Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Suppression of a SARS-CoV-2 outbreak in the Italian municipality of Vo’

An Author Correction to this article was published on 15 January 2021

An Author Correction to this article was published on 15 January 2021

This article has been updated

Abstract

On 21 February 2020, a resident of the municipality of Vo’, a small town near Padua (Italy), died of pneumonia due to severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection1. This was the first coronavirus disease 19 (COVID-19)-related death detected in Italy since the detection of SARS-CoV-2 in the Chinese city of Wuhan, Hubei province2. In response, the regional authorities imposed the lockdown of the whole municipality for 14 days3. Here we collected information on the demography, clinical presentation, hospitalization, contact network and the presence of SARS-CoV-2 infection in nasopharyngeal swabs for 85.9% and 71.5% of the population of Vo’ at two consecutive time points. From the first survey, which was conducted around the time the town lockdown started, we found a prevalence of infection of 2.6% (95% confidence interval (CI): 2.1–3.3%). From the second survey, which was conducted at the end of the lockdown, we found a prevalence of 1.2% (95% CI: 0.8–1.8%). Notably, 42.5% (95% CI: 31.5–54.6%) of the confirmed SARS-CoV-2 infections detected across the two surveys were asymptomatic (that is, did not have symptoms at the time of swab testing and did not develop symptoms afterwards). The mean serial interval was 7.2 days (95% CI: 5.9–9.6). We found no statistically significant difference in the viral load of symptomatic versus asymptomatic infections (P = 0.62 and 0.74 for E and RdRp genes, respectively, exact Wilcoxon–Mann–Whitney test). This study sheds light on the frequency of asymptomatic SARS-CoV-2 infection, their infectivity (as measured by the viral load) and provides insights into its transmission dynamics and the efficacy of the implemented control measures.

Main

As of 23 May 2020, 5,105,881 confirmed cases and 333,446 deaths of COVID-19 have been reported worldwide2. In Italy, COVID-19 has caused more than 32,616 confirmed deaths. The causative agent (SARS-CoV-2), a close relative of SARS-CoV4, was detected in the human population in Wuhan city, Hubei province (China) around the beginning of December 20195,6. In Hubei province and in the rest of mainland China, recent reports suggest that strategies based on the isolation of cases and their contacts, along with drastic social distancing measures that include the quarantine of whole cities and regions, the closure of schools and workplaces and the cancellations of mass gatherings had a considerable effect on the control of the epidemic7,8. However, the long-term effectiveness of these interventions remains unclear9. In Europe, similar interventions have been implemented to control the transmission of SARS-CoV-2. Recent analyses suggest that control is likely to be achieved across Europe10. In Italy, interventions have successfully controlled the transmission of SARS-CoV-2 in all regions, but uncertainties remain about the ability to avoid a resurgence of transmission as interventions are relaxed11. Effective long-term control of transmission in Europe and worldwide depends on an improved understanding of the mechanisms of SARS-CoV-2 transmission, particularly on the relative contribution of asymptomatic, presymptomatic and symptomatic transmission12. This is particularly important given that, in the absence of a vaccine or effective treatment, alternative public health interventions are being evaluated to allow the population to maintain essential societal and economic activities, while controlling the spread of SARS-CoV-2, limiting mortality and maintaining healthcare demand within capacity.

In this study, we present the results of two surveys of the resident population of Vo’, conducted less than 2 weeks apart, to investigate population exposure to SARS-CoV-2 before and after the lockdown. We present an analysis of population demography, the prevalence of infection, viral load and the frequency of symptomatic, asymptomatic and presymptomatic infections. We assessed the risk of SARS-CoV-2 infection associated with comorbidity and therapies for underlying conditions, characterized chains of transmission, studied the transmission dynamics of SARS-CoV-2 and assessed the impact of the lockdown. Our analyses show that viral transmission could be effectively and rapidly suppressed by combining the early isolation of infected people with community lockdown. The experience of Vo’ shows that, despite the silent and widespread transmission of SARS-CoV-2, transmission can be controlled and represents a model for the systematic suppression of viral outbreaks under similar epidemiological and demographic conditions.

During the two surveys, we collected nasopharyngeal swabs from 2,812 and 2,343 study participants, which corresponded to 85.9% and 71.5% of the eligible study population, respectively (Fig. 1). All age groups were homogeneously sampled with age-specific percentages ranging from 57.1% to 95.4% in the first survey and 40.1% to 80.4% in the second survey (Extended Data Table 1). Statistical analysis showed that, while the recruited and non-recruited populations are different in terms of age distribution (P < 0.001 for the first and second surveys, Fisher’s exact test), there was no statistically significant bias in the composition of the different age groups enrolled in the two surveys (P = 0.31, exact Wilcoxon–Mann–Whitney test) (Extended Data Fig. 1). Notably, no additional infections were reported in Vo’ despite the escalating epidemic in the surrounding regions.

Fig. 1: Study description.
figure1

a, Map showing the location of Vo’ and the Veneto region (grey area) within Italy, produced using shapefiles from GADM (https://gadm.org/) and Italian National Institute of Statistics (ISTAT; https://www.istat.it/it/archivio/222527 and https://www.istat.it/it/archivio/104317#accordions). b, Flow chart summarizing the key statistics on the two sequential nasopharyngeal swab surveys conducted in Vo’ to assess the transmission of SARS-CoV-2 before and after the implementation of interventions. c, Summary of the key events in the study period.

