The Effects Of Particulate Matter On Emergency Room Visits In Beijing,Based On Generalized Additive Model

BACKGROUND Over the past ten years,air pollution has become a serious issue in China.Particulate matters（PMs）are the most important air pollutants for most cities in China,which seriously endanger human health.PMs can be divided into respirable particulate matter(aerodynamic diameter ≤10 μm,PM₁₀),coarse particulate matter（aerodynamic diameter <10 μm and >2.5 μm,PM_C）,fine particulate matter(aerodynamic diameter ≤2.5 μm,PM_2.5)and ultrafine particulate matter（aerodynamic diameter <0.1 μm,UFP）.Numerous studies have evaluated the short-term effects of PMs air pollution in developed countries.However,because of the different level of effects,such as PMs levels,characteristics of PMs,life style,population susceptibility,and weather patterns and so on,epidemiological evidence reported in western countries cannot be directly generalized to Chinese populations.In China,related research started very late.Besides,the evidence of the short-term effects of PM_2.5 and PM_C is more limited than that of PM₁₀.Further,a few studies have examined association between cerebrovascular disease and PMs but reported inconsistent results.Some studies reported that the health risk of PMs differed by season or the characteristics of individuals,although there are contrasting findings.There are some studies that reported short-term effects of PMs by using hospital emergency room visits（ERVs）as health indicator in China.However,these studies only used emergency data from one hospital and PMs data from a fixed monitoring station,which might make the results less representative due to the large population size and wide residential distribution in a city.Limited studies have examined the associations between PMs of different particle sizes and ERVs for respiratory and cardio-cerebrovascular diseases.OBJECTIVES This study aimed to evaluate the acute effects of PMs of different particle sizes(PM_2.5,PM_C and PM₁₀)on ERVs for respiratory and cardiocerebrovascular diseases,including upper respiratory tract infection（LRTI）,lower respiratory tract infection（URTI）,chronic obstructive pulmonary disease（COPD）,asthma,cerebrovascular disease（CD）,ischemic heart disease（IHD）,heart rhythm disturbances（HRD）and heart failure（HF）,and to determine whether the effects differed across patient characteristics and seasons,and to explore the exposureresponse relationships.METHODS Daily data of ERVs from Jan 1 to Dec 31,2013,were obtained from ten general hospitals located in urban areas in Beijing.Daily 24-hour average concentration data for PM_2.5,PM₁₀,SO2,O3,NO2 and CO and meteorological data,including daily mean temperature and relative humidity,were collected over the same period.PM_C concentrations were estimated by subtracting PM_2.5 from PM₁₀ at collocated monitoring stations.A time-series design was employed to explore the associations between PMs concentrations and respiratory and cardio-cerebrovascular ERVs.The core model was generalized-additive model（GAM）with the quasi-Poisson link function.The natural cubic smoothing spline function of calendar time was incorporated in the model to adjusting for long-term trends and seasonality in daily ERVs.Day of the week（DOW）,public holidays and influenza outbreaks were added in the model as categorical variables.The natural cubic smoothing spline functions of daily mean temperature and relative humidity were incorporated in the model,allowing for nonlinear relationships with the outcomes.Stratified analysis was conducted according to gender,age and season（warm seasons: May to October;cold seasons: November to April）.The effects of PMs were estimated for single-day delayed effects（lag0^lag5）and cumulative delayed effects（lag0-1,lag0-3 and lag0-5）.The optimal lag time was determined according to the maximum effect value.And the optimal lag was reported in subgroup analysis.The exposure-response relationships between PMs concentrations and the log-relative risk of overall respiratory and cardiocerebrovascular ERVs were investigated.RESULTS1.Descriptive statisticsDuring the study period,the mean daily concentrations of PM_2.5,PM_C and PM₁₀ were 102.1±73.6 μg/m3,47.2±39.0 μg/m3 and 133.1±88.7 μg/m3,respectively.In total,there were 92464 respiratory ERVs and 56221 cardio-cerebrovascular ERVs.The mean daily count was 253±73 for respiratory ERVs and 154±30 for cardio-cerebrovascular ERVs.2.Associations between PM and respiratory ERVsThe effects of the three particles on overall respiratory ERVs were statistically significant at different lay days（P<0.05）.The optimal lag time of single-day delayed effects of PM_2.5,PM_C and PM₁₀ on overall respiratory ERVs were lag3、lag2 and lag0.We estimated increases of 0.29%（95%CI: 0.18%⁰.39%）,0.32%（95%CI: 0.10%⁰.53%）and 0.23%（95%CI: 0.14%⁰.33%）in overall respiratory ERVs associated with every 10 μg/m3 increase in PM_2.5 at lag3,PM_C at lag2,PM₁₀ at lag0,respectively.The optimal lag time of cumulative delayed effects of PM_2.5,PM_C and PM₁₀ on overall respiratory ERVs were lag0-5,the corresponding effect values were 0.58%（95%CI: 0.41%⁰.75%）,0.69%（95%CI: 0.32%¹.05%）and 0.42%（95%CI: 0.28%⁰.56%）.According to cause-specific analysis,for LRTI ERVs,the