Objectives: Colonoscopy in children routinely includes the practice of obtaining multiple biopsy samples even in the absence of gross mucosal abnormalities. The aim of our investigation was to examine the level of agreement between endoscopic and histological findings during pediatric colonoscopy. We also investigated the predictors of agreement and abnormal histology.
Methods: We performed an institutionally approved retrospective review of consecutive patients who underwent diagnostic colonoscopy during a 6-month period. Descriptive analyses and regression models were used to determine agreement rates, as well as potential predictors of both agreement and abnormal histology.
Results: Of 390 included colonoscopies, endoscopists (n = 26) reported abnormal gross findings in 218 (56%) and pathologists (n = 4) found histopathology in 195 (50%). Considering histology as the criterion standard, endoscopy had a sensitivity of 90% and a specificity of 78%. Reports of grossly normal endoscopic findings were highly associated with agreement (odds ratio [OR] 1.9, P = 0.001). A known diagnosis of inflammatory bowel disease was a strong predictor of abnormal histology (OR 6.4, P < 0.0001). Abdominal pain as a procedural indication was a strong predictor for normal histology (OR 0.4, P < 0.0001).
Conclusions: The results of our study suggest good agreement between endoscopic and histological findings, especially when an endoscopist reports normal-appearing colonic mucosa. We identified predictors of abnormal histology to include known inflammatory bowel disease, whereas abdominal pain was found to be a negative predictor. Future studies are needed to determine evidence-based protocols for obtaining biopsies during colonoscopy in children.