key: cord-0819482-b5287qd4
authors: Mizumoto, Kenji; Kagaya, Katsushi; Chowell, Gerardo
title: Effect of the Wet Market on the coronavirus disease (COVID-19) transmission dynamics in China, 2019-2020
date: 2020-06-02
journal: Int J Infect Dis
DOI: 10.1016/j.ijid.2020.05.091
sha: 8f7c974231e2b7798d42604622e646724e73fc24
doc_id: 819482
cord_uid: b5287qd4

Abstract Objectives The novel coronavirus (SARS-CoV-2) originating from Wuhan rapidly spread throughout China. While its origin remains uncertain, accumulating evidence links a wet market for the early SARS-CoV-2 spread in Wuhan. Similarly, the influence of the marketplace on the early transmission dynamics is yet to be investigated. Methods Using the daily series of COVID-19 incidence stratified according to 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, the reporting probability and the early effects of public health interventions. Results We estimated R at 0.24 (95%CrI: 0.01-1.38) for market-to-human transmission, and 2.37 (95%CrI: 2.08-2.71) for human-to-human transmission during the early spread in China (2019-2020). Moreover, we estimated that the reporting rate for cases stemming from market-to-human transmission was 2-34 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 R estimate tied to market-to-human transmission has substantial uncertainty, but it was significantly lower compared to the reproduction number driving human-to-human transmission. Our results also suggest that asymptomatic and subclinical infections constitute a substantial component of the COVID-19 morbidity burden.

A novel coronavirus (SARS-CoV-2) originating from Wuhan rapidly spread around the world to give rise to the most important pandemic event in recent history. The cumulative number of confirmed cases has reached 3.5 million cases including 250000 deaths as of May 17, 2020 [1] . Early mean estimates of the reproduction number based on the epidemic's growth rate have been estimated in the range 1.4-3.5, comparable with estimates for seasonal, 2009 pandemic flu, SARS and MERS [2] [3] [4] [5] [6] [7] . Considering the explosive spread and significant severity associated with the novel coronavirus, the WHO declared this public health emergency a pandemic on March 11, 2020 , and the world is coming together to fight the spread of the coronavirus.

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 [8] [9] . The virus first infected multiple individuals working or visiting the Huanan Seafood Wholesale Market at an early stage, initiating multiple chains of transmission that ensured sustained transmission in the human population [10] . While the detailed origin of the outbreak remains uncertain, significant evidence strongly links the Huanan Seafood Wholesale J o u r n a l P r e -p r o o f Market in Wuhan for the early spread of the novel coronavirus (COVID-19) among humans [9] .

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 [11] .

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

Daily series of laboratory-confirmed COVID-19 cases were extracted from a recently published study [9] . 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 at an early stage [9] , and those arising from human-to-human transmission. Because the corresponding incidence curve is subject to reporting delays, the last 12 epidemic days J o u r n a l P r e -p r o o f were excluded from our analysis. Thus, the study period is set from December 8 th , 2019

to January 9 th , 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-tohuman route are denoted by im and ih, respectively.

We employed a discrete-time integral equation to capture the daily incidence series with contact history. Let fs denote the probability mass function of the serial interval of length s days, which is given by , where E[cm(t)] represents the expected number of new cases with onset day t infected through market-to-human transmission, p is the conditional probability of extinction J o u r n a l P r e -p r o o f within one generation, given by 1/(Rm+1) [12] [13] , where Rh and Rm 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.

To account for the type of infection dependent probability of occurrence, θ j [14] , we assume that the number of observed cases for infection type j on day t, hj(t), occurred according to a Bernoulli sampling process, with the expected values E(cj;Ht-1), where E(cj; Ht-1) denotes the conditional expected incidence, on day t, given the history of observed data from day 0 to day (t-1), denoted by Ht-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 

where α1 and α2 scale the intensity of public health interventions (where α1 and α2 is expected to be smaller than 1) and period1 and period2 define the study periods The number of expected newly observed cases should be updated as

We assume the incidence, hj is the result of the Poisson sampling process with the expected value of E[hj]. The likelihood function for the time series of observed cases of primary infection through the market-to-human transmission and secondary infection through human-to-human transmission that we employ to estimate the reproduction number and other relevant parameters is given by:

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

The serial interval is assumed to follow a lognormal distribution with the mean and SD at 4.7 and 2.9 days, respectively, based on ref. [15] . The maximum value of the serial interval was fixed at 16 days as the cumulative probability distribution of the lognormal distribution up to 16 days reaches 0.99.

