Additive-Multiplicative Rates Model for Recurrent Event Data with Intermittently Observed Time-Dependent Covariates

Area of Science:

• Biostatistics
• Epidemiology
• Survival Analysis

Background:

• Recurrent event data analysis often uses proportional rates and additive rates models, which have restrictive assumptions on covariate effects.
• The additive-multiplicative rates model offers greater flexibility by allowing both additive and multiplicative covariate effects but requires continuous covariate monitoring.
• Intermittent observation of time-dependent covariates in practice limits the applicability of existing additive-multiplicative rates model estimation methods.

Purpose of the Study:

• To develop a semiparametric estimation method for the additive-multiplicative rates model accommodating intermittently observed time-dependent covariates.
• To compare the performance of the proposed method against simpler imputation techniques like last covariate carried forward and linear interpolation.
• To apply the novel method to an epidemiologic study investigating the impact of time-varying streptococcal infections on pharyngitis risk in school children.

Main Methods:

• Development of a semiparametric regression model for recurrent event data with intermittently measured covariates.
• Simulation studies to evaluate the statistical properties and performance of the proposed estimator.
• Application of the method to real-world epidemiologic data on childhood pharyngitis and streptococcal infections.

Main Results:

• Simulation results demonstrate the effectiveness of the proposed semiparametric estimator compared to traditional imputation methods.
• The study successfully applied the new method to analyze the risk of pharyngitis associated with time-varying streptococcal infections.
• An R package ('rectime') is provided for implementing the proposed methodology.

Conclusions:

• The proposed semiparametric method provides a viable and more accurate approach for analyzing recurrent events when time-dependent covariates are observed intermittently.
• This advancement extends the utility of the additive-multiplicative rates model in practical epidemiological and biostatistical research.
• The availability of the R package facilitates the adoption and application of this method in future studies.