Constructing an Air Quality Health Index for children: A case study in Shanghai, China

Background: The Air Quality Health Index (AQHI) based on the association between excess mortality risk and air pollutants was established by Canadian scientists in 2008 and it has been widely used for predicting multiple air pollutants related health risks. However, it remains unclear whether AQHI is a better indicator in predicting other health risks like respiratory diseases for vulnerable populations. Objective: This study aimed to propose a case on constructing an AQHI based on the association between air pollution and hospital outpatient visits for respiratory diseases among children and to determine whether the index adequately predicts early risk of respiratory diseases in children. Method: Data of air pollutants, including particulate matter of less than 2.5 mu m in aerodynamic diameter (PM2.5), sulfur dioxide (SO2), nitrogen dioxide (NO2), and ozone (O-3) were collected from Shanghai Environmental Monitoring Center from January 1, 2015 to December 31, 2019. Daily number of hospital outpatient visits for pediatric (0-12 years old) respiratory diseases were also obtained. Time-series analysis with a generalized additive model (GAM) during warm (Apr. to Sep.) and cool periods (Oct. to Mar.) was conducted to estimate the associations between respiratory-related hospital outpatient visits in children and the concentrations of air pollutants including PM2.5, SO2, NO2, and O-3 in Shanghai from 2015 to 2018. The sum of excess risk (ER) of hospital outpatient visits in warm and cool periods was used to construct the AQHI for children (AQHI(c)). As AQHI(c) was established using the data from 2015 to 2018, we examined the validity of the index with data from 2017 to 2019. We also compared the predictive power of AQHI(c) and the currently used Air Quality Index (AQI) with the data of daily hospital outpatient visits from 2017 to 2019. Result: According to one- and two-pollutant models of GAM, the concentration-response coefficients of PM2.5, SO2, NO2, and O-3 were selected to construct the AQHI(c) in the warm period, while only SO2, NO2, and PM2.5 were included for the construction of AQHI(c) in the cool period as O-3 was negatively correlated with the hospital outpatient visits. There were almost linear exposure-response correlations between AQHI(c) and daily hospital outpatient visits. AQHI(c) and AQI showed similar results with the annual data in terms of model fit statistics. When the data was divided into warm and cool periods, the power of AQHI(c) was slightly stronger than that of AQI in predicting the air pollution-related health risks. Conclusion: AQHI(c) we developed might comprehensively reflect the combined effects of air pollution in Shanghai and it could be a more valid prediction index for evaluating air pollution-related health risks in children.