Analysis of infection prevalence

A total of 73 out of the 2,812 participants who were tested at the first survey were positive, which gives a prevalence of 2.6% (95% CI: 2.1–3.3%) (Table 1). The second survey identified 29 total positive cases (prevalence of 1.2%; 95% CI: 0.8–1.8%), 8 of which were new cases (prevalence of 0.3%; 95% CI: 0.15–0.7%) (Fig. 2). One of the eight new infections detected in the second survey was a hospitalized participant who tested positive, then negative, then positive again. It is unclear whether this was a case of SARS-CoV-2 re-infection or whether the second test was a false negative. The frequency of the symptoms in the participants who were positive for SARS-CoV-2 infection was systematically recorded, with fever and cough being the most common (Extended Data Fig. 1). Notably, a total of 29 out of the 73 participants (39.7%; 95% CI: 28.5–51.9%) who tested positive at the first survey were asymptomatic (that is, did not show symptoms at the time of swab sampling nor afterwards; see the definition of symptomatic in the Methods). A similar proportion of asymptomatic infection was also recorded at the second survey (13 out of 29, 44.8%; 95% CI: 26.5–64.3%); of the eight new cases, five were asymptomatic (Table 2, Extended Data Fig. 2). No infections were detected in either survey in 234 tested children ranging from 0 to 10 years of age, including those living in the same household as infected individuals (Extended Data Table 3). The prevalence of infection oscillated between a central estimate of 1.2% and 1.7% up to 50 years of age (Extended Data Fig. 1). Older participants showed a threefold increase in the infection prevalence (Table 2, Extended Data Fig. 1). Of the 81 participants who were positive for SARS-CoV-2 across the two surveys, 13 required hospitalization (16.0%). Their age distribution was as follows: 1 (7.7%) aged 41–50 years, 1 (7.7%) aged 51–60 years, 4 (30.8%) aged 61–70 years, 5 (38.5%) aged 71–80 years and 2 (15.4%) aged 81–90 years.

Table 1 Participants positive for SARS-CoV-2 at the first and second surveys
Fig. 2: SARS-CoV-2 prevalence statistics.
figure2

a, The prevalence of SARS-CoV-2 infection at the first survey (x = 73 positive out of n = 2,812 tested) and the second survey (x = 29 positive out of n = 2,343 tested). The error bars represent the 95% exact binomial CI. b, The number of SARS-CoV-2 infections detected in the sampled population of the residents of Vo’ in the first survey (x = 73) and the second survey (x = 29, of which 8 were new infections).

Table 2 Participants testing positive stratified by sex and age

A substantial fraction of infected participants (58.9%; 95% CI: 46.8–70.3%, presymptomatic, symptomatic and asymptomatic combined over all ages) cleared the infection between the first and second surveys, that is, had a negative test at the second survey after a positive test at the first survey (Extended Data Table 2). For all infections identified in the study, clearance was confirmed by an additional negative test that was conducted independently by the local health authority (data not shown). The time to viral clearance (the time from the earliest positive sample for the participants with more than one sample in the first survey to a negative sample in the second survey) ranged from 8 to 13 days and was on average 9.3 days, with a standard deviation of 2.0 days. The minimal duration of the positivity window (the time from the earliest positive sample in the first survey to a positive sample in the second survey) ranged from 3 to 13 days and was on average 9.1 days, with a standard deviation of 2.3 days. In particular, 61.4% (95% CI: 45.5–75.6%) of symptomatic and 55.2% (95% CI: 35.7–73.6%) of asymptomatic individuals with SARS-CoV-2 infections cleared the virus during the study period (that is, had a negative test after a positive result at the first survey); the highest rate (100%) was observed in the age groups of symptomatic individuals aged 31–40 and 41–50 years (Extended Data Table 2). SARS-CoV-2 positivity overall (that is, the first and second surveys combined) and at the first survey was more frequently associated with individuals who were 71–80 years of age (compared to those 21–30 years of age; P = 0.012 and P = 0.017, respectively) (Extended Data Fig. 1). Being male was associated with SARS-CoV-2 positivity in the second survey (P = 0.04) (Table 2). Analyses of the association between common comorbidities such as diabetes, hypertension, vascular diseases, respiratory diseases in asymptomatic and symptomatic people and the use of treatment for a number of different conditions with symptomatic infection showed no significant association (Supplementary Tables 3, 4).

Role of asymptomatic transmission

The analysis of viral genome equivalents inferred from cycle threshold data from real-time reverse transcription PCR (RT–PCR) assays indicated that asymptomatic and symptomatic participants did not differ when data from viral PCR templates recovered from the nasopharyngeal swabs of asymptomatic and symptomatic participants were compared (P = 0.62 and 0.74 for gene E and gene RdRp, respectively; exact Wilcoxon–Mann–Whitney test) (Extended Data Fig. 3). We found that the viral load tends to peak around the day of symptom onset and, for most of the participants, tends to decline after symptom onset (Extended Data Fig. 3). The relative risk of contracting the infection from having close contacts with an infected relative, including those living in the same household, gives an odds ratio of 84.5 (95% CI: 16.8–425.4) (Extended Data Table 4, Supplementary Text 3). Two out of the eight participants with new infections that were detected in the second survey either shared a household or had a contact history with asymptomatic individuals (Supplementary Table 1).

