key: cord-337818-mhmc3ts8 authors: Majek, O.; Ngo, O.; Jarkovsky, J.; Komenda, M.; Razova, J.; Dusek, L.; Pavlik, T. title: Modelling the first wave of the COVID-19 epidemic in the Czech Republic and the role of government interventions date: 2020-09-11 journal: nan DOI: 10.1101/2020.09.10.20192070 sha: doc_id: 337818 cord_uid: mhmc3ts8 In the Czech Republic, the first COVID-19 cases were confirmed on 1 March 2020; early population interventions were adopted in the following weeks. A simple epidemiological model was developed to help decision-makers understand the course of the epidemic and perform short-term predictions. In this paper, we present the use of the model and estimated changes in the reproduction number (decrease from > 2.00 to < 1.00 over March and April) following adopted interventions. More than 27 million of COVID-19 cases and over 800 thousand deaths have been reported globally so far. [1] Population interventions including restrictions limiting public gatherings and social contact have proved crucial in the fight against COVID-19. [2] In the Czech Republic, first COVID-19 cases were confirmed on 1 March 2020. A series of early measures was adopted over the following weeks in accordance with the Public Health Protection Act and the Act on the Security of the Czech Republic (Table 1) , leading to rather favourable results after the first wave of epidemic (87 cases and 3 deaths per 100,000 population at the end of May, compared to 269 cases and 32 deaths per 100,000 population in the entire EU/EEA and UK). [3] A simple epidemiological model was developed at the Institute of Health Information and Statistics of the Czech Republic to help decision-makers understand the course of the epidemics including an estimation of the effective reproduction number, [4] and to facilitate short-term predictions. In this paper, is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted September 11, 2020. . https://doi.org/10.1101/2020.09.10.20192070 doi: medRxiv preprint In the Czech Republic, only selected laboratories are allowed to perform testing. Test results are then, with minimum delay, reported to the central Information System of Infectious Diseases (ISID), and subsequently validated by the respective regional public health authority. Therefore, ISID allows us to quickly obtain and analyse key data for evaluation of the course of epidemics and to publicly share the current status. [5] Data on basic epidemiological characteristics (cumulative confirmed cases, active cases, incidence, etc.) are available as open data (on-line at https://onemocneniaktualne.mzcr.cz/api/v2/covid-19). We developed an original epidemiological model, maintaining the simplicity of statistical models while also considering the mechanics of transmission, which allowed us to better understand the course of the epidemics and to produce more realistic predictions. [6] Our model uses classical S(E)IR approach [7, 8] with the following compartments: S (susceptible), I (infected, set of compartments), Rsubcl (subclinical cases) and R (removed, laboratory-confirmed COVID-19 cases; see Figure 1 ). Individuals identified as cases imported from abroad as well as individuals infected within the community (in line with the estimated reproduction number) enter the state I1. They stay in the individual compartments I1 to I7 always for one day. Individuals in states I4 to I7 can infect others, the number of newly infected individuals depends on the reproduction number. To be able to consider the testing effectiveness (i.e. the delay between referring the patient for testing and the availability of test results), the average length of stay in the I8+ compartment was calibrated to the ISID data. For simplicity, it was assumed that the isolation or end of patients' infectiousness always comes on the second day after the onset of symptoms, limiting further patient's infectiousness. In line with the general testing policy applicable in the Czech Republic in the respective time period, testing was assumed only in symptomatic is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted September 11, 2020. . individuals. It was assumed that 10 % of infectious individuals were subclinical and would not be included in the ISID statistics (see Table 2 for the complete set of parameters). were derived using the random search method; details are given in the Supplement S1. 5-10% of the best fitting simulations were utilised for parameter estimation (later model editions used 5%). Of these simulations, subset of simulations predicting the highest numbers of cases (see Table 3 for details on the size of these subsets) were used in the early predictions as a precautionary approach. Parameter values were estimated as means of parameter values from those accepted simulations; the reproduction number was also estimated with standard deviations to allow estimation of an indicative 95% confidence interval. The 95% confidence interval bands were subsequently applied as reproduction number values on the recent (retrospective) and on the near future (prospective) period for estimating of a 'sensitivity interval', i.e., the predicted interval of the potential numbers of cases at individual target dates (Table 3 ). Calculations were performed in Microsoft Excel. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted September 11, 2020. . 1 precautionary approach: high-risk decile of best-fitting simulations, confidence interval bands used for the sensitivity interval 2 precautionary approach: high-risk quartile of best-fitting simulations, confidence interval bands used for the sensitivity interval The results of the first model edition were published on 24 March 2020 and predicted a sharp increase till the end of March (more than 3 thousand of cases, compared to 1,047 cases observed on 21 March, Table 3 ). The real (observed) number of cases indeed exceeded 3 thousand on 30 March. The basic reproduction number for Czech population was estimated to be 2.64, with a partial decline since 12 March (1.84) , reflecting the introduction of the state of emergency and the first nationwide restrictive measures (Table 1 ). The reproduction number following the broader restrictive measures since 16 March was assumed to be 1.2. In reality, the effect of the restrictive measures was even more dramatic, which lead to a substantial overestimation of the number of cases predicted for the end of April. Further estimates of the reproduction number lead to downward corrections (1.00 since 16 March, following the restriction of free movement, Table 1 The basic reproduction number of COVID-19 was previously estimated to be 2. we, therefore, needed to assume their values ( Table 2 ) and apply them in the model structure. The values we assumed were consistent with previously published estimates of the serial interval of 4-5 days (e.g. [10] ). Besides, substantial uncertainty exists around the proportion of subclinical cases. Nevertheless, the use of an alternative proportion of subclinical cases (i.e., 30% as suggested by Nishihura [17] instead of 10% used originally in our model) did not lead to a substantial change of the is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted September 11, 2020. . https://doi.org/10.1101/2020.09.10.20192070 doi: medRxiv preprint conditions. Indeed, the model has shown a satisfactory predictive validity; however, great uncertainty is associated with the future values of the reproduction number, which is to a great degree affected by adopted policies and the compliance of the target population. The described model allowed us to analyse the course of the epidemic, including the estimation of the basic reproduction number, and to perform useful short-term predictions, which facilitated the estimation of the necessary readiness of the healthcare system in the days and weeks after the prediction. The comparison of the predicted and observed numbers of cases is incorporated in the early warning system, which is currently used by policy-makers both on the national and regional levels. The Czech data on COVID-19 epidemic have also demonstrated the potential of early implementation of government measures in slowing the spread of the COVID-19 epidemic. . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted September 11, 2020. . https://doi.org/10.1101/2020.09.10.20192070 doi: medRxiv preprint The effect of travel restrictions on the spread of the 2019 novel coronavirus (COVID-19) outbreak European Centre for Disease Prevention and Control Situation dashboard -COVID-19 cases in Europe and worldwide 2020 Unraveling r 0: Considerations for public health applications Complex Reporting of the COVID-19 Epidemic in the Czech Republic: Use of an Interactive Web-Based App in Practice Wrong but useful-what covid-19 epidemiologic models can and cannot tell us Nowcasting and forecasting the potential domestic and international spread of the 2019-nCoV outbreak originating in Wuhan, China: a modelling study Early dynamics of transmission and control of COVID-19: a mathematical modelling study. The lancet infectious diseases Incubation period of 2019 novel coronavirus (2019-nCoV) infections among travellers from Wuhan, China Serial interval of novel coronavirus (COVID-19) infections. International journal of infectious diseases Feasibility of controlling COVID-19 outbreaks by isolation of cases and contacts. The Lancet Global Health Modelling the COVID-19 epidemic and implementation of population-wide interventions in Italy Preliminary estimation of the basic reproduction number of novel coronavirus (2019-nCoV) in China, from 2019 to 2020: A datadriven analysis in the early phase of the outbreak Estimation of the reproductive number of novel coronavirus (COVID-19) and the probable outbreak size on the Diamond Princess cruise ship: A data-driven analysis Clinical characteristics of coronavirus disease 2019 in China Transmission interval estimates suggest pre-symptomatic spread of COVID-19 Estimation of the asymptomatic ratio of novel coronavirus infections (COVID-19) We would like to acknowledge all the employees of the Public Health Offices, Institute of Health Information and Statistics, National Public Health Institute and Ministry of Health of the Czech Republic involved in the COVID-19 information support for their great work, which was also vital for preparation of this article. The authors would also like to thank Dr Jaroslav Janošek for his insightful comments.