Regression models for multiple outcomes in large epidemiologic studies.

摘要: In situations in which one cannot specify a single primary outcome, epidemiologic analyses often examine multiple associations between outcomes and explanatory covariates or risk factors. To compare alternative approaches to the analysis of regression models, I used generalized estimating equations (GEE) multivariate extension linear incorporate dependence among from same subject provide robust variance estimates coefficients. applied methods hospital-population-based study complications surgical anaesthesia, using GEE model fitting quasi-likelihood score Wald tests. specification, allowed each covariate differ, yielding coefficient for outcome combinations; obtained covariances set outcome-specific coefficients 'sandwich' estimator. address problem inference, simultaneous that make adjustments test statistic p-values confidence interval widths, control type error coverage, respectively. second assumed common association covariate, eliminates multiplicity by use global association. an approach multiplicity, empirical Bayes shrink toward pooled mean is similar effect coefficient. models can flexible framework estimation testing outcomes.