key: cord-320636-mvtux07x
authors: Pullano, G.; Di Domenico, L.; Sabbatini, C. E.; Valdano, E.; Turbelin, C.; Debin, M.; Guerrisi, C.; Kengne-Kuetche, C.; Souty, C.; Hanslik, T.; Blanchon, T.; Boëlle, P.-Y.; Figoni, J.; Vaux, S.; Campese, C.; Bernard-Stoecklin, S.; Colizza, V.
title: Underdetection of COVID-19 cases in France in the exit phase following lockdown
date: 2020-08-12
journal: nan
DOI: 10.1101/2020.08.10.20171744
sha: 
doc_id: 320636
cord_uid: mvtux07x

A novel testing policy was implemented in May in France to systematically screen potential COVID-19 infections and suppress local outbreaks while lifting lockdown restrictions. 20,736 virologically-confirmed cases were reported in mainland France from May 13, 2020 (week 20, end of lockdown) to June 28 (week 26). Accounting for missing data and the delay from symptom onset to confirmation test, this corresponds to 7,258 [95% CI 7,160-7,336] cases with symptom onset during this period, a likely underestimation of the real number. Using age-stratified transmission models parameterized to behavioral data and calibrated to regional hospital admissions, we estimated that 69,115 [58,072-77,449] COVID-19 symptomatic cases occurred, suggesting that 9 out of 10 cases with symptoms were not ascertained. Median detection rate increased from 7% [6-9]% to 31% [28-35]% over time, with regional estimates varying from 11% (Grand Est) to 78% (Normandy) by the end of June. Healthcare-seeking behavior in COVID-19 suspect cases remained low (31%) throughout the period. Model projections for the incidence of symptomatic cases (4.5 [3.9-5.0] per 100,000) were compatible with estimates integrating participatory and virological surveillance data, assuming all suspect cases consulted. Encouraging healthcare-seeking behavior and awareness in suspect cases is critical to improve detection. Substantially more aggressive and efficient testing with easier access is required to act as a pandemic-fighting tool. These elements should be considered in light of the currently observed resurgence of cases in France and other European countries.

As countries in Western Europe gradually relaxed lockdown restrictions, robust surveillance and detection systems became critical to monitor the epidemic situation and maintain activity at low levels 1 . The need is to rapidly identify and isolate cases to prevent onward transmission in the community and avoid substantial resurgence of cases. In France, the surveillance strategy implemented by authorities to exit lockdown on May 11, 2020 was multifold 2,3 and based on an expanded case definition for COVID-19 suspect cases to guide clinical diagnosis 4 ; recommendations to the general population to seek healthcare even in presence of mild symptoms; prescription of diagnostic tests to suspect cases by general practitioners for systematic and comprehensive testing; isolation of confirmed cases and tracing of their contacts.

The specific characteristics of COVID-19 epidemic, however, hinder the identification of cases 5 . Large proportions of asymptomatic infectious individuals 6 , and presence of mild or paucisymptomatic infections that easily go unobserved 7, 8 present serious challenges to detection and control [8] [9] [10] . This may potentially result in substantial underestimates of the real number of COVID-19 cases in the country. Here we estimated the rate of detection of COVID-19 symptomatic cases in France in May-June 2020 after lockdown, through the use of virological and participatory syndromic surveillance data coupled with mathematical transmission models calibrated to regional hospitalizations. The study focused on mainland France where the epidemic situation was comparable across regions, and excluded Corsica reporting a very limited epidemic activity and overseas territories characterized by increasing transmission 11 .

