Are interpretations of video defecographies reliable and reproducible?

Video defecography is a dynamic investigation which can influence surgical decision making in constipated patients. A study was therefore undertaken to assess the inter and intraobserver variability in video defecography. Specifically, we sought to assess the interpretation of video defecographies by a group of observers with the same training, guidelines and standards. To determine interobserver variation, four independent observers, two blinded to the patient's history, reviewed 100 randomly sequenced video defecographies performed in constipated patients. The presence or absence of sigmoidocele, rectocele, intussusception or prolapse was noted. Adequate or improper function of the puborectalis, anal canal opening, anorectal angle (ARA) and grade of emptying of the rectum were also assessed. Two weeks after the initial assessment, intraobserver variation was determined by a repeat blinded review of unlabelled randomly sequenced studies. The results of interobserver accuracy for sigmoidoceles, rectoceles, intussusception, rectal prolapse, rectal emptying, opening of the anal canal, puborectalis contraction and straightening of the ARA and rectal emptying were 89.5%, 46.0%, 87.5%, 97.5%, 86.5%, 88.5%, 83.0%, and 80.0%, respectively. The intraobserver variations were 88.5%, 83.8%, 80.5%, 94.5%, 77.0%, 84.8%, 80.5% and 85.5%, respectively. Prior knowledge of the patient's history did not significantly influence the outcome. In summery, video defecography has an overall accuracy of 83.3% and as such is a valid tool in assessing constipated patients.