Statistics from Altmetric.com
Functional constipation is a common worldwide complaint in infants and children.1 The aetiology of constipation is multi-factorial and seldom caused by structural, endocrine or metabolic disease.
Careful history taking and physical examination are usually sufficient to make a diagnosis. Criteria for a definition of functional constipation vary widely and are mostly based on a variety of symptoms, including decreased frequency of bowel movements, faecal incontinence and a change in consistency of stools.
Traditionally, treatment starts with education of the parents and children. Demystification and understanding of the problem helps to enlist cooperation and to improve compliance.2 When not adequately treated, constipation in children may lead to faecal incontinence and, subsequently, to psychological problems and social isolation.3
Most guidelines for the treatment of functional constipation are based on reviews of the literature that do not apply a systematic literature search, do not incorporate quality assessment of studies, or use a language restriction.4–6 On the other hand, a previous Cochrane review evaluating the effect of stimulant laxatives on constipation could not include any study because of the strict inclusion criteria set by the authors.7
In this systematic review, we aim to investigate and summarise the quantity and quality of all current evidence for the effect of laxatives and dietary measures on functional childhood constipation in comparison to placebo, no treatment or alternative treatments.
The Medline and Embase databases were searched from inception to December 2007. The keywords used to describe the study population were: “constipation”, “obstipation”, “coprostasis”, “encopresis”, and “soiling”. These words were combined with keywords referring to the different types of intervention groups that were investigated in the present review.
For the retrieval of controlled trials we used the keywords described in the Cochrane Handbook8 and the International Epidemiological Association.9 Additional strategies for identifying studies included searching the reference lists of review articles and the included studies. No language restriction was applied. The full search strategy is available from the authors.
Two reviewers (MP, MYB) independently screened all abstracts of identified published articles for eligibility. For this purpose, three specific criteria were used: (1) the study population consisted of children aged 0–18 years; (2) the study was a randomised controlled trial (ACT), a comparative clinical trial (CCT) or a crossover study; and (3) one of the aims of the study was to evaluate the effect of a medicamentous treatment or dietary intervention on functional constipation with or without faecal incontinence.
All potentially relevant studies, as well as the studies for which the abstracts did not provide sufficient information for inclusion or exclusion, were retrieved as full papers.
Full papers were additionally screened as to whether they fulfilled the following criteria: (4) the intervention consisted of osmotic, bulk-forming, stimulant or emollient laxatives, lubricating agents or dietary measures and were compared to placebo, no treatment or alternative treatment; and (5) outcome measures at least were either establishment of normal bowel habit (increase of defecation frequency and/or decrease of faecal incontinence frequency) or treatment success as defined by the authors of the study.
Excluded were papers concerning children with mental handicaps or psychiatric diseases (eg, eating disorders), as well as studies investigating children with organic causes of constipation and children with exclusively non-retentive faecal incontinence.
Any disagreements regarding the inclusion of articles were resolved through consensus when possible or by arbitration of a third person (MT).
Two reviewers (MP and either MT or MYB) independently rated the methodological quality of the included studies using a standardised list developed for RCTs, the Delphi list10 (table 1). Disagreement between the two reviewers was resolved by consensus when possible, or a third person (MYB or MT) made the final decision.
Two reviewers (MP and either MT or MYB) independently performed a structured data extraction from the original reports. Extracted information included (if available) items referring to study design, setting and participants (diagnosis, age, gender, severity of disease), as well as interventions and outcome measures. Disagreements were resolved by consensus when possible, or a third person (MYB or MT) made the final decision.
The inter-assessor reliability on the methodological quality was calculated using Kappa scores.11
In the present review the outcome measure was “treatment success” as defined by the authors of the included study. In addition, the establishment of normal bowel habit defined as an increase of defecation frequency and/or decrease of faecal incontinence frequency was considered as an outcome measure.
When the participants, interventions and outcome measures were sufficiently similar, data were statistically pooled using a random effects model. Heterogeneity was quantified by χ2, which can be interpreted as the percentage of the total variation between studies that is attributable to heterogeneity rather than to chance. A p value of less than 0.10 was used as cut-off point to indicate heterogeneity.
As most studies in this systematic review were highly diverse with regard to the participants, interventions and outcome measures, we often refrained from statistically pooling the data and used a best evidence synthesis to summarise the data. Methodological quality scores were calculated as a percentage of the maximum quality score on the Delphi list. High quality is defined as a score of ⩾60% (ie, ⩾6 points).
In the best evidence synthesis the level of evidence was ranked12 13 (table 2). Studies with a small study sample (<5 children per arm) were excluded, and in this synthesis only significant associations (ie, p<0.05) are considered as associated.
