Aliment Pharmacol Ther 31, 938–949
Background Comparison of quality of life (QoL) across disease areas requires the use of appropriate tools. Although many studies have investigated QoL in constipation, most used disease-specific tools that are inappropriate for cross-comparisons.
Aims To identify studies of QoL in constipation and to compare these results with other chronic conditions.
Methods A comprehensive literature search identified studies in constipation that used a generic QoL tool. Results were statistically pooled where possible and compared with published results using the same tools in other chronic conditions.
Results A total of 13 qualifying studies were identified, 10 in adults and three in children. Results from eight studies using the SF-36/12 tools were pooled; the remaining five were narratively reported. Mental and physical components of QoL scores were consistently impaired in both adult and child populations, with the greatest impact being seen in secondary care studies. Mental health effects predominated over physical domains. The magnitude of impact was comparable with that seen in patients with allergies, musculoskeletal conditions and inflammatory bowel disease.
Conclusions The impact of constipation on QoL is significant and comparable with other common chronic conditions. Improving management may prove to be an effective way of improving QoL for a substantial number of patients.
Chronic functional constipation is a common condition, with around 15–17% of adults1, 2 reporting symptoms consistent with the Rome I or II diagnostic criteria. Studies in children yield considerable variation in prevalence estimates, with a range of estimates from 1% to 30% (median 10.4%) being identified in the literature.3 Although rarely associated with life-threatening complications, the impact of constipation on sufferers may be considerable: a recent review4 identified a growing evidence base that these patients have a significantly impaired health-related quality of life (HRQoL) compared with unaffected populations, as assessed by objective questionnaires.
Measurement of HRQoL requires the application of an objective and reproducible series of measures to characterize physical, mental, social and functional aspects of an individual’s life. The information this yields may be used to inform individual patient management5 or, more commonly, to provide insight into the typical impact of one or more related conditions within a defined group of patients.6 More challenging to achieve, but potentially more useful, is the comparison of HRQoL across several unrelated conditions.7 This process is not only clinically valuable but also increasingly emphasizes the importance of the decision-making process for resource allocations within health care systems.
Meaningful comparison of HRQoL across differing disease areas requires careful selection of the assessment tool to be used. Many of the questionnaires used in QoL research are specific to individual diseases, or groups of diseases. For example, when considering constipation, one may choose to investigate using a very specific tool, such as the Patient Assessment of Constipation – Quality of Life questionnaire,8 which has been shown to be internally consistent, reproducible, valid and responsive to improvements over time. This makes the tool especially valuable for tracking individual patients longitudinally, but of limited value when comparing a group of patients with constipation with a similar group with irritable bowel syndrome. In this circumstance, a broader tool such as the Elderly Bowel Symptom Questionnaire would be more appropriate.9 However, if the intention is to compare the impact of constipation with a nongastrointestinal problem, then neither of these approaches would be helpful. In this circumstance, a generic HRQoL tool is required – some of the most commonly used being the Short Form 36 version 2 (SF-36v2),10 the Health Utilities Index mark 3 (HUI3)11 or the EuroQol 5D (EQ-5D).12
These generic tools are relatively insensitive to specific clinical issues in any given medical condition and are therefore not ideal for following up individual patients. Their strength, however, lies in their ability to detect the global impact of illness and/or disability on a multidimensional construct of psychological, social and physical aspects of QoL. For this reason, they are ideally placed to give meaning to questions on the relative impact of disparate diseases on HRQoL.
The aim of this review was to identify published studies that used generic HRQoL tools in the field of adult or child constipation; to extract data allowing a rational assessment of the impact of constipation on QoL; to pool these results where appropriate and to carry out a narrative review otherwise; and to compare these results with those arrived at using the same tools in other chronic, nonlife-threatening conditions.
A primary search for QoL studies was carried out in PubMed, using the following terms:
Laxatives [MeSH] OR Constipation [MeSH]
Quality-of-life [MeSH] OR HRQL [TW] OR HRQOL [TW]
Additional electronic searches using similar strategies were carried out using EMBASE and Cochrane Library.
All abstracts that appeared likely to comply with the inclusion criteria were obtained as full text, together with relevant review articles. Reference lists of all sourced papers were scrutinized to identify any missed studies.
