Permutation tests for joinpoint regression with applications to cancer rates

Area of Science:

• Epidemiology
• Biostatistics

Background:

• Analyzing cancer incidence and mortality trends requires accurate identification of trend changes.
• Existing methods may not adequately handle complex data variations.

Purpose of the Study:

• To develop and evaluate a robust statistical method for detecting significant changes in cancer trend data.
• To apply this method to U.S. prostate cancer incidence and mortality rates.

Main Methods:

• Application of a joinpoint regression model with a grid-search method to identify trend changes.
• Utilizing permutation tests and Monte Carlo methods for significance testing.
• Extension of methods to accommodate non-constant variance and autocorrelated errors.

Main Results:

• The study successfully identifies significant joinpoints in cancer trend data.
• The methods are validated through simulations and applied to real-world prostate cancer data.
• The approach accounts for variations like Poisson variation and autocorrelation.

Conclusions:

• The joinpoint regression and permutation testing approach provides a reliable method for analyzing cancer trend changes.
• This methodology enhances the understanding of cancer incidence and mortality dynamics.
• The study demonstrates the utility of the developed tests on U.S. prostate cancer data.