Predictions of restoration deterioration.

This study evaluated the application of a mixture model involving a Weibull distribution function to predict the median times for restorations of three dental restorative materials to achieve unsatisfactory rating scores for six clinical factors. The accuracy of the method was assessed graphically against the known actuarial long-term deterioration observations of 1813 amalgam, 1774 anterior resin composite and 474 glass polyalkenoate (ionomer) cement restorations, assessed over periods of up to 20, 18 and 14 years, respectively. Of the six clinical factors investigated (which included marginal fracture), only four had sufficient long-term unsatisfactory rating score data to enable their median times to be predicted. These predicted times were: for amalgam restorations, surface roughness 32.5 years and surface tarnishing 16.0 years; for resin composites, marginal staining 25.4 years and colour mismatch 14.2 years; and for glass polyalkenoate (ionomer) cements, marginal staining 17.6 years and colour mismatch 3.6 years. The known and predictive unsatisfactory rating score results were generally in close agreement. However, it was not possible to predict median times for unsatisfactory rating scores associated with very slowly deteriorating restoration factors. The actual replacement rates of the amalgam restorations were too low to obtain their median survival time. However, for the faster failing resin composites this time was 7.9 +/- 0.5 years, and for the glass polyalkenoate (ionomer) cements 2.2 +/- 0.2 years. The relationship of restoration deterioration to restoration replacement and dental health requires further analysis.