Modeling age-specific cancer incidences using logistic growth equations: implications for data collection.

Area of Science:

• Epidemiology
• Mathematical Biology
• Biostatistics

Background:

• Cancer registries and surveillance systems generate vast data for epidemiological modeling.
• Current models like time series and age-period-cohort (APC) primarily focus on analysis and prediction, with less emphasis on optimizing data collection.
• There is a need to explore modeling approaches that can inform the design of more efficient cancer data collection initiatives.

Purpose of the Study:

• To model age-specific cancer incidence rates using logistic growth equations.
• To evaluate the performance of these models under various data completeness scenarios.
• To derive insights for improving cancer registry and surveillance data collection strategies.

Main Methods:

• Utilized the China Cancer Registry Report 2012 as the data source.
• Employed 3-parameter logistic growth equations to model age-specific incidence rates for all and top 10 cancers.
• Performed modeling using full age-span fitting, multiple 5-year segment fitting, and single-segment fitting.
• Assessed model performance using adjusted goodness of fit (combining sum of squared residuals and relative errors).

Main Results:

• Logistic growth models demonstrated excellent fit (goodness of fit R > 0.96) for most cancers when using full age-span data.
• Models derived from urban resident data showed performance greater than or equal to those from rural data.
• Models based on multiple 5-year segments maintained high performance (R > 0.89) even when up to 75% of segments were excluded.
• Single-segment models showed higher goodness of fit for older age groups.

Conclusions:

• Logistic growth models effectively describe age-specific cancer incidence rates for most cancers.
• These models offer a promising approach to inform and potentially reduce the data collection volume for cancer registries and surveillance.
• The findings suggest that optimized data collection strategies can be developed based on robust mathematical modeling.