Reconstructing transmission chains

From the inferred transmission pairs, we estimated a serial interval distribution over the whole study period with a mean of 7.2 days (95% CI: 5.9–9.6). We found that the lockdown reduced the serial interval from a mean of 7.6 days (95% CI: 6.4–8.7) before the lockdown to a mean of 6.2 days (95% CI: 2.6–10.7) after the lockdown (Extended Data Fig. 4). We also found that the lockdown substantially reduced transmission, with the reproduction number dropping from an initial value of 2.44 (95% CI: 1.30–3.91) before the lockdown to 0.41 (95% CI: 0.21–0.64) after the lockdown.

Modelling point prevalence data

We used the prevalence estimates obtained in Vo’ at the first and second surveys to calibrate a modified susceptible–exposed–infectious–recovered compartmental model of SARS-CoV-2 transmission that incorporates symptomatic, presymptomatic and asymptomatic infections, virus detectability (in swabs) before and after the infectious period and the impact of the lockdown (Extended Data Fig. 5). We assumed that presymptomatic, symptomatic and asymptomatic infections transmit the virus. We estimated that on average 41% of the infections are asymptomatic, that the mean infectious period is approximately 3.6–6.5 days, and that the lockdown reduced SARS-CoV-2 transmissibility on average by between 82% and 98%, depending on the assumed initial value of \({R}_{0}^{1}\) and on the duration of virus detectability (Supplementary Table 5). The model suggests that on average up to 86.2% (range: 82.2–91.6%) of the population would have been infected in the absence of interventions and that with the lockdown, 4.9% (range: 2.9–8.1%) of the population of Vo’ was infected by SARS-CoV-2 (Fig. 3). These estimates are in line with the attack rates that were recently estimated for the Veneto region11. The model suggests that shorter values of the average duration of virus detectability beyond the infectious period better capture the central point prevalence estimates (Extended Data Fig. 6, Supplementary Table 5). Our results suggest that SARS-CoV-2 was introduced into the Vo’ population at the beginning of February 2020.

Fig. 3: SARS-CoV-2 dynamics of the mitigated and counterfactual unmitigated epidemic in Vo’ and the relative final size estimates.
figure3

a, The prevalence of SARS-CoV-2 infection inferred from the observed prevalence data for symptomatic, presymptomatic and asymptomatic infections in the first and second surveys using \({R}_{0}^{1}\) (the reproduction number before the lockdown) = 2.4 and 1/σ (the average duration of positivity beyond the duration of the infectious period) = 4 days. The dashed vertical line represents the time that the lockdown started. The points represent the observed prevalence data, the 95% CI is the exact binomial CI. The solid lines represent the mean and the shading represents the 95% credible interval obtained with the model from 100 samples from the posterior distribution of the parameters. b, The incidence of the epidemic fitted to the prevalence data (blue) and of the unmitigated epidemic (red), obtained assuming the same initial reproduction number value \({R}_{0}^{1}\) = 2.4 throughout the whole epidemic and 1/σ = 4 days. The dashed vertical line represents the time that the lockdown started. The solid lines represent the mean and the shading represents the 95% credible interval obtained with the model from 100 samples from the posterior distribution of the parameters. c, The mean epidemic final size (the proportion of the population infected at the end of the epidemic) of the counterfactual unmitigated epidemic (red) and of the epidemic fitted from the prevalence data with the lockdown (blue). The error bars represent the range (minimum to maximum) of the mean final size obtained from n = 100 independent samples drawn from the posterior distribution of the parameters, calculated over the models with DIC (deviance information criterion) < 36.4.

Discussion

The results of the two surveys carried out in Vo’ provide important insights into the transmission dynamics of SARS-CoV-2. Our finding that 42.5% (95% CI: 31.5–54.6%) of all confirmed SARS-CoV-2 infections across the two surveys were asymptomatic is in accordance with other population surveys13. Among confirmed SARS-CoV-2 infections, we did not observe significant differences in the frequency of asymptomatic infection between age groups (Supplementary Fig. 10; P = 0.96, Fisher’s exact test). Among symptomatic participants, older age groups tended to show higher frequencies of SARS-CoV-2 infection (Extended Data Table 2). Recent studies have found that the clinical progression of infection in children is generally milder than in adults14,15,16. We found that none of the children under 10 years of age who took part in the study tested positive for SARS-CoV-2 infection at either survey, despite at least 13 of them living together with infected family members (Extended Data Table 3). This agrees with a recent study conducted in Iceland13 and is particularly intriguing given the very high observed odds ratio for adults to become infected when living together with family members who are positive for SARS-CoV-2. However, this result does not mean that children cannot be infected by SARS-CoV-2, but suggests that children may be less susceptible than adults. The pathogenesis of SARS-CoV-2 infection in young children is not well understood15. Notably, nasopharyngeal swabs are tested for the presence of SARS-CoV-2 and can only detect active infection, not exposure. A cross-sectional serological survey would clarify the actual infection rates of the whole population, including children’s exposure, to SARS-CoV-2.

The contribution of asymptomatic infections to SARS-CoV-2 transmission is supported by the viral load data (Extended Data Fig. 3), by the model fit to the observed prevalence data (Extended Data Fig. 6, Supplementary Table 5) and by the observation that two out of the eight participants with new infections that were detected in the second survey reported contacts with asymptomatic individuals (Supplementary Text 3). The extent to which symptoms may promote viral shedding remains to be determined, but the decreasing trend in viral load post-symptom onset suggests that presymptomatic transmission may play an important part17. Asymptomatic transmission and presymptomatic transmission pose clear challenges for the control of COVID-19 in the absence of strict social distancing measures or active epidemiological surveillance comprising, for instance, a test, trace and isolate strategy.

