The Leech method for diagnosing constipation: intra- and interobserver variability and accuracy.
BACKGROUND: The data concerning the value of a plain abdominal radiograph in childhood constipation are inconsistent. Recently, positive results have been reported of a new radiographic scoring system, "the Leech method", for assessing faecal loading. OBJECTIVE: To assess intra- and interobserver variability and determine diagnostic accuracy of the Leech method in identifying children with functional constipation (FC). MATERIALS AND METHODS: A total of 89 children (median age 9.8 years) with functional gastrointestinal disorders were included in the study. Based on clinical parameters, 52 fulfilled the criteria for FC, six fulfilled the criteria for functional abdominal pain (FAP), and 31 for functional non-retentive faecal incontinence (FNRFI); the latter two groups provided the controls. To assess intra- and interobserver variability of the Leech method three scorers scored the same abdominal radiograph twice. A Leech score of 9 or more was considered as suggestive of constipation. ROC analysis was used to determine the diagnostic accuracy of the Leech method in separating patients with FC from control patients. RESULTS: Significant intraobserver variability was found between two scorers (P=0.005 and P<0.0001), whereas there was no systematic difference between the two scores of the other scorer (P=0.89). The scores between scorers differed systematically and displayed large variability. The area under the ROC curve was 0.68 (95% CI 0.58-0.80), indicating poor diagnostic accuracy. CONCLUSIONS: The Leech scoring method for assessing faecal loading on a plain abdominal radiograph is of limited value in the diagnosis of FC in children.