Clinical question

What is the diagnostic value of the gastrointestinal endoscopy in children with chronic idiopathic constipation?

Studies considered in this section

Studies were considered if they:

• included neonates, infants or children up to their 18th birthday with chronic idiopathic constipation undergoing gastrointestinal endoscopy
• were not case reports
• were published in English.

No restrictions were applied on the publication date or country.

Overview of available evidence

The searches identified 139 articles but no articles were retrieved for detailed assessment.

GDG interpretation of the evidence

No published evidence was found for the diagnostic value of the gastrointestinal endoscopy in children with chronic idiopathic constipation. Gastrointestinal endoscopy is an invasive procedure with associated morbidity and mortality. In the very rare circumstances when this test will be indicated because of suspicion of organic pathology, this will happen only after less invasive tests have shown positive results, for example positive blood tests for coeliac disease. Therefore the GDG concluded that gastrointestinal endoscopy should not be used to investigate children with idiopathic constipation.

Clinical question

What is the prevalence of hypothyroidism and coeliac disease in children with chronic constipation?

Previous NICE guidelines

A similar clinical question was looked at in the NICE clinical guideline for coeliac disease¹¹ where the question addressed was: ‘What are the signs and symptoms which indicate a diagnosis of coeliac disease?’ including both gastrointestinal symptoms and non-gastrointestinal symptoms.

• The guideline recommended: 'Consider offering serological testing for coeliac disease to children and adults with any of the following:

  –

  ‘persistent or unexplained constipation’

  –

  (other conditions not related to constipation were also listed)

• ‘Offer serological testing for coeliac disease to children and adults with any of the following signs and symptoms:

  –

  ‘failure to thrive or faltering growth (in children)’

  –

  (other conditions not related to constipation were also listed).

Studies considered in this section

Studies were considered if they:

• included neonates, infants or children up to their 18th birthday with chronic idiopathic constipation
• were not case reports
• were published in English.

No restrictions were applied on the publication date or country.

Overview of available evidence

The searches identified 92 articles (50 on coeliac disease, 42 on hypothyroidism) and 18 articles were retrieved for detailed assessment (12 on coeliac disease, 6 on hypothyroidism). Of these, four studies on coeliac disease were identified for inclusion in this review: two prospective cohorts and two retrospective case series. None of these studies investigated the prevalence of coeliac disease in children with idiopathic constipation but rather looked at the associations between coeliac disease and symptoms of constipation in a variety of populations of children. No studies were identified for inclusion that considered the prevalence of hypothyroidism in children with idiopathic constipation.

Narrative summary

A prospective cohort conducted in Italy²² (2001) [EL=2+] estimated the prevalence of coeliac disease (CD) in people with Down's syndrome and defined the clinical characteristics of CD among 1202 people with Down's syndrome (609 males). Of these, 1110 were children (15 months to 18 years and 92 were adults (18 to 46 years). CD was diagnosed according the Revised European Society of Paediatric Gastroenterology, Hepatology and Nutrition (ESPGHAN) criteria. Participants were selected for intestinal biopsy on the basis of antiendomysium antibodies (EMA) positivity, antigliadin antibodies immunoglobulin (AGA IgA) positivity, or both in children younger than 2 years. Down's syndrome was confirmed by karyotype in all cases. All participants were receiving a diet containing gluten.

Group 1 consisted of 55 patients, including 47 children (36 males, aged 4 to 46 years) who were diagnosed with CD. Their clinical features were compared with those observed in 55 IgA AGA-positive/EMA-negative patients (group 2: 33 males, aged 3 to 40 years) and in 57 IgA AGA-negative/EMA-negative Down's syndrome patients (group 3: 34 males, aged 4 to 38 years). Group 2 and group 3 patients were selected randomly from among the screened patients to be age and gender matched to group 1. A detailed questionnaire was completed to obtain information about familial gastroenterologic history with special attention to:

• feeding habits

  –

  breast milk or formula

  –

  age of introduction of gluten-containing foods

• gastrointestinal function, particularly the features of CD such as:

  –

  chronic diarrhoea

  –

  vomiting

  –

  failure to thrive

  –

  anorexia

• presence of autoimmune or neoplastic conditions.

Weight and height were evaluated using Down's syndrome percentile charts.

Constipation was present in significantly more patients in group 1 (29.1%) when compared to patients in groups 2 and 3 (14.5% and 8.8% respectively, P < 0.05). However, other signs and symptoms were also present in significantly more patients in group 1 when compared to patients in groups 2 and 3:

• growth failure: 52.7% versus 10.9% versus 7%; P < 0.001
• diarrhoea: 41.8% versus 1.8% versus 6.9%; P < 0.001
• vomiting: 20% versus 1.8% versus 1.7%; P < 0.001
• anorexia: 18.2% versus 1.8% versus 3.4%; P < 0.01.

It should be noted that the parents of eight EMA positive children and two EMA positive adults did not give permission for intestinal biopsy to be performed and were not included among the 55 CD patients.

A prospective cohort study conducted in the UK²³ (2004) [EL=2+] established the prevalence of undiagnosed CD in the general population at age 7 years and looked for any associated clinical features in 5470 children aged 7.5 years (gender not reported) participating in the Avon longitudinal study of parents and children (ALSPAC, a population based birth cohort study established in 1990). CD was diagnosed based on a two stage screening. First, a sensitive initial radioimmunoassay for antibodies to tissue transglutaminase (endomysial antigen) (tTG antibodies) was conducted. If positive to previous, serum IgA antiendomysial antibodies (IgA-EMA) were measured by indirect immunofluorescence. Children with tTG antibodies less than the 97.5th centile were defined as antibody negative. Details of gastrointestinal symptoms including constipation were collected by routine questionnaire at age 6.75 years.

Of 5470 children tested, 54 children were IgA-EMA positive (1.0%, 95% confidence interval [CI] 0.8 to 1.4) and 5333 children were tTG antibody negative controls. An additional 137 children were tTG antibody positive, but IgA-EMA negative. Questionnaires were returned for 4324 children (79%). Of 4285 tTG antibody negative controls who returned their questionnaires, 435 (10%) reported any constipation at age 6.75 years. Of 42 IgA-EMA positive children who returned their questionnaires, 6 (14%) reported any constipation at age 6.75 years (odds ratio [OR] 1.48, 95% CI 0.62 to 3.52).

Other symptoms reported at age 6.75 years were not significantly more frequent in IgA-EMA positive children than in tTG antibody negative controls:

• any diarrhoea: 21 (50%) versus 1450 (34%); OR 1.96, 95% CI 1.06 to 3.59
• any vomiting: 23 (55%) versus 1933 (45%); OR 1.47, 95% CI 0.80 to 2.71
• any stomach pains: 28 (66%) versus 2557 (60%); OR 1.35, 95% CI 0.71 to 2.57

However, significantly more IgA-EMA positive children than tTG antibody negative controls reported multiple (3 or more) gastrointestinal symptoms (17 [40%] versus 931 [22%]; OR 2.45, 95% CI 1.33 to 4.5).

