Premium
Nested case–control data utilized for multiple outcomes: a likelihood approach and alternatives
Author(s) -
Saarela Olli,
Kulathinal Sangita,
Arjas Elja,
Läärä Esa
Publication year - 2008
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.3416
Subject(s) - covariate , computer science , missing data , statistics , stability (learning theory) , nested case control study , selection (genetic algorithm) , control (management) , econometrics , data mining , mathematics , machine learning , cohort , artificial intelligence
Suppose a nested case–control design has been applied for collecting covariate data when studying a specific disease. With possible new outcomes of interest it would be sensible to utilize the previously selected control group instead of (or in addition to) a new control selection, given that the same covariate data were relevant and available, and that their measurements had adequate stability and quality. We formulate this problem in the framework of the competing risks survival model. In this approach covariate information collected for all outcomes can be utilized in the analysis. We not only propose likelihood‐based parameter estimation but we also review alternative methods based on weighted partial/pseudolikelihoods. The methods discussed here are closely related to the analysis of a case–cohort design, where the control group is not tied to cases of a specific disease. The different methods are compared in a simulation study. Copyright © 2008 John Wiley & Sons, Ltd.