|
Abstract:
|
The possible association between PM2.5 and lung cancer mortality can be partitioned into components, within similar observational units and across different observational units. Within units covariates are very similar and across units covariates can be, and usually are, very different. Hence, there is a need to understand the possible effect of PM2.5 on mortality taking into account within and between observational units. Our idea is to use Local Control Analysis (LCA) to estimate these two components and determined how much of the variation in estimates can be attributed to know the important covariates. For the purpose of analysis, we calculated Local Treatment Difference (LTD) for LTD approach and slope and intercept for Local Linear Regression (LLR) approach, to determine if the treatment (PM2.5) effect vary significantly across clusters. For that evaluation we used recursive partitioning. The benefit of this study are twofold. First, we use a reliable strategy (LCA) for observational data. Second and importantly, there is subgroup heterogeneity in the effect of PM2.5 on lung cancer mortality and this heterogeneity is largely explained by factors other than air quality.
|