IgA-EMA were more common in girls (OR 2.12, 95% CI 1.20 to 3.75). IgA-EMA positive children were shorter and weighed less than those who tested negative for tTG antibody (P < 0.0001). It should be noted that since ALSPAC is an observational study based on analysis of anonymous samples, confirmatory biopsy for coeliac disease was not possible. No data regarding clinical symptoms at 6.75 years were available for 21% of the total sample. It is unclear how the symptom 'constipation' was defined.

A multicentre, hospital based retrospective case series²⁴ conducted in Italy (2004) [EL=3] evaluated the prevalence of CD in immigrant children, the clinical findings in these patients and the possible relationship between immigration, dietary habits and CD in childhood. This included 1881 Italian children and young people (891 males, age 6 months to 16 years, mean age 7.9) and 36 immigrant children and young people (15 males, age 6 months to 15 years, mean age 7.3) consecutively diagnosed as having CD between January 1999 and December 2001. CD was diagnosed based on the revised criteria of the European Society of Paediatric Gastroenterology and Nutrition (ESPGAN).

Clinical pattern and presenting symptoms at diagnosis were classified and grouped in three categories:

• ‘classical forms’ included the following symptoms

  –

  chronic diarrhoea

  –

  weight loss

  –

  abdominal distension

  –

  vomiting

• ‘atypical forms’ included:

  –

  iron-deficiency anaemia

  –

  short stature

  –

  delayed puberty

  –

  recurrent oral aphthae

• ‘silent forms’ included:

  –

  serological screening of first degree relative

  –

  loss of Kerckring folds at endoscopy.

Two out of nine children (25%) presenting with atypical forms of CD had abdominal pain with constipation. None of the children diagnosed with ‘classical forms’ (n=25, 69.4%) or with ‘silent forms’ (n=2, 5.5%) was reported to have experienced constipation. Clinical patterns in Italian children were similar to those of immigrant children but presenting symptoms at diagnosis were not reported for Italian children. It is unclear how the symptom ‘constipation’ was defined in the first place.

One retrospective case series conducted in Ireland²⁵ (1972) [EL=3] assessed the incidence of constipation in 112 children diagnosed with CD. Of the total population 12 children had constipation (six boys, age 6 to 102 months). CD was diagnosed based on clinical variables (undernutrition and retarded growth) and jejunal biopsy (grade 2/3 or grade 3 jejunal mucosal damage). Growth retardation was assessed using the graphs of Tanner and Whitehouse (1959) and subsequently confirmed by catch-up growth following treatment with gluten-free diets. Jejunal mucosal damage was assessed according to the authors' classification: grade 0 indicating normal mucosa; grade 1 indicating mild non-specific change; and grades 2 and 3 corresponding to moderate and severe villous atrophy respectively. Constipation was defined as the passage of stools of harder consistency than normal or the clinical observation of impaction of abnormal amounts of hard (usually pale) faeces in colon and rectum.

Twelve children (10.7%) had been constipated at some stage before diagnosis: 66.7% of those children had had constipation alternating with diarrhoea and 25% additionally presented anorexia and failure to thrive. It is unclear whether the authors used a validated classification system for jejunal mucosal damage.

Evidence statement

There is no published evidence on the prevalence of hypothyroidism and coeliac disease in children with idiopathic constipation.

One prospective cohort study [EL=2+] showed that the prevalence of constipation as a symptom in patients, both adults and children, with Down's syndrome and subsequently diagnosed with CD was 29.1%. Constipation was present in significantly more patients diagnosed with CD compared to controls. Faltering growth, diarrhoea, vomiting and anorexia were also present in significantly more patients diagnosed with CD when compared to controls.

One prospective cohort [EL=2+] showed that 14% of children who tested positive to serum IgA antiendomysial antibodies had constipation. However, constipation was not associated with positivity to serum IgA antiendomysial antibodies, and neither were diarrhoea, vomiting or stomach pains. Having multiple (3 or more) gastrointestinal symptoms was associated with positivity to serum IgA antiendomysial antibodies.

One retrospective case series [EL=3] showed that the prevalence of constipation as a symptom in children with CD was 10.7% and that 66.7% of those children had constipation alternating with diarrhoea and 25% presented with constipation, anorexia and faltering growth.

One retrospective case series [EL=3] showed that 25% of children presenting with atypical forms of coeliac disease had abdominal pain with constipation. This corresponded to 5.6% of the total sample of children with CD.

GDG interpretation of the evidence

The GDG noticed that none of the studies investigated the prevalence of coeliac disease in children with idiopathic constipation but rather looked at the associations between coeliac disease and symptoms of constipation in a variety of populations of children. No studies were identified for inclusion that considered the prevalence of hypothyroidism in children with idiopathic constipation.

The GDG therefore concluded that there is no published evidence on the prevalence of hypothyroidism and coeliac disease in children with idiopathic constipation, hence the recommendation of not testing as a routine but only in the ongoing management of intractable constipation and when requested by specialist services. In some children who do not respond to sustained optimal medical management it is the GDG's experience that an atypical presentation of hypothyroidism or CD could be the cause of the constipation, therefore testing would be justified.

From their own clinical experience (and also from the evidence in the case of CD) the GDG believes that if other symptoms, for example faltering growth, are present in the history, this may suggest an underlying disorder like CD or hypothyroidism as the cause of the constipation, and in those cases testing would also be justified.

Clinical question

What is the diagnostic value of the anorectal manometry in children with chronic idiopathic constipation?

Studies considered in this section

Studies were considered if they:

• included neonates, infants or children up to their 18th birthday with chronic idiopathic constipation undergoing anorectal/rectal manometry and also undergoing rectal biopsy as the gold standard method to diagnose Hirschsprung's disease (HD)
• were not case reports
• were published in English.

No restrictions were applied on the publication date or country.

Overview of available evidence

The searches identified 480 articles and 27 articles were retrieved for detailed assessment. Of these, five studies were identified for inclusion in this review: two prospective case series and three retrospective case series.

Narrative summary

A retrospective case series conducted in Finland²⁶ (2009) [EL=3] reported on the value of anorectal manometry (ARM) with reference to operative rectal biopsy in the diagnosis/exclusion of HD in children under 1 year and on the prognostic significance of a normal rectoanal inhibitory reflex (RAIR) in these patients. The case series included 81 patients under 1 year who presented with delayed passage of meconium, abdominal distension and vomiting or constipation who underwent ARM (49 boys, median age at time of ARM and biopsy 2 months [range 0.1 to 11 months]). The records of all patients who met the inclusion criteria were reviewed.

All children underwent both ARM and operative rectal biopsy. The RAIR was present in 40 children. None of those children had HD, 39 had normal histology and 1 had hypoganglionosis. The RAIR was absent in 41 children, 33 of whom had HD and 8 had normal histology. The operative rectal biopsy was 100% accurate in diagnosing HD for all variables (sensitivity, specificity, positive predictive value and negative predictive value). Both the sensitivity and the negative predictive value was 100% for the ARM, but its specificity was 83% and its positive predictive value was 80%.

