INTERPRETING MULTIPLE LOGISTIC REGRESSION COEFFICIENTS IN PROSPECTIVE OBSERVATIONAL STUDIES

摘要: Multiple logistic models are frequently used in observational studies to assess the contribution of a risk factor disease while controlling for one or more covariates. Often, covariates correlated with factor, resulting multiple coefficients that difficult interpret. This paper highlights problem assessing magnitude coefficient and proposes supplemental procedure usual analysis describing relationship between disease. An example is given, along results not apparent when considered alone. Conclusions presented important biologic if effect influenced by correlation covariate.