PM2.5 in Beijing - temporal pattern and its association with influenza

Abstract

Background: Air pollution in Beijing, especially PM2.5, has received increasing attention in the past years. Despite Beijing being one of the most polluted cities in the world, there has still been a lack of quantitative research regarding the health impact of PM2.5 on the impact of diseases in Beijing. In this study, we aimed to characterize temporal pattern of PM2.5 and its potential association with human influenza in Beijing.

Methods: Based on the data collected on hourly ambient PM2.5 from year 2008 to 2013 and on monthly human influenza cases from 2008 and 2011, we investigated temporal patterns of PM2.5 over the five-year period and utilized the wavelet approach to exploring the potential association between PM2.5 and influenza.

Results: Our results found that ambient PM2.5 pollution was severe in Beijing with PM2.5 concentrations being significantly higher than the standards of the World Health Organization, the US EPA, and the Chinese EPA in the majority of days during the study period. Furthermore, PM2.5 concentrations in the winter heating seasons were higher than those in non-heating seasons despite high variations. We also found significant association between ambient PM2.5 peak and human influenza case increase with a delayed effect (e.g. delayed effect of PM2.5 on influenza).

Conclusions: Ambient PM2.5 concentrations were significantly associated with human influenza cases in Beijing, which have important implications for public health and environmental actions.

Figure 1

PM _2.5 concentrations (μg/m ³ ) from 2008 to 2013 in Beijing. Top-24-hour averages of PM_2.5 measurements superimposed with standards by the WHO (25 μg/m³ for the 24-hour average, black line), the USEPA (35 μg/m³ for the 24-hour average, redline), and the Chinese State Environmental Protection Administration (SEPA, 75 μg/m³, implemented from 2012, green line; the 75 μg/m³ standard, or Class II standard, was developed specific for residential, commercial, industrial, and heavy-traffic areas) (ref, Cao et al. Chinese and US air quality standard evolution paper). Middle - Monthly variations in average, 97.5 and 2.5 percentiles of PM_2.5 from 2008 to 2013 in Beijing. Bottom – Annual averages of PM_2.5 concentrations from 2008 to 2013, superimposed with the standards by the WHO (10 μg/m³ for the annual average, black line), the US EPA (15 μg/m³ for the annual average, redline, the EPA standard was reduced to 12 μg/m³ after 2013), and the SEPA (35 μg/m³, implemented from 2012).

Figure 2

Distribution of PM _2.5 concentrations and days with 24-hr averages exceeding the standards. Top – frequency distribution of days with varying 24-hr average concentrations for the whole study period; black line – the WHO standard (25 μg/m³), redline – the EPA standard (35 μg/m³), green – the SEPA standard (75 μg/m³). Bottom – distribution of days each month with the 24-hr averages exceeding the standards by the WHO (red), the EPA (green) and SEPA (blue). The gap between December 2008 and February 2009 was due to missing measurements.

Figure 3

Monthly variations in PM _2.5 in Beijing for each year between 2008 and 2013. The light blue shaded area covers from the mid-November to mid-March, typically winter season in Beijing during which centralized heating is provided. Overall, the PM_2.5 concentrations are higher in the winter season. During 2008, the year of the Beijing Olympics, the Chinese government imposed regulatory controls on emissions, which may be a mechanistic explanation for why concentrations in 2008 are generally lower in each month than concentrations in other years.

Figure 4

Wavelet power spectrum analysis of monthly PM _2.5 concentrations from year 2008 to 2013 in Beijing. The color code for power values from dark blue (low value) to dark red (high value). The dotted white lines show the maxima of the undulations of the wavelet power spectrum and the dotted-dashed show the α = 5% significant levels and the cone of influence which indicates the region not influenced by edge (following the algorithm/software by Cazelle et al. 2007).

Figure 5

The association between monthly PM _2.5 (μg/m ³ ) and reported influenza cases (in five urban districts) from 2008 to 2011 in Beijing. Top – monthly PM_2.5 and reported influenza cases. Bottom: Wavelet coherence between the standardized PM_2.5 measurements and reported influenza cases in Beijing. The 5% significance level against red noise is shown as a thick contour. The relative phase relationship is shown as arrows, suggesting that PM_2.5 leads influenza by about 90^o pointing down, equivalent to 1-2 months delay in occurrence.