key: cord-313193-q5zeoqlb
authors: Carrat, F.; de Lamballerie, X.; Rahib, D.; Blanche, H.; Lapidus, N.; Artaud, F.; Kab, S.; Renuy, A.; Szabo de Edelenyi, F.; Meyer, L.; Lydie, N.; Charles, M.-A.; Ancel, P.-Y.; Jusot, F.; Rouquette, A.; Priet, S.; Saba Villaroel, P. M.; Fourie, T.; Lusivika-Nzinga, C.; Nicol, J.; Legot, S.; Druesne-Pecollo, N.; Essedik, Y.; Lai, C.; Gagliolo, J.-M.; Deleuze, J.-F.; Bajos, N.; Severi, G.; Touvier, M.; Zins, M.
title: Seroprevalence of SARS-CoV-2 among adults in three regions of France following the lockdown and associated risk factors: a multicohort study.
date: 2020-09-18
journal: nan
DOI: 10.1101/2020.09.16.20195693
sha: 
doc_id: 313193
cord_uid: q5zeoqlb

Aim To estimate the seroprevalence of SARS-CoV-2 infection in May-June 2020 after the lockdown in adults living in three regions in France and to identify the associated risk factors. Methods Participants in a survey on COVID-19 from an existing consortium of three general adult population cohorts living in the Ile-de-France (IDF) or Grand Est (GE), two regions with high rate of COVID-19, or in the Nouvelle-Aquitaine (NA), with a low rate, were asked to take a dried-blood spot (DBS) for anti-SARS-CoV-2 antibodies assessment. The primary outcome was a positive anti-SARS-CoV-2 ELISA IgG result against the spike protein of the virus (ELISA-S). The secondary outcomes were a positive ELISA IgG against the nucleocapsid protein (ELISA-NP), anti-SARS-CoV-2 neutralizing antibodies titers >=40 (SN), and predicted positivity obtained from a multiple imputation model (MI). Prevalence estimates were adjusted using sampling weights and post-stratification methods. Findings Between May 4, 2020 and June 23, 2020, 16,000 participants were asked to provide DBS, and 14,628 were included in the analysis, 983 with a positive ELISA-S, 511 with a positive ELISA-NP, 424 with SN>=40 and 941 (Standard Deviation=31) with a positive MI. Adjusted estimates of seroprevalence (positive ELISA-S) were 10.0% (95%CI 9.1%;10.9%) in IDF, 9.0% (95%CI 7.7%; 10.2%) in GE and 3.1% (95%CI 2.4%; 3.7%), in NA. The adjusted prevalence of positive ELISA-NP, SN and MI were 5.7%, 5.0% and 10.0% in IDF, 6.0%, 4.3% and 8.6% in GE, and 0.6%, 1.3% and 2.5% in NA, respectively. A higher seroprevalence was observed in younger participants and when at least one child or adolescent lived in the same household. A lower seroprevalence was observed in smokers compared to non-smokers. Interpretation At the end of the lockdown the prevalence of anti-SARS-CoV-2 IgG or neutralizing antibodies remained low in the French adult population, even in regions with high reported rates of COVID-19.

To estimate the seroprevalence of SARS-CoV-2 infection in May-June 2020 after the lockdown in adults living in three regions in France and to identify the associated risk factors.

Participants in a survey on COVID-19 from an existing consortium of three general adult population cohorts living in the Ile-de-France (IDF) or Grand Est (GE) -two regions with high rate of COVID-19, or in the Nouvelle-Aquitaine (NA) -with a low rate, were asked to take a dried-blood spot (DBS) for anti-SARS-CoV-2 antibodies assessment. in IDF, 6.0%, 4.3% and 8.6% in GE, and 0.6%, 1.3% and 2.5% in NA, respectively. 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 September 18, 2020. . https://doi.org/10.1101/2020.09. 16.20195693 doi: medRxiv preprint

