key: cord-0747896-slqc8nle authors: Ansumali, S.; Kumar, A.; Agrawal, S.; Shashank, H. J.; Prakash, M. K. title: A steady trickle-down from metro districts and improving epidemic-parameters characterize the increasing COVID-19 cases in India date: 2020-09-29 journal: nan DOI: 10.1101/2020.09.28.20202978 sha: ca93d69a78f3be3378de05f25ae0b1513f29bb86 doc_id: 747896 cord_uid: slqc8nle Background. By mid-September of 2020, the number of daily new infections in India have crossed 95,000. To facilitate an intuition for the spatio-temporal development of the pandemic and to help resource deployment planning, we analyze and describe how the disease burden almost-predictably shifted from large metropolitan districts to sub-urban districts. Methods. We gathered the publicly available granular data from 186 different districts (equivalent of counties) on their COVID-19 infections and deaths during the 15 April to 31 August 2020 period. These districts presented an active case burden of 559,566 and a cumulative 2,715,656 infections as of August 31. The epidemiological data of these districts was fit to a susceptible-asymptomatic-infected-recovered-dead (SAIRD) model and the underlying epidemic parameters for each of these districts during the course of 4 months was estimated. We validated these parameters against known epidemiological characteristic distributions and analyzed them to understand their changes in space-time during the pandemic. Findings. The center of the burden of the current-active infections which on May 15 was in the large metro districts with most international access shifted continuously and smoothly shifted towards districts which could be accessed by domestic airports and by trains. A linear trend-analysis showed a continuous improvement in most epidemic parameters consistently across the districts with four categories of accessibility from an international travel perspective - large metro, metro, urban and sub-urban districts. The reproduction numbers improved from 1.77{+/-}0.58 on May 15 to 1.07{+/-} 0.13 on August 31 in large metro districts (p-Value of trend 0.0001053); and from 1.58{+/-}0.39 on May 15 to 0.94{+/-}0.11 on August 31 in sub-urban districts (p-Value of trend 0.0067). The recovery rate per infected person per day improved from 0.0581{+/-}0.009 on May 15 to 0.091{+/-}0.010 on August 31 in large metro districts (p-Value of trend 0.26.10^{-12}); and from 0.059{+/-}0.011 on May 15 to 0.100{+/-}0.010 on August 31 in sub-urban districts (p-Value of trend 0.12.10{-16}). The death rate of symptomatic individuals which includes the case-fatality-rate as well as the time from symptoms to death, consistently decreased from 0.0025{+/-}0.0014 on May 15 to 0.0013{+/-}0.0003 on August 31 in large metro districts (p-Value of trend 0.0010); and from 0.0018{+/-}0.0008 on May 15 to 0.0014{+/-}0.0003 on August 31 in sub-urban districts (p-Value of trend 0.2789. Interpretation. As the daily infections continue to rise at a national level, it is important to notice a `local-flattening' in larger metro districts, and a shift of the pandemic-burden towards smaller sized districts in a clear hierarchical fashion of accessibility from an international travel perspective. The pandemic burden shifting towards remotely accessible regions, with possibly lesser health care facilities, is a call for attention to the re-organization of resources. and analyzed them to understand their changes in space-time during the pandemic. Findings The center of the burden of the current-active infections which on May 15 was in the large metro districts with most international access shifted continuously and smoothly shifted towards districts which could be accessed by domestic airports and by trains. A linear trendanalysis showed a continuous improvement in most epidemic parameters consistently across the districts with four categories of accessibility from an international travel perspective -large metro, metro, urban and suburban districts. The reproduction numbers improved from 1.77 ± 0.58 on May 15 to 1.07 ± 0.13 on August 31 in large metro districts (p-Value of trend 0.0001053); and from 1.58 ± 0.39 on May 15 to 0.94 ± 0.11 on August 31 in sub-urban districts (p-Value of trend 0.0067). The recovery rate per infected person per day improved from 0.0581 ± 0.009 on May 15 to 0.091 ± 0.010 on August 31 in large metro districts (p-Value of trend 0.26 × 10 −12 ); and from 0.059 ± 0.011 on May 15 to 0.100 ± 0.010 on August 31 in sub-urban districts (p-Value of trend 0.12 × 10 −16 ). The death rate of symptomatic individuals which includes the case-fatality-rate as well as the time from symptoms to death, consistently decreased from 0.0025 ± 0.0014 on May 15 to 0.0013 ± 0.0003 on August 31 in large metro districts (p-Value of trend 0.0010); and from 0.0018 ± 0.0008 on May 15 to 0.0014 ± 0.0003 on August 31 in sub-urban districts (p-Value of trend 0.2789). Interpretation As the daily infections continue to rise at a national level, it is important to notice a 'local-flattening' in larger metro districts, and a shift of the pandemic-burden towards smaller sized districts in a clear hierarchical fashion of accessibility from an