key: cord-0931871-yuiivakf
authors: Wang, Qingan; Zhao, Yu; Zhang, Yajuan; Qiu, Jiangwei; Li, Juan; Yan, Ni; Li, Nan; Zhang, Jiaxing; Tian, Di; Sha, Xiaolan; Jing, Jinyun; Yang, Chan; Wang, Kairong; Xu, Rongbin; Zhang, Yuhong; Yang, Huifang; Zhao, Shi; Zhao, Yi
title: Could the ambient higher temperature decrease the transmissibility of COVID-19 in China?
date: 2020-12-03
journal: Environ Res
DOI: 10.1016/j.envres.2020.110576
sha: 17bf1a78c74ae1cc14948c1274b4739500d6cfcb
doc_id: 931871
cord_uid: yuiivakf

BACKGROUND: Existing literatures demonstrated that meteorological factors could be of importance in affecting the spread patterns of the respiratory infectious diseases. However, how ambient temperature may influence the transmissibility of COVID-19 remains unclear. OBJECTIVES: We explore the association between ambient temperature and transmissibility of COVID-19 in different regions across China. METHODS: The surveillance data on COVID-19 and meteorological factors were collected from 28 provincial level regions in China, and estimated the instantaneous reproductive number (R(t)). The generalized additive model was used to assess the relationship between mean temperature and R(t). RESULTS: There were 12745 COVID-19 cases collected in the study areas. We report the effect of temperature on R(t) is likely to be negative but not of statistical significance, which holds for most of included regions except for those in North China. CONCLUSIONS: We found little statistical evidence for that the higher temperature may reduce the transmissibility of COVID-19. Since intensive control measures against the COVID-19 epidemics were implemented in China, we acknowledge this may impact the underlying effect size estimation, and thus cautiousness should be taken when interpreting our findings.

of temperature on R t is likely to be negative but not of statistical significance, which holds for 26 most of included regions except for those in North China. 27 Conclusions: We found little statistical evidence for that the higher temperature may reduce 28 researchers found there was an inverse relationship between temperature and SARS (Cai et al., 54 2007). it is tempting to assume that such association could apply to COVID-19 as well, which 55 may provide the region-specific prevention measures. 56 Earlier research on this issue in China including over 400 cities (Wang et al., 2020) 57 found there was a non-linear relationship between temperature and cumulative number of 58 cases, they found that as the temperature rise, the transmissibility rise first and then fall. 59 Another study also found a non-linear relationship between temperature and daily cases of 60 COVID-19, but they suggested that the infected cases would continue to increase in spite of 61 rising temperature, which implies that the rising temperature may only reduce the growth rate 62 of disease (Xie and Zhu, 2020). In addition, Yao et al. investigated the association between 63 the basic reproduction number and the weather conditions, however, they did not find any 64 evident result between temperature and transmissibility of COVID-19 (Yao et al., 2020) . Ran 65 et al. (Ran et al., 2020) explored the association between temperature and COVID-19 66 transmissibility based on an ecological study, they found that the overall nonlinear association 67 between basic reproductive number and temperature was of statistical significance. Thus, how 68 ambient temperature may influence the transmissibility of COVID-19 remains unclear. 69 Studies in China on this issue mostly considered the city of Wuhan (or Hubei province), 70 this may cause some bias in analysis since there are numerous confirmed cases in Wuhan (or 71 Hubei) compared to other regions in China, such difference of confirmed case scale between 72

Hubei province and other regions in China may misestimated the effect of temperature on 73 COVID-19 morbidity. In addition, the medical resources and control interventions between 74

Hubei and the other regions in China exist different, which may also affect the analysis result 75 of association between air temperature and COVID-19 morbidity. 76 In this study, we aim to explore the association of temperatures with risks of 77 J o u r n a l P r e -p r o o f demographic and geographic characteristics, China was divided into seven regions according 79 to Chinese Geographical Division. The generalized additive model was used to assess the 80 relationship between mean temperature and R t . Furthermore, we perform the sensitive 81 analysis to test the robustness of the results. We further discussion the public health relevance 82 of our estimates. 83 Guizhou, Sichuan, Tibet and Yunnan). 92 We selected a total of 28 Chinese provincial regions as study place in this work. Since 93 Wuhan (in Hubei province, China) was the epicenter of COVID-19 in China, where relatively 94 intensive control measures were implemented, we avoid including COVID-19 data in Hubei 95 province. Moreover, there were enormously larger number of cases in Wuhan than other cities, 96 so it may be inappropriate for comparison among these regions. Besides, we excluded these 97 cities from analysis, and they include Hong Kong, Hainan, and Taiwan for unavailable 98 meteorological data, and Tibet for lack of COVID-19 case. 99