This study has informed the policy adopted by the Veneto region, where swabs are available to all contacts of positive symptomatic cases. This testing and tracing approach has had a tremendous impact on the course of the epidemic in Veneto compared to other Italian regions. In this context, the control strategy applied to Vo’ serves as a model to suppress SARS-CoV-2 transmission across spatial scales. Enhanced community surveillance, the early detection of SARS-CoV-2 transmission and the timely implementation of interventions are key to control COVID-19 and reduce its substantial public health, economic and societal burden worldwide.

Methods

Study setting

The municipality of Vo’, in the province of Padua, Veneto region, Italy, is located about 50 km west of Venice (Fig. 1a). According to the latest land registry, Vo’ has a population of 3,275 individuals over an area of 20.4 km2. Upon the detection of SARS-CoV-2 in a deceased resident of Vo’ on 21 February 2020, the same day where the first COVID-19 case was detected in Vo’ and 1 day after the first locally acquired COVID-19 infection was identified in Italy, we conducted an epidemiological study to investigate the prevalence of SARS-CoV-2 infection in the population. Sampling was conducted on the majority of the Vo’ population at two time points: the first during the days immediately after the detection of the first cases (21–29 February 2020), and the second one at the end of the 2-week lockdown (7 March 2020) (Fig. 1c). For each resident, we collected information on the sampling dates, the results of SARS-CoV-2 testing, demographics (for example, age and sex), residence, health record (including symptoms and COVID-19-related hospitalization dates, previous conditions and therapy taken for other illnesses), household size and contact network. The data were collated using Microsoft Excel and the data set spreadsheet is available at https://github.com/ncov-ic/SEIR_Covid_Vo. No statistical methods were used to predetermine sample size. The experiments were not randomized and the investigators were not blinded to allocation during experiments and outcome assessment.

The definition of symptomatic is as follows: a participant who required hospitalization and/or reported fever (yes/no or a temperature above 37 °C) and/or cough and/or at least two of the following symptoms: sore throat, headache, diarrhoea, vomit, asthenia, muscle pain, joint pain, loss of taste or smell, or shortness of breath.

Laboratory methods

Upper respiratory tract samples were collected by healthcare professionals with a single flocked tapered swab used for the oropharynx then nasal mid-turbinates and immediately put into a sterile tube containing transport medium (eSwab, Copan Italia Spa). Sampling was performed according the US Centers for Disease Control and Prevention guidelines18. In brief, for oropharyngeal sampling, the swab was inserted into the posterior pharynx and tonsillar areas and rubbed over both tonsillar pillars and posterior oropharynx, avoiding touching the tongue, teeth and gums; for deep nasal sampling, the swab was inserted into both nostrils for about 2 cm while gently rotating against the nasal wall several times. Samples were stored at 2–8 °C until testing, which was performed within 72 h from collection. As a measure of the correct execution of the sampling, each PCR contains an internal control designed to amplify the human housekeeping gene encoding RNase P. Reactions that failed to show the internal positive control were classified as invalid and repeated. Total nucleic acids were purified from 200 μl of nasopharyngeal swab samples and eluted in a final volume of 100 μl by using a MagNA Pure 96 System (Roche Applied Sciences). Detection of SARS-CoV-2 RNA was performed by an in-house real-time RT–PCR method, which was developed according the protocol and the primers and probes designed by Corman et al.19 that targeted the genes encoding envelope (E) (E_Sarbeco_F, E_Sarbeco_R and E_Sarbeco_P1) and RNA-dependent RNA polymerase (RdRp: RdRp_SARSr-F, RdRp_SARSr-R, RdRP_SARSr-P1 and RdRp_SARSr-P2) of SARS-CoV-2. Real-time RT–PCR assays were performed in a final volume of 25 μl, containing 5 μl of purified nucleic acids, using One Step Real Time kit (Thermo Fisher Scientific) and run on ABI 7900HT Fast Sequence Detection Systems (Thermo Fisher Scientific). The sensitivity of the E and RdRp gene assays was 5.0 and 50 genome equivalent copies per reaction at 95% detection probability, respectively. Both assays had no cross-reactivity with the endemic human coronaviruses HCoV-229E, HCoV-NL63, HCoV-OC43 and HCoV-HKU1 and with MERS-CoV. All tests were performed at the Clinical Microbiology and Virology Unit of Padova University Hospital, which is the Regional Reference Laboratory for emerging viral infections. After an initial period of dual testing by the National Reference Laboratory at the Italian Institute of Health (Istituto Superiore di Sanità), which demonstrated 100% agreement of results, the Regional Reference Laboratory received accreditation as Reference Laboratory for COVID-19 testing.

Assessment of genome equivalents

Cycle threshold (Ct) data from real-time RT–PCR assays were collected for E and RdRp genes. Ct data for the E gene were available for 30 symptomatic, 5 presymptomatic and 23 asymptomatic infections, and for the RdRp gene for 27 symptomatic, 9 presymptomatic and 26 asymptomatic infections. Genome equivalent copies per ml were inferred according to linear regression performed on calibration standard curves. The interpolated Ct values were further multiplied by 100, according to the final dilution factor (1:100). Linear regression was calculated in Python3.7.3 using modules scipy 1.4.1, numpy 1.18.1 and matplotlib 3.2.120. Genome equivalent distributions from the two genes, for positive symptomatic, asymptomatic and presymptomatic participants were compared with the exact Wilcoxon–Mann–Whitney test. Both viral load genome equivalents and raw Ct data are provided in the data set.

