A gap time model based on a multiplicative marginal rate function that accounts for zero-recurrence units

Abstract

In this article, we propose an alternative gap time model based on a multiplicative marginal rate function, which is formulated considering each gap time conditional on the previous recurrence times. In this formulation, the gap times are treated equally and the relation between successive events is no longer a problem. Furthermore, this article considers the inclusion of a proportion of zero-recurrence units (for which the event of interest will not occur) into the model to analyze recurrent event data. Inference aspects of the proposed model are discussed through maximum likelihood approach. A simulation study is carried out to examine the performance of the estimation procedure. The model is applied to hospital readmission data among colorectal cancer patients.

A parametric model to jointly characterize rate, duration, and severity of exacerbations in episodic diseases

“…As an example, [18] showed that ignoring the durations can lead to serious bias in estimating the rate of COPD exacerbations. Moreover, the model addresses the issue of the excessive presence of individuals without any exacerbations during follow-up (zeroinflation), which has been frequently discussed in the literature [13][14][15]. Importantly, the proposed framework can be implemented with standard statistical software; to enhance its accessibility, an annotated SAS macro that implements the model in a generic fashion, along with a manual and a simulated dataset, is available at our website (http:// resp.…”

“…Common analytic approaches include marginal models [7][8][9], modeling time to the first exacerbation in a survival analysis framework [10], using models for count outcomes [11,12], or employing models for recurrent events [13,14]. Recent developments in the analysis of exacerbation trials include the use of random effect models to account for between-individual variability in exacerbation rates [13,14], and the use of a mixture of a random effect or a gap-time model with a logistic regression model to allow for the excessive presence of individuals without any exacerbations ("zero inflation") during follow-up [13][14][15]. Another approach could be using marginal regression analysis of recurrent point processes (and their extensions) for recurrent events (see, e.g., [16]).…”

Bayesian modelling of zero‐inflated recurrent events and dependent termination with compound Poisson frailty model

“…Recurrent event data provide a typical example of multivariate survival data that commonly arise in a wide variety of situations, such as demography and reliability, engineering and medical studies (Louzada, Macera, & Cancho, 2017). Examples include repeated tumour occurrences in bladder cancer patients (Byar, 1980), multiple hospital readmissions following surgery for colorectal cancer (González Ruiz et al, 2005), and successive exacerbations of respiratory disorders among patients with cystic fibrosis (Therneau & Hamilton, 1997).…”