Statistics and Its Interface

Volume 7 (2014)

Number 2

Joint modeling of survival data and mismeasured longitudinal data using the proportional odds model

Pages: 241 – 250

DOI: http://dx.doi.org/10.4310/SII.2014.v7.n2.a9

Authors

Juan Xiong (Department of Statistical and Actuarial Sciences, University of Western Ontario, London, On., Canada)

Wenqing He (Department of Statistical and Actuarial Sciences, University of Western Ontario, London, On., Canada)

Grace Y. Yi (Department of Statistics and Actuarial Science, University of Waterloo, Ontario, Canada)

Abstract

Joint modeling of longitudinal and survival data has been studied extensively, where the Cox proportional hazards model has frequently been used to incorporate the relationship between survival times and covariates. Although the proportional odds model is an attractive alternative to the Cox proportional hazards model by featuring the dependence of survival times on covariates via cumulative covariate effects, this model is rarely discussed in the joint modeling context. To fill up this gap, we investigate joint modeling of survival and longitudinal data where the proportional odds model is employed to feature survival data, and longitudinal covariates are postulated using measurement error models. An estimation method based on the expectation maximization algorithm is developed. In addition, the impact of naive analyses, which fail to address errors occurring in longitudinal measurements, is assessed. The performance of the proposed method is evaluated through simulation studies, and a real example is invoked for illustration.

Keywords

joint modeling, proportional odds model, measurement error, survival analysis, timevarying covariates

2010 Mathematics Subject Classification

62N01

Full Text (PDF format)