Reconstructing transmission chains

We used data on contacts traced within the community and on household contacts derived from household composition data (reported in the data set) to impute chains of transmission and transmission clusters. We used the R package epicontacts21,22 to visualize the reconstructed transmission chains. We provide the algorithms used to infer the serial interval (the time from symptom onset of the infector to symptom onset of the infectee) distribution and the effective reproduction number (the average number of secondary infections generated by the identified infectors) in Supplementary Information Text 1 and 2, respectively. In brief, we inferred the date of symptom onset for the participants who tested positive but with a missing onset date from the observed time-lags from symptom onset to confirmation (for the participants who tested positive at multiple sampling times, we used the first sampling time). We then used the observed and inferred dates of symptom onset alongside the contact information to infer transmission pairs within the sampled population. In turn, reconstructed transmission pairs were used to characterize the serial interval in the whole study period as well as during the pre-lockdown and post-lockdown periods. Central effective reproduction number estimates were calculated as the average number of secondary infections generated by observed or imputed infectors, having assigned the infector stochastically when more than one or no potential infectors were identified. The 95% CIs were estimated by bootstrapping. All details are provided in Supplementary Information Text 1 and 2.

Mathematical modelling

The first survey occurred between 21 and 29 February 2020 and the second survey occurred on 7 March 2020. In the model, we assumed that prevalence was taken on the weighted average of the first sample collection date, that is, on 26 February 2020 and on 7 March 2020. The flow diagram of the model is given in Extended Data Fig. 5. We assumed that the population of Vo’ was fully susceptible to SARS-CoV-2 (S compartment) at the start of the epidemic. Upon infection, infected people incubate the virus (E compartment) and have undetectable viraemia for an average of 1/ν days before entering a stage (TP compartment) that lasts an average of 1/δ days, in which people show no symptoms and have detectable viraemia. We assume that a proportion p of the infected population remains asymptomatic throughout the whole course of the infection (IA compartment) and that the remaining proportion 1 − p develops symptoms (IS compartment). We assume that symptomatic (IS), asymptomatic (IA + pTP) and presymptomatic ((1 − p)TP) people contribute to the onward transmission of SARS-CoV-2 and that symptomatic, asymptomatic and presymptomatic people transmit the virus for an average of 1/δ + 1/γ days. We further assume that the virus can be detected by swab testing beyond the duration of the infectious period; this assumption is compatible with the hypothesis that transmission occurs for viral loads above a certain threshold but the diagnostic test can detect the presence of virus below the threshold for transmission. Compartments TPS and TPA, respectively, represent symptomatic and asymptomatic people who are no longer infectious but have a detectable viral load, and hence test positive. Eventually, the viral load of all infections decreases below detection and people move into a test negative (TN) compartment. We assume a step change in the reproduction number on the day that lockdown started. Before the implementation of quarantine, the reproduction number is given by \(\,{R}_{0}^{1}=\beta \left(\frac{1}{\gamma }+\frac{1}{\delta }\right)\), and we assume that it drops to \({R}^{2}=w{R}_{0}^{1}\) after the start of the lockdown, where 1 − w represents the per cent reduction in \({R}_{0}^{1}\) due to the intervention. We let Ti denote the number of participants swabbed on survey i (i = 1, 2) and let PAi, PPi and Psi, respectively, denote the number of swabs testing positive among asymptomatic, presymptomatic (that is, those showing no symptoms at the time of testing but develop symptoms afterwards) and symptomatic participants, respectively. We assume that the number of positive swabs among symptomatic, presymptomatic and asymptomatic infections on survey i follows a binomial distribution with parameters Ti and πXi, where πXi represents the probability of testing positive on survey i for X (where X = A, P, S). For symptomatic participants, πSi is given by \({\pi }_{{\rm{S}}i}=\,\frac{{I}_{{\rm{S}}}({t}_{i})+{{\rm{T}}{\rm{P}}}_{{\rm{S}}}({t}_{i})}{N}\), for asymptomatic participants, πAi is given by \({\pi }_{{\rm{A}}i}=\,\frac{p{\rm{T}}{\rm{P}}({t}_{i})+{I}_{{\rm{A}}}({t}_{i})+{{\rm{T}}{\rm{P}}}_{{\rm{A}}}({t}_{i})}{N}\), and for presymptomatic participants, πPi is given by \({\pi }_{{\rm{P}}i}=\,\frac{(1-p){\rm{T}}{\rm{P}}({t}_{i})}{N}\), assuming perfect diagnostic sensitivity and specificity. The likelihood of the model is given by the product of the binomial distributions for symptomatic, presymptomatic and asymptomatic participants at times ti, i = 1, 2. Inference was conducted in a Bayesian framework, using the Metropolis–Hastings Markov chain Monte Carlo (MCMC) method with uniform prior distributions23. We fixed the average generation time (equal to 1/ν + 1/δ + 1/γ) to 7 days19 and let the model infer 1/ν and 1/δ. We explored the following values of \({R}_{0}^{1}\): 2.1, 2.4, 2.7, which are compatible with a doubling time of 3–4 days, as observed in Vo’ and elsewhere in the Veneto region. We assumed that seeding of the infection occurred on 4 February 2020. We explored different scenarios on the average duration of viral detectability beyond the infectious period and fixed 1/σ to be 2, 4, 6, 8, 10 and 12 days. We estimate the number of infections introduced in the population from elsewhere at time t0 (4 February 2020), the proportion of asymptomatic infections p, the average durations 1/ν, 1/δ and 1/γ and the per cent reduction in \({R}_{0}^{1}\) due to the interventions (1 − w)100%.

