key: cord-0741214-eflwztji authors: He, Xi; Lau, Eric HY; Wu, Peng; Deng, Xilong; Wang, Jian; Hao, Xinxin; Lau, Yiu Chung; Wong, Jessica Y; Guan, Yujuan; Tan, Xinghua; Mo, Xiaoneng; Chen, Yanqing; Liao, Baolin; Chen, Weilie; Hu, Fengyu; Zhang, Qing; Zhong, Mingqiu; Wu, Yanrong; Zhao, Lingzhai; Zhang, Fuchun; Cowling, Benjamin J; Li, Fang; Leung, Gabriel M title: Temporal dynamics in viral shedding and transmissibility of COVID-19 date: 2020-03-18 journal: nan DOI: 10.1101/2020.03.15.20036707 sha: 4c6d07b2e6870aedf4a320b5edfc7b497a9da311 doc_id: 741214 cord_uid: eflwztji We report temporal patterns of viral shedding in 94 laboratory-confirmed COVID-19 patients and modelled COVID-19 infectiousness profile from a separate sample of 77 infector-infectee transmission pairs. We observed the highest viral load in throat swabs at the time of symptom onset, and inferred that infectiousness peaked on or before symptom onset. We estimated that 44% of transmission could occur before first symptoms of the index. Disease control measures should be adjusted to account for probable substantial pre-symptomatic transmission. variation can be summarized by the incubation period distribution and the serial interval distribution respectively. If the observed mean serial interval is shorter than the observed mean incubation period, this indicates that a significant portion of transmission may have occurred before infected persons have developed symptoms. A large proportion of presymptomatic transmission would likely reduce the effectiveness of control or preventive measures that are directly or indirectly initiated by symptom onset, such as isolation, contact tracing for quarantine, and enhanced hygiene or use of face masks for symptomatic persons. Failing to consider pre-symptomatic transmission would likely overestimate the effect of such measures when devising control strategies. As a comparison, the Severe Acute Respiratory Syndrome (SARS) coronavirus was notable because infectiousness increased around 7-10 days after illness onset. 5, 6 Once containment measures were effectively implemented, including isolation of cases and quarantine of their close contacts, it was possible to reduce substantially onwards transmission of infection ( Figure 1a) . 7, 8 In contrast, seasonal influenza is characterized by increase in infectiousness shortly around or even before illness onset. 9 For SARS-CoV-2, there remains very limited information to date on the dynamics of contagiousness over time. In this study, we compared clinical data on virus shedding with separate epidemiologic data on incubation periods and serial intervals between cases in transmission chains, to draw inferences on infectiousness profiles. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint Among 94 laboratory-confirmed COVID-19 patients admitted to Guangzhou Eighth People's Hospital, 47/94 (50%) were male and the median age was 47 years. 61/93 (66%) of the patients were moderately ill (with fever and/or respiratory symptoms and radiographic evidence of pneumonia) but none were classified as "severe" or "critical" on hospital admission, although 20/94 (21%) of the patients deteriorated to a severe or critical condition during hospitalization (Supplementary Table 1 ). A total of 414 throat swabs were collected from these 94 patients (median = 4 swabs per patient), from the day of illness onset up to 32 days after onset. We detected high viral loads soon after illness onset, which gradually decreased towards the detection limit at about 21 days after onset ( Figure 2 ). There was no obvious difference in viral loads across sex, age groups and disease severity (Supplementary Figure 1 we inferred that infectiousness started from 2.5 days before symptom onset and reached its peak at 0.6 days before symptom onset (Figure 1b) . The proportion of transmission before symptom onset (area under the curve) was 44%. Infectiousness was estimated to decline relatively quickly within 7 days of illness onset. Viral load data was not used in the estimation but showed similar monotonic decreasing pattern after symptom onset. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint On sensitivity analysis, instead of estimating the start of infectiousness from the model, we fixed the start of infectiousness at 4 and 7 days before symptom onset respectively. We estimated that peak infectiousness was reached at about 0-2 days before symptom onset, and the proportion of transmission occurring before illness onset was 52% and 46% respectively (Supplementary Figure 2) . Finally, simulation showed that when infectiousness started at the same time as symptom onset, there would be a small proportion of short serial intervals (e.g. < 2 days) (Supplementary Figure 3) . The proportion would become larger if infectiousness were to start before symptom onset, where the resulting serial intervals would resemble more those observed (Figure 1b) . Here we used detailed information on the timing of symptom onsets in transmission pairs to infer the infectiousness profile of COVID-19. We found substantial transmission potential for patients who have not yet shown symptoms, with 44% of transmission prior to symptom onset. This should however be interpreted in the context of most cases being isolated after symptom onset, preventing some post-symptomatic transmission that might otherwise have occurred. Nevertheless it is clear that pre-symptomatic transmission is most likely occurring at significant frequencies. Our modeling analysis suggests that viral shedding may begin 2-3 days before showing first symptoms. After symptom onset, viral loads decreased monotonically, consistent with a . 