key: cord-345623-ba3we97o authors: BOELLE, Pierre-Yves title: Excess cases of Influenza like illnesses in France synchronous with COVID19 invasion. date: 2020-03-17 journal: nan DOI: 10.1101/2020.03.14.20035741 sha: doc_id: 345623 cord_uid: ba3we97o Several French regions where COVID19 has been reported currently show a renewed increase in ILI cases in the general practice based Sentinelles network. Here we computed the number of excess cases by region and found correlation with the number of reported COVID19 cases so far. These data suggest larger circulation of SARS-CoV-2 in the French population than apparent from confirmed cases. Influenza Like illnesses cases are defined as cases with fever of sudden onset (>39°C) with respiratory signs (cough, running nose) and myalgia of any age. Acute Respiratory Infection (AIR) include diseases with respiratory signs and are only monitored in the > 65 y.o. Cases are reported in real time by participating GPs (approx. 600 GPs) all over France. Additionally, Sentinelles GPs may swab 1 ILI case and up to 2 IRA cases per week for viral characterization (3) . Influenza virus, respiratory syncytial virus, human rhinovirus and human metapneumovirus are routinely looked for, and SARS-CoV-2 has been added since February 2020. We computed the expected number of ILI consultations using the superposition of a seasonal (4) and an influenza epidemic component (5) , as detailed in the appendix. Excess cases were computed as the difference between observed cases and expected numbers. The rate of increase in excess cases was determined, assuming exponential growth. Confirmed COVID19 cases were obtained from the santé Publique France website, March 7 Bulletin(6). In week 10 of 2020, 93 swabs were collected in ILI patients and 23 in ARI patients. Two (2) swabs in each category were positive for SARS-CoV-2. The 2 positive swabs in ARI patients had been collected in In Bourgogne-Franche Comté (BFC) and Grand Est (GRE) regions, out of a total of 7 swabs in these 2 regions. Assuming a positivity rate for SARS-CoV-2 of 2/7 in these regions, we estimated that 760 (CI95% [219, 1706]) ARI consultations in the > 65 y.o. could have been caused by COVID19 during week 10 (2/7 of 2600 ARI visits). One of the positive ILI case was linked to an existing cluster in the east of France. . 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 17, 2020. . The modelling approach allowed computing excess cases in 11 out of 13 regions. The model did not provide a good fit in the remaining 2 regions (see appendix). The overall ILI incidence showed renewed increase with 33 [-8, 64] consultations/100000 in excess during week 9 and 84 [447, 108] consultations/100000 in excess in week 10. This changed with regions: 4/11 regions displayed positive excess (credible interval excluding 0) in week 9 and 7/11 regions in week 10. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint Several processes may be at play in this observation, including at least characteristics of the influenza season, change in population behaviour or increase in COVID19 incidence. H1N1pdm and B influenza viruses co-circulated in 2019-20 in France. Seasons with the type B virus can lead to prolonged periods for decrease in incidence but a renewed increase has never occurred before (1). The school holidays occurred gradually in France during the end of February and early March. While this may change the dynamics of influenza (7), the effect is generally notable in the ascending phase of the epidemic rather than after the peak. The current situation with COVID19 may change the health-seeking behaviour of patients and, to some extent, of the contributing general practitioners although they conform to a case definition. Preliminary assessment of behavioural data in crowdsourced surveillance system is under way (grippenet.fr(8)) but, as of today, no massive trend towards more consultations has been observed. As information regarding the coronavirus risk is widespread, a uniform increase over all regions could have been expected in this scenario, but we acknowledge that it may have led to increased consultation rates in regions where COVID19 is the most reported. 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 17, 2020. . https://doi.org/10.1101/2020.03.14.20035741 doi: medRxiv preprint The overlap between COVID19 and flu symptoms makes it possible that the excess of ILI cases is due to COVID19 cases, and the presence of SARS-CoV-2 positive swabs in the patients demonstrate this possibility. The correlation between the number of confirmed cases and the computed excess is also consistent with this scenario. In this case, the excess in cases is compatible with an exponential growth with rate estimated at approximately 0.7/week. Interestingly, this estimate is very similar from one region to the next in the four most affected regions (GRE, IDF, OCC, HDF). This estimate may prove useful to help calibrate models for studying the impact of the COVID19 pandemic. Estimating the true number of COVID19 cases remains difficult in the presence of these uncertainties. We note that the extrapolated number of COVID19 presenting as ARI was 760 in the >65 y.o. in the 2 eastern regions (GRE and BFC), which was at least twice more than the number of reported cases at the time (250+129), all ages confounded. As we enter a period of generalized circulation of the SARS-CoV-2 virus, surveillance based on clinical description and swabbing by GPs will prove essential to help us assess the situation. The number of expected ILI consultations reported to the Sentinelles network was modelled as the superposition of 2 components: an "epidemic" component due to influenza infections and a background seasonal signal not due to influenza (4) . More precisely, we fitted the influenza epidemic part using Richard's model (5) . Richard's model is a 4 parameters equation describing the cumulated number of cases at time t as follows : The seasonal part was modelled as In a second step, we fitted excess cases by an exponential increase ( ) = 0 exp ( ), and assumed that number of ILI cases for weeks 8 to 10 was Poisson distributed with mean ( ) + ( ) + ( ). We tied the number of excess cases at 0 in week 1 of 2020 for this estimation. Estimations were performed with STAN. Code is available upon request. Example data used for estimating excess cases in a French region. The orange points were used to estimate the seasonal part; the first peak in the black part was fit using Richard's model plus the seasonal part, and excess was computed over the last 2 weeks as indicated in figure S3 . 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 17, 2020. . https://doi.org/10.1101/2020.03.14.20035741 doi: medRxiv preprint 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 17, 2020. . https://doi.org/10.1101/2020.03.14.20035741 doi: medRxiv preprint Influenza epidemics observed in primary care from 1984 to 2017 in France: A decrease in epidemic size over time. Influenza Other Respir Viruses First cases of coronavirus disease 2019 (COVID-19) in France: surveillance, investigations and control measures Baseline characteristics and clinical symptoms related to respiratory viruses identified among patients presenting with influenza-like illness in primary care A routine tool for detection and assessment of epidemics of influenza-like syndromes in France Early Outbreak of 2009 Influenza A (H1N1) in Mexico Prior to Identification of pH1N1 Virus COVID-19 : point épidémiologique du 7 mars 2020 Estimating the impact of school closure on influenza transmission from Sentinel data Participatory Syndromic Surveillance of Influenza in Europe the Sentinelles syndromic and viral and surveillance group" including (but not limited to) Cécile Souty 1 , Titouan Launay 1 , Clément Turbelin 1 , Caroline Guerrisi 1 , Chiara Poletto 2 , Vittoria Colizza 2 , Sylvie van der Werf 4 5 -Institut Pasteur, Centre Coordonnateur du Centre National de Référence des virus des infections respiratoires (dont la grippe) Sorbonne Paris Cité, Unité de Génétique Moléculaire des Virus à ARN