Air Pollution Mediates the Association between Human Mobility and COVID-19 Infection

Received: September 22, 2021
Revised: December 6, 2021
Accepted: December 20, 2021

 Copyright The Author(s). This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are cited.

Cite this article:

Lo, K., Tung, N.T., Wu, C.D., Thao, H.N.X., Dung, H.B., Thuy, T.P.C., Chuang, H.C. (2022). Air Pollution Mediates the Association between Human Mobility and COVID-19 Infection. Aerosol Air Qual. Res. 22, 210249. https://doi.org/10.4209/aaqr.210249

HIGHLIGHTS

• National restrictions in Taiwan may decrease human mobility.
• A decrease in mobility was associated with air pollution declines.
• Declines in air pollution levels was associated with a decrease in COVID-19 cases.
• Air pollution mediated the associations of human mobility with COVID-19 cases.

Keywords: Air pollution, Particulate matter, Restriction, SARS-CoV-2, Transmission

1 INTRODUCTION

Since the primary outbreak of coronavirus disease 2019 (COVID-19) in Wuhan, China in late 2019, the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has spread worldwide and infected over 200 million people (WHO, 2021). There are approximately 211 million confirmed cases and 4.4 million deaths due to COVID-19 infection according to a World Health Organization (WHO) report (WHO, 2021). Despite most patients having a mild respiratory illness after an infection, about 26% of patients experience severe pulmonary abnormalities (Huang et al., 2020). Besides the association of COVID-19 transmissibility with human mobility, as evaluated in previous studies, it is worth mentioning that emerging evidence has shown the impact of air pollution on transmission of COVID-19 (Cao et al., 2021; Shao et al., 2021). Most countries implemented national restrictions to avoid the exponential growth of hospital admissions and to progressively slow down COVID-19 transmission (Flaxman et al., 2020). National lockdowns successfully reduced human activities and air pollution levels (Sahraei et al., 2021). However, the emergence of mutant SARS-CoV-2 variants from the United Kingdom (UK) and South Africa led to unexpected resurgences in the pandemic around the world in 2021.

The alpha variant (B.1.1.7 lineage) of SARS-CoV-2 spread to Taiwan and caused a severe threat to public health in early May 2021 (Taiwan CDC, 2021a). Additionally, the low vaccination rate (8.35%) also was a potential risk for virus transmission (Taiwan CDC, 2021a). To prevent rapid SARS-CoV-2 transmission in Taiwan, the Taiwan Central Epidemic Command Center (CECC) announced a Level 3 national restriction on 19 May 2021 (Taiwan CECC, 2021c). Restrictions included prohibition of outdoor gatherings of more than 10 people, reductions of unnecessary mobility and activities, and the closure of entertainment venues (i.e., cinemas, museums, galleries, and sport centers) and educational facilities (i.e., libraries, schools, colleges, and universities) (Taiwan CECC, 2021b). However, the efficiency of the reduction in human mobility due to the social distancing measures of cities on preventing SARS-CoV-2 transmission remains unclear. We hypothesized that air pollution mediated the association between mobility and confirmed COVID-19 cases. The objective of this study was to examine the effects of the reduction in human mobility on daily confirmed COVID-19 cases in Taiwan and the mediating effects of air pollution.

 
2 METHODS

 
2.1 Data of COVID-19 Cases and Human Mobility

Daily confirmed COVID-19 case data (1–30 May 2021) in each city of Taiwan were obtained from the National Center for High-Performance Computing (https://scidm.nchc.org.tw). Daily mobility data in each city of Taiwan during the same period were collected from the Mobility Trends Reports (https://covid19.apple.com/mobility). Mobility data included walking, driving, and using public transport. Residential cities for confirmed cases were matched with commuter zones to access their mobility, and daily residential mobility is expressed as average minutes in transit.

