Effects Of Major Air Pollutants On Mortality Of The Residents In Beijing And Their Spatial Variation

BackgroundIn recent years, haze frequently occurred in our country which attracted wide attention all over the world. Hazard of air pollution to the public’s health has become a hotspot in environmental epidemiological studies. Ecological time-series analyses indicated that air pollution associated with the increase in mortality among general population. However, due to restrictions on air quality monitoring in this country, studies on association between air pollution and residents’health were obviously insufficient, which restricted further exploration spatial variation in health effects caused by air pollution.ObjectiveThis study aims to investigate association between major air pollutants and mortality in local residents by time-series analysis and spatiotemporal analysis, and explore the factors related to their spatial variation.MethodsEcological design was used to study the association between changes of daily concentrations of air particulate matter less than lOμm in aerodynamic diameter (PM10), nitrogen dioxide (NO2) and carbon monoxide (CO) with non-accidental (overall) mortality (ICD-10 codes of A00-R99), cardiopulmonary mortality (ICD-10 codes of I00-I99 and J00-J98), and cardio-and cerebrovascular mortality (ICD-10 codes of I00-I99) in permanent residents of 16 districts (counties) of Beijing, during January 1,2009 to December 31,2010, adjusted for meteorological factors (including ambient temperature, relative humidity), temporal factors, and so on. Data of air pollutants were collected during that period from 16 districts of Beijing. Then, firstly, a new predicting model was fitted with general linear model (GLM) to estimate daily concentrations of various air pollutants in the districts where no air pollutants monitoring data were available. Secondly, the association between daily concentrations of air PM10, NO2 and CO and daily number of non-accidental (overall) deaths and deaths due to cardiopulmonary, cardio-and cerebrovascular diseases in permanent residents city-wide was studied with generalized additive model (GAM) linked to quasi-Poisson distribution. Thirdly, a generalized additive mixed model (GAMM) linked to quasi-Poisson distribution, considering spatial factors, with data of daily concentrations of air PMio, NO2 and CO in 16 districts of Beijing was fitted to assess the association between concentrations of these three pollutants and number of deaths in residents and compare the difference of the results between GAMM and GAM. Finally, district-specific GAM models were fitted for each district to analyze spatial variation in the associations between daily number of non-accidental (overall) deaths, cardiopulmonary, cardio-and cerebrovascular deaths of the residents. And, robust regression model was used to explore influences of the three air pollutants on spatial variation of the effects adjusted for some district-specific indicators.Results(1) Based on GAM, acute exposure to air PM10, NO2 and CO could increase daily number of non-accidental (overall) deaths and deaths due to cardiopulmonary, cardio- and cerebrovascular diseases in local residents. An interquartile increase in concentrations of air PMio (80μg/m3, lag 01) was associated with increases in daily number of non-accidental (overall) deaths and deaths due to cardiopulmonary, cardio-and cerebrovascular diseases by 2.58 percent (95%CI:1.69%～3.48%),2.97 percent (95%CI:1.83%～4.12%) and 2.87 percent (95%CI:1.66%～4.09%), respectively. An interquartile increase in concentrations of air NO2 (25.1μg/m3, lag 01) was associated with increases in daily number of non-accidental (overall) deaths and deaths due to cardiopulmonary, cardio-and cerebrovascular diseases by 4.25 percent (95%CI:3.17%～5.35%),4.80 percent (95%CI: 3.40%～6.22%) and 5.02 percent (95%CI:3.54%～6.52%), respectively. And, an interquartile increase in concentrations of air CO (0.9 mg/m3, lag 01) was associated with increases in daily number of non-accidental (overall) deaths and deaths due to cardiopulmonary, cardio-and cerebrovascular diseases by 3.35 percent (95%CI: 2.41%～4.30%),4.03 percent (95%CI:2.83%～5.24%) and 4.11 percent (95%CI: 2.84%～5.40%), respectively.(2) Based on GAMM adjusted for the random effect of district indicators, acute exposure to air PM10, NO2 and CO could significantly increase the number of non-accidental (overall) deaths and deaths due to cardiopulmonary, cardio-and cerebrovascular diseases.(3) The effects of air pollutants on residents’deaths derived from GAMM was lower than that derived from GAM, with a narrower 95% CI.(4) There existed a spatial variation in the effects of air pollutants on residents’deaths. District-specific effects of air PM10 on daily non-accidental (overall) deaths, cardiopulmonary, cardio-and cerebrovascular deaths in local residents ranged from minus 5.56 percent to 32.55 percent, from minus 7.00 percent to 39.86 percent, and from minus 12.73 percent to 46.22 percent, respectively. District-specific effects of NO2 on daily non-accidental (overall) deaths, cardiopulmonary, cardio-and cerebrovascular deaths ranged from minus 3.55 percent, to 11.62 percent, from minus 6.07 percent to 13.46 percent, and from minus 4.06 percent to 15.55 percent, respectively. And, district-specific effects of CO on daily non-accidental (overall) deaths, cardiopulmonary, cardio-and cerebrovascular deaths ranged from minus 1.86 percent to 13.49 percent, from minus0.11 percent to 14.91 percent, and from minus 0.58 percent to 17.12 percent, respectively.(5) Spatial variation in the effects of air pollutants on residents’mortality was associated with local public finance expenditure for health-care and environmental protection, and housing condition.ConclusionsAcute exposure to air pollutants can increase risk of daily non-accidental (overall) deaths, cardiopulmonary, cardio-and cerebrovascular deaths in local residents. There existed spatial variation in the effects of PM10, NO2 and CO on non-accidental (overall) deaths, cardiopulmonary, cardio-and cerebrovascular deaths in local residents, and such variation was associated with some district-specific indicators, such as health-care provided locally, environmental situation and housing condition.