COVID-19 epidemic management in France in the post-lockdown phase involved the creation of a centralized database collecting data on virological testing (SI-DEP, Information system for testing) to provide indicators to monitor the epidemic over time 2, 12 . 20,736 virologically-confirmed cases were reported from May 13 (week 20) to June 28 (week 26) in mainland France. After imputation of missing data (see Methods), an estimated 9,326 [95%CI 9,234-9,403] cases with symptoms resulted in the study period (Figure 1) . The average delay from symptom onset to testing decreased from 20.7 days in week 20 (w20) to 7.1 days in w26. Accounting for this delay (see Methods), we estimated that 7,258 [95% CI 7,160-7,336] confirmed symptomatic cases had onset in the study period, showing a decreasing trend over time (1, 663 in w20, 892 in w26). The test positivity rate decreased in the first weeks and stabilized around 1.2% (average over w24-w26). A digital participatory system was additionally considered for COVID-19 syndromic surveillance in the general population 11 , including those who do not consult a doctor. Called COVIDnet.fr, it was adapted from the platform GrippeNet.fr (dedicated to influenza-like-illness surveillance since 2011 13, 14 ) to respond to the COVID-19 health crisis in early 2020. It is based on a set of volunteers who weekly self-declare their symptoms, along with socio-demographic information. Based on symptoms declared by approximately 7,500 participants each week, the estimated incidence of COVID-19 suspect cases decreased from about 1% to 0.8% over time (Figure 1) , according to the expanded suspect case definition recommended by the High Council of Public Health for testing 4 (Methods). 162 out of 524 suspect cases (31%) consulted a doctor in the study period. Among them, 89 (55%) received a prescription for a test, resulting in screening for 50 individuals (56% of those given the prescription). Week of symptom onset for symptomatic confirmed cases was estimated based on patients' declarations (see panel b) through a Gamma distribution fitted to the data with a maximum likelihood approach. Missing data about presence/absence of symptoms and declaration of onset were imputed by region and by week, by sampling from a multinomial distribution according to the observed breakdown among cases with complete information (see Methods). Test positivity rate was computed on cases with complete information. Data for weeks 20-26 were consolidated in w27. (b) Breakdown of virologically-confirmed symptomatic cases by week of testing according to declared onset of symptoms, along with estimated mean time from onset to testing. (c) Incidence of COVID-19 suspect cases (estimates by week and 3-week moving average (thick line)), along with percentage of those seeking healthcare, estimated from participatory surveillance system COVIDnet.fr. (d) Number of COVID-19 suspect cases of the participatory cohort seeking healthcare, and among them those receiving a prescription, and performing a virological test given the prescription. COVIDnet.fr estimates were adjusted on age and sex of participants. (e) Estimated change in individuals attending their workplace locations over time and by region based on Google location history data 15 

We used stochastic discrete age-stratified epidemic models 17,18 based on demography, age profile 19 , and social contact data 20 of the 12 regions of mainland France, to account for age-specific contact activity and role in COVID-19 transmission. Disease progression is specific to COVID-19 17, 18 and parameterized with current knowledge to include presymptomatic transmission 21 , asymptomatic 6 and symptomatic infections with different degrees of severity (paucisymptomatic, with mild symptoms, with severe symptoms requiring hospitalization) 8, 22 . The model was shown to capture the transmission dynamics of the epidemic in Île-de-France and was used to assess the impact of lockdown and exit strategies 17, 18 . Full details are reported in the Methods section.

Intervention measures were modeled as modifications of the contact matrices, accounting for a reduction of the number of contacts engaged in specific settings, and were informed from empirical data. Lockdown data came from Refs. 17, 18 . The exit phase was modeled considering region-specific attendance at school based on . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted August 12, 2020. . https://doi.org/10.1101/2020.08. 10.20171744 doi: medRxiv preprint Ministry of Education's data 23 , partial maintenance of telework according to estimated presence in workplaces from mobile phones location history data 15 (Figure 1) , reduction in adoption of physical distancing over time based on survey data 16 (Figure 1) , partial reopening of activities, senior protection 17 . A sensitivity analysis was performed on the reopening of activities and senior protection, as data were missing for an accurate parameterization. Testing and isolation of detected cases were implemented by considering a 90% reduction of contacts for the number of virologically-confirmed COVID-19 cases 17, 18 . Region-specific models were calibrated to regional hospital admission data (Figure 2 ) through a maximum likelihood approach in the phase before lockdown, during lockdown, and in the exit phase. Further details are reported in the Methods section. (a-c) Hospital admissions over time, data (points) and simulations (median and 95% probability range), for Île-de-France (a), Pays de la Loire (b), Normandie (c). Hospital admission data up to w27 (consolidated in w28) were used to calibrate the models. (d-f) Projected number of new symptomatic cases over time (median and 95% probability range) and estimated number of virologically-confirmed symptomatic cases by week of onset (points), for the same regions above. The estimated detection probability of symptomatic cases (%) is also shown (red points, median and 95% probability ranges, right y axis).