 
2.2 Air Pollution

Town-level exposure to air pollution (i.e., particulate matter (PM) of < 10 µm in aerodynamic diameter (PM₁₀), PM_2.5, nitrogen dioxide (NO₂), and carbon monoxide (CO)) from 1 to 23 May 2021 was estimated using the hybrid kriging/land-use regression (LUR) method (Wu et al., 2018; Chen et al., 2020). Air pollution data were obtained from Taiwan Environmental Protection Administration air quality monitoring stations (https://airtw.epa.gov.tw/). Land-use predictors with a Spearman’s correlation coefficient of ≥ 0.4 with effects on air pollution were entered into a stepwise linear regression. Moreover, a set of pollutant levels was created by a leave-one-out kriging interpolation method and added to the model to improve its robustness.

 
2.3 Statistical Analysis

We examined associations of air pollution with 7-day-lag 6611 confirmed COVID-19 cases and with human mobility using linear regression models. Next, the mediating effects of air pollution on the association of mobility with confirmed COVID-19 cases were estimated using the PROCESS (Hayes, 2013). The direct effect was defined as the effect of mobility on confirmed COVID-19 cases after controlling for the indirect (mediated) effects of air pollution. The indirect effect was defined as the effect of mobility on confirmed COVID-19 cases through the mediating effect of air pollution. The total effect was determined as the indirect effect plus the direct effect. Co-variables of age and sex were adjusted in the linear regression and mediating analyses. Results of the linear regression are presented as beta (β) coefficients and 95% confidence intervals (CIs). Results of the mediation effect are presented as beta coefficients, standard errors, and 95% CIs. Data analyses were conducted using SPSS ver. 20.0 (SPSS, Chicago, IL, USA). Values of p ≤ 0.05 were accepted as statistically significant.

 
3 RESULTS AND DISCUSSION

 
3.1 Human Mobility and Daily Confirmed COVID-19 Cases

During the study period, we observed that mobility through walking, driving, and using public transport gradually decreased by 37.76%, 37.87%, and 65.42%, respectively. Our results showed that mobility reductions began from 7 May, which was before the nationwide restrictions on 19 May. This could have been due to public awareness of the beginning of COVID-19 infections in the community. Thus, people began to work from home and to reduce their outdoor activities before the official restriction announcement in Taiwan. There were 6611 confirmed COVID-19 cases in total during the study period. We observed that the number of COVID-19 cases increased steadily to the highest number of 699 cases on 22 May, then the confirmed cases gradually decreased after 27 May 2021 (Fig. 1). Increasing numbers of recent studies have investigated associations between mobility restrictions and declines in confirmed COVID-19 cases (Wang et al., 2020a; Nouvellet et al., 2021). Our findings suggest a trend between human mobility and confirmed COVID-19 cases. Therefore, we performed further analysis to examine the direct effects of human mobility on confirmed COVID-19 cases in Table 3.

Fig. 1. Daily distributions of human mobility and COVID-19 confirmed cases during 1 to 30 May 2021 in Taiwan.

 
3.2 Air Pollution was Associated with Increases in Confirmed COVID-19 Cases

Table 1 depicted the minimum, maximum, and average concentrations of air pollution during the study period. We observed a decline in air pollutant levels after the nationwide restrictions on 19 May. Notably, concentrations of PM₁₀, PM_2.5, NO₂, and CO respectively decreased by 42%, 37%, 35%, and 25%. We next examined the 7-day-lag effects of confirmed COVID-19 cases on air pollution. The 7-day lag effect on the newly confirmed COVID-19 cases and mortality was reported in previous studies (Wang et al., 2020a; Chung and Chan, 2021; Dales et al., 2021). Previous findings showed that 10 µg m^–3 increase in PM_2.5 was associated with the confirmed cases of COVID-19, and the estimated strongest relative risk were observed at 7-day lag (Wang et al., 2020a). We observed that 1-unit increases in PM₁₀, PM_2.5, NO₂, and CO were respectively associated with 0.042 (95% CI: 0.036, 0.048), 0.025 (95% CI: 0.011, 0.039), 0.077 (95% CI: 0.067, 0.086), and 2.650 (95% CI: 2.280, 3.020) increases in confirmed COVID-19 cases (Table 1). The association between air pollution exposure and COVID-19 cases was identified in various countries. For instance, a multi-city study in China showed that both single-day and cumulative lag effects of short-term exposure to PM_2.5 and PM₁₀ were associated with an increased risk of COVID-19 infection (Wang et al., 2020a). Another study suggested that exposure to short-term NO₂ was positively associated with the transmissibility of COVID-19 in China (Yao et al., 2021). A previous report showed that short-term exposure to PM₁₀, PM_2.5, NO₂, and O₃ was associated with increases in daily COVID-19 infections (Zhu et al., 2020a). Furthermore, a study in the United States found that daily ozone (O₃) concentrations were associated with new confirmed cases of COVID-19 (Adhikari and Yin, 2020). These results are consistent with our findings that exposure to air pollution might increase the risk of confirmed COVID-19 cases.

