The reproducibility of the mandibular cortical index

To assess the reproducibility and diagnostic validity of Mandibular Cortical Index (MCI) when used by minimally trained observers.

Four 'experts' and 45 final-year dental students classified the appearance of the lower border of the mandibular cortex in 30 panoramic radiographs using the MCI. Experts viewed the original radiographs. The students received instruction in the MCI and viewed slides of the radiographs in a lecture theatre.

For intra-observer agreement, the experts had significantly higher overall values of weighted kappa, indicating substantial agreement in MCI assessment, whereas the students showed moderate agreement. For inter-observer agreement, there was fair agreement between the experts and poor agreement between the students. Using the experts' MCI assessment as the 'gold standard', the mean sensitivity of the students in diagnosis of C3 was 0.71 (maximum 0.95, minimum 0.25) and mean specificity was 0.56 (maximum 0.9, minimum 0.20).

The MCI has important limitations in terms of intra- and inter-observer agreement. Minimal training in its use, such as might be given in a lecture format to dentists, was ineffective and associated with poor inter-observer agreement and limited diagnostic validity in identifying signs of osteoporosis. More lengthy training and experience in using the MCI would be needed for it to be effective as a diagnostic tool in general dental practice.