international travel perspective. The pandemic burden shifting towards remotely accessible regions, with possibly lesser health care facilities, is a call for attention to the re-organization of resources. In the COVID-19 pandemic that began in December 2019, each month 2 witnessed the critical rise of infections in completely different countries. The 3 initial declaration of national emergency which was a "one-size fits all" was 4 quickly irrelevant in many countries as each state or county perceived the 5 need for stringency differently. However, now the focus is it identify the next 6 hotspot [1, 2] . The differences arose mainly because even within the first wave phases of the pandemic, rising, stable or declining, perceived the threat dif-10 ferently from others. As the rise of infections in different geographical regions 11 is asynchronous, the critical care burden shifts dynamically with each region 12 attempting to achieve its 'local-flattening' of the peak at a different time. Epidemiological models have focused on making predictions of the rise of 14 infections, and a peak of critical care burden on a country or a state basis. The data shows predictable shifts to remote regions. The infections spread-32 ing to newer and under-catered locations presents both an opportunity and 33 a challenge, an asynchronous peak of critical requirements which can be ad- CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted September 29, 2020. . https://doi.org/10.1101/2020.09.28.20202978 doi: medRxiv preprint Evidence before this study We searched pubmed and medRxiv up to 20 September 2020 with the keywords "COVID-19", "time variation or longitudinal", "cantons or states", "hotspots", "diffusion" "active-cases". The published research thus far focused mainly on reporting either the time variation of reproduction number at the state-level, or the predictions of rise of infections at the canton level. However, from these constantly updating predictions with new hotspots appearing, there is no continuity of knowledge and possibility to obtain a comprehensive picture of the development of the pandemic. A study correlating the number of infections in different regions in China to the traffic-flow out of Wuhan, and another describing the development of hotspots along the highways in Brazil describe some aspects of this geographical spread. Firstly, a similar analysis for with granular Indian data has been missing. Further, the questions of how the center of the active-case burden diffused and the spatio-temporal trends in epidemiological parameters underlying the large number of infections have not been the focus of these earlier studies. Our study begins with the understanding that the different geographic regions such as districts (or counties) may be in different phases of the pandemic, and mixing of population within them happens more likely than from outside. Considering this geographical heterogeneity, and the ease of access to these different districts, we develop a summary how the pandemic evolved in the regions of different accessibility. Predictable trends in the case-burden diffusion are identified. Authorities should understand the shift in the dynamics of the critical-care requirements, not by geographically contiguous regions, but by ease of travel or by common economic interests which guided such ease of travel. Strategies should be driven by knowing where the pandemic burden is likely to move, before it happens. The trend of the case-burden predictably shifting towards suburban districts can be useful in re-organizing the available resources as per the needs. best describes the observations. The (pre-or) asymptomatic to symptomatic 80 conversion rate (δ) was assumed to be a constant for each district. The estimates of δ for the 186 districts were 0.34 ± 0.02, which corresponds 93 6 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted September 29, 2020. . Infections continue to rise, but with lower Reproduction numbers. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. 12 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Scientists are racing to model the next moves of a coronavirus 259 that's still hard to predict Digital 262 technologies in the public-health response to covid-19 Estimating 266 the effects of non-pharmaceutical interventions on covid-19 in europe The effect of travel 270 restrictions on the spread of the 2019 novel coronavirus (covid-19) out-271 break Population 273 flow drives spatio-temporal distribution of covid-19 in china Sub-276 stantial undocumented infection facilitates the rapid dissemination of 277 novel coronavirus (sars-cov2) Estimating hidden asymptomatics, herd 280 immunity threshold and lockdown effects using a covid-19 specific model Minimal and adaptive numerical strategy for critical re-284 source planning in a pandemic The incubation period of coro-287 navirus disease 2019 (covid-19) from publicly reported confirmed cases: 288 estimation and application Es-292 timates of the severity of coronavirus disease 2019: a model-based anal-293 ysis The hidden geometry of complex, network-295 driven contagion phenomena Geographic and temporal development of plagues