Multiplicative rates model for recurrent events in case-cohort studies

Area of Science:

• Biostatistics
• Epidemiology
• Public Health Research

Background:

• Large prospective cohort studies incur high costs due to extensive covariate and follow-up data collection.
• Rare events and expensive variables can render traditional cohort studies infeasible.
• Existing case-cohort designs address time-to-event data but not recurrent events.

Purpose of the Study:

• To propose novel case-cohort sampling schemes specifically designed for recurrent event data.
• To develop a statistical methodology for analyzing recurrent events under these new sampling schemes.
• To evaluate the performance of the proposed methods through simulations and a real-world case study.

Main Methods:

• Development of case-cohort sampling strategies tailored for recurrent event data.
• Application of a multiplicative rates model for recurrent event analysis.
• Utilizing a weighted estimating equations approach for parameter estimation.
• Consistency and asymptotic normality of the proposed estimators were theoretically established.

Main Results:

• The proposed weighted estimating equations approach yields consistent and asymptotically normal estimators.
• Simulation studies demonstrated good performance of the proposed estimator in finite samples.
• The methods were illustrated by examining the link between measles and acute lower respiratory tract infections (ALRI) in Brazilian children.

Conclusions:

• The proposed case-cohort sampling schemes offer a cost-effective approach for studying recurrent events in large cohorts.
• The developed statistical methods provide a valid framework for parameter estimation in recurrent event analysis.
• This research has significant implications for epidemiological studies involving recurrent health outcomes.