Studies were appraised to determine whether they complied with the following inclusion criteria:
- (i) study applying a generic HRQoL tool to patients with constipation;
- (ii) patients with diagnosis of constipation consistent with Rome II/III criteria;
- (iii) detailed results of HRQoL scores presented in the paper;
- (iv) reference to healthy comparator group scores in the paper.
Studies were specifically excluded if:
- (i) only disease-specific assessment tools were used;
- (ii) constipation was secondary to other diagnoses;
- (iii) the study had not been fully published in a peer-reviewed journal.
No limitations were placed on language of publication or age of participants.
Data extraction and analysis
In studies which used the SF-36 or SF-36v2 tools, mean scores for each of the eight domains were extracted (see Table 1). These were transformed into norm-based scores, using appropriate national reference values.13–16 Norm-based scoring standardizes the raw results for each domain to a consistent scale with a mean of 50 and standard deviation of 10 when applied to a healthy population. This allows cross-comparisons between domains.13 These eight results were then aggregated using published weighting factors to yield two composite scores for each study – the Physical Component Score (PCS) and the Mental Component Score (MCS). This facilitates the comparison of results across disease entities.13 In addition to calculating these values for each study and patient group individually, we calculated a mean value for all studies combined. This mean value included results from one study carried out using the SF-12. This comprises a subset of the SF-36 that yields the same two composite measures (PCS + MCS). This is scored using the same normalizing metric as the SF-36, which therefore allows meaningful inclusion of these results into the pooled estimates of MCS and PCS.13
|Study||Study population||Control group||HRQoL tool used||N (constipated)||N (control)|
|Wald et al.17||Multinational population sample aged 16+ with constipation on questionnaire||Multinational population sample aged 16+ with no constipation on questionnaire||SF-36||1435||1435|
|Tuteja et al.18||Workforce survey aged 24–77 with constipation on questionnaire||Workforce survey aged 24–77 with no constipation on questionnaire||SF-36||140||583|
|Rao et al.19||Patients aged 19+ referred to hospital clinic with slow transit constipation||Healthy control group with no bowel symptoms on questionnaire||SF-36||38||44|
|Patients aged 19+ referred to hospital clinic with dyssynergic defecation||76|
|Irvine et al.20||National population sample aged 18+ with constipation on questionnaire||National population sample aged 18+ with no constipation on questionnaire||SF-36||163||705|
|Chan et al.21||Patients aged 18+ referred to hospital for assessment of constipation||Healthy volunteers recruited through advertisement||SF-36||80||18|
|O’Keefe et al.22||Local population sample aged 65+ with constipation on questionnaire||Local population sample aged 65+ with no GI symptoms on questionnaire||SF-36||126||173|
|Mason et al.23||Women aged 21–67 with constipation attending hospital clinic for biofeedback training||Sample of UK age-matched population||SF-36||22||298|
|Koloski et al.6||Local population sample aged 18+ with constipation on questionnaire||Local population sample aged 18+ with no GI disorders on questionnaire||SF-12||227||2683|
|Glia et al.24||Patients aged 17+ referred to hospital clinic with constipation||Sample of Swedish population||PGWB||84||4624|
|Bengtsson et al.25||Women aged 18–65 recruited for a clinical trial: constipated and treated with Na Picosulphate||Sample of sex-matched Swedish population||PGWB||35||2300|
|Women aged 18–65 recruited for a clinical trial: constipated and treated with other laxatives||51|
|Youssef et al.26||Patients aged 5–18 referred to hospital clinic with constipation||Healthy controls recruited through community health clinics||PedsQL||80||46|
|Clarke et al.27||Patients aged 8–18 referred to hospital clinic with constipation||Healthy controls recruited at Scout jamboree||PedsQL||51||79|
|Faleiros et al.28||Patients aged 5–12 referred to hospital clinic with constipation||Sample of Brazilian age-matched population||CHQ-PF50||57||314|
In addition to the results for the total dataset, pooled values were separately estimated for community- and hospital-based studies.
