Reliability of intraoperative frozen sections in the management of Hirschsprung's disease

Background: The use of a 1-stage pull-through for Hirschsprung's disease (HD) is dependent on accurate identification of the normally innervated bowel on intraoperative frozen sections (IOFS). The authors wished to determine the incidence and sources of error during this process. Methods: All HD patients undergoing IOFS over a 15-year period were reviewed. Results: Three hundred four patients underwent a total of 700 IOFS. In 9 cases (3%), there was discrepancy between IOFS and permanent sections. Two of these were false-positive (ganglion cells incorrectly believed to be present at IOFS); both required a second operation as a result of the error. Seven were false-negative (presence of ganglion cells not recognized at IOFS); none required a subsequent operation, but 2 had a significantly more extensive colonic resection than was necessary. Responsible factors included sampling from the transition zone, freezing artifact, and misinterpretation of ganglion cells in very young patients owing to pathologist inexperience. There was significant variability in the error rate among the 11 pathologists. However, the numbers were too small for statistical analysis to determine whether there was a correlation between the rate of errors and the volume of cases done or years of experience. Conclusions: Error in reading of IOFS is rare but can have significant repercussions in patient care. Multiple factors, including technical issues and pathologist experience, may have a role in contributing to these errors. (C) 2004 Elsevier Inc. All rights reserved.