A nonparametric method for estimating interaction effect of age and period on mortality

Area of Science:

• Biostatistics
• Epidemiology
• Statistical Modeling

Background:

• Accurate analysis of mortality data is crucial for public health.
• Understanding age and period-specific effects is essential for epidemiological studies.
• Existing models may not fully capture complex mortality patterns.

Purpose of the Study:

• To introduce a new statistical model for analyzing mortality data.
• To develop an effective estimation strategy for the proposed model.
• To apply the model to analyze rectum cancer mortality data.

Main Methods:

• A novel statistical model with the structure E[log qij] = mu + alpha i + beta j + rho ij was developed.
• Linear restrictions were imposed on model parameters.
• A combined technique of ANOVA and nonparametric smoothing was used for parameter estimation.

Main Results:

• The proposed model and estimation strategy were successfully applied to analyze mortality data.
• The methods allowed for the estimation of age effects, period effects, and interaction effects.
• The analysis provided insights into rectum cancer mortality trends in Japanese males and females.

Conclusions:

• The new model and estimation strategy offer a robust approach to analyzing mortality data.
• The methodology is applicable to various epidemiological datasets.
• This work contributes to a better understanding of cancer mortality patterns.