Accuracy of and reasons for frozen sections: a correlative, retrospective study.

Previous studies on the accuracy of frozen sections (FS) were not based on the reasons for which the FS was performed. By omitting this important information, those studies reported FS accuracy of 94% to 97.4%. The current study analyzes the accuracy of FS diagnoses based on the reasons for which they were conducted. Of 482 specimens examined by FS in 1986, 41 were performed for evaluation of section margins of tumors, 29 for identification of unknown tissue, and 43 for detection of lymph node metastases. All 113 of these examinations proved to be 100% accurate. The remaining 369 FS were performed for diagnosis of an unknown pathologic process. Of these, 83.47% were precisely diagnosed; in 10.30%, the pathologic process was correctly, but not precisely, diagnosed; in 3.79%, the diagnosis was deferred; and the remaining 2.44% were incorrectly diagnosed (with no harmful consequences to the patients). By eliminating the cases of deferred diagnoses, the accuracy rate increased to 86.76%. Therefore, we suggest that the pathologist and the surgeon should not draw any conclusion from an inconclusive interpretation of FS and should proceed as though an FS had not been performed. When combining results of "precise diagnoses" with those of "correct pathologic process," the overall accuracy rate increased to 97%. The investigators conclude that FS should be used for a general diagnosis of a pathologic process rather than for an exact or precise diagnosis. Finally, we suggest that a similar survey of FS be periodically performed in every pathology department as part of its quality assurance program.