A marginal cure rate proportional hazards model for spatial survival data

Area of Science:

• Biostatistics
• Dental Research
• Epidemiology

Background:

• Dental studies generate complex, spatially referenced time-to-event data, often involving tooth loss due to periodontal disease.
• These datasets are characterized by spatial variations, heavy censoring, and inherent clustering due to the dependence between nearby teeth.

Purpose of the Study:

• To develop and validate a novel statistical model for analyzing multivariate time-to-event data in dental research.
• To accurately identify risk factors associated with tooth loss and predict individual patient outcomes, such as remaining tooth lifespan.

Main Methods:

• A proportional hazards model with a surviving fraction was employed to handle clustered, correlated, and censored time-to-event data.
• Spatial frailties, modeled as linear combinations of positive stable random effects, were incorporated to account for the dependence between nearby teeth.

Main Results:

• The proposed model allows for predictions that incorporate the survival status of neighboring teeth while maintaining interpretable marginal proportional hazards.
• Simulation studies and application to a real dental dataset demonstrated the model's effectiveness in identifying tooth loss risk factors.

Conclusions:

• The developed statistical framework offers an advantageous approach for analyzing spatially referenced dental data compared to existing methods.
• This model enhances the ability to identify critical risk factors for tooth loss and provides more accurate predictions of tooth survival for individual patients.