0096 Marginal structural models for risk or prevalence ratios and differences
针对队列和横断面研究中常见结局变量，提出边际结构log-binomial模型来估计风险比或患病率比及其差值，解决了传统log-binomial模型不收敛的问题。
<sec><st>Objectives</st> Occupational epidemiologists often analyse binary outcomes in cohort and cross-sectional studies using multivariable logistic regression models, yielding estimates of adjusted odds ratios. When the outcome is common the adjusted odds ratio will not closely approximate the covariate-adjusted risk or prevalence ratio. Consequently, investigators may decide to directly estimate the risk or prevalence ratio using a log-binomial regression model; however, such models tend to be unstable and may not converge. </sec> <sec><st>Method</st> A marginal structural log-binomial model can be used to estimate risk and prevalence ratios and differences. The approach reduces problems with model convergence typical of log-binomial regression by shifting all explanatory variables except the exposures of primary interest from the linear predictor of the outcome regression model to a propensity score model for the exposure. The approach also facilitates evaluation of departures from additivity in the joint effects of two exposures. </sec> <sec><st>Results</st> We illustrate the proposed approach using data from several illustrative occupational studies of common outcomes. </sec> <sec><st>Conclusions</st> The proposed approach facilitates analysis of risk or prevalence ratios and differences in cohort and cross-sectional studies with common outcomes. </sec>