Projected number of cases decreased over time in all regions, in agreement with the decreasing tendency reported in hospital admissions in the study period (Figure 2) . Overall, 69,115 [58,072-77,449, 95% probability range] new infections were predicted in mainland France in weeks 20-26 (from 22,882 [18, 221 ] in w20 to 2,922 [2,530-3,248] in w26). Île-de-France was the region with the largest predicted number of cases (8,126 [4,848-10,305 ] to 944 [712-1,088] from w20 to w26), followed by Grand Est and Hauts-de-France (Table 1) .

Projections were substantially higher than virologically-confirmed cases (Figures 2 and 3) . The estimated detection rate for symptomatic infections in mainland France in the period w20-w26 was 11% [9-13%], suggesting that 9 out of 10 new cases with symptoms were not identified by the surveillance system. Estimated detection rate increased over time (7% [6] [7] [8] [9] % in w20, 31% [28] [29] [30] [31] [32] [33] [34] [35] % in w26). By the end of June,

. CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted August 12, 2020. . https://doi.org/10.1101/2020.08.10.20171744 doi: medRxiv preprint 9 regions had a median detection above or equal 25% (Figure 3) , and 2 regions detected a number of cases within the probability ranges of model projections ( Table 1) . All regions except Pays de la Loire displayed increasing trends in the estimated detection rate.

We compared the projected incidence of COVID-19 symptomatic cases in w26 (4.5 [3.9-5.0] per 100,000) with the value obtained from confirmed cases (1.38 per 100,000) and two estimates based on COVIDnet.fr data (Figure 3 ). The first estimate applies the measured test positivity rate to the number of self-reported COVID-19 suspect cases (estimate #1, yielding 8.6 [6.2-11.5] per 100,000); the second additionally assumes that only 55% would be confirmed as suspect case by a physician and prescribed a test (according to COVIDnet.fr data on consulting participants, estimate #2, yielding 4.7 [3.4-6.3] per 100,000). Our projections are in line with plausible estimates from COVIDnet.fr.

Sensitivity analysis showed that findings were robust against elements of the contact matrices that could not be informed by empirical data, and against current epidemiological uncertainties. Including in the analysis also asymptomatic cases led to higher detection rates, 43% [39] [40] [41] [42] [43] [44] [45] [46] [47] [48] [49] [50] % in w26 compared to 31% [28] [29] [30] [31] [32] [33] [34] [35] % for symptomatic cases only. This however assumes that asymptomatic cases were detected by the virological surveillance system in the week of infection, as no additional information was available to adjust for the possible delay. . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted August 12, 2020. . https://doi.org/10.1101/2020.08.10.20171744 doi: medRxiv preprint Table 1 . Number of virologically-confirmed symptomatic cases, number of projected symptomatic cases, estimated detection rate, estimated trend in detection rate, population per region. Regions are ranked by decreasing number of confirmed cases in w20. The trend is estimated comparing the average of the estimated detection rate in the weeks of June (w23-26) with the average in the weeks of May (w20-w22). 

Despite a test positivity rate in mainland France well below WHO recommendations (5%) 24 , a substantial proportion of symptomatic cases (9 out of 10) remained undetected in the first 7 weeks following the end of lockdown. More than 60,000 symptomatic infections were not ascertained by the surveillance system from May 11 to June 28, 2020, according to our estimates.

Surveillance improved substantially over time. Detection rate was estimated to be 7% [6] [7] [8] [9] % at the national level in mid-May, in line with estimates for the same period from a seroprevalence study in Geneva, Switzerland 25 . By the end of June, it increased to 31% [28] [29] [30] [31] [32] [33] [34] [35] %, leaving about 2/3 of cases with symptoms undetected. Two regions (Occitanie, Normandy) reported cases compatible with model projections. These figures suggest that the new surveillance framework was progressively strengthened with increasing resources and capacity over time. Detection became also faster, with a 66% reduction of the delay from symptom onset to testing. At the same time, increasing performance benefited from a concurrent decrease of the epidemic activity in all regions.