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 March 18, 2020. . https://doi.org/10.1101/2020.03. 15.20036707 doi: medRxiv preprint recent study of 17 patients from another hospital in Guangdong. 10 We could not identify patient subgroups which demonstrated a different viral shedding pattern and our results is likely applicable to COVID-19 patients generally. Whereas a recent study from Wuhan reported that virus was detected for a median of 20 days (up to 37 days amongst survivors) after symptom onset, 11 infectiousness may decline significantly after 10 days. 12 Together this supports our findings that the infectiousness profile may more closely resemble that of influenza than of SARS (Figure 1a ), although we did not have data on viral shedding before symptom onset for COVID-19. 9, 13 Our results are also supported by reports on asymptomatic and pre-symptomatic transmission. 14, 15 For a reproductive number of 2.5, 2 contact tracing and isolation alone are less likely to be successful if more than 30% of transmission occurred before symptom onset, unless >90% of the contacts can be traced. 16 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 March 18, 2020. . https://doi.org/10.1101/2020.03. 15.20036707 doi: medRxiv preprint and social distancing would likely be the key instruments for disease control in the community. Our study has several limitations. First, the infectiousness profile was inferred from infector-infectee pairs and could be affected by the isolation of some cases, as well as behavioral changes after symptom onset such as enhanced personal hygiene and selfisolation. Second, we reviewed available information on known serial intervals, but the observations we identified may have been biased towards shorter intervals because reporting of more recent exposure to sick persons before symptom onsets is more prominent. This may have shifted the inferred infectiousness profile towards negative (i.e. transmission before symptom onset). Nevertheless in our dataset of transmission pairs we were able to capture relatively long serial intervals (e.g. > 10 days), hence the impact on the inferred infectiousness was likely limited. Our simulation also shows how infectiousness starting before symptom onset would be more consistent with the observed serial interval distribution. In conclusion, we found that viral shedding of laboratory-confirmed COVID-19 patients peaked on or before symptom onset, and a substantial proportion of transmission likely occurred before first symptoms in the index case. More inclusive criteria for contact tracing to capture potential transmission events 2-3 days before symptom onset should be urgently considered for effective control of the outbreak. . 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 Guangzhou Eighth People's Hospital in Guangdong, China was designated as one of the specialized hospitals for treating COVID-19 patients at both city and provincial levels on 20 Jan 2020. Since then many COVID-19 cases were admitted via fever clinics, the hospital emergency room or after confirmation of cases from community epidemiological investigation carried out by the Guangzhou Center for Disease Control and Prevention, or transferred from other hospitals. The first confirmed COVID-19 case was admitted on 21 Jan 2020, but in the initial phase suspected cases have also been admitted. We identified all suspected and confirmed COVID-19 cases admitted from 21 Jan 2020 to 14 Feb 2020, and collected throat swabs from them. Patients included those who travelled from Wuhan or Hubei to Guangzhou, and local cases, ranging from asymptomatic, mild to severe cases on admission. The samples were tested by N-gene-specific quantitative reverse-transcriptasepolymerase-chain-reaction (RT-PCR) assay as previously described. 17 To understand the temporal dynamics of viral shedding and exclude non-confirmed COVID-19 cases, we 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 March 18, 2020. We analyzed two separate data sets -clinical and epidemiologic -to assess presymptomatic infectiousness. First, we assessed longitudinal viral shedding data from laboratory-confirmed COVID-19 patients starting from symptom onsets, where viral shedding during the first few days after illness onset can be compared with the inferred infectiousness. Second, the serial intervals from clear transmission chains, combined with information on the incubation period distribution, were used to infer the infectiousness profile as described below. . 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 March 18, 2020. . https://doi.org/10.1101/2020.03. 15.20036707 doi: medRxiv preprint We present SARS-CoV-2 viral loads in the throat swabs of each patient by day of symptom onset. A smoothing spline was fitted to the Ct values to summarize the overall trend. We also compared the viral load by disease severity, age, sex and travel history from Hubei. We fitted a gamma distribution to the transmission pairs data to estimate the serial interval distribution. We used a published estimate of the incubation period distribution to infer infectiousness with respect to symptom onset, from the first 425 COVID-19 patients in Wuhan with detailed exposure history. 