0097 Matching and Counter-Matching on Propensity Scores in Nested Case-Control Studies
提出在巢式病例对照研究中用倾向性评分降维匹配，并引入反匹配策略提高效率，对职业队列研究中的混杂控制有用。
<h3>Objectives</h3> Occupational cohort studies often collect information on a primary exposure of interest and a small set of potential confounders. A nested case-control study may be undertaken to collect additional information on covariates that are potential confounders. <h3>Method</h3> We describe and illustrate an approach to matched nested case-control studies in such settings. The approach reduces problems of sparse data typical of matched nested case-control studies by reducing the set of matching on explanatory variables to a single dimension defined by a propensity score model for the exposure. We then consider additional uses of propensity score for matching. We first consider varying control: case ratio in matched designs to increase efficiency when exposure prevalence is rare. Next we consider a second propensity score that include additional covariates not in the set of strong confounders selected for matching. Counter matching on this score can further increase efficiency. <h3>Results</h3> We illustrate the proposed approach using data from a cohort study of underground miners. <h3>Conclusions</h3> The proposed approach can improve efficiency of nested case control designs.