Despite this positive trend, our findings highlight a critical need for improvement. Some regions remained with limited diagnostic exhaustiveness by the end of June. This is particularly concerning in those regions predicted to have a large number of weekly infections, such as Île-de-France where approximately only 3 out of 10 cases with symptoms were detected by the end of June, and Grand Est (1 out of 10). Novel recommendations since the end of lockdown require that all patients with symptoms suggestive of COVID-19 (as well as contacts of a confirmed case) be screened for SARS-CoV-2 2 . Almost all cases (92% since May 25) clinically diagnosed by sentinel general practitioners as possible COVID-19 cases were prescribed a test 11 . However, only 31% of individuals with COVID-19-like symptoms consulted a doctor in the study period according to participatory surveillance data. Overall, these figures suggest that a large number of symptomatic COVID-19 cases were not screened because they did not seek medical care despite recommendations. A similar evidence emerged from a large-scale serological study in Spain where only . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted August 12, 2020. . https://doi.org/10.1101/2020.08.10.20171744 doi: medRxiv preprint between 16% and 20% of symptomatic participants with antibodies against SARS-CoV-2 reported a previous virological screening 26 . By combining estimates from virological and participatory surveillance, together with measured rates for test recommendations by general practitioners (e.g., due to misclassification of selfreported symptoms), we extrapolated an incidence of symptomatic cases from crowdsourced data that is compatible with model projections. This finding further supports consultation for all COVID-19 suspect cases. Large-scale communication campaigns should reinforce recommendations to raise awareness in the population and strongly encourage healthcare-seeking behavior especially in patients with mild symptoms. At the same time, investigations to identify reasons for not consulting could be quickly performed through the participatory surveillance system.

Red tape might have contributed to low testing rates. Prescription for a test was deemed compulsory in the new testing policy to prevent misuse of diagnostic resources 2 , however the path involving consultation, prescription, and lab appointment may have discouraged mildly affected individuals not requiring medical assistance. To facilitate access, some local initiatives emerged recently that increase the number of drivethrough testing facilities, mail test vouchers to promote massive screening in certain regions (e.g. in Île-de-France 27 ), offer temporary mobile testing facilities (buses, pavilions) to increase proximity with the population 28 . These initiatives are particularly relevant for counteracting socio-economic inequalities in access to information and care in populations vulnerable to COVID-19 29 and may be established in the longterm. Given the non-uniform detection rate estimated within the country, learning from specific regional realities may aid to improve detection. The recent change in screening policy no longer requiring a prescription for testing 30 could further improve access.

Screening rates remained overall well below the objective fixed by authorities for the post-lockdown phase (average weekly number of tests in May-June was 250,000 vs. target of 700,000), and the delay from onset to screening was still very long (7 days) by the end of June, despite substantial reduction over time. The large demand for testing currently observed in certain regions, mainly as a result of imminent travels and protocols imposed by certain countries and air companies, is reportedly causing long waiting lists at overwhelmed testing sites 31 . Given pre-symptomatic transmission, notification to contacts should be almost immediate to allow the effective interruption of transmission chains 21 . For testing to be an actionable tool for surveillance and, most importantly, for control of COVID-19 transmission, screening rates should be radically increased and delays suppressed. The risk would otherwise be a rapid and uncontrolled resurgence of cases with potential transmission in the community 10 , as currently reported in some French areas (e.g. Mayenne district in Pays de la Loire region) 32 and countries in Europe 33 . Such risk is expected to increase if the reported relaxation in preventive behaviors persists, due to adhesion fatigue 16 .

Aggressive and efficient testing will become increasingly more important in the Fall months, as other respiratory viruses, such as influenza, RSV, rhinoviruses, will start to circulate and influenza-like-illness incidence levels will become comparable with those of COVID-19. Reviewing the testing strategy over Summer, while at low COVID-19 epidemic activity, is an important opportunity to strengthen French response system for next season.

