Self-controlled case series method with smooth age effect.

Area of Science:

• Epidemiology
• Biostatistics
• Pharmacovigilance

Background:

• The self-controlled case series (SCCS) method is vital for vaccine safety surveillance, assessing associations between vaccines and adverse events.
• Existing parametric and semiparametric SCCS methods have limitations, including potential bias from age mis-specification and computational challenges with large datasets.

Purpose of the Study:

• To introduce a novel penalized likelihood approach using splines for SCCS analysis.
• To address the limitations of existing parametric and semiparametric SCCS methods, particularly concerning age effect modeling and computational efficiency.

Main Methods:

• Utilized a penalized likelihood framework with cubic M-splines to model age-specific relative incidence.
• Employed integrated splines for estimating cumulative relative incidence.
• Conducted simulation studies to compare the spline-based approach with traditional parametric and semiparametric SCCS methods.

Main Results:

• The proposed spline-based method demonstrated performance equivalent to existing approaches in small sample sizes.
• The new method proved efficient and effective for large datasets, overcoming computational limitations.
• The approach was successfully applied to real-world data on febrile convulsions and pediatric vaccines.

Conclusions:

• The spline-based penalized likelihood approach offers a robust and computationally efficient alternative for SCCS analyses.
• This method enhances the reliability of vaccine safety studies, particularly with large-scale epidemiological data.
• The findings support the broader adoption of spline-based modeling in public health research and adverse event monitoring.