Analysing the temporal effects of age, period and cohort

Area of Science:

• Epidemiology
• Biostatistics
• Public Health

Background:

• Longitudinal trends in health data are crucial for understanding disease patterns.
• Analyzing age, birth cohort, and year of diagnosis simultaneously presents identifiability challenges in epidemiological studies.

Purpose of the Study:

• To address the identifiability issue when analyzing multiple temporal effects (age, birth cohort, diagnosis year) simultaneously.
• To propose methods for summarizing complex longitudinal trends and understanding temporal interactions.
• To illustrate how risk factor trends impact population rates using modeling approaches.

Main Methods:

• Partitioning temporal effects into linear and curvature components.
• Developing models that incorporate risk factor effects to analyze population-based rates.
• Utilizing lung cancer incidence and mortality data for conceptual illustration.

Main Results:

• The proposed partitioning approach offers a way to summarize trends when temporal effects are not simultaneously identifiable.
• Models incorporating risk factors can help elucidate their impact on population rates.
• The methods provide a framework for analyzing subgroup and temporal interactions.

Conclusions:

• The study provides a methodological framework for analyzing complex longitudinal trends in epidemiology.
• Understanding the interplay of age, cohort, and diagnosis year is essential for accurate trend analysis.
• The presented methods can enhance the interpretation of risk factor impacts on population health outcomes.