Angela Chen
Title: Stratified Regression Analysis of Zero-Truncated Recurrent Event Data with Applications in Pediatric Mental Health Care
Date: July 25th, 2025
Time: 10:30am
Location: LIB 2020
Supervised by: X. Joan Hu & Rhonda Rosychuk
Abstract:
This dissertation is motivated by a pediatric mental health care (PMHC) program in which records of mental health-related emergency department (MHED) visits were extracted from population-based administrative databases. We formulate these MHED visits as recurrent events and develop statistical methods to analyze the zero-truncated data.
The dissertation begins with the PMHC program as a motivating example, followed by descriptive analyses of the MHED data that inspire subsequent methodological development. A literature review is also provided to contextualize our research within the broader field.
A particular interest is in understanding how the event occurrence depends on the occurrences in the past. We first consider a stratified Cox regression model with time-independent coefficients and propose a procedure for estimating the model parameters using the zero-truncated recurrent event data integrated with population census information. We also compare the proposed estimator with a maximum likelihood estimator based on zero-truncated data only.
Recognizing that covariate effects may vary with time, we extend the model to allow time-varying coefficients and introduce a corresponding estimation procedure using the same integrated data. To further investigate how event patterns evolve over time, we consider a stratified regression model with an indicator for the late time period as an additional covariate. With appropriate reformulation, the previously developed method is adapted for this setting.
We establish the consistency and asymptotic normality of the proposed estimators, and evaluate their finite-sample performance through simulation studies. Throughout the dissertation, the MHED data are used to motivate the methods and demonstrate their applications.