1. Introduction
The World Health Organization formally announced a global pandemic from a new coronavirus in March 2020 [
1]. The illness which is caused by the virus called SARS-CoV-2 produces respiratory symptoms that are fatal in some cases. Current evidence suggests that the virus is spread through close contact with other humans, primarily through coughing or sneezing [
2]. The coronavirus disease (COVID-19) continues to spread, with emerging variants of the virus. Millions of cases and deaths have been reported globally. In the United States alone, millions of cases have been recorded, with thousands of deaths resulting from complications related to the disease [
3].
To alleviate the damages caused by the COVID-19 pandemic, numerous studies have examined the possible factors contributing to the transmission of the disease. Pioneer studies focused on person-to-person transmission and analyzed how factors such as population mobility play a significant role in virus transmission [
4,
5]. Other studies explored how contaminated environmental surfaces aid the transmission of the virus from an infected to an uninfected person [
6,
7]. Research on the impact of short-term exposure to pollutants such as particulate matter (PM), carbon monoxide (CO), sulfur dioxide (SO
2), nitrogen dioxide (NO
2), and ozone (O
3) on the transmission of COVID-19 is still emerging, and comparatively scarce.
Before the COVID-19 pandemic, past studies have linked pollutants to adverse health conditions [
8,
9]. Nitrogen dioxide (NO
2), for example, has been linked to chronic obstructive pulmonary disease and asthma [
10]. Other studies suggest that particulate matter such as PM
10 and PM
2.5 could act as a platform for respiratory virus transmission [
11,
12,
13,
14]. Air quality and environmental conditions have also been associated with lung infections caused by viruses [
15]. There are also indications that air pollution weakens the immune system and may increase one’s chances of contracting respiratory viral infections such as COVID-19 [
16,
17]. A recent study suggests that a 1 μg/m
3 increase in PM
2.5 is associated with an 8% increase in the COVID-19 death rate [
18].
Though air quality varies with weather conditions, some cities experience poor air quality for extended periods due to environmental disasters such as volcanic eruptions and wildfires. This paper examines the relationship between short-term exposures to pollutants such as PM2.5, NO2, CO, and overall air quality (AQI) and confirmed cases of COVID-19 during the 2020 wildfire season in California. We model the effect of these pollutants using time series methods. We apply Generalized Additive Models (GAMs) for estimating the relationship while controlling for meteorological factors, day-fixed effects, and county fixed effects. In one of the sensitivity analyses, we use a Feasible Generalized Least Squares Model (FGLS) model to check the robustness of our findings. The study focuses on 20 counties impacted by the wildfires between February and December 2020.
1.1. The 2020 California Wildfires and COVID-19
The peak of the COVID-19 pandemic in the United States in 2020 coincided with the most significant wildfire season recorded in modern California’s history [
19,
20]. The 2020 California wildfire season began in February and lasted for months with the major fire outbreaks occurring in the months from August to December [
21]. By the end of the fire season in December, the fires had destroyed 4,257,863 acres of land [
21,
22]. Though particulate matter, oxides of nitrogen, and sulfur pollution declined sharply in many parts of the world during the pandemic, wildfire days in California recorded higher amounts of PM
2.5 pollution [
23,
24,
25]. This highlights the need to investigate if higher daily concentrations of pollutants are associated with increased COVID-19 cases in counties that experienced wildfires in California.
Past research has linked wildfire smokes to adverse health outcomes [
26,
27,
28]. The Environmental Protection Agency describes wildfire smoke as a public health concern comprising a mix of gaseous pollutants such as carbon monoxide (CO) and particulate matter. Studies indicate that particulate matter, which contains solid and liquid suspension, poses a significant risk to public health during a wildfire [
29]. Furthermore, elevated CO levels outdoors during a fire can be of particular concern for those with some preexisting conditions such as heart disease. This paper analyzed data from 20 counties impacted by the wildfires. We use 1 February 2020, to 31 December as our study period. This was selected for two reasons; COVID-19 data is available from February 2020, and the 2020 California fire season lasted until December 2020.
1.2. Long-Term and Short-Term Air Pollution and COVID-19
Both long and short-term exposure to air pollutants may be a complex factor in increasing SARS-CoV-2 transmission and lethality [
30]. A study in 71 provinces across Italy suggests that chronic air quality was highly correlated with COVID-19 cases suggesting that chronic exposure to air pollution may predispose people to the disease [
2]. Similarly, another study found that long-term exposure to high amounts of PM
2.5 is associated with increased mortality from COVID-19 [
18]. A study in Spain examined the spatial spread of COVID-19 using a mixed longitudinal ecological design. Their results suggest that chronic exposure to NO
2 and PM
10 are predictors of the spatial spread of the virus [
31]. Another study employed an ecological analysis to examine the relationship between chronic exposure to pollutants and reported cases of COVID-19 in Canada. They applied a negative binomial regression model and found positive associations between long-term exposure to PM
2.5 and COVID-19 incidence [
32].
There is still relatively less research on the impact of short-term exposure to pollutants on COVID-19 transmission. Recent studies report mixed findings and use different methodologies. A study using Generalized additive models (GAM) found positive associations between moving average concentrations of pollutants such as NO
2, PM
2.5, O
3, and COVID-19 cases in China [
33]. Another examined the spatial relationship between PM
10 and PM
2.5 and COVID-19 deaths. Their results suggest a positive relationship between pollutants and COVID-19 deaths [
34]. However, another research found negative relationships between COVID-19 and these pollutants including NO
2 and SO
2 in California [
35]. This study, however, employed fundamental techniques such as Spearman and Kendall correlation for their statistical analysis.