Analysis of associations

We applied logistic regression to test the association between SARS-CoV-2 positivity (overall and at the first and second surveys separately) with the age group (10 years of age bands, from 0 to >81 years of age) and sex (male and female). We used Fisher’s exact test for comparing two binomial proportions to assess whether there is an association between the presence of symptoms for 41 confirmed COVID-19 cases who are resident in Vo’ and the different types of comorbidities and treatments used. The analyses were repeated on the subset of patients who became negative at the second time point (results not shown). To increase the power of the data, we increased the sample size by including an additional 11 confirmed COVID-19 cases who were resident in other villages close to Vo’. None of these scenarios provided significant associations at the 5% level.

Ethical approval statement

The first sampling of the Vo’ population was conducted within the surveillance programme established by the Veneto region and did not require ethical approval; the second sampling was approved by the Ethics Committee for Clinical Research of the province of Padua. Study participation was by consent. For participants under 18 years of age, consent was provided by a parent or legal guardian.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this paper.

Data availability

The data set is available at https://github.com/ncov-ic/SEIR_Covid_Vo. Queries can be addressed to A.C. (a.drcrisanti@imperial.ac.uk; andrea.crisanti@unipd.it) or I.D. (i.dorigatti@imperial.ac.uk).

Code availability

The code is available at https://github.com/ncov-ic/SEIR_Covid_Vo.

Change history

References

  1. 1.

    Crisanti, A. & Cassone, A. In one Italian town, we showed mass testing could eradicate the coronavirus. The Guardian https://www.theguardian.com/commentisfree/2020/mar/20/eradicated-coronavirus-mass-testing-covid-19-italy-vo (2020).

  2. 2.

    World Health Organization. Novel coronavirus (COVID-19) situation. WHO https://covid19.who.int/ (2020).

  3. 3.

    Saini, V. Coronavirus: voices from a quarantined Italian town. EU Observer https://euobserver.com/coronavirus/147552 (2020).

  4. 4.

    Zhou, P. et al. A pneumonia outbreak associated with a new coronavirus of probable bat origin. Nature 579, 270–273 (2020).

    ADS  CAS  Article  Google Scholar 

  5. 5.

    Volz, E. et al. Report 5: phylogenetic analysis of SARS-CoV-2. Imperial College London https://www.imperial.ac.uk/media/imperial-college/medicine/mrc-gida/2020-02-15-COVID19-Report-5.pdf (2020).

  6. 6.

    Centre for Health Protection of the Hong Kong Special Administrative Region Government. CHP closely monitors cluster of pneumonia cases on Mainland. The Government of the Hong Kong Special Administrative Region https://www.info.gov.hk/gia/general/201912/31/P2019123100667.htm (2020).

  7. 7.

    Prem, K. et al. The effect of control strategies to reduce social mixing on outcomes of the COVID-19 epidemic in Wuhan, China: a modelling study. Lancet Public Health 5, e261–e270 (2020).

    Article  Google Scholar 

  8. 8.

    Lai, S. et al. Effect of non-pharmaceutical interventions to contain COVID-19 in China. Nature https://doi.org/10.1038/s41586-020-2293-x (2020).

  9. 9.

    Anderson, R. M., Heesterbeek, H., Klinkenberg, D. & Hollingsworth, T. D. How will country-based mitigation measures influence the course of the COVID-19 epidemic? Lancet 395, 931–934 (2020).

    CAS  Article  Google Scholar 

  10. 10.

    Flaxman, S. et al. Estimating the effects of non-pharmaceutical interventions on COVID-19 in Europe. Nature. https://doi.org/10.1038/s41586-020-2405-7 (2020).

  11. 11.

    Vollmer, M. A. C. et al. Report 20: using mobility to estimate the transmission intensity of COVID-19 in Italy: a subnational analysis with future scenarios. Imperial College London https://www.imperial.ac.uk/media/imperial-college/medicine/mrc-gida/2020-05-04-COVID19-Report-20.pdf (2020).

  12. 12.

    Mizumoto, K., Kagaya, K., Zarebski, A. & Chowell, G. Estimating the asymptomatic proportion of coronavirus disease 2019 (COVID-19) cases on board the Diamond Princess cruise ship, Yokohama, Japan, 2020. Euro Surveill. 25, 200018 (2020).

    Article  Google Scholar 

  13. 13.

    Gudbjartsson, D. F. et al. Spread of SARS-CoV-2 in the Icelandic population. N. Engl. J. Med. 382, 2302–2315 (2020).

    CAS  Article  Google Scholar 

  14. 14.

    Cai, J. et al. A case series of children with 2019 novel coronavirus infection: clinical and epidemiological features. Clin. Infect. Dis. https://doi.org/10.1093/cid/ciaa198 (2020).

  15. 15.

    Zimmermann, P. & Curtis, N. Coronavirus infections in children including COVID-19: an overview of the epidemiology, clinical features, diagnosis, treatment and prevention options in children. Pediatr. Infect. Dis. J. 39, 355–368 (2020).

    Article  Google Scholar 

  16. 16.

    Bi, Q. et al. Epidemiology and transmission of COVID-19 in 391 cases and 1286 of their close contacts in Shenzhen, China: a retrospective cohort study. Lancet Infect. Dis. https://doi.org/10.1016/S1473-3099(20)30287-5 (2020).

  17. 17.

    He, X. et al. Temporal dynamics in viral shedding and transmissibility of COVID-19. Nat. Med. 26, 672–675 (2020).

    CAS  Article  Google Scholar 

  18. 18.

