Objectives: Intraoperative frozen section analysis of surgical margins is w
idely used in head and neck cancer surgery. This study evaluates frozen sec
tion accuracy relative to permanent controls and final margins from the ent
ire specimen, the rate at which frozen sections impact intraoperative manag
ement, and the resultant cost. Study Design: Retrospective. Methods: From 1
997 to 1999 the frozen section results, permanent controls, and final tumor
margins from 80 consecutive patients undergoing 420 intraoperative frozen
section margins for head and neck malignancy were reviewed. Results: A 98.3
% accuracy rate (sensitivity, 88.8%; specificity, 98.9%) was found compared
with permanent sections of the same tissue. However, 40% (8 of 20) of pati
ents with positive final margins on the resection specimen, and 100% (15 of
15) with close (<5 mm) margins were not detected by frozen section analysi
s. The overall accuracy of frozen section in the evaluation of close or pos
itive final margins was 71.3% (sensitivity, 34.3%; specificity, 100%). In a
ddition, 5% (4 of 80) of patients potentially benefited from intraoperative
frozen section by virtue of immediate margin revision. The estimated cost
of intraoperative frozen section averaged as much as $3123 per patient, wit
h a cost-benefit ratio of 20:1. Conclusions: Intraoperative frozen section
margins are accurate, but they are costly and cannot reliably eradicate pos
itive final margins. Patients with early-stage lesions and those undergoing
re-resection for recurrence or salvage surgery after radiation failure der
ived the greatest potential benefit from frozen section margins.