Patients who had HD were significantly younger at the time of investigation than those who did not. The operative rectal biopsy was adequate and diagnostic in all cases. There was one case of rectal bleeding following biopsy which required suturing in theatre. In the case of patients diagnosed with HD the histology from bowel resected at pull-through operation was consistent with pre-operative diagnosis in all cases.

A retrospective case series conducted in Korea²⁷ (2007) [EL=3] evaluated the incidence and clinical aspects of allergic proctitis (AP) in patients with symptoms that mimic HD. In addition, the authors determined the sensitivity and specificity of ARM and suction rectal biopsy used for evaluation of HD. One hundred and five infants younger than 6 months (61 boys, mean age 2.1 ± 0.9 months) with severe abdominal distension that mimicked HD were referred to department of paediatrics and division of paediatric surgery and underwent all triple tests including barium enema, ARM and rectal suction biopsy. Some patients had associated symptoms like constipation, poor oral intake, vomiting, poor weight gain and diarrhoea. HD was finally diagnosed with full thickness biopsy. The RAIR was absent in 48 children, 34 of whom had HD and 10 had normal histology. In this group four children were diagnosed with other pathologies (two with AP and two with intestinal neuronal dysplasia [IND]). The RAIR was present in 57 children, 5 of whom had HD and 43 had normal histology. In this group nine children were diagnosed with other pathologies (five with AP and four with IND).

The diagnostic variables for the ARM in HD were:

• sensitivity 87.18% (CI 73.29 to 94.90)
• specificity 78.79% (CI 67.49 to 86.92)
• positive predictive value 70.83%
• negative predictive value 91.23%.

The diagnostic variables for the rectal suction rectal biopsy in HD were:

• sensitivity: 92.31 % (CI 76.68 to 97.35)
• specificity: 100% (CI 94.50 to 100)
• positive predictive value: 100%
• negative predictive value: 95.65%.

A prospective case series conducted in Singapore²⁸ (1989) [EL=3] assessed the accuracy of ARM in the diagnosis of HD using histological aganglionosis as the reference point for final diagnosis. The case series included 50 children referred consecutively to one of the authors for anorectal manometric studies. All children underwent both manometry and biopsy.

Forty-five patients had concordant results (both on manometry and biopsy) and demographic data are only reported for these patients (31 boys, age birth to 11 months). Specimens not including the submucosal layer were considered inadequate and repeat full-thickness operative rectal biopsies were taken.

The RAIR was absent in 16 children, 15 of whom had HD and 1 had normal histology. The RAIR was present in 34 children, 4 of whom had HD and 30 had normal histology. Diagnostic variables for the ARM in the total sample (n=50) were:

• accuracy 90%
• sensitivity 79%
• specificity 97%
• positive predictive value 94%
• negative predictive value 88%.

Diagnostic variables for the ARM in neonates (n=10) were:

• accuracy 90%
• sensitivity 86%
• specificity 100%
• positive predictive value 100%
• negative predictive value 75%.

Diagnostic variables for the ARM in infants (n=18) were:

• accuracy 94.4%
• sensitivity 90%
• specificity 100%
• positive predictive value 100%
• negative predictive value 89%.

Five children (10%) required repeat full-thickness biopsy for inadequate sampling. No complications were encountered with manometry in all 50 children studied.

A retrospective case series conducted in Taiwan²⁹ (1993) [EL=3] evaluated the possibility of using ARM for screening for HD. The case series included 39 patients (age 3 days to 9 years) with constipation or suspected HD. All children underwent both anorectal manometry and rectal suction biopsy. The RAIR was absent in eight patients, five of whom had HD and three normal histology. The final diagnosis of HD was made by the patient's clinical history, barium enema and rectal suction biopsy. Three children showed inconclusive results with manometry due to poor tracing of internal sphincter contraction as a result of oversedation (n=2) and anal stenosis (n=1). Diagnostic variables for the ARM were: accuracy 90%, sensitivity 100%, specificity 86%, positive predictive value 83% and negative predictive value 100%.

A prospective case series conducted in Belgium³⁰ (1990) [EL=3] ascertained the traps and limitations of testing the RAIR, how frequently they occur and the possible explanations for equivocal or false results. The case series included 261 patients referred for ARM in order to confirm or exclude HD. All patients had presented with constipation varying from slight to intractable, with highly differing durations ranging from neonatal ileus to chronic constipation in adults. Ninety-four patients (36%) were under 6 months, 106 (41%) were age 6 months to 6 years, 47 (18%) were age 6 to 15 years and 5% comprised 2 adolescents and 12 adults (gender not reported for all patients). All children underwent ARM.

A confident interpretation of the RAIR occurred in 232 children, with RAIR present in 207 and absent in 25. The result of this first manometric evaluation was verified either by biopsy or by repeated manometry in 54 cases. In other cases the clinical evolution did not warrant further investigation. This review only includes children who underwent both manometry and biopsy. In these, the RAIR was present in two children who had HD and was absent in four children who had a normal histology. The RAIR was equivocal (‘?absent’) in nine children, four of whom had HD and five who had normal histology. The RAIR was equivocal (‘?present’) in eight children, two of whom had HD and six who had normal histology. The incidence of false results at first manometry was significantly higher in neonates compared to children older than 1 month (5 out of 22 [22.7.8%] versus 4 out of 239 [1.7%]). The incidence of equivocal results at first manometry was also higher in neonates compared to children older than 1 month (4 out of 22 [18.2%] versus 25 out of 239 [10.4%]). The result of a rectal biopsy was not known at the time of manometry in any case.

Authors reported that the following factors prevented the examiners from reaching a definite conclusion when measuring the RAIR:

• low anal tone (eight cases)
• restlessness of patient (seven cases)
• reflex external sphincter contraction partially or completely masking possible RAIR (four cases)
• presence of megarectum (three cases)
• artefacts (one case)
• unstable RAIR (six cases).

Details of both the manometry and biopsy results were reported only in cases where the RAIR was equivocal in the first manometry and in those children where the result proved to be false (either negative or positive). Considering this, it is not possible to calculate the sensitivity, specificity, positive and negative predictive values of the ARM. The incidence of false results in manometry performed by different examiners is reported in the paper, but there are missing data not accounted for and therefore we do not report it here.

Evidence statement

A retrospective case series [EL=3] showed that the anorectal manometry (ARM) had the same sensitivity and negative predictive value (100%) as the operative rectal biopsy in diagnosing Hirschsprung's Disease (HD) but its specificity and positive predictive value were lower (83% versus 100% and 80% versus 100% respectively)

A retrospective case series [EL=3] showed that the ARM performed worse in all diagnostic variables than the suction rectal biopsy in diagnosing HD (sensitivity: 87.18%, CI 73.29 to 94.90 versus 92.31%, CI 76.68 to 97.35; specificity: 78.79%, CI 67.49 to 86.92 versus 100%, CI 94.50 to 100; positive predictive value 70.83% versus 100% and negative predictive value 91.23% versus 95.65%)

A prospective case series [EL=3] showed that the diagnostic variables for the ARM in diagnosing HD were: accuracy 90%, sensitivity 79%, specificity 97%, positive predictive value 94% and negative predictive value 88 %. ARM was less accurate and less sensitive in neonates compared to infants and its negative predictive value was also lower. Specificity and positive predictive value were the same for both age groups (100%).

