key: cord-294690-fpjhkb4g
authors: Sharifi, Hamid; Jahani, Yunes; Mirzazadeh, Ali; Ahmadi Gohari, Milad; Nakhaeizadeh, Mehran; Shokoohi, Mostafa; Eybpoosh, Sana; Tohidinik, Hamid Reza; Mostafavi, Ehsan; Khalili, Davood; Hashemi Nazari, Seyed Saeed; Karamouzian, Mohammad; Haghdoost, Ali Akbar
title: Estimating the number of COVID-19-related infections, deaths and hospitalizations in Iran under different physical distancing and isolation scenarios: A compartmental mathematical modeling
date: 2020-04-25
journal: nan
DOI: 10.1101/2020.04.22.20075440
sha: 
doc_id: 294690
cord_uid: fpjhkb4g

Background: Iran is one of the countries that has been overwhelmed with COVID-19. We aimed to estimate the total number of COVID-19 related infections, deaths, and hospitalizations in Iran under different physical distancing and isolation scenarios. Methods: We developed a Susceptible-Exposed-Infected-Removed (SEIR) model, parameterized to the COVID-19 pandemic in Iran. We used the model to quantify the magnitude of the outbreak in Iran and assess the effectiveness of isolation and physical distancing under five different scenarios (A: 0% isolation, through E: 40% isolation of all infected cases). We used Monte-Carlo simulation to calculate the 95% uncertainty intervals (UI). Findings: Under scenario A, we estimated 5,196,000 (UI 1,753,000 - 10,220,000) infections to happen till mid-June with 966,000 (UI 467,800 - 1,702,000) hospitalizations and 111,000 (UI 53,400 - 200,000) deaths. Successful implantation of scenario E would reduce the number of infections by 90% (i.e. 550,000) and change the epidemic peak from 66,000 on June 9th to 9,400 on March 1st. Scenario E also reduces the hospitalizations by 92% (i.e. 74,500), and deaths by 93% (i.e. 7,800). Interpretation: With no approved vaccination or therapy, we found physical distancing and isolation that includes public awareness and case-finding/isolation of 40% of infected people can reduce the burden of COVID-19 in Iran by 90% by mid-June.

Evidence before this study:

Iran has been heavily impacted by the COVID-19 outbreak, and the virus has now spread to all of its provinces. Iran has been implementing different levels of partial physical distancing and isolation policies in the past few months. We searched PubMed and preprint archives for articles published up to April 15, 2020 that included information about control measures against COVID-19 in Iran using the following terms: ("coronavirus" OR "2019-nCoV" OR "COVID-19") AND "Iran" AND ("intervention" OR "prevention" OR "physical distancing" OR "social distancing").

We found no studies that had quantified the impact of policies in Iran.

Given the scarcity of evidence on the magnitude of the outbreak and the burden of COVID-19 in Iran, we used multiple sources of data to estimate the number of COVID-19 infections, hospitalizations, and deaths under different physical distancing and isolation scenarios until mid-June. We showed that implementing no control measures could lead to over five million infections in Iran; ~19% of whom would be hospitalized, and ~2% would die. However, under our most optimistic scenario, these estimates could be reduced by ~90%.

With no effective vaccination or treatment, advocating and enforcing physical distancing and isolation along with public education on prevention measures could significantly reduce the burden of COVID-19 in Iran. Nonetheless, even under the most optimistic scenario, the burden of COVID-19 would be substantial and well beyond the current capacity of the healthcare system in Iran.

The COVID-19 was declared a pandemic on March 11, 2020 and the disease is now expected to spread worldwide. The risk is relatively low for the general population, although people aged 65 years and over, those with suppressed immune systems, and people with underlying medical conditions (e.g., cardiovascular or respiratory diseases) are at increased risk of adverse outcomes.

The case fatality rate of the infection is estimated to be around 2% (95% CI: 2-3%) and as of April 15, 2020, a total number of 2,049,888 confirmed cases, 510,486 recovered cases, and 133,572 death have been reported worldwide. 1 Iran is one of the hardest hit countries by COVID-19 and has been struggling with controlling the disease for over two months. The first confirmed cases of COVID-19-related deaths were reported on January 21 in the city of Qom; 200 Km away from Iran's capital city of Tehran. As of April 15, 2020, a total number of 76,389 confirmed cases, 49,933 recovered cases, and 4,777 death have been reported and COVID-19 has spread to all of its provinces; 2 figures that are highest among the Eastern Mediterranean Region countries. 3

