key: cord-0952423-sp4nvbbh
authors: Azuma, Kenichi; Kagi, Naoki; Kim, Hoon; Hayashi, Motoya
title: Impact of climate and ambient air pollution on the epidemic growth during COVID-19 outbreak in Japan
date: 2020-08-12
journal: Environ Res
DOI: 10.1016/j.envres.2020.110042
sha: 83814abfe5e8f6f0be5e47fd56ce7192b962f6b1
doc_id: 952423
cord_uid: sp4nvbbh

Coronavirus disease 2019 (COVID-19) rapidly spread worldwide in the first quarter of 2020 and resulted in a global crisis. Investigation of the potential association of the spread of the COVID-19 infection with climate or ambient air pollution could lead to the development of preventive strategies for disease control. To examine this association, we conducted a longitudinal cohort study of 28 geographical areas of Japan with documented outbreaks of COVID-19. We analyzed data obtained from March 13 to April 6, 2020, before the Japanese government declared a state of emergency. The results revealed that the epidemic growth of COVID-19 was significantly associated with increase in daily temperature or sunshine hours. This suggests that an increase in person-to-person contact due to increased outing activities on a warm and/or sunny day might promote the transmission of COVID-19. Our results also suggested that short-term exposure to suspended particles might influence respiratory infections caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Further research by well-designed or well-controlled study models is required to ascertain this effect. Our findings suggest that weather has an indirect role in the transmission of COVID-19 and that daily adequate preventive behavior decreases the transmission.

We collected data during the COVID-19 outbreak in Japan (May 22, 2020), as a symptoms. In addition, residential area, occupation, history of traveling abroad, data 2 sources from the MHLW and local governments, and so forth have been also recorded. 3 The COVID-19 cases infected in Japan were included in this study. The COVID-19 4 cases in the Diamond Princess cruise ship docked at Yokohama were excluded. The Thus, cases returning from foreign country that entered to Japan within 5 days before 8 onset of the disease, cases identified as positive by the RT-PCR test at the airport 9 quarantine depot, and cases living in a foreign country were excluded from our study. 10 We identified the date of onset of the symptoms in each case from collected data. The 11 median of the time interval between the date of onset and the date identified as positive 12 by the RT-PCT test was 6 days from January to April 12. The mean value was 13 approximately 6.5 days. Thus, in the absence of a date of onset of symptoms, we 14 estimated the date of onset by subtracting 6 days from the date identified as a positive. 15 Furthermore, we estimated the date of infection from the date of onset. As described 16 above, the incubation period for COVID-19 was approximately 5 days. Thus, we 17 deduced the date of infection by subtracting 5 days from the date of onset.