For sensitivity analyses, we examined the effect of varying the mean of the serial interval on Rj by varying the mean serial interval from 2.7 to 6.7 days. 

The [16, 17] .

Our method is a powerful tool to estimate the underlying cases including asymptomatic and mild symptoms, and our results suggests that those proportions constitute a large fraction of the epidemic's magnitude. In fact, the result of careful screening, detailed examination and follow-up studies to confirm that the people did not turn symptomatic at a later stage of a total of 565 Japanese returnees evacuated from Wuhan city by government-chartered planes during January 29-31, 2020, revealing a ratio of 4 asymptomatic to 9 symptomatic cases, and the estimated asymptomatic proportion was reported to be 30.8% (95% confidence interval: 7.7-53.8) [18] [19] . For the COVID-19 outbreak that unfolded aboard the Diamond Princess cruise ship, the proportion of truly asymptomatic infections was estimated at 17.9% (95%CrI: 15.5-20.25) among a largely senior population [20] . In addition, one study reported that the majority of seasonal coronavirus infections are asymptomatic by most symptom definitions and only 4% of individuals experiencing a seasonal coronavirus infection episode sought medical care for their symptoms [21] . Taken together, this evidence further underscores the need to account for a significant fraction of asymptomatic cases. 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 2-34 fold higher, suggesting that contact history with the Huanan Seafood Wholesale Market played a key role in identifying cases with COVID-19, in the context of the influenza season when many influenza patients with non-specific symptoms, fever and respiratory symptoms, visit medical facilities, which further complicates an accurate diagnosis of COVID-19 based on nonspecific symptoms [22] . Indeed, during the initial disease stages, infected individuals with COVID-19 exhibit clinical features that are similar to those of other common respiratory diseases [23] .

Our results are not free from limitations. We cannot rule out the possibility that some cases with contact history with the Huanan Seafood Wholesale Market got infected by relatively lower compared to human-to-human transmission. The power of our approach lies in the ability to infer epidemiological parameters with quantified uncertainty from limited data collected by surveillance systems.

The authors declare no conflicts of interest.

No ethical approval is requested. routes are classified into market-to-human transmission and human-to-human transmission. Infected through market-to-human route and human-to-human route is indicated by im and ih, respectively. Reported cases of primary case infected through market-to-human route and of secondary or later cases though human-to-human route is indicated by cm and ch, respectively. A virus with substantial human-to-human transmission potential could yield subsequent transmissions caused by human-to-human transmission.

J o u r n a l P r e -p r o o f 

Novel Coronavirus (2019-nCoV) situation reports

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

statement-on-the-meeting-of-the-international-healthregulations-(2005)-emergency-committee-regarding-the-outbreak-of-novelcoronavirus

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

Early dynamics of transmission and control of COVID-19: a mathematical modelling study

The reproductive number of COVID-19 is higher compared to SARS coronavirus

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

Incubation Period and Other Epidemiological Characteristics of 2019 Novel Coronavirus Infections with Right Truncation: A Statistical Analysis of Publicly Available Case Data

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

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

Estimation of the Asymptomatic Ratio of Novel Coronavirus Infections (COVID-19)

Estimating the asymptomatic proportion of coronavirus disease 2019 (COVID-19) cases on board the Diamond

Direct Measurement of Rates of Asymptomatic Infection and Clinical CareSeeking for Seasonal Coronavirus. medRxiv 2020.01.30

Chinese National Influenza Center

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

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.