Models were region-based and did not consider a possible coupling between regional epidemics caused by mobility. This choice was supported by stringent movement restrictions during lockdown 34 , and by the limited mobility increase in May-June 35 , before important inter-regional displacements took place at the start of summer holidays in July. Foreign importations were neglected 9,36 as France reopened its borders with EU Member States on June 15, and the Schengen area remained closed till July. COVIDnet.fr cohort is not representative of the general population 14 , however the agreement found with sentinel incidence trends for influenza-like-illness suggests that these limitations have little effect once results are adjusted for lack of representativeness 13 . Underdetection may also proceed from the imperfect characteristics of RT-PCR (reverse transcription-polymerase chain reaction) tests used to identify infected cases 37 . Some cases tested for SARS-. CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted August 12, 2020. . https://doi.org/10.1101/2020.08.10.20171744 doi: medRxiv preprint

CoV-2 could have been falsely negative. This would affect the analysis presented in the manuscript and would be in line with our conclusion that a large part of cases may have been undetected. Asymptomatic infections were not considered in the main analysis, as we lacked information on the likely time of infection. The median duration from first to last positive nasopharyngeal swab was estimated to be 19 days in asymptomatic patients in China, with the longest duration at 45 days 38 . No such analysis has been performed in France yet. Assuming that asymptomatic infections were rapidly identified through contact tracing yielded higher detection rates than estimated for symptomatic cases only. This is due to a proportionally higher fraction of asymptomatic cases among the confirmed ones. Though limited by the underlying assumption, this result further strengthens the main conclusion that detection of symptomatic index cases is the key aspect that requires fundamental improvement.

Our findings identify critical needs of improvement to increase case ascertainment in France and the performance of the response system to monitor and control COVID-19 epidemic. Substantially more aggressive and efficient testing needs to be achieved to act as a pandemic-fighting tool. These elements should be considered in light of the resurgence of cases currently observed in some regions in France and in other countries with similar response systems.

Virological surveillance data and analysis. The centralized database SI-DEP for virological surveillance 12 collects detailed information on patients tested in France, including (i) date of test, (ii) result of test (positive or negative), (iii) location (region), (iv) absence or presence of symptoms, (v) self-declared delay between onset to test in presence of symptoms. The delay is provided with the following breakdown: onset date occurring 0-1 day before date of test, 2-4 days before, 5-7 days before, 8-15 days before, or >15 days before. The SI-DEP database provided complete information for 13,887 (62%) out of 23,053 laboratory-confirmed COVID-19 cases tested between week 20 (May 11-May 17) and week 27 (June 29-July 5), with an increasing trend of complete information over time. The study referred to the period from w20 to w26; data of w27 were used to account for the delays. Data were consolidated in w27.

For cases with complete information, we estimated the number of symptomatic laboratory-confirmed COVID-19 cases by date of onset, using the information on the date of test and the time-interval of onset-to-test delay declared by the patient. We fitted a Gamma distribution to these data with a maximum likelihood approach, obtaining a shape parameter equal to 0.64, and expected value of delay equal to 8 days. Given a symptomatic confirmed case tested on a specific date, we assigned the onset date by sampling the onset-to-testing delay from the fitted distribution, conditional to the fact that the delay lies in the corresponding time-interval declared by the patient. To account for the changes in the distribution of self-declared delays over time, we also fitted the distribution to three periods of time, obtaining no significant difference.

Cases with missing data were imputed by sampling from a multinomial distribution with probabilities equal to the rate of occurrence of the labels reported for cases with complete information. Imputation was performed by region and by week. Onset date was then estimated for the imputed symptomatic cases.

Participatory surveillance data and analysis. COVIDnet.fr is a participatory online system for the surveillance of COVID-19, available at www.covidnet.fr. It was adapted from GrippeNet.fr to respond to the COVID-19 health crisis in March 2020. GrippeNet.fr is a participatory system for the surveillance of influenzalike-illness available in France since 2011 through a collaboration between Inserm, Sorbonne Universite and Sante publique France 13, 14, 39 , supplementing sentinel surveillance. The system is based on a dedicated website to conduct syndromic surveillance through self-reported symptoms volunteered by participants resident in France. Data are collected on a weekly basis; participants also provide detailed profile information at enrollment. In addition to tracking influenza-like-illness incidence 13, 39 , GrippeNet.fr was used to estimate . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted August 12, 2020. . https://doi.org/10.1101/2020.08.10.20171744 doi: medRxiv preprint vaccine coverage in specific subgroups 40 individual perceptions toward vaccination 41 and health-seeking behavior 42 . It was also used to assess behaviors and perceptions related to other diseases beyond influenza 43 .