Serological surveys help determine the extent of infection by a viral agent in a population and identify associated risk factors. 1 In addition, characterizing the distribution of antibodies against this agent can help evaluate the portion of the population that is immunized to quantify herd immunity. However, despite the ongoing COVID-19 pandemic, there are still very few serologic surveys describing the factors associated with SARS-CoV-2 seroprevalence, and only one study explored the distribution of neutralizing antibodies against SARS-CoV-2 in a general adult population in a very low prevalence area. 2 Serologic surveys of SARS-CoV-2 have been performed between January 2020 and July 2020 in the general population in Iceland, 3 Switzerland, 4 Spain, 5 UK, 6, 7 Italy, 8

Belgium, 9 Germany, 2 China, 10 Brazil, 11 Canada, 12 and the US. 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 September 18, 2020. . https://doi.org/10.1101/2020.09. 16.20195693 doi: medRxiv preprint

The present report combined data collected from questionnaires in the SAPRIS ("SAnté, Perception, pratiques, Relations et Inégalités Sociales en population générale pendant la crise COVID-19") survey in France, with serological results from the SAPRIS-SERO study.

The SAPRIS survey has been described elsewhere. 16 Briefly, the survey was created in March 2020 to evaluate the main epidemiological, social and behavioral challenges of the SARS-CoV2 epidemic in France in relation to social inequalities in health and healthcare. It is based on a consortium of prospective cohort studies including three general population-based adult cohorts and two child-cohorts (not presented in this study). [17] [18] [19] All participants from the original cohorts with regular access to electronic (internet) questionnaires were invited to participate in the SAPRIS survey (supplementary figure 1). Two self-administered questionnaires covering the lockdown and the post-lockdown periods were sent as of April 1, 2020 and returned before May 27, 2020. Variables collected in the questionnaires included socio-demographics, household size and composition, COVID-19 diagnosis, SARS-CoV-2 RT-PCR test, a detailed description of the subject's symptoms in the two weeks before each questionnaire, comorbidities, healthcare use and treatment, employment, daily life, child care, alcohol, tobacco and cannabis use, social and sexual life, preventive measures, risk perception and beliefs.

The goal of the SAPRIS-SERO study (#NCT04392388) including participants enrolled in the SAPRIS survey, was to quantify and follow the cumulative incidence of SARS-CoV-2 infection in the French population using serological tests and to assess the determinants of infection. Self-sampling dried-blood spot (DBS) kits were . 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 Eluates were processed with a commercial Elisa test (Euroimmun®, Lübeck, Germany) to detect anti-SARS-CoV-2 antibodies (IgG) directed against the S1 domain of the spike protein of the virus (ELISA-S). The volume of eluate used corresponded to the amount of serum and dilution recommended in the manufacturer's instructions. All samples with an ELISA-S test optical density ratio ≥ 0.7 were also tested with an ELISA test to detect IgG antibodies against the SARS-CoV-2 nucleocapsid protein (Euroimmun®, Lübeck, Germany, ELISA-NP) and with an in-house micro-neutralization assay to detect neutralizing anti-SARS-CoV-2 antibodies (SN), as described elsewhere. 20 Briefly, we used VeroE6 cells cultured in 96-well microplates, 100 TCID50 of the SARS-CoV-2 strain BavPat1 (courtesy of Pr. 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 September 18, 2020. . https://doi.org/10.1101/2020.09. 16.20195693 doi: medRxiv preprint We randomly selected 16,000 of the participants of the SAPRIS survey for this study who agreed to be tested and who were residents from one of the three French administrative regions: Ile-de-France (IDF) or Grand Est (GE), i.e. the two regions with the highest reported cumulated rates of COVID-19 at the end of the lockdown period, or Nouvelle-Aquitaine (NA), a region with a low reported rate. 21 Ethical approval and written or electronic informed consent were obtained from each participant before enrolment in the original cohort. The SAPRIS survey was approved by the Inserm ethics committee (approval #20-672 dated March 30, 2020). The SAPRIS-SERO study was approved by the Sud-Mediterranée III ethics committee (approval #20.04.22.74247) and electronic informed consent was obtained from all participants for DBS testing.

