key: cord-0689011-1us7hhwq
authors: Sam Leo, Jason Lorenzo; San Tam, Wilson Wai; Seow, Wei Jie
title: Association between air quality, meteorological factors and COVID-19 infection
date: 2021-03-17
journal: Environ Res
DOI: 10.1016/j.envres.2021.111024
sha: c68e56c41914b423a61d10c697ac5d608f66d91e
doc_id: 689011
cord_uid: 1us7hhwq

The coronavirus disease (COVID-19) has become a global pandemic affecting many countries, including Singapore. Previous studies have investigated the relationship of air pollutant levels and meteorological factors with respiratory disease risk and hospital admission rates. However, associations between air pollutant concentrations and meteorological factors with COVID-19 infection have been equivocal. This study aimed to assess the association between core air pollutant concentrations, meteorological variables and daily confirmed COVID-19 cases in Singapore. Data on air pollutant levels (particulate matter [PM(2.5), PM(10)], ozone [O(3)], carbon monoxide [CO], nitrogen dioxide [NO(2)], sulphur dioxide [SO(2)], pollutant standard index [PSI]) and meteorological factors (rainfall, humidity, temperature) was obtained from the Singapore National Environmental Agency (NEA) from 23 January 2020 to 6 April 2020. The daily reported COVID-19 case numbers were retrieved from the Singapore Ministry of Health (MOH). Generalized linear models with Poisson family distribution and log-link were used to estimate the model coefficients and 95% confidence intervals for the association between air pollutant concentrations and meteorological factors (8-day and 15-day moving averages (MA)) with COVID-19 case numbers, adjusting for humidity, rainfall and day of week. We observed significantly positive associations between NO(2), PSI, PM(2.5) and temperature with COVID-19 cases. Every 1-unit increase (15-day MA) in PSI, 1 μg/m(3) increase (15-day MA) in PM(2.5), NO(2) and 0.1°C increase in temperature were significantly associated with a 35.0% (95% CI: 29.7% to 40.5%), 22.6% (95% CI: 12.0% to 34.3%), 34.8% (95% CI: 29.3% to 40.4%) and 28.6% (95% CI: 25.0% to 32.4%) increase in the daily number of COVID-19 cases respectively. On the contrary, PM(10), O(3), SO(2), CO, rainfall and humidity were significantly associated with lower daily numbers of confirmed COVID-19 cases. Similar associations were observed for the 8-day MAs. Future studies could explore the long-term consequences of the air pollutants on COVID-19 infection and recovery.

The coronavirus disease outbreak first emerged in Wuhan city (Hubei, China) in December 2019 and has since become a global pandemic (Dong et al., 2020) . Runny nose, cough, sore throat, breathing difficulty, loss of taste or smell, and fever are some of the common symptoms of COVID-19 infection. The incubation period is around 5 to 14 days (Lauer et al., 2020) . The first reported case of COVID-19 in Singapore occurred on 23 January 2020. As of 30 April 2020, the number of confirmed cases had increased to 16,169. In that period, there were only a total of 6 fatalities due to . Elderly people are more prone to developing complications due to COVID-19 (Wang et al., 2020) . With the rising daily COVID-19 case counts in Singapore, and to prevent further spread of the disease in the community, the government implemented a series of strict measures called the 'circuit breaker', essentially a partial lockdown from 7 April 2020 to 1 June 2020. During this period, all schools and non-essential places of work were suspended, with only essential businesses continuing operation.

Air pollution is one of the leading causes of overall mortality globally and accounts for more than 3 million deaths annually (World Health Organization, 2016) . Previous studies have investigated the association between air pollutants and other meteorological factors on hospital admission rates (Liu et al., 2020; Qi et al., 2020; Zhu et al., 2020) . A recent study investigated the short term effects of air pollutants on the number of hospital admissions in Bangkok, and showed that the increased exposure to air pollutants such as NO 2 or CO was associated with increased hospital admissions (Phosri et al., 2019) . A recent study in Singapore also investigated the correlation between meteorological factors and the COVID-19 confirmed cases in Singapore (Pani, Lin, & RavindraBabu, 2020) . Other studies have also investigated the effects of meteorological variables on respiratory diseases and found an inverse relationship between temperature and respiratory disease risk. For instance, lower J o u r n a l P r e -p r o o f temperature was associated with an increased risk of invasive pneumococcal disease (Ciruela et al., 2016) . A recent study in Brazil reported that the largest number of respiratory related problems occurred during the winter, with temperature inversely associated with the number of hospitalizations (da Silva et al., 2019) .

Several studies have investigated the association between air pollutant concentrations and meteorological factors with the number of daily COVID-19 incident cases. However, the findings are conflicting. Most studies reported significant positive associations between PM 10 , PM 2.5 , NO 2 and O 3 , but inverse associations between temperature and daily COVID-19 case counts (Liu et al., 2020; Marquès et al., 2021; Qi et al., 2020; Tobías and Molina, 2020) .