A retrospective case series [EL=3] showed that the diagnostic variables for the ARM in diagnosing HD were: accuracy 90%, sensitivity 100%, specificity 86%, positive predictive value 83% and negative predictive value 100%.

A prospective case series [EL=3] showed that the incidence of both false and equivocal results for ARM were significantly higher in neonates than in children older than 1 month. Different factors prevented the examiners from reaching a definite conclusion when measuring the RAIR:

• low anal tone
• restlessness of patient
• reflex external sphincter contraction partially or completely masking possible RAIR
• presence of megarectum
• artefacts
• unstable RAIR.

GDG interpretation of the evidence

The GDG understands from the evidence that ARM is not a reliable test to diagnose HD and that there are many factors which can confound its results. The GDG is aware that ARM is used as a research tool in some centres. However, if there is a strong clinical suspicion for HD then a rectal biopsy should be performed without delay, because this is the gold standard test to diagnose HD.

Clinical question

What is the diagnostic value of plain abdominal radiography to diagnose chronic idiopathic constipation in children?

Studies considered in this section

Studies were considered if they:

• included neonates, infants, or children up to their 18th birthday with chronic idiopathic constipation
• were not case reports
• were published in English.

No restrictions were applied on the publication date.

Overview of available evidence

A search was conducted for all radiological investigations (plain abdominal radiography, abdominal ultrasound and transit studies). This search identified 646 articles and 72 articles were retrieved for detailed assessment. Of these, one systematic review (including six studies), two case control studies and one retrospective case series were identified for inclusion in this review.

Narrative summary

A robust systematic review conducted in the Netherlands³¹ (2005) [EL=III] evaluated the additional diagnostic value of plain abdominal radiography in the diagnosis of constipation in children. Six studies (three case series, two case–control studies and one retrospective re-examination of abdominal radiographs) were included. All were hospital based, controlled, observational studies investigating the relationship between faecal loading on plain abdominal radiography and symptoms and signs related to constipation in otherwise healthy children aged from 1 to 18 years. Some studies included children with soiling or encopresis, while others excluded this group.

In the six studies included, three different scoring systems were used for assessing impaction on abdominal radiography: three studies used Barr-score; two studies used revised Barr-score (Blethyn); and one study used the authors' own scoring system (Leech).

The ability of the abdominal radiography to discriminate between clinically constipated and non constipated children was evaluated in four studies with variable results. One study reported only an accuracy of 80% (95% CI 50 to 100). Results from the other three studies were:

• sensitivity: 76% (95% CI 58 to 89) versus 60% (95% CI 46 to 72) versus 80% (95% CI 65 to 90)
• specificity: 75% (95% CI 63 to 85) versus 43% (95% CI 18 to 71) versus 90% (95% CI 74 to 98)
• likelihood ratio (LR): 1.0 (95% CI 0.5 to 1.6) versus 3.0 (95% CI 1.6 to 4.3) versus 8.0 (95% CI 0.7 to 17.1).

The ability of the clinical examination to discriminate between radiographically constipated and non constipated children was evaluated in one study and reported a sensitivity of 77% (95% CI 70 to 84) a specificity of 35% (95% CI 27 to 44) and a LR of 1.2 (95% CI 1.0 to 1.4).

One study found a significant association between a history of hard stool and faecal impaction on abdominal radiography (LR 1.2, 95% CI 1.0 to 1.4) whereas another study found a significant association between a finding of absent rebound tenderness and faecal impaction on abdominal radiography (LR 1.1, 95% CI 1.0 to 1.2). The association between stool present on rectal examination and faecal impaction on abdominal radiography was significant in one study (LR 1.6, 95% CI 1.2 to 2.0) but not in a second one (LR 1.5, 95% CI 0.8 to 2.3). The interobserver reliability ranged from moderate to excellent (k = 0.63 to 0.95) in five studies and from poor to moderate (k = 0.28 to 0.60) in one study. The intraobserver reliability was only evaluated in three studies and ranged from moderate (k = 0.52) to excellent (k ≥ 0.85).

A diagnostic case control study conducted in the Netherlands³² (2006) [EL=III] assessed the intra- and interobserver variability and determined diagnostic accuracy of the Leech method in identifying children with functional constipation. The study, which was carried out at a tertiary gastroenterology outpatient's clinic, included 89 non-selected consecutive children (median age 9.8 years), with a patients group of 52 constipated children. The 37 control children fulfilled the criteria for functional abdominal pain (FAP) (n=6) and for ‘functional non-retentive faecal incontinence’ (FNRFI) (n=31).

The mean Leech score (using the first score) was significantly higher in constipated children than in the control group (10.1 versus 8.5; P = 0.002). The mean colonic transit time (CTT) was significantly longer in constipated children than in the control group (92 hours versus 37 hours; P < 0.0001). The Leech method showed a sensitivity of 75% and a specificity of 59%. The positive predictive value and the negative predictive value were 72% and 63% respectively. The CTT showed a sensitivity of 79% and a specificity of 92% (with a cut-off point of 54 hours as used in the study). Using a cut-off point of 62 hours (as in the literature) the sensitivity decreased to 71% whereas the specificity improved to 95%. The positive predictive value was 69% and the negative predictive value was 97%.

The area under the curve receiver operator characteristic (ROC) was significantly smaller for the Leech method compared to the CTT (0.68, 95% CI 0.58 to 0.80 versus 0.90, 95% CI 0.83 to 0.96; P = 0.00015).

Two scorers produced significantly higher or lower scores in their repeat scoring of the same radiograph using the Leech method (intraobserver variability). Scorer 3 produced the largest difference (−1.6 [−2.0 to −1.3]; P < 0.0001) while the second score of scorer 2 was on average 0.7 points lower (0.03 [−0.4 to −0.5]; P = 0.0005). The two scores of scorer 1 were not systematically different (0.7 [0.2 to 1.2]; P = 0.89). Differences between repeated scores of the same scorer showed large variability, even after accounting for a systematic error (scorer 1: SD 2.2, limits of agreement −6.0 to 5.0; scorer 2: SD 2.2 limits of agreement −7.0 to 7.0 and scorer 3: SD 1.5 limits of agreement −5.0 to 3.0). These ‘limits of agreement’ are large in comparison to the scale on which the Leech score is measured. Analysis of interobserver variability of the Leech method showed that scorer 3 scored consistently lower than scorer 1 (mean of differences 2.7; P < 0.000) and scorer 2 (mean of differences 2.9; P < 0.0001). No systematic differences were found between scorer 2 and scorer 1. In 5% of cases the Leech scores of the same patient produced by different scorers could differ by four points or more. It should be noted that positive and negative predictive values (PPV, NPV) depend upon disease prevalence and reference to these is not helpful in case–control studies.

