Approximate maximum likelihood estimation in cure models using aggregated data, with application to HPV vaccine...

Area of Science:

• Biostatistics
• Public Health
• Epidemiology

Background:

• Increasing rates of vaccine-preventable communicable diseases necessitate improved childhood immunization strategies.
• Estimating 'never-vaccinator' proportions is crucial for targeted public health interventions.
• Privacy concerns often restrict access to individual patient data (IPD), hindering traditional survival analyses.

Purpose of the Study:

• To develop and validate statistical methods for analyzing aggregated survival data.
• To accommodate a 'cured fraction' within survival models using only summary statistics.
• To address the challenge of analyzing vaccination uptake data when IPD is unavailable.

Main Methods:

• Proposed a novel statistical methodology for the analysis of aggregated survival data.
• Utilized a polynomial approximation of the mixture cure model log-likelihood function.
• Validated the method through simulation studies and application to a real-world human papillomavirus (HPV) vaccination dataset.

Main Results:

• The proposed statistical methodology effectively analyzes aggregated survival data.
• The method successfully accommodates a cured fraction, providing estimates of 'never-vaccinators'.
• Demonstrated applicability to real-world vaccination uptake studies, such as HPV vaccination.

Conclusions:

• The developed methods offer a viable approach for analyzing complex survival models with aggregated data.
• These techniques can overcome data privacy barriers and other concerns limiting IPD access.
• The methodology can be generalized for various public health research scenarios requiring survival analysis without individual-level data.