    Centers for Disease Control and Prevention. Interim guidelines for collecting, handling, and testing clinical specimens from persons for coronavirus disease 2019 (COVID-19). CDC https://www.cdc.gov/coronavirus/2019-ncov/lab/guidelines-clinical-specimens.html (accessed 18 May 2020).

  19. 19.

    Corman, V. M. et al. Detection of 2019 novel coronavirus (2019-nCoV) by real-time RT-PCR. Euro Surveill. 25, 2000045 (2020).

    PubMed Central  Google Scholar 

  20. 20.

    Python language reference, version 2.7 (Python Software Foundation, 2020).

  21. 21.

    R Core Team. R: A Language and Environment for Statistical Computing. http://www.R-project.org/ (R Foundation for Statistical Computing, 2020).

  22. 22.

    Nagraj, V. P. et al. epicontacts: handling, visualisation and analysis of epidemiological contacts. F1000Res. 7, 566 (2018).

    CAS  Article  Google Scholar 

  23. 23.

    Robert, C. The Metropolis–Hastings algorithm. Wiley StatsRef https://doi.org/10.1002/9781118445112.stat07834 (2015).

Download references

Acknowledgements

We thank the Mayor of the municipality of Vo’, G. Martini, for his unreserved support throughout the study; a special thanks to the population of Vo’ who volunteered en masse to this study; M. Perilli and S. Guglielmo for assistance in data collection and consistency check; and F. Bosa and G. Rupolo from the Italian Red Cross for the support in patient samplings. This work was supported by the Veneto region and was jointly funded by the UK Medical Research Council (MRC; grant MR/R015600/1), the UK Department for International Development (DFID) under the MRC/DFID Concordat agreement, the Abdul Latif Jameel Foundation and is also part of the EDCTP2 programme supported by the European Union. I.D. acknowledges research funding from a Sir Henry Dale Fellowship funded by the Royal Society and Wellcome Trust (grant 213494/Z/18/Z). C.C. acknowledges funding from the Wellcome Trust (grant 203851/Z/16/Z). L.C.O. from the Imperial College COVID-19 Response Team and G.C.-D. acknowledge research funding from The Royal Society. L.B., E.L. and S.T. acknowledge research funding from the European Union's Horizon 2020 research and innovation programme, under grant agreement no. 874735 (VEO).

Author information

Affiliations

Authors

Consortia

Contributions

A.C. conceived the project with input from E.L. and I.D. I.D. conceived the modelling with input from N.M.F. and C.A.D. E.L. coordinated data collection, curation and analyses. E.F. coordinated the diagnostic team and facilities. C.C. and G.C.-D. are joint second authors. E.F., L.B., C.D.V., L.R., R.M., A.L., D.A., M.S., E.D.C., M.C.V., F.S., F.O., V. Besutti, M.P., S.G.P., G.M. and M.T. performed laboratory testing on swabs and validated the results. E.L., S.T., V. Baldo, A.S., N.N. and S.C. analysed the data, contributed to the discussion and commented on the manuscript. A.R.B., I.D. and C.A.D. performed the statistical analyses. C.C., L.C., N.M.F. and I.D. developed the mathematical model. G.C.-D., K.A.M.G., C.A.D. and I.D. performed cluster analysis. E.L., M.N., F.C., G. Castelli, E.N., B.L., L. Fava and M.D. performed data collection, direct contacting of study participants at follow up and consistency check on metadata. S.M., R.S., G. Carretta, D.D. and L. Flor organized sampling logistics. S.M. and R.S. performed swab samplings. The Imperial College COVID-19 Response Team contributed to the discussion and background understanding of COVID-19 epidemiology. A.C. and I.D. wrote the manuscript, with contribution from E.L., L.B., V. Baldo and C.A.D.

Corresponding authors

Correspondence to Ilaria Dorigatti or Andrea Crisanti.

Ethics declarations

Competing interests

The authors declare no competing interests.

Additional information

Peer review information Nature thanks Gabriel Leung, Malik Peiris and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data figures and tables

Extended Data Fig. 1 Summary statistics, frequency of symptoms and prevalence by age.

a, Age distributions (in years) of the participants enrolled in the first and second surveys. b, Frequency of individual symptoms (fever x = 29, cough x = 19, sore throat x = 9, headache x = 9, diarrhoea x = 3, malaise x = 2 and conjunctivitis x = 1) among participants with confirmed SARS-CoV-2 infection across the whole study period (that is, the first and second surveys aggregated; n = 80 participants). The error bars represent the 95% exact binomial CI. c, Age distribution of the population recruited and not recruited in the first survey. d, Age distribution of the population recruited and not recruited in the second survey. e, SARS-CoV-2 prevalence by age at the first and second surveys combined (positive x = 0, 5, 6, 9, 7, 23, 21, 25 and 6, tested n = 374, 460, 431, 527, 805, 935, 733, 580 and 310, respectively, in age groups 0–10, 11–20, 21–30, 31–40, 41–50, 51–60, 61–70, 71–80 and 81+ years) and at the first (positive x = 0, 3, 4, 7, 5, 16, 15, 19 and 4, tested n = 217, 250, 240, 286, 439, 496, 384, 318 and 182, respectively, in age groups 0–10, 11–20, 21–30, 31–40, 41–50, 51–60, 61–70, 71–80 and 81+ years) and second (positive x = 0, 2, 2, 2, 2, 7, 6, 6 and 2, tested n = 157, 210, 191, 241, 366, 439, 389, 262 and 128, respectively, in age groups 0–10, 11–20, 21–30, 31–40, 41–50, 51–60, 61–70, 71–80 and 81+ years) surveys separately. The error bars represent the 95% exact binomial CI.