Using data from Los Angeles and Ventura counties in the US, another study applied a generalized linear model (GLM) to examine the same relationship [
36]. They also found negative relationships between daily SARS-CoV-2 cases and pollutants PM
2.5 and PM
10. Though they applied a dynamic emission model to further strengthen their analysis, the study focused only on two counties, and they did not control for the daily SARS-CoV-2 test which is highly correlated to the number of confirmed cases recorded across counties. A more recent study found positive associations between exposure to PM
2.5 in the short-term and COVID-19 cases and deaths using data collected during the 2020 wildfire season in 92 western U.S. counties [
25]. However, this study did not examine other air pollutants like CO and NO
2. They also did not adjust for the confounding effects of SARS-CoV-2 tests in their models.
Since the daily levels of pollutants are higher on wildfire days than on non-wild fire days [
25], the 2020 California wildfire season provides an opportunity to estimate the relationship between short-term exposure to pollutants and confirmed cases of COVID-19. This paper extends the current literature on the impacts of short-term pollution on health by exploiting the wildfire season in California and applying different time series methodologies to further investigate this relationship. We first use Generalized additive models in the main analyses to analyze the short-term impact of pollutants such as PM
2.5, CO, NO
2, and overall air quality index (AQI) on confirmed COVID-19 cases. We then use a Feasible Generalized Least Squares Model in one of the four sensitivity analyses to check the robustness of our findings.
4. Discussion
In this paper, we employed time series methods to explore the relationship between air pollution and daily confirmed COVID-19 cases. We observed significant associations for all the pollutants examined in this study. As demonstrated in previous studies, the effect of air pollution can linger for several days after incidents [
33,
44]. Our choice for moving averages (lag 0–7, lag 0–14, or lag 0–21) is based on previous studies and official statements on the COVID-19 incubation period issued by the US Centers for Disease Control and Prevention [
46,
47]. Our findings show that PM
2.5, CO, and AQI are all significantly and positively associated with confirmed COVID-19 cases in all the moving average lags. However, the GAM results show a negative relationship between NO
2 and COVID-19 cases. These results remained robust in all the sensitivity analyses with the GAM. In the sensitivity analyses using the FGLS model, we find positive associations between all the pollutants and confirmed COVID-19 cases. These findings suggest that air pollution could play an important role in COVID-19 transmission.
Many studies have shown that air pollution is correlated to respiratory infections caused by microorganisms [
55,
56]. Air quality and environmental conditions have been associated with lung infections caused by viruses [
15]. Some studies suggest that particulate matter such as PM
10 and PM
2.5 could act as a platform for virus transmission [
11,
12,
13,
14]. There are also indications that air pollution weakens the immune system and may increase one’s chances of contracting respiratory viral infections such as COVID-19 [
16,
17]. We made comparisons between the findings in this study and previous studies to check for similarities and differences. In one study [
57], short-term exposure to higher PM
2.5 was correlated with higher confirmed cases of acute lower respiratory infection using an observational cross-over design. Another study that combined a generalized Poisson regression model and a distributed lag nonlinear model (DLNM) found significant associations between atmospheric particulate matter (PM
2.5 and PM
10,) and hospitalizations for respiratory diseases in China [
58].
Other studies specific to COVID-19 have also found significant associations between air pollution and the disease. Using distributed lag models, a study found that increased exposure to PM
2.5 was associated with increased COVID-19 cases and deaths across 92 counties in the US [
25]. Another study using a GAM found significant relationships between confirmed cases of COVID-19 and six air pollutants in China [
33]. As in our study, these two studies consistently found positive associations between pollutants (PM
2.5, CO, and NO
2) and confirmed COVID-19 cases. Unlike our study, other studies done in the US have found negative associations between confirmed cases of COVID-19 and the pollutants analyzed in our study [
35,
36,
51]. However, the methodologies employed in these studies were very different from the ones used in our study.
Some studies suggest that exposure to certain concentrations of NO
2 could reduce susceptibility to respiratory viral infections [
59,
60]. This could be the reason for the negative relationship observed in the results in the GAM models used in this study. Additional research is needed to understand the biological mechanisms behind this phenomenon observed not just in our study but in others [
35,
36,
51].
This study has several limitations. For one, it focused on associations and not causal effects of air pollution and indicators of air pollution on confirmed cases of COVID-19. For example, more people could have stayed indoors during the wildfires which may have increased their chances of contracting the virus. Studies indicate that close spaces with poor air circulation and inhalation of aerosols from infected persons increase the spread of respiratory viruses [
61,
62].
This study also focused on 20 out of the 53 counties impacted by the 2020 California wildfires. While this was in part due to the incompleteness of data on some pollution and meteorological indicators, a complete sample of the counties affected would have improved the validity of our findings. We also did not account for economic or social factors that could have increased COVID-19 risk. Many studies indicate that certain socioeconomic groups have been disproportionately impacted by COVID-19 [
63]. Lastly, this study did not consider sub-group analysis in terms of demographics such as gender, occupation, or race. Future studies should check for possible heterogeneity across socio-demographic groups.
5. Conclusions
This study suggests that there is a significant relationship between air pollution and confirmed cases of COVID-19. Short-term exposure to increased concentrations of PM2.5, CO, and higher values of AQI is associated with an increased risk of COVID-19. These results remained robust in all the sensitivity analyses done in this study, Our findings also suggest that short-term exposure to a higher concentration of NO2 is related to decreased risk of COVID-19 infection. This finding calls for further research to understand this phenomenon.
This study has obvious importance for the management of COVID-19 transmission or future pandemics of respiratory diseases. For a more nuanced public health advisory, policymakers should pay close attention to regions with more predisposition to forest fires and inadvertently, higher measures of air pollution. This is due to the fact that these regions may be disproportionately impacted by respiratory diseases such as COVID-19.