A diagnostic retrospective case series conducted in the Netherlands³³ (2006) [EL=III] assessed the reproducibility of three scoring systems (Barr, Leech and Blethyn) for plain abdominal radiography, in order to determine which one is most useful in clinical practice. Clinical records of 40 consecutive patients (mean age 7 years) referred to hospital for assessment of constipation were reviewed. Patients complained of infrequent defecation, soiling, encopresis or abdominal pain. Masked abdominal radiographs of the children were independently evaluated by two observers, both of whom were experienced paediatric radiologists. Observers assessed each radiograph on two separate occasions, 6 weeks apart.

The Leech score showed the highest reproducibility with high intraobserver agreement for both observers (k = 0.88 and k = 1.00 respectively), and high interobserver agreement (k = 0.91 in the first round and k = 0.84 in the second round). The Barr score showed a fair intraobserver agreement for both observers (k = 0.75 and k = 0.66 respectively) but a moderate interobserver agreement in the first round (k = 0.45). Interobserver agreement improved in the second round (k = 0.71). The Blethyn score showed the lowest reproducibility with low intraobserver agreement for both observers (k = 0.61 and k = 0.65 respectively) and also low interobserver agreement (k = 0.31 in the first round and k = 0.43 in the second round). All k values were statistically significant (P < 0.05).

One diagnostic case control conducted in the USA³⁴ (2005) [EL=III] evaluated the relationship between a history of constipation, faecal loading on X-rays and a history of urinary tract infections (UTIs) in an office practice. The study included 133 children (mean age 5.6 years). Patients were 100 children with a history of UTIs who were already undergoing a voiding cystourethrogram while the 33 controls were children undergoing a plain film of the abdomen for reasons that did not include constipation or UTIs. Faecal load on abdominal radiograph was compared to clinical variables: number of bowel movements per week and stools consistency. The correlation between symptoms of constipation and faecal load on abdominal X-ray was poor (correlation coefficient 0.08).

Evidence statement

One systematic review [EL=III] of six studies found conflicting evidence for the association between a clinical diagnosis of constipation and a radiographic diagnosis of constipation.

One case control study [EL=III] found that the Leech scoring method showed poor diagnostic accuracy and reproducibility.

One retrospective case series [EL=III] showed that the Leech scoring was highly reproducible.

One case control study [EL=III] showed poor correlation between symptoms of constipation and faecal load on abdominal X-ray.

GDG interpretation of the evidence

The GDG is aware that many of the children attending hospital with symptoms of constipation may have a plain abdominal radiography as a routine test to confirm idiopathic constipation and that subsequent treatment is based on the result. However, the evidence shows that the plain abdominal radiography has little or no value to either confirm or refute a diagnosis of idiopathic constipation.

It is the GDG's view that a plain abdominal radiography should only be performed if absolutely necessary and that it is not in the majority of cases of children with chronic constipation. Clinical features obtained from the history-taking and the physical examination would usually allow diagnosis of chronic idiopathic constipation.

The GDG concluded that there may be occasional situations when a plain abdominal radiography is indicated and could be valuable. These include situations when a child has been treated for some time with little success, when there is suspicion that something else is going on that is not functional constipation, in specialist services to track progress in certain circumstances and when a child has been on large doses of laxatives and faecal matter turns soft and with no edges that can be felt on abdominal palpation.

Even when the dose of radiation given per radiography may be small, the GDG believes that it is not necessary to expose children to it when repetitive radiographies are performed, and overuse seems to be common practice. The GDG understands that abdominal radiography appearances are open to misinterpretation, usually over-estimating faecal loading or missing rectal impaction. It is the GDG's view that if radiographies are to be performed at all, a transit study may be most valuable.

It is the GDG's view that when a plain abdominal radiography needs to be performed the reasoning has to be clear and the best possible methodology used with minimal risk.

Clinical question

What is the diagnostic value of the rectal biopsy in children with chronic idiopathic constipation?

Studies considered in this section

Studies were considered if they:

• included neonates, infants, or children up to their 18th birthday with chronic idiopathic constipation undergoing rectal biopsy
• were not case reports
• were published in English.

No restrictions were applied on the publication date or country.

Overview of available evidence

The searches identified 199 articles and 26 articles were retrieved for detailed assessment. Of these, four studies were identified for inclusion in this review: two retrospective cohort studies and two retrospective case series.

Narrative summary

A retrospective cohort conducted in the USA¹⁶ (2003) [EL=II] tested the hypothesis in two cohorts of 315 children that key features in the history, physical examination and radiographic evaluation would enable the avoidance of unnecessary rectal biopsies. Cohort 1 consisted of 265 children presenting with constipation who had undergone rectal biopsy to diagnose Hirschsprung's disease (HD). Cohort 2 was a concurrent selected cohort of 50 children with idiopathic constipation (IC). Only patients with definite information were included, so the number of patients in each analysis varies due to missing data.

Delayed passage of meconium was defined as failure to pass meconium in the first 48 hours of life. These data were available in 59% of cases. Abdominal distension was determined from parental response to questionnaire or data noted during patients' visits. Enterocolitis was defined as diarrhoea associated with fever.

In the group where the onset of constipation occurred when they were under one year, significantly more children with HD reported delayed passage of meconium compared to children with IC (65% versus 13%; P < 0.05). Abdominal distension and vomiting were also reported in significantly more children with HD compared to children with IC (respectively 80% versus 42%; P < 0.05 and 72% versus 21%; P < 0.05). Faecal impaction requiring manual evacuation occurred in significantly more children with IC compared to children with HD (30% versus 6%; P < 0.05). There were no significant differences between children with HD and children with IC regarding enterocolitis. In the group where the onset of constipation occurred after age 1 year significantly more children with HD reported delayed passage of meconium compared to children with IC (81% versus 1%; P < 0.05) and also significantly more children with HD reported abdominal distension compared to children with IC (53% versus 7%; P < 0.05). No children with IC experienced vomiting compared to 23% of children with HD (P < 0.05). There were no significant differences between children with HD and children with IC regarding enterocolitis or faecal impaction requiring manual evacuation.

Data on the onset of symptoms was available for 46 patients with HD and 40 patients with IC. The average age at onset of symptoms for patients with HD was 8 months (range 1 day to 9 years). The distribution of the age of onset of symptoms was:

• 60% during first week of life
• 70% during first month of life
• 87% during first year of life
• 13% after 1 year.

The average age at onset of symptoms for patients with IC was 15 months (range 7 days to 16 years). The distribution of the age of onset of symptoms was:

• 15% during first week of life
• 55% during first month of life
• 68% during first year of life
• 32% after 1 year.

At least 34% of HD patients had the classic triad (delayed passage of meconium plus vomiting plus abdominal distension). At least one feature of the triad was noted in 98% of patients with HD. Only 60% of patients with IC had a history of delayed passage of meconium, vomiting or abdominal distension. All (100%) HD patients compared to 64% of IC patients had one or more of the following: delayed passage of meconium, vomiting, abdominal distension and a transition zone on contrast enema. Thirty-six percent of patients with constipation had none of these features.

A retrospective cohort conducted in Italy³⁵ (2007) [EL=II] described the clinical features of a group of patients with intestinal dysganglionoses (ID, a term comprising HD and intestinal neuronal dysplasia [IND]) along with a group of consecutive patients with IC, to compare them and to find out if the clinical criteria do exist to indicate rectal suction biopsy (RSB) in constipated children. The cohort included 141 patients with ID, with a median age of 20 months and a mean age of 44 months ± 67). A total of 1118 biopsies were performed on 429 patients (mean 2.6 each). In 63 patients (14.7%) biopsies were inadequate for a reliable diagnosis of absent submucosal layer. A diagnosis of ID was received by 143 patients (33.3%). Out of 143 patients, 96 fulfilled the inclusion criteria (49 IND and 47 HD). Forty-five consecutive patients with a diagnosis of IC out of the remaining 286 patients fulfilled the inclusion criteria and were consequently included, giving a total sample of 141.

In case of a negative RSB, idiopathic constipation was diagnosed according to Rome II criteria. Clinical variables (meconium passage, symptoms onset, intestinal obstruction, abdominal distension, reported enterocolitis, failure to thrive, palpable faecal masses and soiling) were retrospectively extracted from patients' notes.

There was failure or delay in the passage of meconium in 87% of children diagnosed with HD compared to 7% of children with IC (P < 0.001). The onset of symptoms occurred at under 1 year in 80% of children with IC compared to 96% of children with HD (P < 0.02). No child with IC experienced intestinal obstruction compared to 49% of children with HD (P < 0.001). Significantly more children with HD experienced abdominal distension and failure to thrive compared to children with IC (85% versus 20%; P < 0.001 and 27.5% versus 11%; P < 0.045, respectively). Significantly more children with IC experienced soiling compared to children with HD (46.5% versus 4%; P < 0.001). There were no significant differences between children with HD and children with IC regarding reported enterocolitis and presence of palpable faecal masses.

A retrospective case series conducted in the UK³⁶ (1998) [EL=III] developed criteria that would reliably and consistently identify children with HD and thereby avoid the trauma and expense of unnecessary rectal biopsies in the others. The case series included 141 children (aged 1 day to 13 years, gender not reported) who had rectal biopsies to exclude HD. Clinical variables (age at diagnosis, bleeding per rectum, anal fissures, severe behavioural and/or emotional problems, soiling and enterocolitis) were retrospectively extracted from patients' case notes. Constipation was defined as a decreased frequency of bowel movements (less than 3 per week) or a difficulty in defecation which is perceived by the parents as a problem, requiring medication (oral or rectal) or manual intervention by the parents.

Seventeen out of 141 children were diagnosed with HD. The age at diagnosis ranged from 1 day to 3 years, but most children were diagnosed when they were neonates (14 children at under 4 weeks, 1 child at 4 to 12 weeks, 1 child at 12 weeks to 1 year and 1 child at over 1 year). Ten children (58.8%) had a history of delayed passage of meconium (more than 48 hours after birth). The age of onset of constipation was under 4 weeks in all 17 children with HD. Eight children (47%) had a history of enterocolitis but no child had experienced bleeding per rectum, anal fissures, severe behavioural and/or emotional problems or soiling.

Of the 141 children, 124 were diagnosed with constipation. The age at diagnosis ranged from 1 day to 13 years, but most children were diagnosed when they were over 1 year (20 children at under 4 weeks, 12 children at 4 to 12 weeks, 14 children at 12 weeks to 1 year and 78 children at over 1 year). Seventeen children (13.7%) had a history of delayed passage of meconium (more than 48 hours after birth). The age of onset of constipation was under 4 weeks in 40 children, between 4 to 12 weeks in 32 children, between 12 weeks to 1 year in 22 children and over 1 year in 25 children. Thirty-seven children (30%) had experienced bleeding per rectum, 14 children (11%) anal fissures, 10 children (8%) severe behavioural and/or emotional problems and 16 (13%) soiling. No child with constipation had a history of enterocolitis. History of onset of constipation was available in 136 of the 141 children (96%). The five children in whom this history could not be obtained from the notes were all older than 1 year (including three teenagers) and none had HD.

A retrospective case series conducted in the UK³⁷ (2003) [EL=III] aimed to review the author's experience of rectal biopsy to exclude HD and the author's clinical criteria to perform rectal biopsy in these children. The case series included 182 patients (118 males, mean age 2.9 years, age range 2 days to 16 years) who presented with chronic constipation or intestinal obstruction and had rectal biopsy to exclude HD. All children underwent either rectal suction biopsy (RSB) (104 children) or full-thickness rectal biopsy (78 children). Clinical variables obtained were: meconium passage, constipation since birth, intestinal obstruction, failure to thrive and chronic abdominal distension.

Twenty-five patients (14%) were diagnosed with HD (mean age 3.64 months, range 2 days to 4 years). The 182 patients provided 355 specimens in which 79% of suction biopsies and 97% of full-thickness biopsies were adequate, including rectal mucosa and submucosal. In 20 children with HD the diagnosis was made at the first attempt by suction rectal biopsy. Repeat biopsies were performed on 14 (8%) of 182 patients because of inadequate initial biopsy, clarification of atypical inervation and confirmation of negative results.

Nineteen out of 104 patients who underwent RSB were under 1 year. Because five children (12 specimens) who were older than 1 year had inadequate suction biopsies at the beginning of the series, it was decided that RSB was not suitable for children over 1 year. Three patients with HD (ages 6 days, 12 days and 6 weeks) had a false negative in acetylcholinesterase staining. In these the diagnoses were later established from repeated biopsies: one full thickness biopsy, one laparotomy and one suction biopsy.

Of the children who passed meconium more than 48 hours after birth, 39% (16 of 41) were diagnosed with HD but only 5% of the children (6 of 114) who passed meconium under 24 hours after birth were diagnosed with HD. Of the children for whom data on passage of meconium was unknown, 6% (3 of 46) were diagnosed with HD.

Of the children who had constipation since birth, 32% (17 of 53) were diagnosed with HD. Of the children who presented with intestinal obstruction, 69% (9 of 13) were diagnosed with HD. Of the children who reported failure to thrive, 22% (4 of 18) were diagnosed with HD. Of the children who reported chronic abdominal distension, 23% (3 of 13) were diagnosed with HD. Figures for patients who may have had more than one symptom were not reported in the paper.

Evidence statement

One retrospective cohort [EL=II] showed that significantly more children with HD reported delayed passage of meconium, abdominal distension and vomiting compared to children with IC. In children under 1 year faecal impaction requiring manual evacuation occurred in significantly more children with IC compared to children with HD, but there were no significant differences between the two groups for children under 1 year regarding this clinical feature. There were no significant differences between children with HD and children with IC regarding enterocolitis. The average age at onset of symptoms for patients with HD was 8 months (range 1 day to 9 years) and for patients with IC it was 15 months (range 7 days to 16 years).

One retrospective cohort [EL=II] showed that significantly more children with HD reported failure or delay in the passage of meconium, intestinal obstruction, abdominal distension and failure to thrive compared to children with IC. Significantly more children with IC experienced soiling compared to children with HD. Symptoms onset occurred at under 1 year in significantly more children with HD compared to children with IC. There were no significant differences between children with HD and children with IC regarding reported enterocolitis and presence of palpable faecal masses.