The search strategy resulted in a total of 736 titles and abstracts. After the eligibility screening, 37 publications met our inclusion criteria. After reading the full-text articles, nine studies were additionally excluded.18 24 28 32 33 38 39 41 45
All randomised controlled trials and the comparative controlled trial were hospital based, of which nine were conducted at a general paediatric department16 19 27 29 31 33 35 47 48 and 11 were conducted in a paediatric gastroenterology department14 20–22 26 36 40 43 44 46 49; two RCTs did not define a setting.23 50 Of the crossover studies four were hospital based, of which two were conducted at a general paediatric department15 30 and two were conducted in a paediatric gastroenterology department.17 25 Only one study was primary care based.37
Methodological quality assessment
The reviewers initially agreed on 85% of the quality items. The inter-observer reliability of the methodological quality assessment (0.70) was high.
The most prevalent shortcomings of the studies were: no concealment of treatment allocation (n = 18 (61%)); no similarity between the intervention groups regarding the most important prognostic indicators (ie, age, sex, duration of disease, severity of disease) (n = 20 (71%)); no blinding of outcome assessor (n = 16 (57%)) and no intention-to-treat analysis (n = 21 (75%)). The overall methodological quality had a mean score of 4.8 (range RCTs 1–10; CCT 3; crossover studies 2–8). Only 10 studies (36%) had a score of ⩾6 points indicating a good methodological quality.
Clinical diversity in the studies included with regard to participants, diagnosis, interventions and outcome measures presented, was large. The lack of a uniform outcome measure made a formal meta-analysis impossible. Most studies, however, reported on either treatment success or defecation frequency. Although the definition of treatment success differed substantially between studies, all studies presented treatment success as the percentage of successfully treated children. We therefore statistically pooled results on treatment success for the comparisons between polyethylene glycol (PEG) and any other laxative, and between PEG and lactulose. In case the presentation of the effect on defecation frequency was comparable we pooled the results on the effect on the number of bowel movements (cisapride compared with placebo and fibre compared with placebo). For all other comparisons, a best evidence synthesis was performed to summarise the results.
Laxatives and dietary measures
PEG compared with placebo
Only one high-quality study42 investigated the effect of PEG in comparison with placebo. Compared with placebo, PEG was more effective in increasing defecation frequency (mean treatment difference 1.64 (95% CI 0.99 to 2.28)). For decrease in faecal incontinence episodes, no significant differences were found (mean treatment difference 0.15 (ns)).
PEG compared with other laxatives
Eight studies comparing PEG to another laxative were included. Of these, one study reported on defecation frequency only.20 The other seven all reported on treatment success23 25 29 31 43 46 48 (pooled risk ratio (RR) 1.47 (95% CI 1.23 to 1.76) (χ2 17.89, p<0.0001)). The number needed to treat (NNT) is 4.0 (95% CI 6.0 to 2.9).
PEG compared with lactulose
Five studies20 23 25 46 48 compared the efficacy of PEG with lactulose. Four of these five studies reported on treatment success25 46 48 and the number of children with soft or normal stools.23 All four studies showed that PEG was more effective than lactulose with regard to these outcome measures (pooled RR for treatment success 1.63 (95% CI 1.40 to 1.90) (χ2 38.95, p<0.0001)). When treating children with constipation 3.3 children need to be treated with PEG in order to get one more treatment success in comparison to treatment with lactulose (95% CI 4.5 to 2.6).
All five studies20 23 25 46 48 evaluated the effect on defecation frequency. Three studies scored as high quality.20 46 48 One high-quality study20 and one low-quality study25 found PEG to be superior to lactulose in increasing the number of bowel movements. Two high-quality studies46 48 and one low-quality study23 reported no significant difference between PEG and lactulose (conflicting evidence).
Youssef et al performed a high-quality, dose-finding study. They compared different doses of PEG (0.25, 0.5, 1.0 and 1.5 g/kg/day) and found that doses of 1.0 and 1.5 g/kg/day were more effective in achieving disimpaction than lower doses.45
In addition lactulose was compared to other laxatives in two low-quality studies. Perkin et al37 compared lactulose to senna and found no significant difference in defecation frequency between the two treatments (limited evidence). Urganci et al44 found lactulose to be less effective compared to liquid paraffin in increasing the defecation frequency (limited evidence).
Based on all the studies on lactulose, we found conflicting evidence for an effect of lactulose on defecation frequency in comparison with PEG, liquid paraffin, and senna with lactulose being less than or equally effective.
