key: cord-0849806-9lw1gb3q authors: Pant, Rashmi; Choudhry, Lincoln Priyadarshi; Rajesh, Jammy Guru; Yeldandi, Vijay V title: COVID-19 Epidemic Dynamics and Population Projections from Early Days of Case Reporting in a 40 million population from Southern India date: 2020-04-21 journal: nan DOI: 10.1101/2020.04.17.20070292 sha: 22022574be820f2b5cc4e2d23ac1fdc089aa3057 doc_id: 849806 cord_uid: 9lw1gb3q India reported its first COVID19 case on 30 January 2020. Since then the epidemic has taken different trajectories across different geographical locations in the country. This study explores the population aggregated trajectories of COVID19 susceptible, infected and recovered or dead cases in the south Indian state of Telangana with a population of approximately 40 million. Information on cases reported from March 2 to April 4 was collated from government records. The susceptible-infected-removed (SIR) model for the spread of an infectious disease was used. Transmission parameters were extracted from existing literature that has emerged over past weeks from other regions with similar population densities as Telangana. Optimisation algorithms were used to get basic reproduction rate for different phases of nonpharmaceutical interventions rolled by the government. Peak accumulation is projected towards end of July with 36% of the population being infected by August 2020 if the population lockdown or social distancing mechanism is not continued. The number of deaths assuming no intervention is projected to be 488000 (95% CI: (329400, 646600)). A draconian enforcement of population lockdown combined with hand and face hygiene adherence would reduce the transmission by at least 99.7% whereas partial social distancing and hygiene would reduce it by 51.2%. Transmission parameters reported should be interpreted with caution as they are population aggregated and do not consider unique characteristics of susceptibility among micro-clusters and vulnerable individuals. More data will need to be collected to optimize transmission parameters and evaluate the full complexity, to simulate real world scenarios in the models. similar population densities as Telangana. Optimisation algorithms were used to 23 get basic reproduction rate for different phases of nonpharmaceutical 24 interventions rolled by the government. Peak accumulation is projected towards 25 end of July with 36% of the population being infected by August 2020 if the 26 population lockdown or social distancing mechanism is not continued. The 27 number of deaths assuming no intervention is projected to be 488000 (95% CI: 28 (329400, 646600)). A draconian enforcement of population lockdown combined 29 with hand and face hygiene adherence would reduce the transmission by at least 30 99.7% whereas partial social distancing and hygiene would reduce it by 51.2%. 31 Transmission parameters reported should be interpreted with caution as they are 32 population aggregated and do not consider unique characteristics of 33 susceptibility among micro-clusters and vulnerable individuals. More data will 34 need to be collected to optimize transmission parameters and evaluate the full 35 complexity, to simulate real world scenarios in the models. The announcement of the novel corona Virus (COVID-19 or SARS-CoV-2) as pandemic was 51 made on January 30 th 2020 [1] . The first case of COVID-19 was detected in India on January 52 30 th , 2020. As of 30 th March 2020, more than 1250 cases had been identified in India, with 32 53 deaths and 102 cases have been discharged after treatment [2] . Many key aspects about the 54 disease dynamics are not known. To improve the understanding about the virus many 55 researchers continue to contribute through peer review journals, blogs, reports and social media 56 platforms [3]. One of the key endeavours among these knowledge products is the quest to 57 quantify the burden of disease through the use of mathematical modelling [4,5,6] so that public 58 health systems can prepare for emergency response. 59 India is a geographically, climatically and culturally diverse country with nearly 1.3 billion 60 population [7] . The population density not only differs from urban to rural areas but also from 61 state to state with Delhi having more than 11,000 people per square kilometre, while Arunachal 62 Pradesh has only 17 people per square kilometre. The country has 53 cities which have more 63 than a million population with a minimum density of 400 persons per square kilometre, 64 according to the census of 2011. The country has 137 airports, including 23 international 65 airports handing more than 6 million international and 20 million domestic passengers every 66 month. The above information indicates the diversity of population distribution that can 67 influence the spread of an infectious disease like COVID-19 and the possibility of import via 68 international passenger influx [8] . China or elsewhere. Thus, we assumed that all those who are infected will be "removed" from 92 the pool of susceptible either due to recovery or death. Because of the short nature of the 93 epidemic elsewhere we assumed that the epidemic is not affected by larger population-level 94 vital dynamics i.e. births, migration etc. 95 The SIR model is represented mathematically by a series of differential equations given below. 