In contrast, a study conducted in Jakarta, Indonesia showed that temperature had a positive correlation with the daily number of COVID-19 cases (Tosepu et al., 2020) . A previous study in Singapore reported that temperature, dew point, relative humidity, absolute humidity, and water vapor were positively correlated with COVID-19 cases (Pani et al., 2020) . This study is the first study to assess the association between air pollutant levels and the number of daily COVID-19 cases in Singapore, controlling for other meteorological factors. Understanding the associations between air pollutant concentrations and climatic factors with COVID-19 infection may be helpful for resource planning in the primary care setting, and for crafting preventive interventions to limit or curb the spread of COVID-19 and other infectious diseases in the community, throughout every stage of a developing acute pandemic.

Air pollutant data for the period of 23 January 2020 to 6 April 2020 was obtained from the National Environmental Agency (NEA) website (https://data.gov.sg/). From 7 April 2020, J o u r n a l P r e -p r o o f there was a spike in the reported number of COVID-19 cases in Singapore. The rise in the reported daily confirmed COVID-19 cases mainly stemmed from the increased testing of the migrant workers in dormitories, which contributed to a large proportion of confirmed cases in April. Air pollution levels were shown to be drastically lower during the COVID-19 lockdown periods in several cities (Anjum, 2020) . Therefore, all data after 7 April 2020 was excluded in order to eliminate the effects from the circuit breaker as well as increased testing in dormitories.

The air pollutant data consisted of hourly nitrogen dioxide (NO 2 ), 24-hour sulphur dioxide (SO 2 ), 8-hour carbon monoxide (CO), 8-hour ozone (O 3 ), hourly Pollutant Standards Index (PSI), hourly particulate matter with a diameter of 2.5 micrometres or less (PM 2.5 ) and hourly particulate matter with a diameter of 10 micrometres or less (PM 10 ). The daily averages of the air pollutants were computed and used for the analysis.

The PSI calculation includes the six main air pollutants, CO, NO 2, O 3 , PM 2.5 , PM 10 and SO 2 .

The calculation for the sub-index of each air pollutant involves transforming the concentrations to a scale from 0 to 500 (NEA, 2014) . The overall value of the PSI is then derived from the maximum sub-index of the six air pollutants (NEA, 2014).

Temperature (°C) and total daily rainfall (mm) were obtained from the NEA website for the period of 23 January 2020 to 6 April 2020 (http://www.weather.gov.sg). Data on humidity (%), measured four times daily, was obtained from Time and Date AS (Aksjeselskap) (https://www.timeanddate.com/weather/singapore). The daily averages of the meteorological factors were computed and used for the analysis.

The daily confirmed number of COVID-19 cases was obtained from the Ministry of Health, Singapore (https://covidsitrep.moh.gov.sg/). The first confirmed case of COVID-19 in Singapore was reported on 23 January 2020 and the data was collected up until 6 April 2020, one day before the start of the circuit breaker. The daily confirmed cases were reported at the end of each day, therefore there were no delays in terms of reporting. There were a total of 1375 COVID-19 cases and 6 fatalities from the period of 23 January 2020 to 6 April 2020.

Descriptive analysis was used to explore the distribution of the air pollutants. The Spearman rank correlation test was used to investigate the bivariate relationship between the air pollutants, meteorological variables, and the daily confirmed number of COVID-19 cases.

Due to the incubation period of COVID-19 that lasts for up to 14 days, a moving average (MA) was applied to capture potential lag effects of the air pollutants (Liu et al., 2020; Qi et al., 2020; Zhu et al., 2020) . Thus, 8 day and 15 day MAs of the air pollutants and meteorological factors were used in the analyses. A generalized linear model (GLM) with a Poisson family distribution and log-link function was used to estimate the association between the MA air pollutant concentrations and the log-transformed daily confirmed number of COVID-19 cases, adjusting for humidity (continuous), rainfall (continuous), and the day of week (categorical). Variables that were selected to be included in the final model were based on the chi-square test. Temperature was not included in the model as the temperature range during the study period was very narrow, as compared to rainfall and humidity. Each model contains one air pollutant to avoid collinearity due to their high correlations (Table A1) . log(Y) = α + β 1 ×X i + β 2 ×Humidity + β 3 ×Rainfall + β 4 ×dow J o u r n a l P r e -p r o o f X i represents each air pollutant used in the analysis. Effect estimates are reported in terms of percentage changes (%) in daily confirmed COVID-19 cases resulting from a unit increase in air pollutant concentrations or meteorological factors (i.e., 1 unit increase in PSI, 1% increase in humidity, 1mm increase in daily rainfall, 1μg/m 3 increase in PM 10 , NO 2 , O3, SO 2 or 0.01mg/m 3 increase in CO). We obtain the percent changes (%) by exponentiating the effect estimates, subtracting them by 1 and then multiplying them by 100. A negative binomial regression model was also used for comparison as part of a sensitivity analysis of the results.

All analyses in this study was performed using the R statistical software (version 3.6). False discovery rates (FDR) were calculated using the Benjamini-Hochberg method to account for multiple comparisons. All statistical analyses were evaluated using two-sided tests at the 0.05 level of significance. to -27.4%) and -83.5% (95% CI: -86.5% to -80.0%) change in the number of daily confirmed COVID-19 cases respectively. Every 1% increase in humidity and 1 mm increase in daily rainfall was associated with -8.50% (95% CI: -9.54% to -7.45%) and -24.8% (95% CI: -32.0% to -16.9%) change in the number of daily confirmed COVID-19 cases respectively.

For CO, each 0.01 µg/m 3 increase (15-day MA) was associated with a -31.3% (95% CI: -33.5

to -29.1) change in the daily number of COVID-19 cases. The results from negative binomial regression models were mostly similar (Table A2 ). 

In this study, we found significant, positive associations between PM 2.5 , PSI, and NO 2 with the daily number of confirmed COVID-19 cases. In addition, we found significant, negative associations between PM 10 , O 3 , SO 2 , CO, daily rainfall and humidity with the daily number of confirmed COVID-19 cases.

This is consistent with results from previous studies conducted in Italy and China, where NO 2 was found to be positively associated with daily COVID-19 cases (Frontera et al., 2020; Zhu et al., 2020) . Prior to the COVID-19 pandemic, there were studies from China and Thailand that found positive associations between NO 2 and the morbidity due to rotavirus as well as the number of hospital admissions (Phosri et al., 2019; Yan et al., 2019) . Furthermore, exposure to air pollutants has been previously reported to have an association with increased respiratory infections (Chen et al., 2007; Huff et al., 2019) . Previous studies that investigated the effects of PM 2.5 have shown that it was positively associated with daily COVID-19 cases in China and the United States (Wu et al., 2020; Zhu et al., 2020) . Potential mechanisms for this association include impaired immune and lung defence mechanisms in the respiratory mucosa and alterations in alveolar macrophage function in response to air pollutant exposures (Chauhan and Johnston, 2003; Huff et al., 2019) .

There are some notable differences in our findings, compared to previous studies. It was previously shown that the NO 2 , CO and SO 2 exposures were associated with a higher risk of J o u r n a l P r e -p r o o f respiratory disease infections (Chen et al., 2007) . However, in our study, we found negative associations between CO and SO 2 with COVID-19 cases. In another study however, SO 2 was shown to have a negative association with daily COVID-19 cases, which was similar to our finding (Zhu et al., 2020) . Our finding of a negative association between O 3 and the daily confirmed number of COVID-19 cases was conflicting with another previous study in China,

where a positive association was reported (Zhu et al., 2020) . In our study, PM 10 was significantly negatively associated with daily confirmed number of COVID-19 cases.

However, previous studies demonstrated a positive association between PM 10 with daily hospitalization cases of respiratory diseases (Davila Cordova et al., 2020; Ferreira et al., 2016; Wu et al., 2020; Zhang et al., 2019) . Higher concentrations of PM 10 have been shown to prolong the spread of the virus in the atmosphere, as evidenced by a recent study conducted in Italy that provided evidence showing that the COVID-19 virus can be detected on outdoor particular matter (Setti et al., 2020a) . A similar paper in Italy had also discovered that in regions that had higher PM 10 concentrations, the spread of the COVID-19 virus was faster (Setti et al., 2020b) . Prior to the start of the circuit breaker on 7 April 2020, there have been several measures put in place progressively to reduce crowd sizes in public areas. These included allowing employees to work from home, and the closure of businesses and schools that had suspected cases or contact with confirmed cases. These preventive measures may have resulted in a reduction in the air pollutant levels due to restricted movement and lower traffic volume. This may partly explain the negative associations that we observed in the association between PM 10 and daily COVID-19 cases.

Reducing air pollutant levels such as NO 2 may help in limiting the spread of COVID-19

infection. Further laboratory research should be done to uncover the fundamental mechanisms between air pollutant concentrations and daily confirmed COVID-19 cases.

From our understanding, this is the first study investigating the association between air J o u r n a l P r e -p r o o f pollutants, climatic factors and number of confirmed COVID-19 cases in a densely populated Asian city with a tropical climate, using a relatively large number of data points spanning over three months. A similar study was done in Singapore that investigated the correlation between the meteorological factors with the daily confirmed COVID-19 cases from 23

January 2020 and 31 May 2020 (Pani et al., 2020) . That study investigated 9 different meteorological variables and its correlation with the daily confirmed number of COVID-19

cases, while our study looked at humidity, rainfall and temperature along with the air pollutant concentrations. However, it should be noted that this study has some limitations.