One retrospective case series [EL=III] showed that most children were diagnosed with HD when they were neonates compared to most children with IC who were diagnosed when they were over 1 year. The age of onset of constipation was under 4 weeks in all children with HD. Significantly more children with HD had a history of delayed passage of meconium (more than 48 hours after birth) compared to children with constipation. Forty-seven percent of children with HD had a history of enterocolitis but no child had experienced bleeding per rectum, anal fissures, severe behavioural and/or emotional problems or soiling. No child with constipation had a history of enterocolitis, but symptoms like bleeding per rectum, anal fissures, severe behavioural and/or emotional problems or soiling were reported in most of them.

One retrospective case series [EL=III] showed that delayed passage of meconium (more than 48 hours after birth), constipation since birth, intestinal obstruction, failure to thrive or chronic abdominal distension were present in significantly more children diagnosed with HD compared to children diagnosed with constipation.

GDG interpretation of the evidence

Rectal biopsy is primarily indicated to confirm or refute the diagnosis of Hirschsprung's disease (HD) in children with relevant clinical features. The GDG is aware that many children are undergoing rectal biopsies which have been inappropriately requested from a clinical point of view. Parental pressure to establish a diagnosis, particularly when the child's symptoms do not improve with medical treatment, cannot be addressed by performing a rectal biopsy in children without clinical features of HD. The GDG understands from the evidence that there are clear features in a child's history that are good predictors of HD and that, if discovered, would increase the chances of a positive biopsy result. Clinicians should take time to elicit these features when taking a history and also make sure that there are no issues of treatment adherence that could explain why the child is not getting better.

Clinical question

What is the diagnostic value of transit studies in children?

Studies considered in this section

Studies were considered if they:

• included neonates, infants, or children up to their 18th birthday with chronic idiopathic constipation undergoing transit studies to aid diagnosis
• were not case reports
• were published in English.

No restrictions were applied on the publication date or country.

Overview of available evidence

A search was conducted for all radiological investigations (plain abdominal radiography, abdominal ultrasound and transit studies). A total of 646 articles were identified and 72 articles were retrieved for detailed assessment. Of these, 20 studies were identified for inclusion in this review: 11 diagnostic case control studies, 4 diagnostic prospective case series and 5 diagnostic retrospective case series.

Narrative summary

Evidence statement

One diagnostic case control study [EL=II] showed that the colonic transit time with radiopaque markers was more accurate at detecting children with functional constipation compared to the plain abdominal radiography read using the Leech score. One diagnostic case control [EL=III] showed a better reproducibility for the colonic transit time with radiopaque markers in detecting the presence of faecal retention compared to the plain abdominal radiography read using the Barr score.

Seven diagnostic case controls [EL=III] and one diagnostic prospective case series [EL=III] showed that collectively children with constipation have longer colonic transit times compared to children without constipation.

One diagnostic case control [EL=III] showed that colonic transit time was not significantly different in children with severe brain damage and constipation compared with children with no brain damage and functional faecal retention

Four diagnostic case controls, three diagnostic prospective case series [EL=III] and one diagnostic retrospective case series [EL=III] showed an association between clinical variables and length of colonic transit time. One diagnostic retrospective case series [EL=II] showed no significant association between clinical variables and length of colonic transit time.

GDG interpretation of the evidence

The GDG concluded that transit studies may be of value to inform clinical and surgical decision making in a small number of children with intractable constipation following referral to specialist services. It is the GDG's view that transit studies can help in demystifying constipation as a ‘psychological’ problem and facilitate communication with parents.

There is no clear evidence of what is ‘normal’ and the fact that a test comes back as ‘normal’ does not necessarily mean that the child is not constipated. The GDG believes that the results of the transit studies should be interpreted in the context of the clinical picture, the population and the clinical setting.

Different methods to measure transit time are used in different centres and there is no evidence to confirm which one is better.

Clinical question

What is the diagnostic value of the abdominal ultrasound in children with chronic constipation?

Studies considered in this section

Studies were considered if they:

• included neonates, infants, or children up to their 18th birthday with chronic idiopathic constipation undergoing abdominal ultrasound
• were not case reports
• were published in English.

No restrictions were applied on the publication date or country.

Overview of available evidence

A search was conducted for all radiological investigations (plain abdominal radiography, abdominal ultrasound and transit studies). A total of 646 articles were identified and 72 articles were retrieved for detailed assessment. Of these, four diagnostic case control studies and one diagnostic prospective case series were identified for inclusion in this review on abdominal ultrasound.

Narrative summary

A diagnostic case–control study conducted in the UK⁵⁷ (2004) [EL=III] investigated the accuracy of the transverse diameter of the rectum on ultrasonography as an additional parameter for diagnosing constipation in children with lower urinary tract dysfunction. Forty-nine children aged 5 to 13 years were enrolled in the study. Cases were 23 patients with a positive history of voiding dysfunction and constipation and controls were 26 urological patients without lower urinary tract dysfunction and a normal defecation pattern. The study was conducted at a hospital clinic.

The mean rectal diameter was significantly larger in constipated children than in children with a normal defecation pattern (4.9 cm [SD 1.01, 95% CI 4.4 to 5.3] versus 2.1 cm [SD 0.64, 95% CI 1.8 to 2.4]; P < 0.001). There was no significant difference in age between the two groups (P = 0.20) or in the period between the last time a stool was passed prior to the rectal measurement (P = 0.16). In all patients with voiding dysfunction and constipation the rectal examination confirmed stool in the rectum. It should be noted that none of the patients had a sensation to defecate during the investigation.

A diagnostic case–control study conducted in the UK⁵⁸ (2005) [EL=III] established normal values for the rectal crescent (diameter) in healthy children, compared them with the rectal crescent in children with constipation and explored whether pelvic ultrasound can help in establishing a diagnosis of megarectum. The study was conducted at a tertiary referral centre and 177 children were enrolled. Ninety-five children (median age 6.5 years) with a history of constipation of at least 6 months duration were compared to 82 children (median age 5.5 years) with no history of constipation or other anorectal or gastrointestinal problems and no previous anorectal surgery.

The median rectal crescent was significantly larger in children with constipation compared to healthy children (3.4 cm, range 2.10 to 7.0 cm, interquartile range [IQR] 1.0) versus 2.4 cm, range 1.3 to 4.2 cm, IQR 0.72, P < 0.001). A receiver operating characteristic analysis indicated good discrimination between rectal diameters of children with constipation and healthy children (area under the curve 0.847, 95% CI 0.791 to 0.904). The cut-off point for establishing the diagnosis of megarectum was set at 3.0 cm. There were no significant differences between the two groups in terms of age, weight and height (p values 0.114, 0.198 and 0.131 respectively). Results were adjusted for confounders (age, height and weight). Age and rectal diameter were significantly related (P < 0.0001): the older the child, the bigger the rectal diameter. It should be noted that time to last evacuation was not ascertained and authors acknowledged that this may influence the size of the rectal crescent.

