Prediction of amalgam restoration longevity

Area of Science:

• Dental Materials Science
• Biostatistics
• Predictive Modeling

Background:

• Dental amalgam restorations are widely used.
• Predicting the longevity of dental restorations is crucial for treatment planning and resource allocation.
• Actuarial methods are a traditional approach for survival analysis.

Purpose of the Study:

• To evaluate the predictive accuracy of the Weibull distribution model for dental amalgam restoration longevity.
• To compare Weibull distribution predictions against traditional actuarial methods.
• To assess the influence of data duration on prediction accuracy.

Main Methods:

• Utilized a dataset of 1345 amalgam restorations from 100 Royal Australian Air Force members monitored for at least 10 years.
• Applied maximum likelihood estimates to fit a Weibull distribution.
• Compared Weibull survival curves with actuarial survival curves derived from the same data.
• Assessed prediction accuracy using 6-year actuarial survival data.

Main Results:

• Weibull distribution curves closely approximated actuarial survival curves.
• Weibull estimates based on less than 5-6 years of survival data showed poor long-term prediction.
• Using 6-year data, predicted and observed survival curves showed less than 10% disagreement.
• Twenty-five percent of restorations failed within 6 years based on 6-year data, versus 22% in the first 6 years of the full 17-year dataset.

Conclusions:

• The Weibull distribution is a viable method for predicting amalgam restoration longevity when sufficient long-term data is available.
• Short-term survival data (under 5-6 years) significantly limits the Weibull model's ability to predict long-term restoration survival.
• Actuarial methods provide a reliable baseline, but Weibull analysis offers a more refined predictive tool with adequate data.