Premium
Joint analysis of correlated repeated measures and recurrent events processes in the presence of death, with application to a study on acquired immune deficiency syndrome
Author(s) -
Liu Lei,
Huang Xuelin
Publication year - 2009
Publication title -
journal of the royal statistical society: series c (applied statistics)
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.205
H-Index - 72
eISSN - 1467-9876
pISSN - 0035-9254
DOI - 10.1111/j.1467-9876.2008.00641.x
Subject(s) - covariate , repeated measures design , statistics , piecewise , inference , mathematics , random effects model , counting process , survival analysis , proportional hazards model , gaussian , medicine , econometrics , computer science , artificial intelligence , mathematical analysis , meta analysis , physics , quantum mechanics
Summary. In many longitudinal studies, we observe two correlated processes: a repeated measures process and a recurrent events process, both subject to a dependent terminal event. For example, in the ‘Terry Beirn community programs for clinical research on AIDS’ (CPCRA) study, higher CD4 cell counts are associated with lower risk of recurrent opportunistic diseases. They are also correlated with mortality, e.g. higher CD4 cell repeated measures and a lower rate of opportunistic disease imply better survival for patients infected with the human immunodeficiency virus. We propose a joint random‐effects model for the three correlated outcomes. The correlation is modelled by conditioning on shared random effects. Covariate effects can be taken into account in the model. Maximum likelihood estimation and inference are carried out through a Gaussian quadrature technique, assuming piecewise constant baseline hazards for recurrent events and death. The model can be fitted conveniently by Gaussian quadrature tools, e.g. SAS procedure NLMIXED. Simulation studies show that the estimation method yields satisfactory results. We apply this method to the CPCRA data.