The accuracy of telediagnosis of frozen sections is inferior to that of conventional diagnosis of frozen sections and paraffin-embedded sections.

The accuracy of telepathology diagnosis and conventional diagnosis of frozen sections was compared, using the diagnosis established on paraffin-embedded tissue as a reference. Out of a total of 270 cases, remote frozen-section diagnosis was correct in 227 cases (84.1%) and incorrect in 23 cases (8.5%). The latter comprised 12 false positive diagnoses of malignancy (4.4%) and 11 false negative diagnoses (4.1%). A diagnosis was not possible in 20 cases (7.4%). In contrast, the conventional frozen-section diagnosis was correct in 269 cases (99.6%) and incorrect in 1 case (0.4%), the latter being a false negative diagnosis. The average time needed to make a remote diagnosis was 14.2 min (SD 9). Manual examination was not found to be essential for remote frozen-section diagnosis. Overall slide quality was rated as 'satisfactory' to 'fair' by the six pathologists concerned. An improvement in the quality of slides is necessary to guarantee an acceptable level of accuracy of remote frozen-section diagnosis; a shortening of the time needed for diagnosis is a further requirement for the successful implementation of a routine telepathology frozen-section service.