key: cord-1012249-s88c2t7v authors: Kulkarni, H.; Khandait, H. V.; Narlawar, U. W.; Rathod, P. G.; Mamtani, M. title: INDEPENDENT ASSOCIATION OF METEOROLOGICAL CHARACTERISTICS WITH INITIAL SPREAD OF COVID-19 IN INDIA date: 2020-07-26 journal: nan DOI: 10.1101/2020.07.20.20157784 sha: ada2933b68b31a80243315e5c4bb8d6542bef342 doc_id: 1012249 cord_uid: s88c2t7v Whether weather plays a part in the transmissibility of the novel COronaVIrus Disease-19 (COVID-19) is still not established. We tested the hypothesis that meteorological factors (air temperature, relative humidity, air pressure, wind speed and rainfall) are independently associated with transmissibility of COVID-19 quantified using the basic reproduction rate (R0). We used publicly available datasets on daily COVID-19 case counts (total n = 108,308), three-hourly meteorological data and community mobility data over a three-month period. Estimated R0 varied between 1.15-1.28. Mean daily air temperature (inversely) and wind speed (positively) were significantly associated with time dependent R0, but the contribution of countrywide lockdown to variability in R0 was over three times stronger as compared to that of temperature and wind speed combined. Thus, abating temperatures and easing lockdown may concur with increased transmissibility of COVID-19. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.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 July 26, 2020 . . https://doi.org/10.1101 On the other end of the spectrum, a study conducted by Yao et al 12 concluded that there is 56 no association of COVID-19 transmission with temperature or UV radiation in Chinese cities. Data sources: We selected a total of 46 geographical locations across India. For each 69 selected location (either a city, union territory or district), we collected data for a three-month 70 period (March 1, 2020 through May 31, 2020). Following data items were collected for each 71 study location: daily number of confirmed COVID-19 cases, meteorological data, demographic 72 data and overall geographic data. The meteorological data included 3-hourly recordings of 73 temperature, relative humidity, air pressure, wind speed and rainfall. The demographic data 74 included the 2011 census population and the geographical data included area and elevation. The 75 area and population records were combined to estimate the community density. Lastly, temporal 76 data on the lockdown implementation phases and the mobility of the population (estimated 77 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 July 26, 2020 . . https://doi.org/10.1101 5 anonymously from the cellphone use data) was collected to study the potential temporal 78 concurrence with COVID-19 transmission. All data used in this study are publicly available and are completely anonymized. The These indicated percent change from baseline mobility on visits to the following five 88 destinations -retail and recreation, grocery and pharmacy, parks, transit stations and workplaces. Quantification of COVID-19 transmissibility: Using the daily case count data we 90 estimated the basic reproduction rate (R 0 ) in two different ways. First, we estimated the average 91 R 0 over the entire duration of 92 days period of data collection. For this, we used two methods -92 the exponential growth (EG) and the maximum likelihood (ML). Second, we estimated the daily 93 R 0 in a time-dependent (TD) fashion. All estimates of R 0 require a knowledge of serial interval, 94 the time difference between onset of symptoms in an infector and an infectee. We assumed a 95 gamma distributed serial interval with a mean of 3.96 days and a standard deviation of 4.75 days 96 as reported by Du et al. 13 We used the R package R 0 14 to derive all the estimates of R 0 . Finally, we considered the possibility of biased estimates of R 0 owing to the relative lack of testing 98 facilities, especially during the initial period of the epidemic. For this, we used the method of 99 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 July 26, 2020. Representativeness of the study locations: We included 46 locations across India that 120 contained 32 cities, 12 districts and 2 union territories. Figure 1 shows the geographical spread of 121 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 July 26, 2020. . https://doi.org/10. 1101 these locations and the geographic and demographic details for these locations are provided in (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 July 26, 2020. . https://doi.org/10.1101/2020.07.20.20157784 doi: medRxiv preprint 95% confidence intervals (error bars in Figure 2 ) were significantly above unity. These results indicated that over the study period, the average estimates of R 0 were significantly greater than 145 one, confirming the existence of the epidemic; the average R 0 estimates were only moderately 146 above unity; the average R 0 estimates were minimally influenced by potential undertesting; and 147 that the study locations yielded average R 0 estimates consistent with those for the whole country 148 thereby indirectly reaffirming the representativeness of the selected study locations. We also examined the heterogeneity of the average R 0 estimates across the study were: Mumbai 1.16 (95% CI 1.14 -1.18); Delhi 1.25 (95% CI 1.23 -1.28); Chennai 1.20 (95% 157 CI 1.17 -1.23); Ahmedabad 1.10 (95% CI 1.07 -1.13) and Pune 1.22 (95% CI 1.18 -1.26). Temporal changes in R0 estimates: Next, we considered the variability in R 0 estimates 159 over the duration of the study for all locations together. Figure 4 shows that the R 0 estimates 160 were initially high but undulated widely and gradually converged towards the overall estimates 161 shown in Figure 2 with narrow confidence bands later. Thus, the time dependent R 0 estimates 162 showed considerable variation across study time. We examined the association of the time-dependent R 0 estimates with two socio-164 behavioral characteristicsimplementation of a countrywide lockdown and the extent of social 165 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 July 26, 2020. . https://doi.org/10. 