Multilevel modeling in epidemiology with GLIMMIX

Area of Science:

• Epidemiology
• Biostatistics
• Statistical Software

Background:

• Multilevel modeling is a powerful technique for epidemiologic analysis.
• Complexity and lack of appropriate software limit its widespread application.
• Existing SAS software (GLIMMIX macro) is insufficient for complete epidemiologic analysis.

Purpose of the Study:

• To provide additional SAS code to enhance the GLIMMIX macro.
• To enable comprehensive epidemiologic output from multilevel models.
• To offer epidemiologists an easily usable tool for fitting multilevel models.

Main Methods:

• Development of supplementary SAS code for the GLIMMIX macro.
• Application of the enhanced code using data from the EURAMIC study (diet and breast cancer).
• Illustration of multilevel model fitting for epidemiologic research.

Main Results:

• The provided code successfully generates epidemiologic output from GLIMMIX.
• The enhanced tool simplifies the fitting of multilevel models for epidemiologists.
• Demonstrated utility with real-world data on diet and breast cancer.

Conclusions:

• The developed SAS code addresses the need for accessible software in multilevel epidemiologic analysis.
• This tool facilitates the application of complex statistical methods in public health research.
• Empowers epidemiologists to conduct more robust analyses using multilevel modeling.