Table 1.

Literature Review on Air Pollution and COVID-19.

Study Area	Study Period	Statistical Model	Findings
Northern Italy [19]	24 February 2020–13 March 2020	Recursive binary partitioning tree approach	Daily PM10 exceeding 50 µg/m3 with a 15-day lag was a significant predictor for COVID-19 incidence
Chinese cities (Wuhan, Xiaogan and Huanggang) [20]	25 January 2020–29 February 2020	Poison regression adjusting for other air pollutants and meteorological variables in each city	Daily PM2.5 was positively associated with COVID-19 incidence with RR from 1.036 to 1.144. The association with PM10 was negative with RR between 0.915 and 0.964. Results for other pollutants (SO2, CO, NO2, and 8-hour O3) were not consistent among the study sites.
Chinese cities (Wuhan and Xiaogan) [21]	26 January 2020–29 February 2020	Univariate linear regression	PM2.5 and NO2 were positively associated with COVID-19 incidence 4 days later in both cities, while PM10 and CO were inconsistent between cities.
120 Chinese cities [22]	23 January 2020–29 February 2020	Generalized additive model adjusting for meteorological variables with city fixed effects	PM2.5, PM10, NO2 and O3 with a 2-week lag were positively associated with COVID-19 incidence, while SO2 was negatively associated. A 10µg/m3 increase in PM2.5 with a 2-week lag was associated with a 2.24% increase in COVID-19 incidence.
49 Chinese cities [23]	As of 22 March 2020	Multivariate linear regression model adjusting for GDP per capita and hospital beds per capita	Both short-term (01/15/2020 – 02/29/2020) and long-term (2015–2019) exposure to elevated PM2.5 and PM10 were associated with increased COVID-19 fatality. A 0.24% and a 0.61% increase in COVID-19 fatality were associated with 10-µg/m3 increase in short-term and long-term PM2.5, respectively.
7 metropolitan cities and 9 provinces in Korea [24]	3 February 2020–5 May 2020	Generalized additive model adjusting for meteorological variables, location and date	Significantly temporal associations were observed between COVID-19 incidence and daily NO2, CO and SO2, but not with PM2.5, PM10 or O3.
3089 counties in the United States [25]	As of 18 June 2020	Negative binomial fixed model adjusting for 20 covariates	Each 1-µg/m3 increase in long-term PM2.5 exposure (2000–2016 annual average) was associated with 11% increase in COVID-19 mortality.
3223 counties in the United States [26]	As of 11 July 2020	Negative binomial fixed model adjusting for other pollutants as well as county characteristics	HAPs was associated with increase COVID-19 mortality. Each 1-µg/m3 increase in long-term PM2.5 exposure (2000–2014 annual average) was associated with 7% increase in COVID-19 mortality
355 municipalities in the Netherlands [27]	As of 5 June 2020	Linear regression controlling for covariates	Long-term exposure to PM2.5 and NO2 were positively associated with COVID-19 outcomes, including incidence and mortality, but not with SO2. Each 1-µg/m3 increase in long-term PM2.5 exposure (2015–2019) was associated with 9.4 more COVID-19 cases, 3.0 more hospital admissions, and 2.3 more deaths.
71 Italian provinces [28]	As of 27 April 2020	Spatial correlation	Positive correlations were observed between COVID-19 incidence and long-term exposure (2016–2019) to NO2, PM2.5, PM10 and O3.
20 Italian regions and up to 110 provinces [29]	As of 31 March 2020	Multiple linear regression	Both long-term exposure (2017 annual mean) to PM2.5 and PM10 were associated with COVID-19 incidence. Each 1-µg/m3 increase in PM2.5 was associated with 0.26 increase in base-10 transformed COVID-19 incidence.
3108 counties in the United States [30]	As of 31 May 2020	Linear regression with adjusting for county-level covariates	PM2.5 (2016 annual mean) and diesel PM were associated with both COVID-19 incidence and mortality. Additional 23.5 cases and 1.08 deaths were associated with each 1-µg/m3 increase in PM2.5.