Participants are on average older and include a larger proportion of women compared to the general population 14, 44 . Participating population is however representative in terms of health indicators such as diabetes and asthma conditions. Despite these discrepancies, trends of estimated influenza-like-illness incidence from GrippeNet.fr reports compared well with those of the national sentinel system 13, 39 . All analyses were adjusted by age and sex of participants.

To monitor COVID-19 suspect cases in the general population, we used the expanded case definition recommended by the High Council of Public Health for systematic testing and described in their 20 April 2020 notice 4 :

• (Sudden onset of symptoms OR sudden onset of fever) AND (fever OR chills) AND (cough OR shortness of breath OR (chest pain AND age > 5 years old))

• OR (Sudden onset of symptoms OR (sudden onset of fever AND fever)) AND o (age > 5 years old AND (feeling tired or exhausted OR muscle/joint pain OR headache OR (loss of smell WITHOUT runny/blocked nose) OR loss of taste)

o OR ((Age ≥ 80 years old OR Age < 18 years old) AND diarrhea) o OR (Age < 3 months old AND (fever WITHOUT other symptoms))). Figure 3 reports two estimates obtained from COVIDnet.fr cohort data for the incidence of symptomatic cases in w26. They are computed as follows:

• Estimate #1 = (COVIDnet.fr estimated incidence in w26) * (test positivity rate from SI-DEP in w26)

• Estimate #2 = (COVIDnet.fr estimated incidence in w26) * (estimated proportion screened and confirmed as COVID-19 suspect case by a physician, and prescribed a test; estimates from COVIDnet.fr) * (test positivity rate from SI-DEP in w26)

Ethics statement. GrippeNet.fr/COVIDnet.fr was reviewed and approved by the French Advisory Committee for research on information treatment in the health sector (i.e. CCTIRS, authorization 11.565), and by the French National Commission on Informatics and Liberty (i.e. CNIL, authorization DR-2012-024) -the authorities ruling on all matters related to ethics, data, and privacy in the country.

Transmission models summary. We used a stochastic discrete age-structured epidemic model for each region based on demographic, contact 20 , and age profile data of French regions 19 . Four age classes were considered: [0-11), [11] [12] [13] [14] [15] [16] [17] [18] [19] , , and 65+ years old. Transmission dynamics follows a compartmental scheme specific for COVID-19, where individuals were divided into susceptible, exposed, infectious, and hospitalized. We did not consider further progression from hospitalization (e.g. admission to ICU, recovery, death 17 ) as it was not needed for the objective of the study. The infectious phase is divided into two steps: a prodromic phase ( ) and a phase where individuals may remain either asymptomatic ( , with probability =40% 6 ) or develop symptoms. In the latter case, we distinguished between different degrees of severity of symptoms, ranging from paucisymptomatic ( ), to infectious individuals with mild ( ) or severe ( ) symptoms. Prodromic, asymptomatic and paucisymptomatic individuals have a reduced transmissibility = 0.55, as estimated in Ref. 7 . A reduced susceptibility was considered for younger children and adolescents, along with a reduced relative transmissibility of younger children, following available evidence from household studies, contact tracing investigations, and modeling works [45] [46] [47] [48] [49] [50] . A sensitivity analysis was performed on relative susceptibility and transmissibility of younger children, and on the proportion of asymptomatic infections. Models calibration. Models were calibrated regionally to daily hospital admission data through a maximum likelihood approach. The likelihood function is of the form . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted August 12, 2020. is the time window considered for the fit. Calibration involved three steps, each one corresponding to a different epidemic situation: pre-lockdown 17 , during lockdown 17 , post-lockdown.

