Study population and setting
The method of population-based registration of incidence case series was applied and performed for the whole population of Chiba prefecture, Japan from January to May 2020. The first case with reported COVID-19 symptoms was reported in Chiba prefecture on 20 January and was later confirmed positive on 31 January 2020. Out of 735 cases for the final data analysis, there were 9 foreign patients including 4 from Wuhan city, China; 30 participants from other prefectures; 686 participants were residents of 38 cities/towns of Chiba prefecture, and 10 cases did not consent to publish their residential status. On 7 April, Chiba was one of six prefectures proclaimed to be in a state of emergency from 8 April. Data of the daily reported cases confirmed positive with COVID-19 and the daily exposure to species of air pollution, including ozone pollutants, were created and examined.
Population exposure to ozone estimation
The daily data of ozone pollutants were derived from an online database of 20 weather stations throughout Chiba prefecture [7]. All 20 weather stations located in Chiba prefecture, representing different ecological regions, had information available on the following species of ozone; SO2, NO2, CO, PM10, and PM2.5 from 31 December 2013 to 30 April 2020. We calculated the daily mean ozone concentrations including other species of air pollution from the 20 weather stations for the present study. The first positive patient had reported that the onset of symptoms was on 20 January 2020. The daily estimated mean concentrations of ozone from 20 January 2020 to 30 April 2020 was used to merge with the daily incidence cases of COVID-19 in our study population. The estimated mean concentrations of ozone pollutants during the study time from 20 January 2020 to 30 April 2020 was 36.97 (µg/M3), ranging from 20.80 (µg/M3) to 62.80 (µg/M3).
Covariate variables
Covariate variables included age group, sex, relative humidity, ambient temperature, SO2, NO2, CO, PM2.5, and time of every 7 days of the timeline (To minimize effects of the season changing from winter to spring and early summer) that were included in the analysis model in controlling for confounding factors. For PM2.5, the estimated mean concentrations were 37.14 (µg/M3), ranging from 24.60 (µg/M3) to 61.00 (µg/M3). The estimated means of CO and SO2 were less than 3.00 (µg/M3). For CO, the estimated mean was 2.31 (µg/M3), ranging from 1.50 (µg/M3) to 5.00 (µg/M3). For SO2, the estimated mean was 1.78 (µg/M3), ranging from 1.00 (µg/M3) to 3.80 (µg/M3). For relative humidity and ambient temperature, we derived data from the Air Quality Historical Data Platform from January 2020 to date, on the COVID-19 data page [7]. This database was missing data on relative humidity and ambient temperature from 10 to 24 February 2020. For this particular period, we obtained data on relative humidity and ambient temperature from the Narita Airport weather database[8].
COVID-19 outcomes in Chiba prefecture
The first case with reported COVID-19 symptoms was reported in Chiba prefecture on 20 January 2020. The last patient, during this study time, had reported the onset of symptoms on 30 April 2020. By 3 May 2020, there were 866 positive patients reported in the Chiba prefecture. We excluded 131 participants because 99 asymptomatic patients until the date tested positive with COVID-19, 31 patients had not consented to public personal information, and one patient reported onset symptoms after the date tested positive with COVID-19. The remaining 735 participants were eligible for the final data analysis.
Patient information included ID, residence, age, sex, symptom onset date, and confirmation of testing positive with COVID-19. Data has been updated daily by the Chiba Prefecture Health and Welfare Department Disease Control Section regarding the information on ID, age, and sex, symptom onset and features, local and international travel, and social activities. For example, the first case report was published in a Press Release of “About outbreak of patients associated with new coronavirus (January 31st, 2019)” as follows “Patient aged 20s, female, the residence of Chiba; Symptoms and course: Cough and runny nose on 20 January 2020. On 29 January, she had visited a medical institution due to respiratory symptoms. On 30 January, a Public health center conducted a survey based on the recommendations of a contact survey from the Nara prefecture and collected samples for COVID-19 testing. On 16 January, she had traveled from Tokyo to Osaka by a plane. She participated in a tour for Chinese people as a bus guider from 17 to 22 January. On January 22, she moved from Osaka to Tokyo by bus. She had no overseas travel history within 2 weeks before the onset date (January 20) and wearied a mask from the end of the tour from 22 January 2020” [9].
The daily onset of symptoms data was linked with the daily ozone pollutant data. During the period from 20 January 2020 to 30 April 2020, there were 71 days where patients reported the onset of symptoms. The daily number of cases differed from one to another during these 71 days, Figure 1.
Data handling and analysis
We estimated the number of daily person-years by dividing the number of the population of 6,259,000 by 365 (days). Exposure to ozone and other covariate variables was grouped into three levels. For tertile_1, tertile_2, and tertile_3, the estimated mean (µg/M3) of ozone concentrations was 29.2 (20.8-33.5), 36.7 (34.0-39.3), and 45.3 (39.4-62.8); the rounded number of person-years was 410,805, 408,364, and 393,981; the number of incidence cases positive with COVID-19 was 151, 212, and 371, respectively. The incidence rate ratio and 95% confidence interval (IRR, 95% CI) were estimated by the level of ozone concentrations using unconditional logistic regression analysis, adjusted for the age group (0-9, 10-19, 20-29, 30-39, 40-49, 50-59, 60-69. 7-79, and 80+), sex, relative humidity, ambient temperature, SO2, NO2, CO, PM2.5, and time of every 7 days of the timeline. All p-values were two-sided, and p<5% (alpha value) was considered to indicate statistical significance.
Ethical approval
This study uses public data shared by Chiba prefecture, Japan from January to May 2020. This database does not contain any information used to identify patients including name, personal address, date of birth, etc. Patients are only referred to in numerical order, thus the consent of individuals is not required.