1101 distancing as reflected by the cellphone mobility data. When contrasted against the various 166 phases of countrywide lockdown in India (grey shaded regions in Figure 4 ), we found that the 167 median R 0 estimates consistently reduced as lockdown was imposed. Before lockdown began The cellphone-based community mobility data also revealed consistent and interesting Association of time dependent estimates with meteorological data: The time trends 179 for temperature, relative humidity, air pressure, wind speed and rainfall are shown in Figure 180 5A.Over the duration of the study, temperature and wind speed steadily increased; relative 181 humidity and air pressure gradually decreased while rainfall remained steady. As a first step of 182 the association analyses, we estimated the cross-correlation between each meteorological 183 variable and the R 0 estimates. Figure 5B shows the cross-correlograms for lags ranging from -10 184 to 10 days. We found that higher temperature, wind speed and rainfall were correlated inversely 185 while relative humidity and air pressure were correlated positively with time dependent R 0 186 estimates. The best cross-correlation was observed for temperature and humidity on the same day 187 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 July 26, 2020. The results of these analyses are shown in Table 1 . In the full model, we observed that the 198 lockdown phases 3 (only marginally) and 4 and wind speed were the only covariates that were 199 statistically significantly associated with R 0 estimates. In this context, the mobility data (which (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 July 26, 2020. . https://doi.org/10. 1101 estimates for the variables retained in the final model were as follows: temperature: 9.1%, wind 210 speed: 9.9%, lockdown phase 2: 12.2%, lockdown phase 3: 22.5% and lockdown phase 4: 211 27.0%. These results indicate that while the meteorological factors of temperature and wind 212 speed were statistically significant predictors of COVID-19 transmissibility, their contribution to 213 dampening the R 0 estimate was 3-4 times weaker as compared to the countrywide lockdown 214 phases 2-4. Using nationwide data from India over a three-month period, our study made three 217 cardinal observations. First, the average basic reproduction rate (R 0 ) of COVID-19 infection in 218 the period from March 1 through May 31, 2020 ranged from 1.15 to 1.28 even after accounting 219 for the potential undertesting. Second, the COVID-19 transmissibility (quantified using R 0 ) was 220 significantly associated with daily average temperature (inversely), daily average wind speed 221 (positively) and the countrywide intervention of lockdown (inversely). Third, the contribution of 222 lockdown to the variability in time dependent R 0 was three times more than the contribution of 223 temperature and wind speed combined. Together, these results suggest that in India while the 224 meteorological determinants of COVID-19 were independently associated with the 225 transmissibility, their contribution was outweighed by that of the countrywide lockdown. Even though statistically significantly greater than unity, our estimate of R 0 was low. interval with values ranging from as low as 3 days to as high as 9 days. 13, 17-21 We used the serial 230 interval of ~4 days which is on the lower side of the serial interval range and could have partly 231 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 July 26, 2020. . https://doi.org/10.1101/2020.07.20.20157784 doi: medRxiv preprint contributed to the low R 0 observed in this study. Second, the major part of the study period 232 included lockdown and reduced mobility and therefore the R 0 estimate may represent a muted transmissibility. 23, 27 The study duration mark a period of increasing temperature in the Indian 247 peninsula and our results indicate that, in general, high ambient temperatures were associated 248 with lower R 0 estimates such that unit standard deviation increase in air temperature was 249 associated with a 0.08 lower R 0 (Table 1, final model).On the other hand, we observed that a unit 250 standard deviation increase in wind speed was associated with a 0.08 higher R 0 (Table 1, (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 July 26, 2020 . . https://doi.org/10.1101 wind speed is in line with the growing idea that the SARS-CoV-2 virus may be airborne. 33, 34 Of 255 note, incidence of COVID-19 has been shown to be associated with air pollution 1, 35, 36a factor 256 that is significantly influenced by wind speed. 37 Our study cannot directly answer these 257 interesting hypotheses, which should be tested in future studies. Nevertheless, a head-to-head 258 comparison indicated that the lockdown period was associated with three times stronger 259 contribution to the variability in R 0 as compared to that of air temperature and wind speed (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 July 26, 2020. . https://doi.org/10. 1101 Texas, USA and is also the President of the not-for-profit Lata Medical Research Foundation (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 July 26, 2020. (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 July 26, 2020 . . https://doi.org/10.1101 328 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 July 26, 2020 . . https://doi.org/10.1101 Medical journal of the Islamic Republic of Iran 2020; 34: 34. 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 July 26, 2020 . . https://doi.org/10.1101 Chengdu. Infectious diseases of poverty 2020; 9: 87. 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 July 26, 2020. Italy. The Science of the total environment 2020; 740: 140005. 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 July 26, 2020. 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 July 26, 2020. . https://doi.org/10. 1101 Temperature z-score -0.08 -0.13 --0.03 0.005 Wind speed z-score 0.08 0.03 -0.12 0.003 Short-Term Effects of Ambient Ozone Serial Interval of COVID-19 among Publicly Reported 330 Confirmed Cases The R0 package: a toolbox to estimate reproduction 333 numbers for epidemic outbreaks Correcting under-reported COVID-19 case 337 numbers: estimating the true scale of the pandemic. medRxiv : the preprint server for 338 health sciences Data analysis: A model comparison approach Estimation of the net reproductive number of COVID-19 in Iran Air pollution and COVID-19: Is the connect worth its weight? Dynamic effect analysis of meteorological conditions on air pollution: A case 420 study from Beijing 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 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 July 26, 2020 . . https://doi.org/10.1101 (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 July 26, 2020. (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 July 26, 2020. (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 July 26, 2020. (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 July 26, 2020. indicate the time dependent R 0 and 95% confidence intervals, respectively, for each day during 448 the study. These align to the left axis (colored red). The green curve shows the 5-day rolling 449 average z-score for cellphone-based mobility data and aligns to the right axis (colored green). Shaded boxes in the background indicate different phases of the countrywide lockdown in India. 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 July 26, 2020 . . https://doi.org/10.1101 variable is indicated as a number alongside the lag at which it was observed. Rel, relative. 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 July 26, 2020. . https://doi.org/10. 1101