Extended Data Fig. 2 Symptomatic and asymptomatic infection statistics.

a, Relative proportion of asymptomatic and symptomatic SARS-CoV-2 infections among the total number of positive swabs in the first survey (first survey – total cases; asymptomatic x = 29, symptomatic x = 44, tested n = 73), second survey (second survey – total cases; asymptomatic x = 13, symptomatic x = 16, tested n = 29) and among the number of new positive swabs in the second survey (second survey – new cases; asymptomatic x = 5, symptomatic x = 3, tested n = 8). The error bars represent the 95% exact binomial CI. b, Age distribution and relative proportion of asymptomatic and symptomatic SARS-CoV-2-positive infections among the total number of positive swabs in the first survey (first survey – total cases; asymptomatic x = 0, 2, 0, 3, 3, 6, 6, 8 and 1, symptomatic x = 0, 1, 4, 4, 2, 10, 9, 11 and 3, respectively, in age groups 0–10, 11–20, 21–30, 31–40, 41–50, 51–60, 61–70, 71–80 and 81+ years; tested n = 73) and among the number of new positive swabs in the second survey (second survey – new cases; asymptomatic x = 0, 0, 0, 0, 1, 1, 2, 1 and 0, symptomatic x = 0, 1, 0, 0, 0, 1, 0, 1 and 0, respectively, in age groups 0–10, 11–20, 21–30, 31–40, 41–50, 51–60, 61–70, 71–80 and 81+ years; tested n = 8). The error bars represent the 95% exact binomial CI.

Extended Data Fig. 3 Viral load for asymptomatic, pre-symptomatic and symptomatic infections and viral load dynamics relative to the number of days from symptom onset.

a, The median (solid line), the interquartile range (that is, 25th to 75th percentiles (box)) and the range (that is, minimum to maximum (whiskers)) of gene E genome equivalent copies per ml (log10 scale, y axis) calculated from RT–PCR interpolated values (asymptomatic n = 23, pre-symptomatic n = 5 and symptomatic n = 30). The raw Ct data and the derived values of the genome equivalent copies are provided in the data set. b, The median (solid line), the interquartile range (that is, 25th to 75th percentiles (box)) and the range (that is, minimum to maximum (whiskers)) of gene E genome equivalent copies per ml (log10 scale, y axis) versus the number of days from symptom onset (days, x axis); n = 34 participants. The lines in colour join measurements from the same participant. The solid lines identify the four participants with sequential viral load measurements for both gene E and gene RdRp. c, The median (solid line), the interquartile range (that is, 25th to 75th percentiles (box)) and the range (that is, minimum to maximum (whiskers)) of RdRp genome equivalent copies per ml (log10 scale, y axis) calculated from RT–PCR interpolated values (asymptomatic n = 26, pre-symptomatic n = 9 and symptomatic n = 27). The raw Ct data and the derived values of genome equivalent copies are provided in the data set. d, The median (solid line), the interquartile range (that is, 25th to 75th percentiles (box)) and the range (that is, minimum to maximum (whiskers)) of RdRp genome equivalent copies per ml (log10 scale, y axis) versus the number of days from symptom onset (days, x axis); n = 28 participants. The lines in colour join measurements from the same participant. The solid lines identify the four participants with sequential viral load measurements for both gene E and gene RdRp.

Extended Data Fig. 4 Serial interval distribution and transmission chains.

a, Estimated serial interval distributions for the whole study period (overall) and for the pre-lockdown (before 24 February 2020) and post-lockdown (after 24 February 2020) periods. b, Observed transmission clusters from reported and household contacts. Each node (circle) represents a positive infection, and the edges (the line connecting the nodes) connect positive infections that reported contacts or are household members. The different colours represent different clusters of infection.

Extended Data Fig. 5 Flow chart of the mathematical model fitted to the point prevalence data observed in Vo’ at the first and second surveys.

Further details are provided in the Methods.

Extended Data Fig. 6 SARS-CoV-2 dynamics in Vo’ inferred from the fit of the dynamical model to the observed prevalence of symptomatic, pre-symptomatic and asymptomatic infections in the first and second surveys.

Each sub-panel represents the model fit using the specified values of \({R}_{0}^{1}\) (the reproduction number before the lockdown) and 1/σ (the average duration of positivity beyond the duration of the infectious period). The dashed vertical line represents the time that lockdown started. The points represent the observed prevalence data; the 95% CI is the exact binomial CI. The solid lines represent the mean and the shading represents the 95% credible interval obtained from 100 samples from the posterior distribution of the parameters.

Extended Data Table 1 Age distribution of Vo’ residents and the number of tested participants at the two time points across different age groups
Extended Data Table 2 Age distribution of symptomatic and asymptomatic individuals at the first and second surveys
Extended Data Table 3 Children negative for SARS-CoV-2 living in households with infected relatives
Extended Data Table 4 Results of the second survey for participants living with or reporting close contacts with relatives infected with SARS-CoV-2

Supplementary information

Supplementary Information

This file contains Supplementary Text (Methods and Results) and Supplementary Tables 1-5.

Reporting Summary

Peer Review File

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Lavezzo, E., Franchin, E., Ciavarella, C. et al. Suppression of a SARS-CoV-2 outbreak in the Italian municipality of Vo’. Nature 584, 425–429 (2020). https://doi.org/10.1038/s41586-020-2488-1

Download citation

Further reading

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Search

Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing