Associations between Short-Term Exposure to Ambient Air Pollution and Influenza: An Individual-Level Case-Crossover Study in Guangzhou, China

Abstract

Background: Influenza imposes a heavy burden on public health. Little is known, however, of the associations between detailed measures of exposure to ambient air pollution and influenza at an individual level.

Objective: We examined individual-level associations between six criteria air pollutants and influenza using case-crossover design.

Methods: In this individual-level time-stratified case-crossover study, we linked influenza cases collected by the Guangzhou Center for Disease Control and Prevention from 1 January 2013 to 31 December 2019 with individual residence-level exposure to particulate matter (${\mathrm{PM}}_{2.5}$ and ${\mathrm{PM}}_{10}$), sulfur dioxide (${\mathrm{SO}}_2$), nitrogen dioxide (${\mathrm{NO}}_2$), ozone ($O_3$) and carbon monoxide (CO). The exposures were estimated for the day of onset of influenza symptoms (lag 0), 1-7 d before the onset (lags 1-7), as well as an 8-d moving average (lag07), using a random forest model and linked to study participants' home addresses. Conditional logistic regression was developed to investigate the associations between short-term exposure to air pollution and influenza, adjusting for mean temperature, relative humidity, public holidays, population mobility, and community influenza susceptibility.

Results: $N=\mathrm{108,479}$ eligible cases were identified in our study. Every $10{\text{-}\mu g/m}^3$ increase in exposure to ${\mathrm{PM}}_{2.5}$, ${\mathrm{PM}}_{10}$, ${\mathrm{NO}}_2$, and CO and every ${5\text{-}\mu g/m}^3$ increase in ${\mathrm{SO}}_2$ over 8-d moving average (lag07) was associated with higher risk of influenza with a relative risk (RR) of 1.028 (95% CI: 1.018, 1.038), 1.041 (95% CI: 1.032, 1.049), 1.169 (95% CI: 1.151, 1.188), 1.004 (95% CI: 1.003, 1.006), and 1.134 (95% CI: 1.107, 1.163), respectively. There was a negative association between $O_3$ and influenza with a RR of 0.878 (95% CI: 0.866, 0.890).

Conclusions: Our findings suggest that short-term exposure to air pollution, except for $O_3$, is associated with greater risk for influenza. Further studies are necessary to decipher underlying mechanisms and design preventive interventions and policies. https://doi.org/10.1289/EHP12145.

Figure 1.

(A) Geographical location of the study area shaded in brown, and (B) the spatial distribution of the influenza case residence locations across Guangzhou over 2013–2019. The map outline in this figure was sourced from the China Resource and Environment Science and Data Center (http://www.resdc.cn). The figure was produced using ArcGIS software (version 10.8.1; ESRI).

Figure 2.

Associations between acute exposures to ${\mathrm{PM}}_{2.5}$, ${\mathrm{PM}}_{10}$, ${\mathrm{SO}}_2$, ${\mathrm{NO}}_2$, ${\mathrm{O}}_3$, and CO and influenza at different lag days in 108,479 participants in Guangzhou, 2013–2019. These relative risks were estimated in terms of per ${10\text{-}\mathrm{\mu }\mathrm{g}/\mathrm{m}}^3$ increase in ${\mathrm{PM}}_{2.5}$, ${\mathrm{PM}}_{10}$, ${\mathrm{NO}}_2$, ${\mathrm{O}}_3$, and CO and ${5\text{-}\mathrm{\mu }\mathrm{g}/\mathrm{m}}^3$ increase in ${\mathrm{SO}}_2$ concentration using multivariable conditional logistic models adjusted for mean temperature, relative humidity, public holidays, population mobility, and change in community influenza susceptibility for $N=\mathrm{108,479}$ study participants. Each panel shows multiple models with exposures measured from 0 to 7 d, including both single-lag models (lag 0 to lag 7) as well as moving average lag models (lag01 to lag07). The figure was created using R software (version: 3.5.2; R Development Core Team; package: ggplot2). Numeric data can be found in Excel Table S1. Note: CO, carbon monoxide; ${\mathrm{NO}}_2$, nitrogen dioxide; ${\mathrm{O}}_3$, ozone; ${\mathrm{PM}}_{2.5}$, fine inhalable particles with diameters that are generally $2.5\mathrm{\mu }\mathrm{m}$ and smaller; ${\mathrm{PM}}_{10}$, inhalable particles with diameters that are generally $10\mathrm{\mu }\mathrm{m}$ and smaller; ${\mathrm{SO}}_2$, sulfur dioxide.

Figure 3.

Exposure–response curves showing associations between 8-d moving average (lag07) exposures to ${\mathrm{PM}}_{2.5}$, ${\mathrm{PM}}_{10}$, ${\mathrm{SO}}_2$, ${\mathrm{NO}}_2$, ${\mathrm{O}}_3$, and CO and risk of influenza in single-pollutant models in 108,479 participants in Guangzhou, 2013–2019. DLNM models were adjusted for mean temperature, relative humidity, public holidays, population mobility, and change in community influenza susceptibility. Mean temperature and relative humidity were adjusted by incorporating the moving average from the day of symptom onset up to 7-d lag period using a 3-df natural cubic spline. The two dashed lines in red represent the fifth and ninety-fifth percentiles of air pollutant distributions. The horizontal box plot placed at the bottom of each facet represents the distribution of the six air pollutants concentrations between 2013 and 2019. The center and edges in the box plots show median and 25th and 75th percentiles. In the box plot, black points beyond the upper fence represent outliers. The figure was created using R software (version: 3.5.2; R Development Core Team; package: ggplot2). Numeric data can be found in Excel Table S2. Note: CO, carbon monoxide; df, degrees of freedom; DLNM, distributed lag nonlinear model; ${\mathrm{NO}}_2$, nitrogen dioxide; ${\mathrm{O}}_3$, ozone; ${\mathrm{PM}}_{2.5}$, fine inhalable particles with diameters that are generally $2.5\mathrm{\mu }\mathrm{m}$ and smaller; ${\mathrm{PM}}_{10}$, inhalable particles with diameters that are generally $10\mathrm{\mu }\mathrm{m}$ and smaller; ${\mathrm{SO}}_2$, sulfur dioxide.