 
3.3 Human Mobility Reduction was Associated with Declines in Air Pollution

Associations of population mobility with the three commuting modes and levels of ambient air pollution are shown in Table 2. We found that 1 min spent walking, driving, and using public transport was respectively associated with 0.051 (95% CI: 0.044, 0.059), 0.107 (95% CI: 0.095, 0.118), and 0.107 (95% CI: 0.096, 0.117) unit increases in PM₁₀. We observed that 1 min spent walking, driving, and using public transport was respectively associated with a –0.010 (95% CI: –0.013, –0.006) unit decrease, a 0.023 (95% CI: 0.018, 0.029) unit increase, and a 0.045 (95% CI: 0.040, 0.049) unit increase in PM_2.5. We found that 1 min spent walking, driving, and using public transport was respectively associated with 0.024 (95% CI: 0.019, 0.029), 0.060 (95% CI: 0.052, 0.067), and 0.082 (95% CI: 0.076, 0.089) unit increases in NO₂. Finally, we observed that 1 min spent walking, driving, and using public transport was respectively associated with 0.000 (95% CI: 0.000, 0.000), 0.001 (95% CI: 0.000, 0.001), and 0.001 (95% CI: 0.001, 0.002) unit increases in CO. These results indicated that individuals commuting by driving or public transport were associated with higher levels of PM₁₀, PM_2.5, NO₂, and CO than were those commuting by walking. The effects of human mobility on air pollutants (expressed by beta coefficients) decreased as follows: PM₁₀ > NO₂ > PM_2.5 > CO. Our results suggest that a restriction in human mobility was associated with declines in air pollution levels, which was in line with previous studies (Archer et al., 2020; Zhu et al., 2020b). It was reported that social restriction policies are able to reduce human mobility for disease control (Bonaccorsi et al., 2020; Wang et al., 2020b). A previous study in India showed that approximately 50% decreases in PM₁₀, PM_2.5, and NO₂ levels and a 30% decrease in CO levels occurred in Delhi City during the lockdown period compared to the pre-lockdown period (Mahato et al., 2020). Another study in Europe found that lockdown measures led to significant declines in PM₁₀, PM_2.5, and NO₂ levels (Menut et al., 2020). Together, mobility restrictions by social distancing policies were associated with air pollution declines.