COVID-19) in the EU/EEA and the UK -tenth update

Prise en charge par les médicins de ville des patients atteints de COVID-19 en phase de déconfinement

Testing for COVID-19: A way to lift confinement restrictions

Signes cliniques d'orientation diagnostique du Covid-19

Factors that make an infectious disease outbreak controllable

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

Substantial undocumented infection facilitates the rapid dissemination of novel coronavirus (SARS-CoV2)

Estimates of the severity of coronavirus disease 2019: a model-based analysis

Tracing and analysis of 288 early SARS-CoV-2 infections outside China: A modeling study

Reconstruction of the full transmission dynamics of COVID-19 in Wuhan

Données relatives aux résultats des tests virologiques COVID-19

The potential value of crowdsourced surveillance systems in supplementing sentinel influenza networks: the case of France

Evaluating the Feasibility and Participants' Representativeness of an Online Nationwide Surveillance System for Influenza in France

COVID-19 Community Mobil

Covid-19 : une enquête pour suivre l'évolution des comportements et de la santé mentale pendant l'épidémie. /etudes-et-enquetes/covid-19-une-enquête-pour-suivre-l'évolution-des-comportements-etde-la-sante-mentale-pendant-l-epidemie

Impact of lockdown on COVID-19 epidemic in Île-de-France and possible exit strategies

Expected impact of reopening schools after lockdown on COVID-19 epidemic in Île-de-France

Pyramide des âges 2020 -France et France métropolitaine

The French Connection: The First Large Population-Based Contact Survey in France Relevant for the Spread of Infectious Diseases

Quantifying SARS-CoV-2 transmission suggests epidemic control with digital contact tracing

Déconfinement phase 2 : point de situation au 28 mai. Ministère Educ. Natl

Public health criteria to adjust public health and social measures in the context of COVID-19

Seroprevalence of anti-SARS-CoV-2 IgG antibodies in Geneva, Switzerland (SEROCoV-POP): a population-based study

Prevalence of SARS-CoV-2 in Spain (ENE-COVID): a nationwide, population-based seroepidemiological study

Covid-19 : une campagne de tests à grande échelle débute dans 32 communes d'Ile-de-France

Covid-19 : dépistage gratuit dans un labo itinérant

Coronavirus highlights inequality in France's poorest department

Covid-19 : les tests de dépistages sont possibles sans prescription médicale

Tests Covid-19 : les délais d'attente s'allongent dans les labos surchargés. LCI

COVID-19 country overviews-Week 28

Population mobility reductions during COVID-19 epidemic in France under lockdown

Mobility patterns in Paris before, during, and after lockdown

Assessing the impact of coordinated COVID-19 exit strategies across Europe

Pitfalls in SARS-CoV-2 PCR diagnostics

Clinical and immunological assessment of asymptomatic SARS-CoV-2 infections

Participatory Syndromic Surveillance of Influenza in Europe

Influenza during pregnancy: Incidence, vaccination coverage and attitudes toward vaccination in the French web-based cohort G-GrippeNet

Opinion about seasonal influenza vaccination among the general population 3 years after the A(H1N1)pdm2009 influenza pandemic

Healthcare-seeking behaviour in case of influenza-like illness in the French general population and factors associated with a GP consultation: an observational prospective study

Population perception of mandatory childhood vaccination programme before its implementation

The representativeness of a European multi-center network for influenza-like-illness participatory surveillance

Changes in contact patterns shape the dynamics of the COVID-19 outbreak in China

Age-dependent effects in the transmission and control of COVID-19 epidemics

Coronavirus Infections in Children Including COVID-19: An Overview of the Epidemiology, Clinical Features, Diagnosis, Treatment and Prevention Options in Children

A Case Series of children with 2019 novel coronavirus infection: clinical and epidemiological features

SARS-CoV-2 infection in primary schools in northern France: A retrospective cohort study in an area of high transmission

Cluster of COVID-19 in northern France: A retrospective closed cohort study. medRxiv

This study was partially supported by ANR project DATAREDUX (ANR-19-CE46-0008-03) and EVALCOVID-19 (ANR-20-706 COVI-0007); EU H2020 grants MOOD (H2020-874850) and RECOVER (H2020-101003589); REACTing COVID-19 modeling and surveillance grants. We thank Pascal Crepey, Camille Pelat, Edouard Chatignoux, Juliette Paireau, Daniel Levy-Bruhl for useful discussions. We also thank all participants of COVIDnet.fr system.