) may help in the rational selection of patients for surgery.52 However, the evidence that these measurements can reliably predict outcome is uncontrolled and weak. Two uncontrolled studies evaluated clinical outcome from colectomy in the context of an algorithmic investigative strategy, whereby colectomy is offered in pure STC, or in cases of combined STC and ED when biofeedback has failed.25,28 With this strategy, a favourable outcome from colectomy was reported in 85–97% of patients with no increased likelihood of a poorer outcome in those with coexistent ED and failed biofeedback. Other studies reported widely variable (50–94%), but generally favourable outcome from colectomy adopting a similar workup; again showing concurrent ED did not infer a worse outcome53–55 despite the observation that up to half of the cases operated on had a abnormal balloon expulsion test and one-third had EMG evidence of paradoxical puborectalis contraction.53 A small retrospective study of transit measures in children found that delayed transit predicted a poorer outcome in response to cecostomy and antegrade colonic lavage.56
In summary, available evidence support the following conclusions in respect of transit measures: demonstration of delayed transit does not preclude a coexistent ED (and vice versa); demonstration of distal transit delay is not an accurate predictor of an ED; response to medical therapy appears to be worse in those with total colonic delay as opposed to distal delay or normal transit constipation; the current practice of using a combination of transit measurement and anorectal function studies does select out those patients with an ED who do have a high probability of responding to biofeedback and obviates the need for surgery in this subset (see below). However, there is lack of prospective evidence to indicate that these physiological measures can predict outcome from colectomy.
Clinical utility of anorectal manometry in predicting treatment outcome. There is consistent data showing that demonstration of an ED by anorectal manometry, combined with an additional measure of rectal expulsive ability (balloon expulsion test or defaecography) predicts a high (67–89%) response rate to biofeedback.42,57–59 In addition, there are now four controlled trials confirming that demonstration of dyssynergic defaecation predicts a high probability of a favourable outcome from biofeedback.60–63 Rao et al.61 found, when compared with sham, that patients with dyssynergic defaecation undergoing biofeedback had significant improvement in overall satisfaction, stool frequency, ability to expel a rectal balloon and reversal of manometric dyssynergic features. Chiarioni et al.60 found that when compared with those with STC alone, patients with combined STC and dyssynergic defaecation had a superior clinical response rate (76%vs 8%).
Importantly, anorectal manometry also tests rectal sensitivity, with impaired sensation (particularly hyposensitivity)64 being commonly found in constipated patients. In small (uncontrolled) studies, normalisation of sensory thresholds has been demonstrated in 65–92% of responders to behavioural treatments incorporating ‘sensory biofeedback’,65,66 which was not the case with non-responders.
In summary, anorectal manometry has good test–retest reliability within a laboratory. However, there is a lack of standardisation across laboratories and inter-observer variability is unknown. The extent to which this accounts for widely varying reports of dyssynergic defaecation is unclear. High level evidence exists showing that demonstration of dyssynergic defaecation does predict a good outcome in response to biofeedback, but does not infer a poorer outcome following colectomy in those with coexistent STC.
Clinical utility of tests of evacuation in predicting clinical outcome. Clinical investigation of evacuatory ability is covered in greater detail in Chapter 4 of this Supplement. Nevertheless, it is extremely important to recognise that there is marked geographical variation with regard to which investigation of (simulated) stool expulsion is performed in the clinical work-up of the chronically constipated patient; for example, anorectal manometry and balloon expulsion are almost exclusively favoured in the US, whereas defaecography/evacuation proctography are more commonly performed in Europe. As alluded to above, it is also vital to acknowledge that diagnosis of a particular cause of ED is crucially test-dependent. Manometry and the balloon expulsion test will identify dyssynergic defaecation in around half of patients tested,67 with the remainder classified as having normal evacuation. These tests do not consider a ‘structural’ obstruction to defaecation (e.g. large rectocoele or occluding intussusception). Conversely, only one-quarter of patients are diagnosed as having dyssynergic defaecation on defaecography,67 with only around one-fifth having a (seemingly) normal test.29,68 The majority are considered as having an anatomical or structural impediment to evacuation, although the true clinical impact of these ‘anomalies’ remains unclear. Another issue that demands recognition is that parameters of rectal emptying may be entirely normal, but the presence of a structural abnormality (e.g. large rectal intussusception impinging on the upper anal canal) may still be consistent with symptoms of constipation (e.g. incomplete evacuation, anorectal obstruction, need to digitate, etc.).
Like all other tests, defaecography lacks standardisation, and ‘abnormalities’ are not consistently defined. There is a marked overlap between patients and controls with regard to the presence of structural abnormalities, and normative data for efficiency and completeness of emptying are lacking. In terms of predicting outcomes, ‘repair’ of obstructive features does not always translate to symptomatic relief (see Chapter 7 of this Supplement). Indeed, it has been proposed that such anatomical anomalies simply represent the ‘tip of the iceberg’ in a condition where pathophysiologies are multiple and multidimensional.69
Clinical utility of colonic manometry in predicting clinical outcome. This technique is covered in more detail in Chapters 2 and 3 of this Supplement. Briefly, there is currently no standardisation of colonic manometric techniques, nor terminology, nor clearly agreed minimum duration of study adequate for clinical decision making. A complete absence of HAPCs, an absent meal response or absent response to Bisacodyl (less likely) are emerging as potentially relevant predictors of severe colonic neuromuscular dysfunction. Although there is good evidence in the paediatric setting, it remains unproven in adults. On the basis of current evidence, manometric findings have not been correlated with histopathological findings and therefore stratification into neuropathic and myopathic categories on the basis of manometry remains speculative.