optimal lag time of single-day delayed effects of PM_2.5,PM_C and PM₁₀ were lag3,lag4 and lag0,respectively,the optimal lag time of cumulative delayed effects were lag0-5.For URTI ERVs,the optimal lag time of single-day delayed effects of PM_2.5,PM_C and PM₁₀ were lag0,lag2 and lag0,respectively,the optimal lag time of cumulative delayed effects were lag0-5.For COPD ERVs,the optimal lag time of single-day delayed effects of PM_2.5,PM_C and PM₁₀ were lag1,lag0 and lag1,respectively,the optimal lag time of cumulative delayed effects were lag0-3.No significant associations were found between the three particles and asthma ERVs.Subgroup analysis suggested that the effects of the three particles on URTI and LRTI were higher among women than that among men,and that the effects of the three particles on LRTI and COPD were higher among ≥65 than that among other age groups.PM_2.5 was significantly associated with URTI and LRTI both in warm seasons and cold seasons（P<0.05）.The effects of PM_2.5 on COPD,and the effects of PM_C and PM₁₀ on all the cause-specific respiratory ERVs were statistically significant only in cold seasons（P<0.05）.The exposure-response relationships between the three particles and overall respiratory ERVs were almost consistent,approximately linear without threshold.The curves changed little when NO2,SO2,O3 or CO was included in the model.3.Associations between PM and cardio-cerebrovascular ERVs:The effects of the three particles on overall cardio-cerebrovascular ERVs were statistically significant at different lay days（P<0.05）.The optimal lag time of singleday delayed effects of PM_2.5,PM_C and PM₁₀ on overall cardio-cerebrovascular ERVs were lag3,lag1 and lag2.We estimated increases of 0.14%（95%CI: 0.01%⁰.27%）,0.43%（95%CI: 0.15%⁰.72%）and 0.15%（95%CI: 0.04%⁰.27%）in overall cardiocerebrovascular ERVs associated with every 10 μg/m3 increase in PM_2.5 at lag3,PM_C at lag1,PM₁₀ at lag2,respectively.The optimal lag time of cumulative delayed effects of PM_2.5,PM_C and PM₁₀ on overall cardio-cerebrovascular ERVs were lag0-5,lag0-3 and lag0-5,the corresponding effect values were 0.30%（95%CI: 0.09%⁰.52%）,0.40%（95%CI: 0.03%⁰.76%）and 0.33%（95%CI: 0.12%⁰.53%）.According to causespecific analysis,no significant associations were found between PM_2.5 and CD ERVs.The optimal lag time of single-day delayed effects of PM_C and PM₁₀ on CD ERVs was lag2,and that of cumulative delayed effects was lag0-5.For IHD ERVs,the optimal lag time of single-day delayed effects of PM_2.5,PM_C and PM₁₀ were lag1,lag1 and lag2,respectively,the optimal lag time of cumulative delayed effects of PM_2.5,PM_C and PM₁₀ were lag0-5,lag0-1 and lag0-3,respectively.For HRD ERVs,the optimal lag time of single-day delayed effects and cumulative delayed effects of PM_2.5 were lag1 and lag0-3,respectively.PM_2.5 was significantly associated with HF ERVs at lag0.Subgroup analysis suggested that the effects of the three particles on overall cardiocerebrovascular diseases were statistically significant among men,people aged <65 years and in cold seasons（P<0.05）.The effects of PM_2.5 on CD and HRD ERVs were statistically significant among men（P<0.05）.The effects of PM_C and PM₁₀ on CD ERVs,and the effects of PM_2.5 and PM₁₀ on IHD ERVs were statistically significant among people aged <65 years（P<0.05）.The effects of PM_2.5 and PM₁₀ on HRD ERVs,and the effects of PM_C on IHD ERVs,and the effects of PM_2.5 on HF ERVs were statistically significant among people aged ≥65 years（P<0.05）.The effects of PM_C and PM₁₀ on CD ERVs,and the effects of PM_2.5 on HF ERVs were statistically significant in cold seasons（P<0.05）.The effects of PM_2.5 and PM_C on IHD ERVs were statistically significant in warm seasons（P<0.05）.The exposure-response relationships between the three particles and overall cardio-cerebrovascular ERVs were approximately nonlinear.The curves changed little when NO2,SO2,O3 or CO was included in the model.CONCLUSIONS1.Short-term exposure to PM_2.5,PM_C and PM₁₀ were significantly associated respiratory and cardio-cerebrovascular ERVs in Beijing.The effects of the three particles on respiratory diseases have little difference.PM_2.5,PM_C and PM₁₀ pollution can increase the risk of ERVs for overall respiratory diseases,URTI,LRTI,COPD.The effects of the three particles on cardio-cerebrovascular diseases are different.PM_2.5 pollution can increase the risk of ERVs for overall cardio-cerebrovascular diseases,IHD,HRD and HF.PM_C and PM₁₀ pollution can increase the risk of ERVs for overall cardio-cerebrovascular diseases,CD and IHD.2.Under the environment of high concentrations of PM_2.5,PM_C or PM₁₀,there are higher risk of acute exacerbations of respiratory disease in women,people aged ≥65 years and in cold seasons,there are higher risk of acute exacerbations of overall cardiocerebrovascular diseases in men,people aged <65 years and in cold seasons,while the effects of PMs on cause-specific cardio-cerebrovascular diseases are inconsistent at different gender groups,age groups or seasons.3.The exposure-response relationships were approximately linear without threshold between the three particles and respiratory ERVs,and were approximately non-linear between the three particles and cardio-cerebrovascular ERVs.