Premium
Modeling event count data in the presence of informative dropout with application to bleeding and transfusion events in myelodysplastic syndrome
Author(s) -
Diao Guoqing,
Zeng Donglin,
Hu Kuolung,
Ibrahim Joseph G.
Publication year - 2017
Publication title -
statistics in medicine
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.996
H-Index - 183
eISSN - 1097-0258
pISSN - 0277-6715
DOI - 10.1002/sim.7351
Subject(s) - dropout (neural networks) , estimator , inference , event (particle physics) , count data , nonparametric statistics , computer science , statistics , econometrics , medicine , mathematics , artificial intelligence , machine learning , poisson distribution , physics , quantum mechanics
In many biomedical studies, it is often of interest to model event count data over the study period. For some patients, we may not follow up them for the entire study period owing to informative dropout. The dropout time can potentially provide valuable insight on the rate of the events. We propose a joint semiparametric model for event count data and informative dropout time that allows for correlation through a Gamma frailty. We develop efficient likelihood‐based estimation and inference procedures. The proposed nonparametric maximum likelihood estimators are shown to be consistent and asymptotically normal. Furthermore, the asymptotic covariances of the finite‐dimensional parameter estimates attain the semiparametric efficiency bound. Extensive simulation studies demonstrate that the proposed methods perform well in practice. We illustrate the proposed methods through an application to a clinical trial for bleeding and transfusion events in myelodysplastic syndrome. Copyright © 2017 John Wiley & Sons, Ltd.