Firstly, we only investigated the associations between the core air pollutant concentrations and number of COVID-19 cases for which data was available. Therefore, we were unable to test the associations of other air pollutants and sub-components of PM 2.5 . Secondly, we did not have information on the demographics such as age, gender, ethnicity or other socioeconomic factors of the COVID-19 cases and hence, we were unable to do further stratification by demographic factors. From the period of 23 January 2020 to 6 April 2020, there were a total of 6 fatalities from the COVID-19 virus. We were unable to expand the analysis to assess the association between chronic exposure to air pollutants and the mortality due to COVID-19 in Singapore. The rapid testing in the dormitories took place from the 7th

April 2020 during the partial lockdown (circuit breaker) imposed in Singapore. Therefore, we do not have the data on the total number of tests that occurred in Singapore and were not able to calculate the ratio of COVID-19 cases to tests. In addition, we did not have information on other potential confounders such as smoking status and traffic flow. Lastly, Singapore has a rather stable climate throughout the year, and has a smaller range of the meteorological levels and air pollutant concentrations, namely CO, SO 2 and temperature, as compared to other temperate countries. The temperature range was not able to cover cool weather due to the tropical climate in Singapore, and as such, we were unable to investigate if low temperatures J o u r n a l P r e -p r o o f 15 are associated with the number of COVID-19 cases in Singapore. This may have contributed to differences with the findings from other countries.

Our study found significant associations between some of the core air pollutants with the J o u r n a l P r e -p r o o f : particulate matter with a diameter of 2.5 micrometres or less, PM 10 : particulate matter with a diameter of 10 micrometres or less, FDR: False discovery rate using Benjamini & Hochberg a Models were adjusted for humidity, rainfall and day of week b Models were adjusted for humidity, rainfall and day of week unless the adjustment was the exposure in the analysis c Estimates for CO are shown as a percentage change for a 0.01mg/m 3 increase in CO J o u r n a l P r e -p r o o f

Good in The Worst: COVID-19 Restrictions and Ease in Global Air Pollution

Air pollution and infection in respiratory illness

Outdoor Air Pollution: Nitrogen Dioxide, Sulfur Dioxide, and Carbon Monoxide Health Effects

Invasive pneumococcal disease rates linked to meteorological factors and respiratory virus circulation (Catalonia

Excess of children's outpatient consultations due to asthma and bronchitis and the association between meteorological variables in Canoas City, Southern Brazil

Association of PM(2.5) concentration with health center outpatient visits for respiratory diseases of children under 5 years old in Lima

An interactive web-based dashboard to track COVID-19 in real time

Effects of Particulate Matter and Its Chemical Constituents on Elderly Hospital Admissions Due to Circulatory and Respiratory Diseases

Severe air pollution links to higher mortality in COVID-19 patients: The "double-hit" hypothesis

An update on immunologic mechanisms in the respiratory mucosa in response to air pollutants

The Incubation Period of Coronavirus Disease

From Publicly Reported Confirmed Cases: Estimation and Application

Impact of meteorological factors on the COVID-19 transmission: A multi-city study in China. Science of The Total Environment

Effects of air pollution on the potential transmission and mortality of COVID-19: A preliminary case-study in Tarragona Province

Computation of the Pollutant Standards Index (PSI)

Association of COVID-19 pandemic with meteorological parameters over Singapore. Science of The Total Environment

Effects of ambient air pollution on daily hospital admissions for respiratory and cardiovascular diseases in

COVID-19 transmission in Mainland China is associated with temperature and humidity: A time-series analysis. Science of The Total Environment

SARS-Cov-2RNA found on particulate matter of Bergamo in Northern Italy: First evidence

The Potential role of Particulate Matter in the Spreading of COVID-19 in Northern Italy: First Evidence-based Research Hypotheses

Is temperature reducing the transmission of COVID-19 ? Environmental Research

Correlation between weather and Covid-19 pandemic in Jakarta, Indonesia. Science of The Total Environment

Coronavirus disease 2019 in elderly patients: Characteristics and prognostic factors based on 4-week follow-up

World Health Organization, WHO releases country estimates on air pollution exposure and health impact

Exposure to air pollution and COVID-19 mortality in the United States: A nationwide cross-sectional study. medRxiv : the preprint server for health sciences

Short-Term Effects of Meteorological Factors and Air Pollutants on Hand

Association of particulate matter air pollution and hospital visits for respiratory diseases: a time-series study from

Association between short-term exposure to air pollution and COVID-19 infection: Evidence from China. Science of The Total Environment

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.J o u r n a l P r e -p r o o f

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.