The Susceptible-Exposed-Infected-Removed (SEIR) model provides a mathematical framework to explain the spread of infectious diseases and has previously been used for estimating the epidemiological parameters of several infectious diseases such as measles, Ebola, and influenza. [4] [5] [6] SEIR could also help evaluate the impact of implementing various interventions (e.g., isolation and physical distancing policies) aimed at controlling the growth of the pandemic and flattening the epidemic curve. Physical distancing (also called social distancing) control measures are policies that aim to minimize close contacts within communities and include individual-level (e.g., quarantine, isolation) and community-level (closure of educational and recreational settings, nonessential businesses, and cancellation of public/mass/crowded gatherings) approaches. 7, 8 In Iran, the physical distancing and isolation interventions were scaled up in late February and early March by nationwide closure of schools/universities, cancellation of sports events and Friday/congregational prayers as well as the closure of all non-essential services, tourism sites, and shopping malls (Figure 1 ). Iran also closed its holy shrines in Mashhad and Qom in mid-March. Moreover, while there were no mandatory shelter-in-place or lockdown orders, people were encouraged to stay at home. People were also asked to avoid travelling during the New year Holidays (i.e., Nowruz) from March 19 to 26; however, no restrictions for domestic (via flight, train or bus) or international travels (i.e. no border closure) were imposed. 9 Despite implementing various physical distancing control measures, our understanding of their impact on the magnitude of COVID-19-related new infections, hospitalization, and death remains limited. In this study, we aim to provide an estimate of these epidemiological parameters and approximate the peak date of the epidemic in Iran under different physical distancing and isolation scenarios. These estimates are of particular importance for COVID-19-related health policy, planning, and financing purposes in Iran.

We formed a compartmental model to estimate the total number of COVID-19 patients, hospitalizations and deaths in Iran and its capital city of Tehran ( Figure 2 ). We used an extended Susceptible-Exposed-Infected/Infectious-Recovered/Removed (SEIR) model that divides the target populations (i.e., Iran and Tehran as the capital) into different compartments. The conceptual framework of the COVID-19 transmission model is presented in Appendix AA. In brief, we considered the following comportments: a) susceptible, referring to the total number of individuals . 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)

The copyright holder for this preprint this version posted April 25, 2020. . https://doi.org/10.1101/2020.04.22.20075440 doi: medRxiv preprint (i.e., hosts) who have been susceptible to COVID-19. We assumed the entire population as susceptible in our model; b) exposed, referring to individuals who are exposed to COVID-19 while they are asymptomatic and not yet infectious; c) infected, referring to infected people who demonstrate clinical symptoms after their incubation period and have the potential to transmit the disease to other susceptible individuals; and d) recovered/removed, depending on the severity of the disease. We assumed that the infected people will i) be recovered and immune from re-infection and therefore, no longer transmit the infection, or ii) have mild to moderate clinical symptoms while they follow home-isolation guidelines without requiring hospitalization, iii) have severe clinical symptoms and require hospitalization. These individuals would either be recovered (and then discharged) or fail to respond to treatment and pass away (removed) from the model. Monte

Carlo method was used to build the 95% uncertainty intervals (UI) around the point estimates of the total expected numbers. To do this, we used the statistical distribution of a set of parameters obtained from both the existing evidence and expert opinion (listed in Table 1 Appendix BB). Data were analyzed using Vensim DSS 6.4E software.

Based on the country's official reports and available epidemiological data, January 21, 2020 was considered as the initial day of the COVID-19 outbreak in Iran. We used several parameters as inputs for the model and obtained their values from a comprehensive literature review and published articles in relation to COVID-19, as well as some corresponding parameters and values considered for the similar epidemics, in particular, H1N1 influenza. 10 We first shared the initial values of the parameters with Iran's national and scientific committees and experts, and then made the necessary adjustments. We compared the revised values of these parameters with the literature as well as the pattern of the epidemic in Iran. We then made the final revisions for the values of . 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)

The copyright holder for this preprint this version posted April 25, 2020. . https://doi.org/10.1101/2020.04.22.20075440 doi: medRxiv preprint the parameters used as input for the predicted number of COVID-19 infected cases and its associated deaths using the extended SEIR model. Input parameters and their brief descriptions are presented in Tables 1 and 2 in Appendix BB.