A diagnostic case–control study conducted in Poland⁴³ (2007) [EL=III] determined both whether a new method of ultrasound (US) assessment of stool retention could be used as a method of identifying children with functional chronic constipation and whether children with an enlarged rectum and colon (as seen on US) should be referred for further procedures such as proctoscopy and assessment of colonic transit time. The study was conducted at a gastroenterology outpatient clinic and 225 children were enrolled. One hundred and twenty children (mean age 6.25 years) with chronic constipation were compared to 105 children with normal defecation pattern (mean age 8.25 years).

The diameter of the rectal ampulla measured by US was significantly larger in constipated children than in the control group (mean 43.06 mm ± 9.68 versus 31.83 mm ± 8.24). The diagnosis of megarectum was based on the measurement of the rectopelvic ratio. The rectopelvic ratio for all ages was significantly bigger for the constipated children as compared to the control group (mean 0.22 ± 0.05 versus 0.15 ± 0.04). The cut-off value to diagnose megarectum was 0.189. Children with faecal impaction (as per US) had significantly longer average segmental transit time for the rectum, sigmoid and left colon (P < 0.001, P = 0.0015 and P = 0.0104 respectively). There was no statistically significant difference for the right side of the colon. Children with an overfilled splenic flexure on US had a significantly longer transit time in the left side of the colon (P = 0.0029). A sensitivity of 88.3% was reported for the US compared with proctoscopy in the diagnosis of faecal impaction. No value for specificity was reported.

A diagnostic case–control study conducted in Denmark⁵⁹ (2008) [EL=III] looked into a possible correlation between a dilated rectum measured by US and a faecal mass detected by digital rectal examination, and evaluated whether this method could diagnose constipation according to Rome III criteria. Fifty-one children aged 4 to 12 years were enrolled in the study. Twenty-seven children (mean age 7.0 years) diagnosed with chronic constipation were compared to 24 healthy children (mean age 9.1 years). Constipated children had been referred to an outpatient clinic with either constipation or faecal incontinence, with or without urinary incontinence and with a history of urinary tract infection. All constipated children fulfilled Rome III criteria.

The rectal diameter was significantly larger in children with rectal impaction compared to children without rectal impaction as per digital rectal examination (mean 40.5 mm ± 7.9 [2SD] versus 21.0 mm ± 4.2 [2SD]; P < 0.001). The cut-off value for the presence of rectal impaction was 29.4 mm. The rectal diameter was significantly larger in the constipated children compared to the healthy controls (mean 39.6 mm ± 8.2 [2SD] versus 21.4 mm ± 6.00 [2SD]; P < 0.001). The rectal diameter decreased significantly in children from the constipated group who responded to the laxative treatment (n=15) (mean 39.6 mm ± 8.2 versus mean 26.9 mm ± 5.6; P < 0.01) but still remained significantly greater than in the healthy children (P < 0.05). Eleven children did not respond to treatment and no significant differences were observed in their rectal diameter compared to that before treatment.

Seven of the constipated children (26%) had a rectal diameter smaller than the established cut-off point for rectal impaction, despite the fact that they fulfilled the Rome III criteria for constipation. Two healthy children with rectal impaction had a markedly larger rectal diameter (38 and 31 mm) than the other healthy controls. No correlation was found between the rectal diameter and the age or sex of the children in either group. There was no significant difference in height and weight distribution between the two groups, but the healthy children were significantly older than the constipated children. The intraobserver variability was small, as shown by a low coefficient of variation of the three consecutive measurements (5.8% ± 4.3%). There was no significant correlation between bladder volume at the time of measurement and rectal diameter (r=0.04). It should be noted that all investigations were performed by the same observer, a paediatric intern, who had no prior radiological experience.

A diagnostic prospective case series conducted in the UK⁶⁰ (2008) [EL=III] assessed the correlation between severity of constipation and US findings, the correlation between clinical examination and US findings and the correlation between findings at serial outpatient follow-up visits to assess clinical improvements and US findings. The case series included 500 children, both new referrals and follow-up, attending a constipation outpatient clinic (317 male, median age 8 years, age range 8 months to 18 years). There was a significant correlation between the mean severity symptom score (SSS) score and the mean US total score in all four visits. At the first visit (n=500) mean SSS was 23.5 (SD 11.6), mean US total score was 4.02 (SD 2.8), Pearson's correlation was 0.39; P < 0.001. At the second visit (n=226) mean SSS was 19.9 (SD 12.6), mean US total score 3.49 (SD 2.6), Pearson's correlation 0.49, P < 0.001. At the third visit (n=62) mean SSS was 23.02 (SD 13.7), mean US total score 3.66 (SD 2.6), Pearson's correlation 0.26; P = 0.04. At the fourth visit (n=12) mean SSS was 28.5 (SD 16.8), mean US total score 4.9 (SD 3.2), Pearson's correlation 0.70, P = 0.01.

There was a significant correlation between the US score and the clinical examination of palpable faeces in all four visits. At the first visit (n=500) mean palpable faeces score was 1.42 (SD 1.6), mean US total score 4.02 (SD 2.8), Pearson's correlation 0.89, P < 0.001. At the second visit (n=226) mean palpable faeces score was 1.10 (SD 1.6), mean US total score 3.49 (SD 2.6), Pearson's correlation 0.845, P < 0.001. At the third visit (n=62), mean palpable faeces score was 1.10 (SD 1.6), mean US total score 3.66 (SD 2.6) Pearson's correlation 0.77, P < 0.001. At the fourth visit (n=12) the mean palpable faeces score was 1.92 (SD 1.7), mean US total score 4.9 (3.2), Pearson's correlation 0.91, P < 0.001). It should be noted that no control group was included in the study and that the population size became very small at the fourth visit.

Evidence statement

Four case control studies [EL=III] showed that the rectal diameter as measured by abdominal ultrasound was significantly larger in constipated children than in children with a normal defecation pattern.

Two case control studies [EL=III] showed that abdominal ultrasound made a good discrimination between rectal diameters of children with constipation and healthy children.

One case control study [EL=III] showed that the rectal diameter as measured by abdominal ultrasound was significantly larger in children with rectal impaction as compared to children without rectal impaction as diagnosed per DRE.

One case control study [EL=III] showed that the rectal diameter as measured by abdominal ultrasound decreased significantly in constipated children who responded to laxative treatment but still remained significantly greater than in healthy children.

One case control study [EL=III] showed a good reproducibility for the abdominal ultrasound in measuring the rectal diameter in constipated and healthy children.

One diagnostic prospective case series [EL=III] showed a significant correlation between the severity of constipation and abdominal ultrasound findings, and between clinical examination and abdominal ultrasound findings

GDG interpretation of the evidence

There is no evidence that the abdominal US adds any useful information over and above that ascertained through thorough physical examination and history-taking in the diagnosis of chronic idiopathic constipation. The GDG is aware that the US is used in practice and it is its view that further research may demonstrate its usefulness in follow-up to indicate response to therapy and facilitate prognosis.