Am J Epidemiol 2004; 159:891-899.
Copyright © 2004 by the Johns
Hopkins Bloomberg School of Public Health
ORIGINAL CONTRIBUTIONS |
A Repeated Measures Approach for Simultaneous Modeling of Multiple Neurobehavioral Outcomes in Newborns Exposed to Cocaine in Utero
1 Statistics Research Division, Research Triangle Institute, Rockville, MD.
2 University of Kentucky Chandler Medical Center, Lexington, KY.
Multiple binary outcomes are encountered frequently in epidemiologic research. This work was motivated by the Maternal Lifestyle Study, 1993–1995, where newborns exposed prenatally to cocaine and a comparison cohort were examined for the presence of central and autonomic nervous system (CNS/ANS) signs. Thus, each infant contributed to multiple, possibly interrelated, binary outcomes that may collectively constitute one syndrome (even though specific outcomes that are affected by cocaine are of scientific interest). Because it is neither scientifically appropriate nor statistically efficient to fit separate models for each outcome, here we adopt a multivariate repeated measures approach to simultaneously model all the CNS/ANS outcomes as a function of cocaine exposure and other covariates. This formulation has a number of advantages. First, it implicitly recognizes that all the CNS/ANS outcomes may together constitute one syndrome. Second, simultaneous modeling boosts statistical efficiency by allowing for correlations among the outcomes, and it avoids multiple comparisons. Third, it allows for outcome-specific exposure effects, so that the specific signs that are affected by cocaine exposure can be identified.
binary data; generalized estimating equation; generalized linear mixed model; multiple comparisons; repeated measures; simultaneous inference; syndrome
Abbreviations: Abbreviations: CNS/ANS, central and autonomic nervous system; GEE, generalized estimating equation; GLMM, generalized linear mixed model.