As most studies did not supply figures for standard deviations, we were unable to undertake a formal random effects meta-analysis and these mean estimates were simply derived by study-size weighting. Equally, the absence of variance data from individual studies precluded the significance testing for differences in aggregate PCS and MCS scores for constipated and healthy populations.
For studies that did not use SF-36, data on comparative scores for each patient group were extracted from the text and used to inform a narrative review.
In all, 10 studies were carried out in adult populations: 7 using SF-36,17–23 1 using SF-126 and two using the Psychological General Well Being index (PGWB).24, 25 None used the EQ-5D or HUI3. Three studies were carried out in children, two using the Pediatric Quality of Life Inventory (PedsQL)26, 27 and one using the Child Health Questionnaire – Parent Form 50 (CHQ-PF50)28 (Table 1).
Studies using SF-36/SF-12
The studies using SF-3617–23 yielded scores for each of the eight domains tested that were, with few exceptions, significantly lower in individuals with constipation than in the unaffected comparator groups. When normalized to the relevant reference population and pooled to arrive at a weighted mean for each, it is apparent that the greatest differences are seen in the domains: General Health, Social Functioning and Mental Health (Figure 2).
This result is reflected in the pooled results for the PCS and MCS, which demonstrate greater absolute impact on the domains reflecting mental health status than those connected with physical aspects of QoL (Figure 3).
Mean normalized scores for constipated patients: PCS = 47.5, MCS = 45.8.
Mean normalized scores for healthy controls: PCS = 51.3, MCS = 48.8.
Nominal population mean normalized scores: PCS = 49.97, MCS = 49.90.
Mean normalized scores for constipated patients: PCS = 48.0, MCS = 48.5.
Mean normalized scores for healthy controls: PCS = 51.3, MCS = 49.2.
Mean normalized scores for constipated patients: PCS = 41.9, MCS = 43.9.
Mean normalized scores for healthy controls: PCS = 50.5, MCS = 48.4.
Studies using PGWB
Two studies carried out in Swedish populations used the PGWP.24, 25 This index evaluates HRQoL across six domains – anxiety, depressed mood, positive well-being, self control, general health and vitality – which are then combined to yield an overall score.
In the first study,24 102 consecutive adult patients who had been referred to a hospital clinic for evaluation of severe chronic constipation underwent assessment with the PGWP questionnaire. Data for 84 patients were available. This was a mixed group of patients who were classified according to both faecal transit time and stool frequency. The overall mean score achieved was 85.5 compared with an average score for the Swedish population of 102.9 (maximum possible score 110). Somewhat surprisingly, patients with slow-transit constipation scored significantly better than those with normal transit constipation (median 94 vs. 82), whereas those with fewer than three stools per week scored higher than those with more (median 96 vs. 80).
In the second study,25 86 women with chronic constipation complying with Rome II criteria were recruited for a clinical trial of laxative treatment. Baseline results for PGWB are presented, broken down by the type of treatment the women were using prior to randomization. Amongst 35 women taking sodium picosulphate at baseline, mean score was 92.8 compared with a female average of 101.4. Amongst 51 women using other treatments, the mean score was 85.2.
The PedsQL is a generic questionnaire that is completed by both children and their parent and has been validated in patients aged 5 and over. It encompasses physical, emotional, social and school functioning domains, which are then aggregated to yield an overall score ranging from 0 to 100. Parental and child scores are separately recorded.
In the first study,26 carried out in the United States, 178 children (age 5–18) referred to a paediatric gastroenterology department and 42 healthy controls attending primary care for routine checks or minor problems were screened using the PedsQL. In the patient group, 80 had constipation, 42 inflammatory bowel disease (IBD) and 56 gastro-oesophageal reflux (GERD). The mean aggregate score for constipation was significantly reduced compared with healthy controls (children 70.4 vs. 87.7; P < 0.05; parents 60.6 vs. 80.7; P < 0.001) (Figure 4). Both children and parents scored constipation as having a greater impact on QoL than either GERD (children 70.4 vs. 79.9; P < 0.05; parents 60.6 vs. 76.6; P < 0.05) or IBD (children 70.4 vs. 83.8; P < 0.05; parents 60.6 vs. 77.4; P < 0.05). Amongst children, the impact on QoL was evenly distributed throughout all domains, whereas parents identified school and emotional issues as being the major drivers. Parents consistently rated the impact on QoL as being greater than their children’s own ratings.