in this study since the beginning of the epidemic (set on January 15, 2020; Figure S1 ) as 1 of April 7, 2020. Areas affected by large localized cluster occurrence were excluded 2 from the study. 4 We obtained demographic data from the surveyed areas. This includes population, 5 percentage of inhabitants aged ≥65 years, total land surface area (km 2 ), inhabitable area 6 (km 2 ), taxable income, tax debtor, life expectancy at birth (MIAC, 2019), and health 7 expenditure (Cabinet Office, 2020). We calculated the urban density (using population 8 and inhabitable area) and taxable income per person (using taxable income and tax 9 debtor). 11 The daily meteorological data were obtained from Japan Meteorological Agency 12 (JMA, 2020). This includes mean, minimum, and maximum ambient air temperature; 13 precipitation; sunshine hours; wind speed; vapor pressure; relative humidity; and 14 minimum relative humidity. The mean absolute humidity was calculated by vapor 15 pressure and saturated water vapor pressure at its corresponding temperature. If the 16 vapor pressure was not available, the absolute humidity was determined from relative 15 We used weighted random-effects regression analysis with accommodation of 16 correlated longitudinal data to determine the association between the logarithm of the increase in a continuous exposure variable is associated with an increase in the epidemic 7 growth. 8 We determined the association of epidemic growth with exposure variables using 9 univariate and multivariable analyses for 28 geographical areas, divided to five periods 10 from March 13 to April 6, 2020 with a longitudinal manner. In the multivariable 11 analyses, we examined the association of epidemic growth with exposure variables on 12 the regional climate or air pollutants, adjusting for male inhabitants, inhabitants aged 13 ≥65 years, urban density, taxable income, health expenditure, and life expectancy at 14 birth as the possible specified covariates. After correlations (using Pearson's test) 15 among exposure variables were examined, further multivariable models were examined 16 to determine the robustness of the associations with exposure variables. We used p < 2 We included 28 geographical areas and enrolled 6,529 cases in our analyses after 3 reviewing the inclusion and exclusion criteria ( Figure 1 ). These areas were urban in 4 nature, with a population of >200,000 (Table S1 ). The baseline characteristics of the 5 surveyed areas are displayed in Table 1 (Table 3) . No significant association was 3 found with mean temperature, mean daily minimum temperature, mean daily maximum 4 temperature, precipitation, mean relative humidity, mean daily minimum relative 5 humidity, and air pollutants (NO, Ox, SPM, or PM 2.5 ). Figure S2 shows bubble plots of 6 the RR of COVID-19 on these variables. However, after adjusting for demographic 7 variables with multivariable regression analyses, the associations became insignificant 8 except for the association with the sunshine hours. Conversely, the associations with 9 overall mean temperature [RRR, 1.04; 95% confidence interval (CI), 1.01-1.08], mean 10 daily minimum temperature (RRR, 1.03; 95% CI, 1.00-1.06), mean daily maximum 11 temperature (RRR, 1.04; 95% CI, 1.01-1.08), and SPM (RRR, 1.03; 95% CI, 1.00-12 1.05) became positively significant (Table 3) . Kroll and Seinfeld, 2008). The correlation between mean temperature and sunshine 2 hours was very low (r = 0.108, p = 0.253; Table S1 ). Based on these findings, we further 

In this longitudinal cohort study of 28 geographical areas with 6,529 confirmed -14 -found a significant association with sunshine hours. 1 One assumption for these results is elucidated by the relationship between climatic 2 conditions and human activities or behaviors. A study in Tokyo reported that the use of The WHO has adapted a 1-m social distancing policy, based primarily on the (WHO, 2020b). A recent experimental study indicated that ultraviolet light from the 1 sunlight inactivated SARS-CoV-2 (Bianco et al., 2020). However, our results suggest 2 that an increased person-to-person contact due to increased outing activities on warm 3 and/or sunny days will promote the transmission of the virus. In conclusion, the epidemic growth of COVID-19 was not associated with 9 precipitation, wind speed, humidity, NO, NO 2 , Ox, and PM 2.5 . Conversely, it was 10 significantly associated with increase in daily temperature or sunshine hours during the 11 study period. This suggests that an increase in person-to-person contact due to increased 12 outing activities on a warm and/or sunny day might promote the transmission of 13 COVID-19. Our results also suggested that short-term exposure to suspended particles 3 The authors declare that there is no conflict of interest. We would like to express our deepest gratitude to J.A.G JAPAN Corporation for 7 useful data collection on the COVID-19 Japanese cases. We also thank Nanako Noguchi 8 and Mizuki Harada for help with data collection on climate and ambient air pollution. Figure S1 , Figure S2 , Table S1, and Table S2 can be found Méndez-Arriaga F., 2020. The temperature and regional climate effects on 7.8 (6.5-9.6) 11.7 (10.5-13.9) 10.6 (8.2-12.9) 6.9 (6.0-7.9) 9.8 (8. 3-11.2) Values are expressed as median (interquartile rage). Abbreviations: DA, dry air; Ox, photochemical oxidant; NO, nitrogen monoxide; NO 2 , nitrogen dioxide; PM 2.5 , fine particulate matter; SPM, suspended particulate matter.

J o u r n a l P r e -p r o o f J o u r n a l P r e -p r o o f 

COVID-19 transmission with temperature or UV radiation in Chinese cities

Association between short-term exposure to air 7 pollution and COVID-19 infection: Evidence from China