Semiparametric model for recurrent event data with excess zeros and informative censoring

Zhao, Xiaobing; Zhou, Xian; Wang, Jing Long

doi:10.1016/j.jspi.2011.07.016

Cited by 7 publications

(9 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It supposes that this dependence is due to unmeasurable variables. Wang et al have included a latent parameter in a proportional model, Sun and Kang in an additive‐multiplicative rate model, and Zhao et al in an additive model that takes into account excess of zero when many patients experience no recurrent events. The coefficients are interpreted conditionally to the latent parameter in these models.…”

Section: Marginal Approachmentioning

confidence: 99%

How to analyze and interpret recurrent events data in the presence of a terminal event: An application on readmission after colorectal cancer surgery

Charles-Nelson

Katsahian

Schramm

2019

Statistics in Medicine

View full text Add to dashboard Cite

Recurrent events arise when an event occurs many times for a subject. Many models have been developed to analyze these kind of data: the Andersen-Gill's model is one of them as well as the Prentice-William and the Peterson's model, the Wei Lee and Weissfeld's model, or even frailty models, all assuming an independent and noninformative censoring. However, in practice, these assumptions may be violated by the existence of a terminal event that permanently stops the recurrent process (eg, death). Indeed, a patient who experiences an early terminal event is more likely to have a lower number of recurrent events than a patient who experiences a terminal event later. Thus, ignoring terminal events in the analysis may lead to biased results. Many methods have been developed to handle terminal events. In this paper, we describe the existing methods classifying into conditional or marginal methods and compare them in a simulation study to highlight bias in results if an inappropriate method is used, when recurrent events and terminal event are correlated. In addition, we apply the different models on a real dataset to show how results should be interpreted. Finally, we provide recommendations for choosing the appropriate method for analyzing recurrent events in the presence of a terminal event.

show abstract

Section: Marginal Approachmentioning

confidence: 99%

How to analyze and interpret recurrent events data in the presence of a terminal event: An application on readmission after colorectal cancer surgery

Charles-Nelson

Katsahian

Schramm

2019

Statistics in Medicine

View full text Add to dashboard Cite

show abstract

“…Second, tradeoffs are usually assessed from traits measured on different individuals within a population, leading to between‐individual variation, and the data generally consist of repeated measures (i.e. recurrent events, Zhao et al 2012) of a set of individuals sampled at different ages, leading to within‐individual variation. As tradeoffs are likely to vary with age and among individuals, both a marked individual heterogeneity and age‐specific changes in individual performance can lead to biased estimates of tradeoffs when not properly corrected for (van de Pol and Verhulst 2006).…”

Section: Distinctions Between Binary Data Bernoulli Distribution Anmentioning

confidence: 99%

A standardized approach to estimate life history tradeoffs in evolutionary ecology

Hamel

Yoccoz

Gaillard

2013

Oikos

View full text Add to dashboard Cite

As tradeoff s limit the maximum Darwinian fi tness individuals can reach, measuring reliably the strength of tradeoff s using appropriate metrics is of prime importance to understand the evolution of traits under constraints. Tradeoff s involving phenotypic traits and fi tness components, however, are diffi cult to quantify in free-ranging populations because of confounding eff ects due to environmental variation and individual heterogeneity. Furthermore, although some methods have been used previously to quantify tradeoff s, these methods cannot be applied with respect to binary traits, which are common to describe life histories (e.g. probability of reproduction, nesting success, off spring survival). Here, we demonstrate how to measure reliably the strength of tradeoff s involving binary traits using (auto)correlation estimates obtained from generalized linear (mixed) models. We fi rst propose a standardized approach that accounts for the variation in the nature of the tradeoff s being compared (e.g. continuous/binary traits, repeated/non-repeated measures), and then apply this method to longitudinal data from two contrasting species of large herbivores. Empirical estimates of tradeoff s varied among traits, and between-species comparisons suggested that reproductive tradeoff s between successive breeding attempts might only occur in capital breeders. Th e empirical results we obtained clearly demonstrate that the method we provide allows measuring reliably the strength of tradeoff s under most circumstances, including tradeoff s on binary traits. Our original approach therefore off ers an important fi rst step for comparing the strength and, hence, the relative importance of diff erent tradeoff s, and opens the door to a better understanding of the evolution of life history traits in free-ranging populations.A major goal of life history studies is to understand how natural selection shapes individual fi tness-related traits, such as growth, reproduction, and survival (Roff 1992, Stearns 1992. Without any resource limitation, a ' Darwinian demon ' , would evolve (Law 1979). Nevertheless, resources are usually limited so that individuals have to allocate resources to a given function at the cost of other functions (Williams 1966). According to this principle of energy allocation, the maximum fi tness individuals can reach is limited by tradeoff s that occur among fi tness components during a lifetime (Williams 1966). Although tradeoff s among life history traits have shaped the diversity of life history strategies we currently observe, tradeoff s involving phenotypic traits and fi tness components are diffi cult to measure in freeranging populations for two main reasons. First, tradeoff measures are subject to confounding eff ects of environmental variation in resource availability (van Noordwijk and de Jong 1986). Second, tradeoff s are usually assessed from traits measured on diff erent individuals within a population, leading to between-individual variation, and the data generally consist of repeated measure...

