Ten-year retrospective study of the effectiveness of quantitative percussion diagnostics as an indicator of the level...

Conventional dental diagnostic aids are only partially effective in diagnosing structural defects such as cracks in teeth. A more predictable diagnostic for structural instability in the mouth is needed.

The purpose of this clinical study with an increased population size was to evaluate the effectiveness of diagnosing structural instability by using the quantitative percussion diagnostics (QPD) system and to evaluate the influence of independent variables on the relationship between normal fit error (NFE) and observed structural instability found during the clinical disassembly of teeth.

Twenty-two participants with 264 sites needing restoration were enrolled in an institutional review board-approved 10-year retrospective clinical study. Each site had been tested with the QPD system before being disassembled microscopically with video documentation, and the clinical disassembly results were recorded on a defect-assessment sheet. The NFE data were separately recorded from the preexisting records. The classification of structural pathology based on the disassembly observations for each of the 264 sites was conducted by the clinical researcher (C.G.S.) who was blinded to the NFE values.

The 264 sites from 22 patients were classified as 8 in the none group, 87 in the moderate group, and 169 in the severe group based on the disassembly findings. The NFE data for the sites were analyzed by using the predefined NFE cutoffs that were independently generated from the previous cumulative logistic regression and decision tree model. For the cumulative logistic regression, 235 out of 264 sites were correctly classified with an agreement of 0.89 (adjusted 95% CI: 0.83-0.95). The number of correctly classified sites for the decision tree model was 234, and the agreement was also 0.89 (adjusted 95% CI: 0.83-0.94). For both cumulative logistic regression and decision tree models, the overall misclassification rate was less than 20% for any restoration material or restoration type. Therefore, the overall performance of NFE classification was consistently good, regardless of restoration material or type. In addition, the sensitivity of the severe category was above 90% for any restoration material or type for the decision tree model.

The QPD system was found to be a reliable diagnostic aid for classifying structural damage in the categories of none, moderate, or severe based on clinical disassembly findings under the clinical microscope and NFE values. Furthermore, it was determined that restoration type and restoration design were not significant factors in correlating structural pathology with NFE.