2 We considered that infected cases would become infectious at a certain time point before or after illness onset (tS1). Infectiousness, i.e. transmission probability to a secondary case, would then increase until reaching its peak (Figure 1 ). The transmission event would occur at time tI with a probability described by the infectiousness profile β(tI -tS1) relative to the illness onset date, assuming a gamma distribution. The secondary case would then show symptoms at time tS2, after the incubation period that is assumed to follow a lognormal distribution g(tS2 -tI). Hence the observed serial intervals distribution f(tS2 -tS1) would be the convolution between the infectiousness profile and incubation period distribution. We constructed a likelihood function based on the convolution which was fitted to the observed serial intervals, allowing for the start of infectiousness around symptom onset. Parameters were estimated using maximum likelihood. We also performed sensitivity analyses by fixing the start of infectiousness at day 4 and 7 respectively before symptom onset and inferred the infectiousness profile. 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 March 18, 2020. . As an additional check, we simulated the expected serial intervals assuming the same aforementioned incubation period but two different infectiousness profiles, where infectiousness started on the same day and from 2 days before symptom onset respectively. A recent study isolated live infectious SARS-CoV-2 virus from COVID-19 patients up to 10 days after symptom onset, 12 thus we assumed the same duration of infectiousness. We also assumed that infectiousness peaked on the day of symptom onset. The timing of transmission to secondary cases were simulated according to the infectiousness profile using a lognormal and exponential distribution respectively, where the serial intervals were estimated as the sum of the onset to transmission interval and the incubation period. We present the distribution of the serial intervals over 10,000 simulations. All statistical analyses were conducted in R version 3.6.2 (R Development Core Team, Vienna, Austria). Data collection and analysis were required by the National Health Commission of the People's Republic of China to be part of a continuing public health outbreak investigation. Detailed transmission pairs data in this study are given in the supplementary information and viral shedding data will be available upon request and approval by a data access committee. The data access committee comprises leadership of the Guangzhou Eighth People's Hospital and the Guangzhou Health Commission; there is no restriction to data access. . 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 March 18, 2020. Figure 2 . Viral load (Ct values) detected by RT-PCR in throat swabs from patients infected with SARS-CoV-2 (N=94), overall and stratified by disease severity, sex, age group and link to Hubei province. The detection limit was Ct = 40 which was used to indicate negative samples. The black line showed the trend in viral load using smoothing spline. A Novel Coronavirus from Patients with Pneumonia in China Early Transmission Dynamics in Wuhan, China, of Novel Coronavirus-Infected Pneumonia Nowcasting and forecasting the potential domestic and international spread of the 2019-nCoV outbreak originating in Wuhan, China: a modelling study Pattern of early human-to-human transmission of Wuhan Clinical progression and viral load in a community outbreak of coronavirusassociated SARS pneumonia: a prospective study Estimating variability in the transmission of severe acute respiratory syndrome to household contacts in Hong Kong, China Epidemiology, transmission dynamics and control of SARS: the 2002-2003 epidemic Transmission dynamics of the etiological agent of SARS in Hong Kong: impact of public health interventions Viral Shedding and Transmission Potential of Asymptomatic and Paucisymptomatic Influenza Virus Infections in the Community SARS-CoV-2 Viral Load in Upper Respiratory Specimens of Infected Patients Clinical course and risk factors for mortality of adult inpatients with COVID-19 in Wuhan Clinical presentation and virological assessment of hospitalized cases of coronavirus disease 2019 in a travel-associated transmission cluster. medRxiv Influenza A Virus Shedding and Infectivity in Households Presumed Asymptomatic Carrier Transmission of COVID-19 Potential Presymptomatic Transmission of SARS-CoV-2 Feasibility of controlling COVID-19 outbreaks by isolation of cases and contacts Detectable 2019-nCoV viral RNA in blood is a strong indicator for the further clinical severity This work was supported by Department of Science and Technology of Guangdong Province (Project No #2020B111108001) and a commissioned grant from the Health and Medical Research Fund from the Government of the Hong Kong Special Administrative Region. XH, EHYL, PW, BJC, FL and GML conceived and designed the study. XH, XD, JW, YG, XT, XM, YC and BL were responsible for clinical care and collected all biomaterials. WC and FH carried out laboratory testing. QZ, MZ and YW collected and collated linked clinicalepidemiologic data. LZ, FZ and FL supervised and coordinated all aspects of the study at Guangzhou Eighth People's Hospital. PW, XH, YCL and JYW collected and verified all infector-infectee transmission data. EHYL, BJC and GML wrote the first draft. All authors contributed to data interpretation, critical revision of the manuscript and approved the final version of the manuscript. The authors declare no competing interests.