The impact of seasonality (a sinuses' function) was considered in calculating the transmission probability (beta coefficient) of the disease, indicating the potential for some level of change in transmissibility of the virus from one season to another.(10) Therefore, we assumed that COVID-19 might behave the same as influenza such that the transmission of the virus may tend to reduce by approaching to warm seasons (i.e., spring and summer). We then considered the end of December in winter with the most transmission probability and the end of June in summer with the least transmission probability. The minimum and maximum values of the seasonal changes were considered to be 0.02 and 0.045, respectively. A time-varying state was considered for the effective contact rate (C parameter). We first incorporated value of 14 in the Tehran model and 13 in the national model in the early weeks of the epidemic. [11] [12] [13] [14] [15] After the announcement of the epidemic by the officials, multiple public health measures implemented as a response to the epidemic to reduce contact rates and then transmission rates in public. Approaching the assumed end of the epidemic, we considered value of 5 for this parameter, with some fluctuations due to Nowruz holidays within this period ( Table 2 in Appendix BB). Five possible scenarios were considered for isolation of the infected cases (Table 1) .

. 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.

The copyright holder for this preprint this version posted April 25, 2020. . 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 April 25, 2020. . 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.

The copyright holder for this preprint this version posted April 25, 2020. 

Our modeling exercise showed that with no intervention (i.e., scenario A), more than five million COVID-19 infections would occur in Iran till mid-June, of whom 18.9% would be hospitalized and 2.1% would die. However, under the best-case scenario (i.e., scenario E), the number of infected cases could be reduced by 90%, hospitalizations by 92%, and deaths by 93%. Our projection in scenario C, which is a middle ground scenario, appeared to be aligned with the current statistics from the national reports. Even under scenario C, the burden of the epidemic in Iran will be large, last for several months and might surpass the current capacity of the healthcare system. 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 April 25, 2020. . https://doi.org/10.1101/2020.04.22.20075440 doi: medRxiv preprint approaches towards lifting the physical distancing restrictions is extremely difficult and varies across countries with different economic and healthcare infrastructures, it is critical to follow an evidence-informed approach to avoid the second and further waves of COVID-19 epidemics in Iran. A modeling study from China for example, 15 showed that a stepwise (25% of the workforce working in weeks 1 to 2, 50% of the workforce working in weeks 3 to 4, and 100% of the workforce working and school resuming from week 5 forward) return to work or school at the beginning of April (about five months after the first case reported from China), is much more effective than the beginning of March. This study estimated that just a one-month delay in the stepwise lifting of the physical distancing measures would reduce the magnitude (92% by mid-2020, 24% by end-2020) the epidemic and delay its peak by two months and therefore avoid overwhelming the healthcare systems. 15 In Iran, a recent executive order from the government lifted the restrictions on nationwide business shutdown and allowed most people to return to work only 2.5 months after the identification of the first COVID-19 case in Iran. 9, 20 The government has also planned to reopen schools in lowrisk cities and non-essential low or medium-risk jobs (e.g., all production units in industrial, business, technical service, and distributional sections), as well as removing the shelter-in-place order and resuming domestic and international travels. These decisions are mainly derived from the Iranian government's economic challenges that have been elevated by the comprehensive sanctions imposed by the USA. 21,22 While these concerns are understandable and longer shutdown of an already overstretched economy is a tough decision and would be very taxing on the government and the public, our findings as well as lessons learned from China, 15,23,24 suggest that this approach is not justified by evidence and would most likely risk overwhelming the healthcare systems with soon-to-come second and further waves of COVID-19 epidemic in Iran; costs that . 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.

The copyright holder for this preprint this version posted April 25, 2020. . https://doi.org/10.1101/2020.04.22.20075440 doi: medRxiv preprint might surpass the marginal economic benefits of opening businesses and schools a few months sooner. Nonetheless, it is fortunate that the Iranian CDC is now planning to shift from physical distancing to targeted case-finding, intensify contact tracing and careful isolation of identified cases as well as self-quarantine of symptomatic people.

Our study had three major limitations. First, some of the key parameters (e.g. hospitalization rate, incubation period, transmission probability) that were used in the model were from other countries or expert opinion, but not from empirical data from Iran. To address this limitation, we reported a range of uncertainty intervals. Second, the uncertainty intervals are fairly wide for most of our With no approved vaccination, prophylaxis or therapy, we found physical distancing and isolation that includes public awareness and case-finding/isolation of 40% of the infected cases could reduce the burden of COVID-19 in Iran by 90% by mid-June.

Except for the senior author, Dr. Ali Akbar Haghdoost, who is the Deputy Minister of Health and the Head of National COVID-19 Committee, the rest of the authors declare no conflict of interest, real or perceived.

The authors received no funding for this work.

. 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.

The copyright holder for this preprint this version posted April 25, 2020. . 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.

