Identification of meteorological variables as predictors for forecastinghealth risks of high temperatures / 预防医学

ABSTRACT

Objective@#To identify the most appropriate meteorological variable for forecasting the health risk of high temperatures.@*Methods@#The surveillance on causes of death, meteorological data and surveillance on air quality among registered residents in Ningbo City, Zhejiang Province during the period between May and October from 2013 to 2019 were collected. The association models of daily minimum temperature, average daily temperature, daily maximum temperature, daily minimum heat index, average daily heat index, daily maximum heat index, average daily apparent temperature and torridity index with deaths and years of life lost (YLL) were created using time series analysis and distributed lag non-linear models, and the model fitting effect was evaluated using the minimum Akaike information criterion (AIC) procedure. The most appropriate meteorological variable for forecasting gender-, age- and mortality-specific health risks of high temperatures was identified.@*Results@#A total of 120 628 deaths were reported during the study period, with daily deaths of 94 cases, and daily YLL rate of 19.74 person-years/105. Except for daily minimum heat index and torridity index, the exposure-response relationships between other six meteorological variables and deaths and overall YLL rate all appeared a “J” shape. The lowest AIC values and the optimal model fitting effects were measured for the association models between average daily temperature and whole populations, females, subjects at ages of 65 years and older, and deaths and YLL rates due to circulatory diseases and respiratory diseases.@*Conclusion@#High model fitting effects are observed between average daily temperature and deaths and YLL rates, which are more suitable for forecasting the health risk of high temperature.