A novel method to construct an air quality index based on air pollution profiles.

BACKGROUND

Air quality indices based on the maximum of sub-indices of pollutants are easy to produce and help quantify the degree of air pollution. However, they discount the additive effects of multiple pollutants and are only sensitive to changes in highest sub-index.

OBJECTIVES

We propose a simple and concise method to construct an air quality index that takes into account additive effects of multiple pollutants and evaluate the extent to which this index predicts health effects.

MATERIALS AND METHODS

We obtained concentrations of four criteria pollutants: particulate matter with aerodynamic diameter ≤ 10μm (PM), sulphur dioxide (SO), nitrogen dioxide (NO) and ozone (O) and daily admissions to Hong Kong hospitals for cardiovascular and respiratory diseases for all ages and those 65 years or older for years 2001-2012. We derived sub-indices of the four criteria pollutants, calculated by normalizing pollutant concentrations to their respective short-term WHO Air Quality Guidelines (WHO AQG). We aggregated the sub-indices using the root-mean-power function with an optimal power to form an overall air quality index. The optimal power was determined by minimizing the sum of over- and under-estimated days. We then assessed associations between the pollution bands of the index and cardiovascular and respiratory admissions using a time-stratified case-crossover design adjusted for ambient temperature, relative humidity and influenza epidemics. Further, we conducted case-crossover analyses using the Hong Kong air quality data with the respective standards and classification of pollution bands of the China Air Quality Index (AQI), the United Kingdom Daily AQI (DAQI), and the United States Environmental Protection Agency (USEPA) AQI.

RESULTS

The mean concentrations of PM and SO based on maximum 3-h mean exceeded the WHO AQG by 37% and 50%, respectively. We identified the combined condition of observed high-pollution days as either at least one pollutant > 1.5×WHO AQG or at least two pollutants > 1.0×WHO AQG to characterize the typical pollution profiles over the study period, which resulted in the optimal power=3.0. The distribution of days in different pollution bands of the index was: 5.8% for "Low" (0-50), 37.6% for "Moderate" (51-100), 31.1% for "High" (101-150), 14.7% for "Very High" (151-200), and 10.8% for "Serious" (201+). For cardiovascular and respiratory admissions, there were significant associations with the pollution bands of the index for all ages and those 65 years or older. The trends of increasing pollution bands in relation to increasing excess risks of cardiovascular and respiratory admissions were significant for the proposed index, the China AQI, the UK DAQI and the USEPA AQI (P value for test for linear trend < 0.0001), suggesting a dose-response relation.

CONCLUSIONS

We have developed a simple and concise method to construct an air quality index that accounts for multiple pollutants to quantify air quality conditions for Hong Kong. Further developments are needed in order to support the extension of the method to other settings.