key: cord-0732417-1p941spn authors: Qi, Hongchao; Xiao, Shuang; Shi, Runye; P. Ward, Michael; Chen, Yue; Tu, Wei; Su, Qing; Wang, Wenge; Wang, Xinyi; Zhang, Zhijie title: COVID-19 transmission in Mainland China is associated with temperature and humidity: a time-series analysis date: 2020-03-30 journal: nan DOI: 10.1101/2020.03.30.20044099 sha: 9f7d6699b9e91fe5b81a3d5076708045a4a82de8 doc_id: 732417 cord_uid: 1p941spn COVID-19 has become a pandemic. The influence of meteorological factors on the transmission and spread of COVID-19 if of interest. This study sought to examine the associations of daily average temperature (AT) and relative humidity (ARH) with the daily count of COVID-19 cases in 30 Chinese provinces (in Hubei from December 1, 2019 to February 11, 2020 and in other provinces from January 20, 2020 to Februarys 11, 2020). A Generalized Additive Model (GAM) was fitted to quantify the province-specific associations between meteorological variables and the daily cases of COVID-19 during the study periods. In the model, the 14-day exponential moving averages (EMAs) of AT and ARH, and their interaction were included with time trend and health-seeking behavior adjusted. Their spatial distributions were visualized. AT and ARH showed significantly negative associations with COVID-19 with a significant interaction between them (0.04, 95% confidence interval: 0.004-0.07) in Hubei. Every 1°C increase in the AT led to a decrease in the daily confirmed cases by 36% to 57% when ARH was in the range from 67% to 85.5%. Every 1% increase in ARH led to a decrease in the daily confirmed cases by 11% to 22% when AT was in the range from 5.04°C to 8.2°C. However, these associations were not consistent throughout Mainland China. Starting in December 2019, the severe acute respiratory coronavirus 2 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted March 30, 2020. Daily counts of laboratory-confirmed cases in all provinces in China were collected from the official reports of the National Health Commission of People's Republic of China from December 1, 2019 to February 11, 2020 for Hubei province and from January 20, 2020 to February 11, 2020 for other provinces. Case definitions for suspected cases and laboratory-confirmed cases and the description of the surveillance system have been published online and All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The meteorological data, including daily average temperature (AT) and daily average relative humidity (ARH) of each provincial capital, were retrieved from Weather Underground (https://www.wunderground.com/). The following keywords were applied in the Baidu index (http://index.baidu.com/), the largest search engine in China, using the keywords "Wuhan pneumonia" OR "novel coronavirus" OR "coronavirus disease 2019" OR "coronavirus disease-19" OR "2019 novel coronavirus" OR "2019-nCoV" OR "SARS-CoV-2" OR "COVID-19" OR "SARS-CoV-2" OR "severe acute respiratory coronavirus 2". The province-based Baidu index with massive Internet behavior data recorded, including concerns for COVID-19, was used as a measure of health-seeking behavior that may affect the transmission of coronavirus [13] . A generalized additive model (GAM) was applied to quantify the province-specific associations between meteorological variables and the daily cases of COVID-19 during the study periods, accounting for short-term temporal trend and health-seeking behavior through Baidu index. The distribution of COVID-19 cases was assumed to be a negative binomial given that the variances of the daily counts were larger than their means. Considering the incubation period of COVID-19, the effects of AT, ARH, and covariates were modeled with a 14-day exponential moving average (EMA) to account of their potential lag effects and the interaction between AT and ARH was tested and included in the model if significant. The short-term temporal trend was indicated using natural splines of time with two degrees of freedom. The model is given by: All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. is the daily counts of COVID-19 at day t, Considering the effect of the interaction between AT and ARH, effect plots were shown with the change in log-transformed daily counts over AT (or ARH) given the ARH (or AT) of the 25 th , 50 th , and 75 th percentiles. Additionally, the effects of meteorological factors were parameterized with the incidence rate ratio (IRR) assuming the number of persons at risk stayed stable during the study period. IRRs of AT (or ARH) given the ARH (or AT) of the 25 th , 50 th , and 75 th percentiles in models were estimated and depicted using forest plots and geographic maps. To verify model results, a sensitivity analysis was performed in which the time-series in Wuhan and in Hubei province were restricted to the period the same as that in other provinces − i.e. January 20 to February 11 − with the same GAM fit to the data (the results of sensitivity analysis in Figs. A8-A10). The R software (version 3.5.3, http://cran.r-project.org; R Foundation for Statistical Computing, Vienna, Austria) was used to perform statistical analyses. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted March 30, 2020. . https://doi.org/10.1101/2020.03.30.20044099 doi: medRxiv preprint QGIS Desktop (version 3.4.14, https://www.qgis.org/; Open source geospatial foundation project) from the open-source geospatial foundation project was used to plot the geographic patterns. The daily counts of confirmed cases, AT, and ARH for 31 provinces were summarized in Table A1 . The cumulative confirmed cases varied from one The estimates of regression coefficients of the GAM for Hubei province are listed in Table 1 . There was a significant interaction between AT and ARH (0.04, 95% confidence interval (CI): 0.004-0.07), and the effect of the Baidu index was also statistically significant. Significant interactions between AT and ARH were also found in Zhejiang, Shandong, Hebei, Jilin, and Gansu (Table A2) . (Table1 in here) All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted March 30, 2020. 0.75-0.93), and 0.89 (95% CI: 0.8-1). Every 1°C increase in AT led to a decrease in the daily confirmed cases by 36% to 57% when ARH was in 67% to 85.5% and every 1% increase in ARH led to a decrease in the daily confirmed cases by 11% to 22% when AT was 5.04°C to 8.2°C. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. relationship between ARH and COVID-19 at the different levels of AT. Our study suggests that both daily temperature and relative humidity influenced the occurrence of COVID-19 in Hubei province and in some other provinces. However, the association between COVID-19 and AT and ARH across the provinces was not consistent. We found spatial heterogeneity of COVID-19 incidence, as well as its relationship with daily AT and ARH, among provinces in Mainland China. The study period − and hence time series length − was longer for Hubei province than other provinces. The longer the study period, the more stable the model results are expected to be. Considering the incubation period of COVID-19, we used the 14-day EMA of daily AT and ARH to investigate the effects of ARH and AT on COVID-19. Most of the confirmed cases (and therefore meteorological data) were derived from the period before travel restrictions were enacted for Wuhan city (January 23, 2020); therefore, the (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. A novel finding of this study is the significant interaction between ARH and AT, and COVID-19 transmission. In Hubei province, the interaction between ARH and AT was found to have a positive effect on transmission (0.04, 95% CI: 0.004-0.07). Increased AT (ARH) led to a decreased effect of ARH (AT) on the incidence of COVID-19 in Hubei province. The exact mechanism of the interaction is unclear. One probable reason might be that a combination of low AT and humidity make the nasal mucosa prone to small ruptures, creating opportunities for virus invasion [21] . A previous study observed an interaction between minimum temperature and humidity on influenza [22, 23] There are some limitations in our study. First, some potential risk factors that could impact the incidence of COVID-19 − for example, province-specific social-economic status − were not included in the model. The study period is short, so that it can be assumed that many covariates did not vary substantially during such a short time period. Second, the incidence of COVID-19 in provinces other than Hubei was more likely to be influenced by interventions after the outbreak in Hubei province. In addition, the meteorological data were collected from the Weather Underground for the capital city for each province only, and data accuracy and representativeness can be improved in future studies. Provinces (except for Hubei) had a short study period and included 1 1 many imported cases. In conclusion, meteorological factors influence COVID-19 transmission and spread, potentially with an interactive effect between daily temperature and relative humidity on COVID-19 incidence. There were spatial and temporal heterogeneities in COVID-19 occurrence, which could be attributed to meteorological factors as well as interventions measures across provinces. The reasons for the inconsistency in the impact of meteorological factors on COVID-19 among provinces needs further study. We declare that we have no conflicts of interest. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted March 30, 2020. . https://doi.org/10.1101/2020.03.30.20044099 doi: medRxiv preprint MERS-CoV) under different environmental conditions An experimental study of the survival of turkey coronavirus at room temperature and +4 degrees C Transmission of influenza virus in temperate zones is predominantly by aerosol, in the tropics by contact: a hypothesis Effect of environment and occupational hygiene factors of hospital infection on SARS outbreak Effects and interaction of meteorological factors on influenza: Based on the surveillance data in Shaoyang The Influence of Meteorology on the Spread of Influenza: Survival Analysis of an Equine Influenza (A/H3N8) Outbreak The role of absolute humidity on transmission rates of the COVID-19 outbreak Temperature significant change COVID-19 Transmission in 429 cities The Novel Coronavirus Pneumonia Emergency Response Epidemiology T: The Epidemiological Characteristics of an Outbreak of 2019 Novel Coronavirus Diseases (COVID-19) -China