Nonlinear Concentration-Response Function between Air Pollution and Childrens Hospital Admissions at High Concentration Levels: Evidence from Yangquan, ChinaAbstract Number:1803 Jeremiah Liu*, Zhao Yang, Tian Sang, Yanping Zhang, Baoxin Zhao, and Jinliang Zhang Jeremiah Liu* State Key Laboratory of Environmental Criteria and Risk Assessment, Chinese Research Academy of Environmental Science, China, China, E-mail Address: [email protected] Search for more papers by this author , Zhao Yang Shanxi Medical University, China, China, E-mail Address: [email protected] Search for more papers by this author , Tian Sang Shanxi Medical University, China, China, E-mail Address: [email protected] Search for more papers by this author , Yanping Zhang State Key Laboratory of Environmental Criteria and Risk Assessment, Chinese Research Academy of Environmental Science, China, China, E-mail Address: [email protected] Search for more papers by this author , Baoxin Zhao Taiyuan Center for Disease Control and Prevention, China, China, E-mail Address: [email protected] Search for more papers by this author , and Jinliang Zhang State Key Laboratory of Environmental Criteria and Risk Assessment, Chinese Research Academy of Environmental Science, China, China, E-mail Address: [email protected] Search for more papers by this author AbstractOBJECTIVE: To determine the shape of concentration-response function for air pollution related children’s hospital admissions across a high and wide range of air pollution in Yangquan, China (PM10: 15.00-613.33µg/m3, SO2: 24.00-989.33µg/m3).METHODS: Generalized additive models (GAM) with log link and Poisson error was used to estimate the shape of the concentration response function, and constrained piecewise linear functions to estimate the relative risks, with change points dynamically estimated using structural-change-detection algorithm.RESULTS: We identified three types of Concentration- Response shapes: C, S and J for each of pollutant with each of disease types. Hill slopes were mostly found in the low concentration range (PM10 <111.50µg/m3, SO2<179.17µg/m3) for C, in the intermediate concentration range (PM10: 123.91-197.10µg/m3, SO2: 142.00-326.50µg/m3) for S, and in the high concentration range (PM10>196.33µg/m3 , SO2>239.10µg/m3) for J. S- and J-shape exhibited evident threshold: PM10 related was located at 110µg/m3 for S, and 140µg/m3 for J, while SO2 related was located at 130µg/m3 for S, and 170µg/m3 for J. About risk magnitude, PM10 showed C-shape of association with LRD (54.15% per 10µg/m3), S- and J-shape with ALRI (11.26-11.96% per 10µg/m3), and J-shape with AURI (7.09% per 10µg/m3) in non-heating period. In heating period, PM10 was only associated with AURI, but in all three forms of association (22.89%, 20.22%, 5.53% per 10µg/m3 for C-, S-, and J-shape). For SO2, it showed J-shape with AURI in non-heating period (10.52% per 10µg/m3), and linear shape with AURI in heating period (8.73% per 10µg/m3). In addition, SO2 was associated with ALRI in the form of J-and S-shape in heating period (31.32-38.47% per 10µg/m3).CONCLUSIONS: Our result suggested that health impacts should be estimated across the whole ambient range of air pollutions using threshold models.