The copyright holder for this preprint this version posted April 25, 2020. Iranian press review: Iranians ignore coronavirus travel warning ahead of Nowruz holiday. Available from: https://www.middleeasteye.net/news/iran-press-review-coronavirus-nowruztravel-warning-ignore 19

Ghaffarzadegan N, Rahmandad H. Simulation-based Estimation of the Spread of COVID-19 in Iran. medRxiv. 2020. 20

Iran reports its first 2 cases of the new coronavirus. Available from: https://www.timesofisrael.com/iran-reports-its-first-2-cases-of-the-new-coronavirus/ 21

Healthcare . 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.

The copyright holder for this preprint this version posted April 25, 2020. .

Isolation for the entire period of the epidemic was considered to be 0% (the worst scenario)

Policy interventions were at the level of encouragement and the overall mean of isolation for the entire period of the epidemic was considered to be 10%

Isolation was considered to be 10% from January 21 to February 19, 15% after the initiation of the epidemic from February 20 to March 10, and finally 20% from March 11 to June 19, 2020, which are the results of the minimal possible interventions of the health system, behavior change of the public, and containment strategies

Isolation was considered to be 10% from January 21 to February 19, 15% after the initiation of the epidemic from February 20 to March 10, and finally 30% from March 11 to June 19, 2020, which are the results of the moderate possible interventions of the enhanced health system, social and behavioral change of the public (e.g., social distancing, hand washing), and containment strategies (e.g., closing schools and universities)

Isolation was considered to be 10% from January 21 to February 19, 15% after the initiation of the epidemic from February 20 to March 10, and finally 40% from March 11 to June 19, 2020, which are the results of the maximum possible interventions of the health system, behavior change of the public, and containment strategies. . 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.

The copyright holder for this preprint this version posted April 25, 2020. .

where β(t) indicates transmissibility of the virus, and C(t) indicates the contact rates, and II(t) refers to the total number of infected people who transmit the infection, calculated as Infect (t)+(0.1×Temporary Isolation Units)+(0.02×Hospital) (explained below).

b) Individuals who are exposed (E(t)): refers to individuals who are exposed to the infection, but they are not yet infectious. These individuals are asymptomatic in this period.

The differential equation of this compartment is shown in Equation 2:

Eq. . 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.

The copyright holder for this preprint this version posted April 25, 2020. . https://doi.org/10.1101/2020.04.22.20075440 doi: medRxiv preprint ii) Infected individuals who will have mild to moderate clinical symptoms, but they will be home-isolated without requiring hospitalization (IS(t)), and they will be recovered (isolated box in Fig 1) . Equation iii) Infected individuals who will have severe clinical symptoms requiring hospitalization (hospitalized box in Fig 1) . These individuals will have two possible outcomes: i) some hospitalized cases will be recovered and then discharged (T box in Fig 1) , or ii) some will not respond to the medical care and die (death box in Fig 1) . Equations 6 and 7 show the differential equation of hospital box (H(t)) and T box (T(t)) respectively:

Eq. 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)

The copyright holder for this preprint this version posted April 25, 2020. . The values of these parameters are reported in Appendix BB Table  2 . 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)

The copyright holder for this preprint this version posted April 25, 2020. . https://doi.org/10.1101/2020.04.22.20075440 doi: medRxiv preprint

Selected Findings of the 2016 National Population and Housing Census

Risk estimation and prediction by modeling the transmission of the novel coronavirus (COVID-19) in mainland China excluding Hubei province. medRxiv

Active monitoring of persons exposed to patients with confirmed COVID-19-United States

Modelling of H1N1 flu in Iran

Estimation of the time-varying reproduction number of COVID-19 outbreak in China

Social contacts and mixing patterns relevant to the spread of infectious diseases

Social contacts, vaccination decisions and influenza in Japan

An updated estimation of the risk of transmission of the novel coronavirus (2019-nCov)

Projecting social contact matrices to different demographic structures

A systematic

Values References an infected individual contact with an uninfected individual ) Seasonality distribution: (((Sin (2 × 3.14 × (Time + 110) / 365)) + 1) × ((0.045 -0.02) / 2)) + 0.02Infected individuals with the potential to infect uninfected peopleThe average number of people from susceptible individuals are added to exposed individuals (per day)The average number of people from exposed individuals are added to infected individuals (per day)The average duration taken for an exposed individual becomes an infected individual [7] [8] [9] [10] [11] From Jan 21, 2020, to Jan 30, 2020 14 13 From Jan 31, 2002, to Feb 9, 2020 13 12 From Feb 10, 2020, to Feb 19, 2020 12 11 From Feb 20, 2020, to Feb 29, 2020 10 9 From Mar 1, 2020, to Mar 20, 2020 5 5 From Mar 21, 2020, to Mar 31, 2020 6 * 6 * From Apr 1, 2020, to June 19, 2020 5 5 * Contact rates were assumed to increase due to the Nowruz holidays within these periods