key: cord-1039353-s4ef9qob
authors: Mizumoto, Kenji; Kagaya, Katsushi; Chowell, Gerardo
title: Transmissibility of 2019 Novel Coronavirus: zoonotic vs. human to human transmission, China, 2019-2020
date: 2020-03-20
journal: nan
DOI: 10.1101/2020.03.16.20037036
sha: e0e45b57d8cbbbcfeba3147d6c17ad1373bd14b2
doc_id: 1039353
cord_uid: s4ef9qob

Objectives: The novel coronavirus (2019-nCoV) originating from Wuhan has rapidly spread throughout China. While the origin of the outbreak remains uncertain, accumulating evidence links a wet market in Wuhan for the early spread of 2019-nCoV. Similarly, the influence of the marketplace on the early transmission dynamics is yet to be investigated. Methods: Using the daily series of COVID-19 incidenceincluding contact history with the market, we have conducted quantitative modeling analyses to estimate the reproduction numbers (R) for the market-to-human and human-to-human transmission together with the reporting probability and the early effects of public health interventions. Results: Our mean R estimates for China in 2019-2020 are estimated at 0.37 (95%CrI: 0.02-1.78) for market-to-human transmission, and 3.87 (95%CrI: 3.18-4.78) for human-to-human transmission, respectively. Moreover we estimated that the reporting rate cases stemming from market-to-human transmission was 3-31 fold higher than that for cases stemming from human-to-human transmission, suggesting that contact history with the wet market played a key role in identifying COVID-19 cases. Conclusions: Our findings suggest that the proportions of asymptomatic and subclinical patients constitute a substantial component of the epidemic's magnitude. Findings suggest that the development of rapid diagnostic tests could help bring the epidemic more rapidly under control.

A novel coronavirus (SARS-CoV-2) originating from Wuhan has rapidly spread throughout China, with multiple cases exported to 25 countries around the world. The cumulative number of confirmed cases has reached 75204 including 2006 deaths as of February 19, 2020 [1] . Early mean estimates of the reproduction number have been estimated in the range 1.4-2.5, comparable with estimates for seasonal, 2009 pandemic flu, SARS and MERS [2] [3] [4] [5] . The epidemic continues to spread throughout China at an alarming rate while chains of transmission start to unfold in other countries including Singapore, Korea, and Japan, indicating that the possibility of a pandemic scenario cannot be ruled out.

Evidence suggest that the novel coronavirus likely jumped from a primary reservoir (e.g, horseshoe bats) to an intermediary reservoir, possibly generating an outbreak among wild animals in at least one wet market in Wuhan, China [6] [7] . The virus first infected multiple individuals visiting the Huanan Seafood Wholesale Market, initiating multiple chains of transmission that ensured sustained transmission in the human population [8] . While the detailed origin of the outbreak remains uncertain, significant evidence strongly links the . CC-BY-NC 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 March 20, 2020. . https://doi.org/10.1101/2020.03. 16.20037036 doi: medRxiv preprint Huanan Seafood Wholesale Market in Wuhan for the early spread of the novel coronavirus (COVID-19) among humans [7] .

In this paper we conduct quantitative modeling analyses to quantify the role of the wet marketplace on the early transmission dynamics of the novel coronavirus in China [9] .

For this purpose, we analyzed and modeled data that stratifies the market hazard (market-to-human transmission) and human-to-human transmission.

Daily series of laboratory-confirmed COVID-19 cases were extracted from a recently published study [7] . From December 8, 2019 to January 21, 2020, we analyzed a total of 425 confirmed cases by date of symptoms onset including information on whether the case was linked to the Huanan Seafood Wholesale Market. That is, this unique case series stratifies cases with visiting history to the Huanan Seafood Wholesale Market, which has been purported as source of this large epidemic [9] , and those arising from human-to-human transmission. Because the corresponding incidence curve is subject reporting delays, the last 12 epidemic days were excluded from our analysis. Thus, the study period is set from December 8 th , 2019 to January 9 th , 2020.

. CC-BY-NC 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 March 20, 2020. 

We model two types of infections: a) market hazard (market-to-human transmission, including primary infections arising from zoonotic transmission to some extent) and b) human-to-human transmission chains. A model schematic of the transmission dynamics is provided in Figure 1 . Primary infection (index case) occurs as a result of a spillover event, and primary infections may generate secondary cases. The number of infected individuals stemming from the market-to-human and the human-to-human route are denoted by i m and i h , respectively.

We employed a discrete-time integral equation to capture the daily incidence series with contact history. Let f s denote the probability mass function of the serial interval of length s days, which is given by 

represents the expected number of new cases with onset day t infected through market-to-human transmission, p is the conditional probability of extinction within one generation, given by 1/(R m +1) [10] [11] , where R h and R m represents the average number of secondary infections generated by one single infection generated from the market-to-human or from the human-to-human transmission, respectively.

. CC-BY-NC 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 March 20, 2020. where E(c j ; H t-1 ) denotes the conditional expected incidence, on day t, given the history of observed data from day 0 to day (t-1), denoted by H t-1 . Thus, the number of expected newly observed cases is written as follows:

Subsequently, we also account for the reporting probability that depends on the type of infection. We assume that the number of reported cases by infection type j on day t, h j (t), is the product of type-dependent reporting rate, q j , and the actual number of 

where α 1 and α 2 scale the intensity of public health interventions (where α 1 and α 2 is expected to be smaller than 1) and period 1 and period 2 define the study periods according to timing of the closing of the Huanan Seafood Wholesale Market (December 31 in 2019) and the date on which the novel coronavirus was officially declared as the causative pathogen of the outbreak by China CDC (January 8 in 2020) [8] . After the . CC-BY-NC 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 March 20, 2020. 