In a low-quality, dose-finding study on lactulose, Hejl et al27 investigated the effect of a milk formula with either 4% or 2% lactulose. They reported no significant differences between the two doses regarding all outcome measures.
Cisapride, a prokinetic agent, has been withdrawn from the market because of cardiovascular adverse events. Nevertheless, we found two studies comparing the effect on defecation frequency of cisapride with placebo.26 36 Nurko et al performed a high-quality study and reported no significant difference between cisapride and placebo.36 In a low-quality study Halabi et al found cisapride to be more effective compared to placebo.26
Pooling the data resulted in a weighted, standardised mean difference of 4.0 bowel movements per week in favour of cisapride (95% CI 0.38 to 7.64) (χ2 4.69, p<0.05).
One low-quality study that investigated cisapride added to magnesium oxide found this combination to be more effective than magnesium oxide alone (limited evidence).34
In total, three low-quality studies reported on the effect of senna on constipation. Sondheimer et al compared senna with mineral oil and found senna to be less effective in increasing daily bowel movements (limited evidence).40
Berg et al compared senna to placebo and no medication. They found no significant differences in effect in decreasing the number of faecal incontinence episodes per week between the groups (limited evidence).16
Perkin et al used lactulose as comparison and reported no significant differences in effect on defecation frequency (limited evidence).37
In conclusion, based on all the included studies on senna, we found conflicting evidence for the effect of senna compared to placebo, no medication, mineral oil or lactulose, with senna being less than or equally effective.
Three low-quality studies reported on the effect of mineral oil on the number of bowel movements and episodes of faecal incontinence.40 43 47 Of these, two used a different laxative as control intervention, that is, PEG43 and senna,40 and one study used biofeedback therapy as control intervention.47 Wald et al found no significant difference in the number of children with <1 faecal incontinence episode per week, between mineral oil and biofeedback therapy (limited evidence).47
Sondheimer et al reported that mineral oil resulted in more children with daily bowel movements compared with senna (limited evidence),40 and Tolia et al found mineral oil to be less effective compared with PEG for children having more than one bowel movement per day after treatment (limited evidence).43
Based on all studies on mineral oil, we found conflicting evidence for the effect of mineral oil compared to PEG, senna or biofeedback therapy, with mineral oil being less than, more or equally effective.
Bellomo-Brandao et al compared the effect of erythromycin estolate with the effect of placebo in a low-quality study. They found erythromycin estolate to be more effective than placebo in improving constipation (limited evidence).15
Zoppi et al performed a low-quality study and found calcium polycarbophil to be more effective than placebo in clearing constipation in children (limited evidence).50
In a low-quality study Nolan et al35 compared laxative therapy (ie, Microlax and senna and/or bisacodyl and/or agarol) with standard paediatric behaviour modification. They reported no significant differences between the two treatment groups with regard to a decrease in stool retention (limited evidence).
Infant formula with sn-2 palmitic acid
In their high-quality study Bongers et al found no difference in the defecation frequency of children treated with a new infant formula with a high concentration of sn-2 palmitic acid, a mixture of prebiotic oligosaccharides and partially hydrolysed whey protein (Nutrilon Omneo), and children treated with a standard infant formula (moderate evidence).17
In one high-quality study21 the effect of a cocoa husk supplement on the defecation frequency was investigated, and in a low-quality study30 the effect of glucomannan was investigated. In both studies fibre was compared to placebo, and both found no statistical significant difference in defecation frequency between the treatment groups. The pooled weighted standardised mean difference was 0.35 bowel movements per week in favour of fibre (95% CI −0.04 to 0.74) (χ2 3.11, p<0.10), which is neither significant, nor clinically relevant.
Laxatives used in daily clinical practice are insufficiently tested against placebo in the case of children. This may be because laxatives have already proven to be effective in adults, or because it may be considered unethical to conduct placebo-controlled studies among children. However, these arguments do not hold, when considering that constipation usually has a different aetiology in adults compared with children, and it should be considered unethical to treat children without prior evidence for a beneficial effect of this treatment.
Compared to all other laxatives, the percentage treatment success was higher in children treated with PEG (pooled RR 1.47 (95% CI 1.23 to 1.76) (χ2 17.89, p<0.0001)).
Clinical and statistical heterogeneity between studies was large and the overall methodological quality of the 28 included studies was poor. Only 10 studies were of high methodological quality.
