OBJECTIVETo establish concentration-response relationship model for high particulate air pollution and daily hospital admissions for early warning system.

METHODSThe Poisson generalized additive model was used with natural cubic spline smoothing for air pollutants of PM10, SO2, NO2, CO to determine the shape of concentration-response relationship. And piecewise linear regression was used for risk analysis.

RESULTSAge-specific analysis suggested the non-linear association between particulate air pollution and hospital admissions in all age groups. In respiratory and cardiovascular admissions, the percentage increase was 0.716%-2.145% and 0.65% for 10 microg/m3 increase in PM10, respectively. Cause-specific analysis suggested the non-linear association between particulate air pollution and hospital admissions for all diseases. The significant association was found with COPD and ischaemic heart disease, and the related percentage increase was 2.94% and 1.94%, respectively. Heating and noheating period analysis suggested the effect of PM10 was higher in noheating period than that in heating period.

CONCLUSIONCompared with mortality, particulate air pollution should affect the hospital admissions more greatly, the percentage increase should be higher for hospital admissions than that of mortality (0.25%).