| ObjectivesBased on the analysis of the relationship between mortality data of residents and meteorological data in part cities of China, model fit analysis and preliminary predictive analysis were used to explore the appropriate predictive model for forecasting the health effects of heat-waves on urban residents in China, and provide basic information and scientific basis for warning of health effects of heat-waves.MethodsCollect the mortality data and meteorological data in Haerbin and Chaoyang district of Beijing, and make use of lag model and grey relation method to analyze the relationship between mortality data and meteorological data, and conduct threshold analysis use piecewise-regression model. Combine the results of studies in earlier stage, generalized additive model fitting analysis and preliminary predictive analysis were conducted in multiple cities simultaneously.Results1. The results of investigation data shows that high temperature and heat-wave can cause the increase of incidence of heat related diseases, especially the heatstroke. The middle-age and elderly people are still the susceptible population even though they make some solutions consciously in the period of high temperature.2. The results of the analysis of individual city shows that the influence of high temperature on death indexes are mainly concentrated in lag 0 to 4 days and the influence of PM10 has certain lag. The relative risk of PM10 concentration on an increase of 10μg/m3 up to a high level when lag 10 days and the cumulative relative risk up to the peak value when lag 15 days. Ambient temperature had a high relative risk when lag 0 day and 1 day. The interaction between ambient temperature and PM10 concentration is comparatively obvious when the PM10 concentrature under about 200μg/m3. The changes of annual average mortality had a strong grey relation with temperature and relative humidity. The monthly average mortality showed a strong grey relation with air pressure and the monthly average mortality in summer season had a strong grey relation with air pressure, relative humidity, and wind speed. 3. Threshod analysis of multiple cities shows that thresholds of all-cause mortality risk (℃) in Beijing, Harbin, Chongqing, Fuzhou and Shantou were 25.63 (±0.809), 23.24 (±1.114), 29.29 (±2.886), 36.06 (±0.281), and 31.28 (±1.016) respectively. Increased percentages (95% CI) of mortality caused by temperature rising every 1℃above the threshold were 0.99 (0.46~1.52), 1.12 (0.24~2.00), 0.17 (-0.34~0.68), 21.70 (11.99~32.26), and 2.79 (1.05~4.57) respectively.4. When considering temperature factor only, model fitting analysis of each city determined the predictors of the all-cause of death index respectively, Beijing: lag0+lag1+lag24, Chongqing: lag0,Fuzhou:lag0+lag1521,Shantou: lag0+lag24+lag814, Haerbin: lag0+lag1+lag814+lag1521, when the relative humidity factor was bring into, the predictors for each city are Beijing: lag0+lag1+lag24+lag1521, Chongqing and Fuzhou: none, Shantou: lag0+lag814, Haerbin: lag0+lag1+lag814+lag1521.5. Preliminary predictive analysis shows that the influence of high temperature on death index of residents are mainly concentrated in lag 0 to 4 days. Distribution lag nonlinear model can calculate the relative risk at different temperature and different lags. The index of death of any observation time can be estimated through the change of the temperature and lag days. The max relative risk in Beijing and Chongqing are 1.126 and 1.118 respectively and these are not Corresponding to the highest daily mean temperature. The max relative risk in Shantou, Fuzhou and Haerbin are 1.310, 1.269 and 1.254 respectively and these are Corresponding to the highest daily average temperature.Conclusions1. The health effects of high temperature mainly concentrate in lag 0 to 4 days and as acute effects. The influence of PM10 exists a long lag in the influence of temperature. The interaction between temperature and PM10 is obvious when the PM10 concentration in lower. The long-term trend of death index of residents relate to temperature and relative humidity and in this process air pressure influence the short-term fluctuation.2. The effect of temperature at lag 0 day is significant in different cities when only considering temperature factor, but exist certain differences in the length of lag time. When considering the relative humidity as confounding factor, the model factor has great changes in Chongqing and Fuzhou and this indicate that relative humidity has significant influence on health effect of temperature.3. Distribution lag non-linear model can calculate relative risk values of death index at different temperature and different lag time. The change of temperature can be observed between the observe day and reference day. Combine to determine the lag time the change of relative risk can be calculated. From this we can forecast the relative risk of death index on the observe day. Through preliminary validation, the result indicate that this model can be used to predict the influence of high temperature on health, but the quality of data have large effect on model estimation. |
PDF Full Text Request