The surveillance data of COVID-19 number of cases were collected from the reports 101 released on the official websites of the Health Commissions. We collected case data from 102 January 20 to February 29, 2020, for two reasons: of the SI distribution (in days). The estimation of R t is conducted with a Poisson-distributed 120 likelihood profile for the number of incidences. To set up the model, the distribution of SI 121 used to estimate R t was approximated by a Gamma distribution with mean 5.5 days and 122 standard deviation (SD) 3.3 days . 123

Statistical analyses mainly consisting of the following two steps: (i) as (Zhao, 2020) 125 pointed out that there may exist flawed analytical procedures by directly using the case 126 number as response to modeling the temperature-morbidity relationship for infectious 127 diseases, thus, we estimated instantaneous reproductive number (R t ) as the proxy of the 128 as a function of mean temperature (linear) and other meteorological variables (natural cubic 130 spline) at city-specific level, then combining the city-level results to the region-level with 131 meta-analysis, we calculated the relative risk (RR) of each region with different lag days. The 132 model is described as follows: 133 log = + × Temp + Humidity , df = 3 + Pressure , df = 3 + 134

where is the interception, Temp is the mean temperature on the t-th day. The 136

Humidity , Pressure and Wind are relative humidity, air pressure and wind speed on the 137 t-th day, respectively. The function denotes the natural cubic spline function with degrees 138 of freedom (df) fixed at 3. The DOW is the day of week, a categorical variable with 139 coefficient γ. Considering the autocorrelation of time series data, we added the term 140 $ × log %& to adjust for the autocorrelation. 141

The relative risk (RR) and its 95% confidence interval (95% CI) were employed to 142 measure the association between temperature and the COVID-19 transmissibility (R t ) by 143 using the generalized additive model. We respectively calculated the effects of the current day, 144 lag for 1 day, lag for 2 days, lag for 3 days and their moving averages by using generalized 145 additive model. 146

To test the robustness of the results, we performed the sensitivity analyses as follows: (i) 148 We used daily maximum and minimum temperatures as the interested indicators to explore 149 the temperature-disease relationships, respectively. (ii) The model is also applied for the 150 whole studied areas by using mean, maximum and minimum temperature, respectively. (iii) 151 We extended the endpoint of study period from February 29, 2020 to April 30, 2020. For other regions, the temperatures from high to low followed by East (8.5), Central (7.8), 164

North (-0.2) and Northwest (-1.1). Details of statistics of confirmed cases of COVID-19 and 165 other meteorological variables shows in Table 1 . 166 As table 2 shows, the estimated reproductive number ranges from 1.1 to 1. Figure 2 shows the temperature moving average effect. Similar as in Table 2 , the 182 significant result only found in North China. However, the correlation of temperature and 183 COVID-19 becomes insignificant when we excluded Inner Mongolia (The corresponding  184 values of RR and their 95% CI, see Table S5 ). 185 For sensitivity analysis, the results were similar in term of the effect size of statistical 186 significance, when we replaced mean temperature with maximum and minimum temperature, 187 respectively (see Table S2, Table S3 , Figure S1, FigureS2) . When all regions were analyzed as 188 a whole, the results also suggested that the association between temperature and COVID-19 is 189 insignificant (Table S4) . Hence, we consider our estimates with higher confidence, and 190 unlikely to be altered by different meteorological indicators or varying the study period. That is, the number of incidences for infectious diseases is determined by transmissibility 213 (Wallinga and Lipsitch, 2007) . unlike non-communicable diseases (NCD), using case number 214 as response modeling the relationship between infectious disease and environmental factors 215 may produce non-causal associations . In China, quarantine measures are the 216 most stringent in the world, as well as earliest, the whole city of Wuhan was lockdown on 217 January 23 (Tian et al., 2020) , followed by Hubei and the other provinces. During policy 218 implementation period, most people stay at home during the outbreak and maintain social 219 distance outside, which could effectively reduce the spread of COVID-19 . 220 Thus, from January to February, the rising temperature in China may lead to a non-causal 221 association to the morbidity related to COVID-19. Besides, a counterexample that raises 222 suspicion is that there are some countries more warmer and the prevention measures more 223 looser than China during the same period (Qasim Bukhari and Jameel, 2020), such as India 224 and Brazil (Kumar, 2020; Prata et al., 2020) , the epidemic situation is not more optimistic 225 than China. Third, we should point out that our analysis found insignificant association 226 between the ambient temperature and the transmission of COVID-19, but the RR estimation 227 results (see Table 3 ) suggested that there likely exists a negative correlation between 228 temperature and the transmission of COVID-19. By using global data, Chen et al. (Chen et al., 229 2020b) lately examined the relationship between temperature and the spread of 230 and found that there is a robust and significant negative association between COVID-19 231 J o u r n a l P r e -p r o o f transmissibility and ambient temperature at the country level. Ran et al. (Ran et al., 2020) 232 observed that the ambient temperature was found to have a nonlinear negative association 233 with COVID-19 transmissibility based on the discontinuity regression method. Inconsistent 234 results may be due to the different model frameworks, thus, more multi-data evidences might 235 be needed to further test the association between ambient temperature and COVID-19 236 transmissibility. 237