The main outcome was a positive ELISA-S test. In accordance with the manufacturer's instructions a test was considered to be ELISA-positive with an optical density ratio ≥ 1.1, ELISA-indeterminate between 0.8 and 1.1, and ELISAnegative, <0.8. The secondary outcomes were a positive ELISA-NP (using the same thresholds) and positive SN defined as a titer ≥40. Because test sensitivity and specificity was not 100%, we also used a multiple imputation (MI) method to estimate a participant's positivity in which the likelihood of positivity was based on observed test results and covariates.

The association of seroprevalence was evaluated in relation to age, gender, sociodemographic characteristics, BMI, chronic conditions (according to a pre-specified 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 September 18, 2020. . https://doi.org/10.1101/2020.09.16.20195693 doi: medRxiv preprint list), tobacco and alcohol use before the lockdown. Age groups were categorized according to predefined limits (<40; 40-49; 50-59; 60-69; ≥70 years old) and BMI according to standard cut-offs (<18.5; 18.5-<25; ≥25-<30; ≥30 kg/m 2 ). 22 The association of seroprevalence was also studied in relation to symptoms.

Possible COVID-19 was defined according to the European Centre for Disease Prevention and Control as at least one of the following: cough, fever, dyspnea, and sudden anosmia, ageusia or dysgeusia. 23 Participants who did not report any of these symptoms on either questionnaire, did not have a positive COVID-19 diagnosis, or did not experience cough, fever or feverishness from the beginning of the year were considered to be asymptomatic.

Inverse probability weighting and generalized raking were used to estimate seroprevalence in the adult population. 24 Weights were estimated from each cohort source by logistic regression, with selection or participation as response variables and socio-demographic characteristics as covariates. An initial cohort-specific calibration was performed by generalized raking in relation to the marginal totals of the distribution of age class, gender and socio-professional category in the target population. The weights were then rescaled according to the relative sample size of each cohort, then recalibrated according to the same covariates to provide representative estimates of the adult population. This weighting procedure was performed for each region independently. Confidence intervals for weighted estimates were computed by bootstrapping.

To fit the MI model, participants with all three positive ELISA-S, ELISA-NP and SN test results were classified as "true" positives while those with all three negative . 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 September 18, 2020. . https://doi.org/10.1101/2020.09.16.20195693 doi: medRxiv preprint results or ELISA-S <0.7 were "true" negatives. The Markov Chain Monte Carlo method was used to imputing MI using numerical values from the three serological tests (log-transformed), region, age and gender. The MI was built from 100 imputed data sets and estimates combined with Rubin's rules. 25 Chi-Square test for trend was used on unweighted data to compare symptoms and health care use according to ELISA-S results. Logistic regression models were used on unweighted data with stratification in the source cohort to identify the determinants of a positive ELISA-S (primary outcome). Indeterminate ELISA-S results were grouped with negative results in the primary analysis. Multivariable analysis was performed including region, age, gender and all factors associated with seroprevalence in univariable analysis. A backward elimination procedure was used to identify independent covariates associated with a positive ELISA-S. Contact with a RT-PCR positive household member was not considered to prevent the risk of reverse causation. Multivariable analyses were repeated using secondary outcomes then performed in each region to identify any potential regional effect-modification.

Weighting and multiple imputation used the survey and mice package from R software version 3.6.3 (R Foundation for Statistical Computing, Vienna, Austria).

Other analyses were performed with SAS 9·4 software (SAS Institute Inc., Cary, North Carolina, USA). P<.05 was considered to be statistically significant.

The sponsor and funders facilitated data acquisition but did not participate in the study design, analysis, interpretation or drafting. FC had full access to all data in the study and FC, XL, NB, MT, GS, MZ made the final decision to submit the study for publication. 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 September 18, 2020. 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 September 18, 2020. 

