Flexible and Interpretable Modeling of Overlapping Exposure Risks in Self-Controlled Case Series Analysis

Area of Science:

• Epidemiology
• Biostatistics
• Public Health

Background:

• Self-controlled case series (SCCS) is common for transient exposure-health event research.
• Traditional SCCS models struggle with overlapping exposures and complex multi-exposure interactions.
• Existing methods lack flexibility in modeling intricate exposure relationships.

Purpose of the Study:

• Introduce a novel semiparametric SCCS method using a functional partial-linear single index (PLSI) link function.
• Enable estimation of risks from overlapping exposure periods.
• Accommodate complex interactive effects among multiple exposures for enhanced interpretability and flexibility.

Main Methods:

• Developed a semiparametric SCCS model incorporating a functional PLSI link function.
• Consolidated multiple exposures into a single index for simplified analysis.
• Modeled complex interactions via a nonparametric link function.
• Validated through simulation studies and application to real-world datasets (MMR vaccination, malaria chemoprevention).

Main Results:

• The PLSI-SCCS model accurately estimates overlapping exposure risks.
• Demonstrated superior performance compared to standard methods in simulation studies.
• Successfully applied to real-world data, handling multiple, overlapping exposures effectively.
• The model provides greater interpretability and flexibility in analyzing exposure effects.

Conclusions:

• The PLSI-SCCS model is a robust tool for modern epidemiological and pharmaceutical research.
• Offers a nuanced understanding of exposure effects, especially in complex multi-exposure scenarios.
• Enhances the capability to handle overlapping and interactive exposure data in health event research.