According to the study by , the instantaneous reproductive number R t could 238 be more representative for the transmissibility of COVID-19. In our study, we used R t in place 239 of daily confirmed cases, and did not find the significant association between temperature and the COVID-19 outbreak rate, and high solar radiation was a protective factor, but the 246 temperature is unlikely. 247 A special case is that in North China, we found slightly significant negative association 248 between temperature and R t . However, when we re-estimated the relationship without Inner 249 Mongolia, the weak correlation was disappeared, which indicate this association appears not 250 sufficiently stable. The reason why Inner Mongolia was an exception in our analysis is not 251 clear as the weather-diseases relationship is more complicated and easily affected by the other 252 non-meteorological factors, thus, this phenomenon may need further research. We still prefer 253 to believe that the effect of temperature on COVID-19 is weak in that the exception of one 254 city may not be convinced. 255 Our study has some advantages. First, the association between ambient temperature and 256 transmissibility of COVID-19 in different regions across China was studied. A total of 28 257 J o u r n a l P r e -p r o o f reliable conclusion. Second, for aggregated data, the basic reproductive number was used to 259 avoid the noncausal association between environmental factor and COVID-19. This result 260 was different with that obtained by using the infected cases data. Third, compared to the 261 SARS, the association between temperature and COVID-19 is insignificant across China, 262 which is of important reference value for specific public health strategy. 263

Our study also has some limitations. due to lack of information, our estimation of R t neglected the difference between local and 272 imported cases. We remark the analytical framework adopted in this study can be extended to 273 address this limitation with cases' import or local status available. These issues should be 274 addressed in future. 275 

Investigation of effective climatology parameters on COVID-19 outbreak in 301

How will country-based mitigation measures influence the course of the 303 COVID-19 epidemic? The Lancet

A spatio-temporal analysis for exploring the effect of 305 temperature on COVID-19 early evolution in Spain

Influence of meteorological factors and air pollution on the outbreak of severe 308 acute respiratory syndrome

Epidemiological and clinical characteristics of 99 cases of 2019 novel 310 coronavirus pneumonia in Wuhan, China: a descriptive study

Revisiting the association between temperature and COVID-19 transmissibility 312 across 117 countries

A new framework and software to estimate time-varying reproduction numbers 314 during epidemics

Quantifying SARS-CoV-2 transmission suggests epidemic control with digital 318 contact tracing

The interval between successive cases of an infectious disease

Temperature, humidity, and wind speed are associated with lower Covid-19

Effect of meteorological parameters on spread of COVID-19 in India and air quality 324 during lockdown

Influenza virus transmission is dependent on relative humidity and 326 temperature

Outbreak of pneumonia of unknown etiology in Wuhan, China: The mystery and 328 the miracle

Effects of temperature variation and humidity on the death of COVID-19 in Wuhan

Temperature, humidity, and latitude analysis to predict potential 332 spread and seasonality for COVID19

Quarantine alone or in combination with other public health 334 measures to control COVID-19: a rapid review

Role of temperature and humidity in the modulation of the doubling time of 337 COVID-19 cases

Temperature significantly changes COVID-19 transmission in (sub)tropical 339 cities of Brazil

Will coronavirus pandemic diminish by summer? SSRN

A re-analysis in exploring the association between temperature and COVID-19 342 transmissibility: an ecological study with 154 Chinese cities

Staying at home mobility effects of COVID-19. Covid Economics: Vetted and 344 Real-Time Papers

Coronavirus and Migration: Analysis of Human Mobility and the 346 Spread of Covid-19. Migration Letters

Improved inference of time-varying reproduction numbers during 348 infectious disease outbreaks

Early evaluation of transmission control measures in response to the 2019 novel 350 coronavirus outbreak in China

Correlation of ambient temperature and COVID-19 incidence in Canada. Science of 352 the Total Environment

How generation intervals shape the relationship between growth rates 354 and reproductive numbers

Temperature significantly change COVID-19 transmission in 429 cities 356 medRxiv

The 2019 report of The Lancet Countdown on health and climate change: 358 ensuring that the health of a child born today is not defined by a changing climate

No association of COVID-19 transmission with temperature or UV radiation in 363 Chinese cities

To avoid the noncausal association between environmental factor and COVID-19 when 365 using aggregated data: Simulation-based counterexamples for demonstration. Science of the 366 Total Environment

Analysis of the transmissibility change of 2019-novel coronavirus pneumonia 368 and its potential factors in China from 2019 to 2020

Competing interests 298 The authors declared no conflicts of interest. 299