Participants with a positive ELISA-S had a higher rate of self-reported symptoms than those with negative tests except for skin lesions (table 2) 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 September 18, 2020. 

On univariable analysis, the rate of positive ELISA-S was higher in IDF and GE than in NA (table 3) Multivariable analysis showed an independent and positive association between positive ELISA-S, IDF and GE compared to NA, for younger age, and at least one child or adolescent living in the same household (table 4). A negative association was found with active smoking (vs no smoking). The observed associations were confirmed with MI and were overall consistent with ELISA-NP and SN, although they did not all reach statistical significance due to a smaller number of events (supplementary tables 2-4). When multivariable analysis was performed in each region separately, the associations did not differ between IDF and GE but the pattern in NA was different from that in IDF or GE, with a higher Odds-Ratio (OR) in young age groups in the former (OR= 3. . 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 September 18, 2020. 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 September 18, 2020. 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 September 18, 2020. . https://doi.org/10.1101/2020.09.16.20195693 doi: medRxiv preprint A lower seroprevalence with increasing age was reported in several populationbased serological studies 4,6 and a higher rate of possible COVID-19 with decreasing age was described in an earlier study. 16 Although men are known to be at a higher risk of severe COVID-19, hospitalization and deaths than women, 28 we found an association between seroprevalence and female gender in NA, which was also reported in a recent Italian study. 8 This association was only found in the region with a lower prevalence and may be related to the specific dynamics of transmission in this area. Based on the estimated 5-day median COVID-19 incubation time and the appearance of symptoms within 12 days after infection, 29 participants who developed a possible COVID-19 before March 23 and tested positive were potentially infected before lockdown, probably in the workplace or in the community. This could explain why we did not find any specific association with social health inequalities, while, conversely, univariable analyses showed associations between seroprevalence and working adults with higher incomes and educational levels. As in other studies, univariable analysis identified the size of the household and the number of rooms, but only living with at least one child remained associated with seroprevalence on multivariable analysis, indicating that children could play an important role in household-related transmission. 30

Finally, active smoking was associated with a lower rate of ELISA-S or SN positive results. 7, 31 Smoking status was collected before the peak of the pandemic and thus could not have been affected by preventive behaviours in smokers. Although smoking is a risk factor for severe COVID-19 in infected patients, 32 its role in the risk of infection remains unclear because certain components of the smoke (such as nicotine) regulate ACE2 receptor expression which is involved in SARS-CoV-2 entry into cells. 33, 34 Smoking is also known to be associated with lower serum levels of IgG, 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 September 18, 2020. . https://doi.org/10.1101/2020.09.16.20195693 doi: medRxiv preprint IgA or IgM, 35 but this probably does not explain why smoking was also negatively correlated with SARS-CoV-2 RT-PCR positive results in several studies. 36, 37 Our study has several limitations. First, the primary endpoint is based on a test that does not have a 100% sensitivity and specificity. Thus, certain participants were probably misclassified. We used manufacturer-defined cutoff points for positivity, although the test performance can increased by using other positive and negative cut-off values. 38 However, prevalence correction using these reported test performances or by the manufacturer are not applicable to our study, since the use of 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 September 18, 2020. . https://doi.org/10.1101/2020.09. 16.20195693 doi: medRxiv preprint formally associate the self-reported COVID-19 symptoms with a positive serological result and once again, misclassification may have occurred.

This study has several strengths. In particular, it is based on well-characterized general population cohorts with a very high participation rate. Moreover, serological samples were collected within 1 to 3 months after the period of intense circulation of SARS-CoV-2 and all serological tests were centralized and performed blinded to participants' characteristics or clinical history. Several serological methods were combined, including neutralization, to improve the interpretation of seroprevalence results.

In conclusion, our study shows that the level of seroprevalence remains low in the 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 September 18, 2020. 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 September 18, 2020. 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 September 18, 2020. 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 September 18, 2020. 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 September 18, 2020. 