The major drawback of these studies is the lack of a uniform definition of childhood constipation and treatment success, making the results difficult to compare. In addition, the definition of functional constipation varies over time and between authors. Only defecation frequency was consistently reported in all studies; however, we are fully aware that it is not sufficient to quantify constipation only in terms of number of stools per week. How outcome can affect the results of our review is illustrated by the case of PEG; whereas PEG was found to be more effective on “treatment success” when compared with lactulose, this could not be demonstrated for an effect on the number of bowel movements.
In order to perform proper studies on the effect of an intervention for childhood constipation, a uniform definition is urgently needed. In 1999 experts in the field of paediatric gastroenterology reached the first consensus on defining childhood constipation.53 In 2006 the definition for childhood constipation was redefined since several studies showed that the earlier criteria were too restrictive and excluded several groups of children with constipation.51 52 All these definitions are based on constipation seen in referred children. However, because most children with constipation are seen in primary care, the definitions also need to be validated in primary care.
As in every systematic review, there is a risk that not all relevant studies are included. To minimise this risk, we performed a sensitive literature search without language restrictions.
A large number of outcome measures have been analysed in the included studies. Because it was not feasible to present all these results, we have analysed and presented those outcomes only that enabled a comparison between the studies. In a best evidence synthesis, bias may occur due to misclassification of the methodological quality of the studies. However, because the quality scores of the individual studies were low, that misclassification of an item would not have changed the classification into a high or a low methodological quality.
Only significant effects were assumed to be effective in our best evidence synthesis; this assumption may misclassify the results of studies with a small sample size. Most comparisons were evaluated in only one study, and the methodological quality was low; consequently the level of evidence for the effect of an intervention was low.
The chi-squared test used to detect statistical heterogeneity is of limited value since there are very few studies in the meta-analyses, which imply a low power of this test. For this reason, a p value of less than 0.10 is used to indicate heterogeneity rather than the conventional cut-off point of 0.05.
COMPARISON WITH PREVIOUS REVIEWS
Only Price et al7 performed a systematic review of the literature; they aimed to investigate the effect of stimulant laxative treatment in children with chronic constipation, however, none of the studies found complied with their strict criteria. Although most guidelines provide a review of available studies, none of these reviews provide a summary of the quantity and quality of all current evidence based on a systematic search of the literature. Guidelines on the treatment of functional constipation in children are therefore authority based rather than evidence based.
IMPLICATIONS FOR PRACTICE
There is insufficient evidence to support that laxative treatment of childhood constipation is better than placebo. In comparison to other laxatives, however, PEG is more effective in achieving treatment success. Because of the heterogeneity between the included studies this result should be interpreted with caution. Based on the results of this review we cannot give a recommendation to support one laxative over the other for childhood constipation. Given the lack of evidence for differences in effect of laxatives, adverse effects play an important role in the choice of a laxative.
Two guidelines on the management of childhood constipation were recently published.4 5 6 The main shortcoming of these guidelines was the lack of a systematic review of the available evidence.54 Therefore it remains unclear whether the recommendations of the guidelines are based on personal conviction of the guideline committee or on scientific evidence. Our systematic review of the literature reveals that there is insufficient evidence to recommend one laxative above the other. In future guidelines this can be stated. This will make it clear that recommendations will be based on personal experience and consensus rather then scientific evidence. In addition it will be evident that all available experience should be consulted; this includes experience from primary care. In the guideline committees thus far primary care was under represented.
For future research we recommend large, well-designed, placebo-controlled, randomised trials that evaluate the effect of laxatives (especially PEG and lactulose) on functional constipation in children. Since most children with constipation will first consult their general practitioner, these studies should also be performed in general practice. A well-defined and uniform definition of functional constipation is urgently needed. Dose-finding studies in children are needed in case of the introduction of new laxatives and, since adverse effects may play an important role in the choice of a laxative, it is also necessary to investigate their side effects.
Due to a lack of placebo-controlled trials we found insufficient evidence for an effect of any one laxative or dietary treatment of childhood constipation. Although, PEG achieved more treatment success compared to all other laxatives, the results on defecation frequency were conflicting. Based on the results of this review we cannot give a recommendation to support one laxative over the other for childhood constipation.
What is already known on this topic
Constipation is a common worldwide complaint in infants and children which, if not adequately treated, may lead to faecal incontinence and subsequently to psychological problems and social isolation.
Guidelines on the treatment of functional constipation in children are authority based rather then evidence based.
What this paper adds
Due to a lack of placebo-controlled trials we found insufficient evidence for an effect of any one laxative or dietary treatment of childhood constipation.
A uniform definition of functional constipation in children is urgently needed.
Well-designed trials on the effectiveness of laxative and dietary treatment of childhood constipation still need to be performed.
Competing interests: None.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.