We assume the incidence, h j is the result of the Poisson sampling process with the 

where U indicates parameter sets that are estimated from this likelihood.

The serial interval is assumed to follow a gamma distribution with the mean and SD at 7.5 and 3.4 days, respectively, based on ref. [8] . The maximum value of the serial interval was fixed at 16 days as the cumulative probability distribution of the gamma distribution up to 16 days reaches 0.97.

For sensitivity analyses, we examined the effect of varying the mean of the serial interval on R j by varying the mean serial interval from 5.5 to 9.5 days.

We estimated model parameters and made projections using a Monte Carlo

Markov Chain (MCMC) method in a Bayesian framework. Point estimates and the corresponding 95% credibility intervals were drawn from the posterior probability distributions. All statistical analyses were conducted in R version 3.5.2 (R Foundation . CC-BY-NC 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 March 20, 2020. 

The daily series of nCov incidence in China in 2019-2020 are displayed in 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 March 20, 2020. . CC-BY-NC 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 March 20, 2020. [2, 7] . More importantly, our results indicate a substantial contribution of mild and asymptomatic cases to the epidemic, and this could influence estimates of R derived from a single epidemic curve of observed reported cases, which does not fully capture the transmission dynamics [13, 14] .

Moreover, we also identified a remarkable decrease in the reproduction number, and this decline is in line with the timing of the closing of the Huanan Seafood . CC-BY-NC 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 March 20, 2020. Furthermore, reporting probabilities for cases stemming from market-to-human and human-to-human cases were estimated to be low, but we estimated that the reporting rate for market-to-human route was 3-31 fold higher, suggesting that contact history with the Huanan Seafood Wholesale Market played a key role in identifying cases with COVID-19. Indeed, during the initial disease stages, infected individuals with COVID-19 exhibit clinical features that are similar to those of other common respiratory diseases [15] . Moreover, in the context of the ongoing influenza season, the absence of rapid diagnostic testing for this novel disease complicate case identification for this novel coronavirus [16] . Because polymerase chain reaction (PCR) tests, which are time consuming, are not widely available, there is a crucial need to develop rapid . CC-BY-NC 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 March 20, 2020. . https://doi.org/10.1101/2020.03.16.20037036 doi: medRxiv preprint diagnostic tests in order to improve the accuracy and speed of clinical diagnosis and enhance reporting rates.

Our results are not free from limitations. First, the serial interval for COVID-19 assumed in our analyses was extrapolated from a previous study, and had to exclude recent data from the epidemic curve due to reporting delays. In fact, our sensitivity analysis on the serial interval indicate some influence on our R estimates. Further our R estimates could be further improved using additional data collected by field investigations. Second, we cannot rule out the possibility that some cases with contact history with the Huanan Seafood Wholesale Market got infected by contact with humans. However, our low R estimate for market-to-human transmission indicates that such a proportion is quite low. Third, our method is a powerful tool to estimate the underlying cases including asymptomatic and mild symptoms, and our results suggests 

The authors declare no conflicts of interest.

. CC-BY-NC 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. Lower panel (B) correspond to infections through human-to-human transmission, respectively. Epidemic day 1 corresponds to the day that starts at December 8,2020.

. CC-BY-NC 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 March 20, 2020. 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 March 20, 2020. . https://doi.org/10.1101/2020.03.16.20037036 doi: medRxiv preprint infected through human-to-human transmission, respectively. Epidemic day 1 corresponds to the day that starts at December 8, 2020. . CC-BY-NC 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 March 20, 2020. . https://doi.org/10.1101/2020.03.16.20037036 doi: medRxiv preprint

Novel Coronavirus (2019-nCoV) situation reports

Statement on the meeting of the International Health Regulations (2005) Emergency Committee regarding the outbreak of novel coronavirus

Estimates of the reproduction number for seasonal, pandemic, and zoonotic influenza: a systematic review of the literature

Model parameters and outbreak control for SARS

Discovery of a novel coronavirus associated with the recent pneumonia outbreak in humans and its potential bat origin

Mining coronavirus genomes for clues to the outbreak's origins. Science

Early Transmission Dynamics in Wuhan, China, of Novel Coronavirus-Infected Pneumonia

The epidemiological characteristics of an outbreak of 2019 novel coronavirus diseases (COVID-19) in China

Clinical features of patients infected with 2019 novel coronavirus in Wuhan

The transmissibility of noroviruses: Statistical modeling of outbreak events with known route of transmission in Japan

How to interpret the transmissibility of novel influenza A(H7N9): an analysis of initial epidemiological data of human cases from China

Modeling annualized occurrence, frequency, and composition of ingrowth using mixed-effects zero-inflated models and permanent plots in the Acadian Forest Region of

Transmission of 2019-nCoV Infection from an Asymptomatic Contact in Germany

Study claiming new coronavirus can be transmitted by people without symptoms was flawed

SARS-CoV-2 Viral Load in Upper Respiratory Specimens of Infected Patients