Prof Fabrice Carrat reports personal fees from Imaxio and Sanofi, outside the submitted work.

All other authors declare no competing interest.

This study ANR (Agence Nationale de la Recherche, #ANR-20-COVI-000, #ANR-10-COHO-06), Fondation pour la Recherche Médicale (#20RR052-00), Inserm (Institut National de la Santé et de la Recherche Médicale, #C20-26).

The CONSTANCES Cohort Study is supported by the Caisse Nationale d'Assurance 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 September 18, 2020. 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 September 18, 2020. 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 September 18, 2020. . 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 September 18, 2020. . 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 September 18, 2020. . 7.7 (6.5 ; 9.1) 7.4 (6.7 ; 8.1) 6.0 (5.4 ; 6.7) 5.9 (4.8 ; 7.1) 9.5 (7.8 ; 11.0) 8.5 (7.5 ; 9.4) 6.8 (5.8 ; 7.8) 7.2 (5.2 ; 9.0) Week (first day) 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 September 18, 2020. 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 September 18, 2020. 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 September 18, 2020. . https://doi.org/10.1101/2020.09.16.20195693 doi: medRxiv preprint

Assessing the extent of SARS-CoV-2 circulation through serological studies

Seroprevalence and correlates of SARS-CoV-2 neutralizing antibodies: results from a population-based study in Bonn

Humoral Immune Response to SARS-CoV-2 in Iceland

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

UK Biobank SARS-CoV-2 Serology Study. Weekly Report -21

Population Point Prevalence of SARS-CoV-2 Infection Based on a Statewide Random Sample -Indiana

Incidence and risk factors of illnesses presumably caused by a SARS-CoV-2 infection in the general population during the lockdown period: a multi-cohort study

Cohort Profile: The French E3N Cohort Study

The Nutrinet-Sante Study: a web-based prospective study on the relationship between nutrition and health and determinants of dietary patterns and nutritional status

The French CONSTANCES population-based cohort: design, inclusion and follow-up

Lower prevalence of antibodies neutralizing SARS-CoV-2 in group O French blood donors

COVID-19 : point épidémiologique du 14 mai 2020

Obesity: preventing and managing the global epidemic

Case definition for coronavirus disease 2019 (COVID-19), as of 29

Generalized Raking Procedures in Survey Sampling

Multiple Imputation for Nonresponse in Surveys

Estimating the burden of SARS-CoV-2 in France

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

Factors associated with COVID-19-related death using OpenSAFELY

The Incubation Period of Coronavirus Disease 2019 (COVID-19) From Publicly Reported Confirmed Cases: Estimation and Application

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

Clinical Characteristics of Coronavirus Disease 2019 in China

Nicotine and the renin-angiotensin system

Cigarette Smoke Exposure and Inflammatory Signaling Increase the Expression of the SARS-CoV-2 Receptor ACE2 in the Respiratory Tract

Smoking and immunoglobulin levels

Covid-19 Testing, Hospital Admission, and Intensive Care Among 2,026,227 United States Veterans Aged

The association of smoking status with SARS-CoV-2 infection, hospitalisation and mortality from COVID-19: A living rapid evidence review with Bayesian meta-analyses

Validation of a commercially available SARS-CoV-2 serological immunoassay

Alcohol use before lockdown (in g/dy) <5 [5,10

BMI (kg/m 2 ) <18.5 [18.5

Chronic diseases (Y vs N) Asthma, COPD, other respir. diseases

The authors warmly thank all the volunteers of the Constances, E3N-E4N, and NutriNet-Santé cohorts.We thank the staff of the Constances, E3N-E4N and NutriNet-Santé cohorts that have worked with dedication and engagement to collect and manage the data used for this study and to ensure continuing communication with the cohort participants.We thank the CEPH-Biobank staff for their adaptability and the quality of their work.In the virology department, Dr Nadège Brisbarre and the technical staff for impeccable management of samples and serological assays.