show abstract

“…This "induced-dependent censoring"complicates the statistical modeling of the recurrent event data and may lead to bias with traditional techniques of survival analysis. In order to model the recurrent events data with a possible terminal event, Zhao et al (2012) assume that the event process N i (.) to be a nonstationary Poisson process with an additive rate function:…”

Section: Model Presentationmentioning

confidence: 99%

“…For Model (2.1), regardless the competing risks, generalized estimating equations have been applied to estimate the base line λ 0 (t) and β in Schaubel et al (2006). Also Zhao et al (2012) suppose an alternative semiparametric model for recurrent event data by incorporation a proportion of zero-recurrence subjects into the additive rate model defined in the same way in Eq. 2.2.…”

Section: Divide Above Formula By Hmentioning

confidence: 99%

“…Hence treating death as ordinary censoring could inflate the estimate of the recurrence rate and underestimate the probability of zero-recurrence. More recently (Zhao et al, 2012) study the semi parametric modeling of recurrent event data, when data with excess zero have informative censoring and terminal event maybe occurred. But in all models except (Dauxois and Sencey, 2009) model of recurrent event data, studied without consideration of more risk factors led to fail subject.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Semiparametric Model for Recurrent Event Data Under Two Independent Competing Risks with Excess Zero and Informative Censoring

Sharifi

Hashemi

2022

Sankhya A

View full text Add to dashboard Cite

Recurrent event data are often encountered in longitudinal studies and in many other important areas such as Biomedical science, Econometrics, Reliability, Criminology and so on. Several marginal rates models have been used extensively to analyze recurrent event data, but often fail to fit the data adequately. In addition, the analysis is complicated by excess zeros in data as well as the presence of a terminal event. Furthermore effect of competing risks in recurrent event data analyze is rarely investigated. A semiparametric model with additive rate function is introduced in this paper in which unspecified baseline is applied to analyze recurrent event data with zero-recurrence and terminal event. Moreover multiple specified hazard function is used for analysis of competing risks, which includes a parameter to accommodate excess zeros and a frailty term to take into account for a terminal event. Taylor expansion is used to decompose baseline hazard function for each competing risk separately, and local likelihood procedure is applied to estimate parameters. Finally, a simulation study is presented for two specified baseline hazard functions with two different competing risks which approve the performance of presented model against existing models. An illustrative application based on a real bone marrow transplantation data is prepared.

show abstract

Semiparametric model for recurrent event data with excess zeros and informative censoring

Cited by 7 publications

References 33 publications

How to analyze and interpret recurrent events data in the presence of a terminal event: An application on readmission after colorectal cancer surgery

How to analyze and interpret recurrent events data in the presence of a terminal event: An application on readmission after colorectal cancer surgery

A standardized approach to estimate life history tradeoffs in evolutionary ecology

Semiparametric Model for Recurrent Event Data Under Two Independent Competing Risks with Excess Zero and Informative Censoring

Contact Info

Product

Resources

About