A Nationwide Health Risk Assessment of PM2.5 Exposure by Examining the Regional Concentrations-Response Relations in a Population-Based Epidemiological Study in Japan

在该项目的第二年（2022财年），我们开发了分析方法，以估计日本PM2.5与死亡率之间的短期关联。为了量化这些关联，我们进行了两阶段分析。在第一阶段，利用时间序列准泊松模型估计PM2.5的特定位置的风险。我们应用了一个分布式滞后非线性模型沿着与灵活的函数形式的样条来评估特定于位置的关联。我们控制了各种重要的协变量，包括天气变量，一周中的一天，季节性和长期时间趋势。进行了敏感性分析，以确定我们的研究结果的鲁棒性。为了评估和比较浓度-反应（CR）关系的形状，使用统一的统计设计制定所有局部CR函数。在第二阶段，我们使用随机效应荟萃分析来结合联合收割机的区域和国家的总体估计。

During the second year of the project (FY2022), we developed analytical methods to estimate the short-term associations between PM2.5 and mortality in Japan. To quantify these associations, we conducted a two-stage analysis. In the first stage, the location-specific risks of PM2.5 were estimated by utilizing a time-series quasi-Poisson model. We applied a distributed lag non-linear model along with a flexible functional form of splines to evaluate the location-specific associations. We controlled various important covariates, including weather variables, day of the week, and seasonal and long-term time trends. Sensitivity analyses were carried out to determine the robustness of our findings. To assess and compare the shapes of the concentration-response (CR) relationship, all local CR functions were formulated using a unified statistical design. In the second stage, we used a random-effect meta-analysis to combine the regional and national overall estimates.