96 "Janta curfew" advisory [17]) and setting up of quarantine and isolation beds. 140 All rights reserved. No reuse allowed without permission. the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint (which was not peer-reviewed) is . https://doi.org/10.1101/2020.04.17.20070292 doi: medRxiv preprint the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint (which was not peer-reviewed) is The epidemic curve presented in the different scenarios give a range of the burden of the 209 disease and the scenario of optimisation with the state level data gives a very optimistic 210 scenario. Although the range of scenarios with and without interventions gives us a spread of 211 projections, we however feel that the no interventions scenario may be close to the actual 212 scenario till the second week of April as accumulated cases will reflect undiagnosed 213 infections and unreported deaths in the community. This model considers data till ten days 214 after the shutdown announcement by Government of India on 25 March 2020, which is closer 215 to the incubation period of the disease, indicating that the infections detected in last ten days 216 are not influenced by the shutdown. However, there are other measures like a ban on 217 international travel, campaign on handwashing etc. that was ongoing for almost four weeks, 218 thus the influence of the same on the epidemic progression must be factored in. After the 219 peak around 120 days from the first detected case, the epidemic is expected to show a decline 220 in numbers and be on the downward slope of the curve. In a national COVID model [24] , the authors suggest two types of containment strategy i.e. 233 (i) Port of entry and, (ii) mitigationwithin-country connectivity. One of the arguments for 234 the epidemic response was to have a robust screening at ports of entry and contact tracing 235 program. Our preliminary model for Telangana state does not incorporate strategy (i). The 236 capital city of Hyderabad has one international airport with total traffic of more than 20 237 million in the year 2018-19, including 4 million international travellers. The city has good 238 All rights reserved. No reuse allowed without permission. the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint (which was not peer-reviewed) is The outputs of this model show an expected population-level decline in the burden of 265 reported infections/disease over time. The input data is influenced by the series of measures 266 implemented locally by the authorities, thus its influence over the trajectory of the epidemic 267 cannot be overlooked. As policymakers walk the tightrope of initiating public health 268 interventions to contain the COVID19 epidemic, more granular analyses will be needed, 269 especially in a country as socially and geographically diverse as India. 270 The authors used an open-source program (RStudio-version3.6.3) that is widely used and 272 leaving the codes to be accessed by other researchers on Github (https://github.com/). All the 273 data used in the analysis will be available in the supplementary material. 274 The authors declare that they have no conflict of interest to report. 276 This study was funded by SHARE-INDIA internal research funds. 278 All rights reserved. No reuse allowed without permission. the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint (which was not peer-reviewed) is . https://doi.org/10.1101/2020.04.17.20070292 doi: medRxiv preprint The data used in the analyses is available in Supplementary Table S1 . 280 IHR Emergency Committee on Novel Coronavirus (2019-nCoV) 2. Ministry of Health and Family Welfare Early 289 dynamics of transmission and control of COVID-19: a mathematical modelling study Prudent public health intervention strategies to control the coronavirus disease 293 2019 transmission in India: A mathematical model-based approach The World Bank. Population, total -India | Data Impact of international travel and border control measures on the global 301 spread of the novel 2019 coronavirus outbreak A double epidemic model for the SARS propagation A Bayesian estimate 307 of the underreporting rate for COVID-19 based on the experience of the Diamond 308 Princess cruise ship A contribution to the mathematical theory of 311 epidemics Memory B Cell Responses in Recovered Patients with Severe Acute Respiratory 316 Syndrome: A Six-Year Follow-Up Study Phase-adjusted estimation of the number of coronavirus disease 319 2019 cases in Wuhan, China On the Limited Memory Method for Large Scale 324 Optimization No reuse allowed without permission. the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint (which was not peer-reviewed) is Moving Averages World Health Organisation. Coronavirus disease 2019 (COVID-19):Situation Report-330 46 A spatial model of CoVID-19 334 transmission in England and Wales: early spread and peak timing Available Ffrom Estimating the asymptomatic 337 proportion of coronavirus disease 2019 (COVID-19) cases on board the Diamond 338 Princess cruise ship The 2019 novel coronavirus disease (COVID-19) pandemic: A 341 review of the current evidence. The Indian journal of medical research Prudent public health intervention strategies to control the coronavirus disease 344 2019 transmission in India: A mathematical model-based approach Articles 348 Early dynamics of transmission and control of COVID-19: a mathematical modelling 349 study Real estimates of mortality following COVID-19 infection Directorate of Census Operations. ANDHRA PRADESH VILLAGE AND TOWN 360 DIRECTORY DIRECTORATE OF CENSUS OPERATIONS DISTRICT CENSUS 361 HANDBOOK HYDERABAD-Census of India Tuberculosis and Illicit Drug Use: Review and 363 Update Drop of HIV estimate for India to less than 366 half Public health interventions and epidemic 368 intensity during the 1918 influenza pandemic. Proceedings of the National Academy 369 of Sciences Health Threats Associated with Children Lockdown in Spain 371 during COVID-19 Learning in lockdown: School police, race, and the limits of law Economic impacts of Wuhan 2019-nCoV on 376 China and the world No reuse allowed without permission. the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint (which was not peer-reviewed) is