In the second study,27 carried out in Australia, PedsQL scores for 51 children (aged 8–18) attending surgical and gastrointestinal clinics in for chronic slow transit constipation were compared with 79 healthy controls recruited from a Scout jamboree. Total QoL scores in children with constipation were significantly lower than those of controls (children 72.9 vs. 86.0; P < 0.0001; parents 64.4 vs. 84.3; P < 0.0001). Although decreases in both physical and psychosocial domains were reported, psychosocial factors were impacted to a greater extent. As in the previous study, mean parental scores were lower than those reported by children for the constipated group, although no such difference was observed in the control group.
The reported results of this study correlate well with those of the other study using PedsQL, both in magnitude and trend.
Mean child score: 72.9 vs. 70.4 in Youssef et al.26
Mean parent score: 64.4 vs. 60.6 in Youssef et al.26
Mean child score: 86.0 vs. 87.7 in Youssef et al.26
Mean parent score: 84.3 vs. 80.7 in Youssef et al.26
The third paediatric study28 evaluated 100 consecutive children attending a gastroenterology clinic in Brazil with functional defecation disorders. Using the Rome II criteria, 57 children had a diagnosis of functional constipation, 29 had functional faecal retention and 14 had nonretentive faecal incontinence. Scores on the CHQ-PF50 were compared with documented norms for healthy children in Brazil. The CHQ-PF50 is a 50-item questionnaire completed by the parents of the affected child. It assesses 15 separate domains, which are then aggregated into physical and psychosocial composite scores.
Mean scores amongst all children for any functional defecation disorder were 26.3 for the physical component and 36.0 for the psychological domains. This compares with healthy control values for Brazilian children of 55.0 and 53.0 respectively. The difference is quoted as statistically significant, although no P-value is given. Subanalysis across the three diagnostic subgroups showed no significant differences, although the physical component for children with nonretentive faecal incontinence was numerically lower than for the other diagnoses.
Our literature search identified an extensive evidence base relating to the QoL impact of chronic constipation using a range of validated questionnaires. The results of these studies demonstrated a consistent effect of constipation on both mental and physical components of QoL. Amongst population recruited within secondary care, the magnitude of QoL impairment was substantially greater across all domains than that seen in community-recruited cohorts, reflecting the more intractable nature of the problem in these patients. This difference was most marked in the mental and emotional aspect of the scores.
The generic tools identified in these studies have also been used in other disease areas. The SF36 has been extensively used. Table 2 compares the results of our analysis with those from both a US community reference population with a range of chronic diseases and results of individual hospital-based studies.29–33 The scores found in community studies were comparable to those seen with chronic allergies, dermatitis, diabetes and stable ulcerative colitis, whilst the scores seen in hospital studies were similar to those found in patients with unstable IBD, functional dyspepsia and a range of chronic rheumatological conditions. Similarly, the PGWB studies yielded scores at least as severe as those seen in untreated peptic ulcer,34 GERD35 and patients with mild asthma.36
|Healthy individuals only||Community||1300||55.4|
|Ulcerative colitis (stable)||Hospital||193||49.7|
|Ulcerative colitis (unstable)||Hospital||88||47.4|
|Crohn’s disease (stable)||Hospital||114||45.0|
|Crohn’s disease (unstable)||Hospital||84||41.2|
|Coronary heart disease||Hospital||186||40.1|
|Healthy individuals only||Community||1300||52.9|
|Ulcerative colitis (stable)||Hospital||193||48.7|
|Coronary heart disease||Hospital||186||47.8|
|Crohn’s disease (stable)||Hospital||114||47.1|
|Ulcerative colitis (unstable)||Hospital||88||42.0|
|Crohn’s disease (unstable)||Hospital||84||40.3|
As reported above, in one of the paediatric studies, the scores achieved in the PedsQL were worse for children with constipation than those with GERD and IBD.26 The paediatric data are of interest in that they provide a measure of the impact of constipation on the patients’ families. Constipated children report a modest impairment in mean QoL scores (0.80–0.85 vs. controls). The parents of these children however, report a more marked impact (0.75–0.76 vs. controls). The results are consistent with those seen in a wide range of other chronic diseases affecting children, both in terms of the magnitude of impact and the differential between adult and child assessment5, 37, 38 (Table 3).