 
3.4 Air Pollution Mediated Associations between Mobility and Confirmed COVID-19 Cases

Next, we examined the mediating effects of air pollutants on the association between mobility and confirmed COVID-19 cases as shown in Table 3. In terms of total effects, we observed that 1 min spent walking, driving, and using public transport was respectively associated with –0.244 (95% CI: –0.327, –0.161), –0.131 (95% CI: –0.167, –0.095), and –0.061 (95% CI: –0.083, –0.040) decreases in confirmed COVID-19 cases. We found that the direct effects of walking, driving, and using public transport on confirmed COVID-19 cases after controlling for the indirect (mediated) effects were statistically significant (p < 0.05). In terms of indirect (mediated) effects, we observed that 1 min spent walking, through the mediating effects of PM₁₀ and NO₂, was respectively associated with 0.109 (95% CI: 0.092, 0.131) and 0.087 (95% CI: 0.065, 0.113) increases in COVID-19 cases. However, 1 min spent walking, through the mediating effects of PM_2.5 and CO, was respectively associated with -0.009 (95% CI: –0.013, –0.004) and -0.017 (95% CI: –0.035, –0.004) decreases in confirmed COVID-19 cases. We found that 1 min spent driving, through the mediating effects of PM₁₀, PM_2.5, NO₂, and CO, was respectively associated with 0.084 (95% CI: 0.072, 0.099), 0.011 (95% CI: 0.008, 0.017), 0.083 (95% CI: 0.067, 0.099), and 0.028 (95% CI: 0.018, 0.038) increases in confirmed COVID-19 cases. Next, we observed that 1 min spent using public transport, through the mediating effects of PM₁₀, PM_2.5, NO₂, and CO, was respectively associated with 0.056 (95% CI: 0.048, 0.066), 0.018 (95% CI: 0.012, 0.024), 0.085 (95% CI: 0.074, 0.096), and 0.047 (95% CI: 0.039, 0.054) increases in COVID-19 cases. Importantly, we found that beta coefficients of indirect effects were higher than those of direct effects and total effects. This finding suggests the mediating effects of air pollution on the association between human mobility and confirmed COVID-19 cases. Particulate matter could be another transmission mode of SARS-CoV-2 in the atmosphere. A previous study reported the mediating effects of air pollution on the relationship between the intra-city migration index and COVID-19 infections (Zhu et al., 2020b). It was also demonstrated that exposure to air pollution increased the risk of the COVID-19 mortality (Coker et al., 2020; Lolli et al., 2020). These results show that air pollution plays an essential mediating role in terms of COVID-19 transmission. Previous reports suggest that early social restrictions could effectively control the spread of COVID-19 (Wang et al., 2020b; Zhou et al., 2020), which is in line with our findings. This may be due to the direct and indirect effects of air pollution on COVID-19 transmission. First, PM is considered a “carrier” for the direct transmission method of COVID-19 (Comunian et al., 2020; Farhangrazi et al., 2020; Bourdrel et al., 2021; Tung et al., 2021). Notably, a previous study in Italy found marker genes of SARS-CoV-2 (i.e., N, E, and RdRP) in PM₁₀ samples (Setti et al., 2020). Meanwhile, a study involving 10 cities in Turkey also found SARS-CoV-2 gene (N1 and RdRP) expressions in ambient PM (Kayalar et al., 2021). Another study reported the presence of SARS-CoV-2 RNA in PM_2.5 samples collected from hospital wards (Nor et al., 2021). Second, previous in vivo studies found that exposure to PM_2.5 resulted in increased expressions of transmembrane protease serine type 2 (TMPRSS2) and angiotensin-converting enzyme 2 (ACE2) in lung tissues, which are essential factors facilitating entry of SARS-CoV-2 into host cells (Chuang et al., 2020; Li et al., 2021; Sagawa et al., 2021). Therefore, a reduction in air pollution levels caused by restrictions on human mobility could have resulted in declines in confirmed COVID-19 cases in our study. Taken together, our findings implied that reductions in human mobility (i.e., walking, driving, and using public transport) by restriction policies may be associated with declines in COVID-19 infections through the mediating effects of air pollution.

 
4 CONCLUSION

To conclude, our findings demonstrated that human mobility restrictions caused by the nationwide restrictions could have resulted in decreases in confirmed COVID-19 cases through the mediating effects of air pollution. Therefore, reductions in human mobility and the consequent declines in air pollution could be effective measures for controlling spread of the COVID-19 pandemic.

 
ACKNOWLEDGEMENTS

This study was funded by the Ministry of Science and Technology of Taiwan (MOST 109-2314-B-038-093-MY3).

 
DISCLAIMER

 
The authors declare that they have no conflicts of interest.