|Healthy controls5, 26, 27||522||83.0–87.7|
|Cardiac disease (NYHA class I*)5||26||83.6|
|Acute orthopaedic clinic5||30||78.1|
|Cardiac disease (NYHA class II*)5||26||75.9|
|Cardiac disease (NYHA class III–IV*)5||8||60.9|
|Healthy controls5, 26, 27||838||80.7–87.6|
|Cardiac disease (NYHA class I*)5||47||86.5|
|Cardiac disease (NYHA class II*)5||49||80.1|
|Acute orthopaedic clinic5||43||73.7|
|Cardiac disease (NYHA class III–IV*)5||18||67.9|
Interpretation of the data in Tables 2 and 3 requires a degree of caution, as the exact extent to which a generic tool captures the impact of an individual disease on QoL may vary. Nonetheless, it is probably reasonable to conclude that the impact of constipation is at least comparable to that seen across a wide range of conditions that might normally be considered more ‘serious’.
The threshold for diagnosis of constipation varies substantially between patients and doctors and also to some extent between individual clinicians.39 These studies show, however, that regardless of the criteria used, the impact on patients’ perceived QoL is significant and comparable to that seen with other more recognized causes of impaired health.
In a US survey of 557 constipation patients of all ages,40 52% reported an impact on QoL, while 69% reported that constipation affected their performance at work or at school; 12% reported that constipation had resulted in absence from work or school in the preceding month, with a mean non-attendance period of 2.4 days. This figure relates to a selected subgroup of patients; however, amongst constipated patients as a whole, it has been estimated that a mean period of work absence of 0.4 days/year will be seen, equating to 13.7 million days of restricted activity annually in the United States.41
The extent to which the results of this study reflect the impact of constipation per se, rather than associated symptoms such as abdominal pain, is difficult to ascertain. Examination of the detailed results of the SF-36 studies (Figure 2) does not suggest that pain is a major driver of the results, with emotional and mental factors seeming to be of greater importance. In the PGWB studies, the reduction in score was uniform across all domains (data not shown) with no evidence of physical parameters dominating. Similar results were seen in the paediatric PedsQL studies (Figure 4). In one study,26 scores were explicitly assessed for those with and without abdominal pain: no significant difference was seen. A contrary result was seen in the Brazilian CHQ-PF50 study, where there was a bias towards physical components, with General Global Health and Bodily Pain & Discomfort being particularly affected.
To some extent, this distinction is moot, as constipation constitutes a combination of symptoms that will vary from one patient to another; attempting to distinguish the impact of the individual components has the potential to distract from the more general impact of the syndrome as a whole.
Whilst it is clearly useful to understand the impact of constipation on individuals and groups of patients, these data are of greatest value if there is evidence to show that effective treatment can improve QoL. This has been investigated in a number of studies, which have demonstrated that QoL measures do indeed improve after relief of constipation.42–45 These studies do not necessarily relate to therapies in common use and there are no studies that have been carried out in children. However, whilst there is a clear need to carry out further studies in this area, the apparent impact of treatment on QoL, coupled with the generally low cost of medication, suggests that this is likely to be a cost-effective intervention.
In common with all systematic reviews of the literature, our results are potentially subject to selection bias. Whilst we made every effort to include all published studies in the field, we were obviously unaware of data that were not in the public domain.
Secondly, the studies identified were carried out in populations of differing ages and geographical locations and the diagnosis of chronic constipation was not always made according to consistent criteria. Traditionally, one would deal with this kind of between-studies variation by carrying out a random effects pooling – in this analysis, however, there was insufficient information given in the published papers to allow this to be carried out. Consequently, our aggregate assessment of the SF-36 score was carried out using a simple study-size weighting method. It is therefore possible that our pooled result gives undue weight to individual studies with wide between-subjects variation.
Amongst the community-based SF-36 studies, one study17 was considerably larger than all the others and therefore had the major influence on the estimate of pooled effect. However, this study was carried out to a high standard in seven centres in different countries and yielded results that were comparable with those seen in the smaller studies. Although we were unable to assess heterogeneity formally owing to the lack of variance data, we believe the potential for this study to bias our result was low.
One of the population-based SF-36 studies was carried out exclusively in a population aged 65+.22 Both constipated and healthy patients showed a reduced PCS compared with a multinational study across all adults:17 PCS = 38.7 in elderly constipated vs. 49.1 in constipated adults of all ages; 46.8 in elderly healthy vs. 51.2 in healthy adults of all ages. This finding is perhaps not surprising, given the likelihood of increased co-morbidity in an older age group. Less easy to explain was the observation that the MCS score in the elderly was considerably higher than that seen in adults of all ages: 54.3 vs. 46.2 and 54.1 vs. 48.8 respectively. This study, which was the earliest identified in the review, used an early version of the SF-36 (the Medical Outcomes Survey), which only reported results in six of the eight dimensions of later versions. If these results are excluded from the analysis, the pooled estimate of PCS increases to 48.3, while MCS drops to 48.3.
In general, however, the SF-36 has been designed for scores to be compatible over time, with norm-based scoring allowing direct comparison between the results of all versions. A further issue relates to the geographical spread of included studies. Although the breadth of countries represented in this review offers robustness to the analysis, it also presents potential limitations, as QoL is population-dependent. For the SF-36 studies, this is not an issue: national-specific versions of the questionnaire have been developed and validated for all countries in the identified studies. By then correcting the derived scores using national norms, the overall PCS and MCS results may legitimately be compared internationally.13 For the two PGWB studies,24, 25 the issue does not arise, as both studies were carried out in Sweden. Differences in the recruited populations of these studies, however (hospital vs. community setting), meant that the results could not be meaningfully combined. The paediatric studies were carried out in three different countries (United States,26 Australia27 and Brazil28) using two different questionnaires, with no published means of correcting for these differences. Numerical comparison of these studies was therefore not possible.
Finally, the comparator groups in the studies were not necessarily age-matched healthy individuals; in some cases, they were nonconstipated individuals who had other gastrointestinal problems. Although the country-specific normalization of scores carried out for the SF-36 studies helps to mitigate this bias, it is possible that the magnitude of difference between constipated and healthy individuals has been underestimated.
Published studies are consistent in showing impairment in HRQoL in adults and children with constipation. This impairment is particularly significant compared with healthy adults in the domains of general health, social functioning and mental health. In children, it is notable that parents rate HRQoL lower than their children, particularly in the domains of emotional and social functioning. This may be due to the fact that there is often a degree of denial amongst children that they have a problem at all with constipation. In paediatrics, the brunt of concordance with treatment, effecting a behavioural change in the child, dealing with the educational problems of poor school attendance and ensuring attendance at follow-up appointments, falls upon the parents. Further studies are required to measure the change in parent and child QoL with successful treatment of childhood constipation. Clinical experience suggests that these will be substantial as a result of the improvement in parent-child relationships, which occur following resolution of symptoms of constipation.
The impairment in HRQoL observed in adults with constipation is comparable with that seen in conditions that might be regarded as more ‘serious’, such as osteoarthritis, rheumatoid arthritis, chronic allergies and diabetes. In children, the level of impairment seen is greater than with GERD and IBD. Constipation should therefore not be dismissed as a trivial condition in adults or children.
Given that the population prevalence of constipation is high, the adoption of active strategies to improve its management may be expected to yield significant gains in QoL. Focusing on improving the management of constipation can therefore be an effective way of improving the QoL for a substantial number of adults and children.
Declaration of personal interests: JB has received consultancy payments from Norgine Pharmaceuticals Ltd for data analyses and publications in the field of laxative use. SG has acted as a medical adviser for Norgine Pharmaceuticals Ltd. DC has received research grants from Norgine Pharmaceuticals Ltd and Yakult Ltd. He has received fees from Nutricia Ltd and Heinz Ltd for review articles on constipation. MG is an employee of Norgine Pharmaceuticals Ltd. Declaration of funding interests: This study was funded in full by Norgine Pharmaceuticals Ltd. All data analyses and writing were undertaken by the authors who agreed upon all content prior to publication.