Until the end of 2020, information on the first cases of COVID-19 in Europe and around the world was based on epidemiological surveillance systems that each country had set up in a very short time. These systems made it possible to detect cases just a few weeks after China alerted the WHO on December 31, 2019, to the outbreak of severe pneumonia of unknown origin within its borders. In France, surveillance was implemented as early as January 17, 2020, with the publication of the first case definition, and the first cases were reported on January 24, 2020. It was around the same time that the first cases were reported in other European countries.

In January and February 2020, knowledge of the virus and its circulation outside of China was still limited, and case definitions were based on identified areas where the virus was circulating (where cases were diagnosed); the clinical presentation of the disease was still poorly understood and appeared nonspecific, and diagnostic capabilities were still underdeveloped.

Subsequently, diagnostic techniques evolved, and it became possible to determine whether individuals had been in contact with the SARS-CoV-2 virus in the past with a very limited margin of error. The origins of the epidemic and the early stages of this virus’s global circulation are a subject of major interest for gaining knowledge about emerging diseases and improving our ability to detect them early. This is why the “Constances” team, in partnership with Santé publique France, the Pierre Louis Institute of Epidemiology and Public Health, and the laboratory of the Emerging Viruses Unit in Marseille, quickly explored the possibility of testing samples already available as part of the cohort study described below.

Few similar studies have been conducted in Europe to date. In Italy, a study of serum samples collected from a cohort of 959 participants enrolled in a lung cancer screening and follow-up program identified 111 patients with a positive ELISA test for SARS-CoV-2 between September 2019 and February 2020, of whom 6 patients also had a positive neutralization test: four were collected in October 2019, and the other two in November 2019 and February 2020, respectively ⁽¹⁾. More recently, a reanalysis of 39 virological samples from children suspected of having measles, collected between September 2019 and February 2020, again in Italy, identified a child infected with SARS-CoV-2 whose symptoms had begun in late November 2019, with a perfect match between the viral sequence and that of the Wuhan-HU-1 virus ⁽²⁾. A similar case occurred in France with a patient who tested positive for SARS-CoV-2 via RT-PCR in late December 2019 ⁽³⁾. This patient presented with pneumonia and imaging findings suggestive of SARS-CoV-2 infection and had no specific history of risk exposure or travel abroad. It was the retrospective analysis of preserved samples that made this diagnosis possible. Aside from these reported cases, an environmental study investigating the presence of the virus in wastewater suggests circulation in Italy in late 2019 ⁽⁴⁾. No other studies have been published to date at the European level.

As described in the box below, the value of the Constances cohort lay in the availability of samples collected from individuals in the general adult French population, dating back to the months preceding the emergence of SARS-CoV-2. Another advantage of this cohort was the ability to follow up with participants to assess their potential exposure to viruses (known at the time) and any symptoms they may have experienced during that period.

In our study, 6,020 samples were collected between November and January. Of these, 13 were positive by ELISA and confirmed by seroneutralization. Even assuming that some of these participants with positive seroneutralization results were “false positives,” we can nevertheless estimate that approximately 1 in 1,000 adults in France may have “encountered” the virus before the end of January. In such a scenario extrapolated to the adult population in France (approximately 50 million people), about 50,000 people may have been infected by the end of January, when the first imported cases were officially reported on French territory. With a hospitalization rate and mortality rate due to SARS-CoV-2 infection of 2.7% and 0.49% nationally ⁽⁵⁾, at least 1,350 patients would have been hospitalized, resulting in approximately 250 deaths. In the midst of the seasonal respiratory infection season, and given that approximately 20,000 patients are hospitalized each winter with a diagnosis of influenza-associated pneumonia and an average of 1,500 influenza-related deaths are reported, it cannot be ruled out that these early cases of SARS-CoV-2 infection, even severe ones, went unnoticed. These extrapolations regarding the number of hospitalizations and deaths could even be revised downward if the risk of complications and death following infection were lower with these early infections. The absence of an immediate exponential rise in the incidence curve is also surprising. One explanation could be linked to variations in the virus’s transmissibility, which are known to be related to the virus itself (mutations) and to environmental factors that modulate transmission, either directly or by influencing social contacts. In particular, it is known that a single mutation (D614 to G614) during the first quarter of 2020 worldwide was associated with greater potential transmissibility and comparable severity ^{(6, 7)}. It was this strain that spread across Europe in March.

To validate this hypothesis, it would be extremely helpful if studies using biological samples were conducted and shared in other European countries and/or using other types of specimens. In particular, the possibility of reanalyzing nasopharyngeal samples (more than 50% of which are negative for seasonal viruses) collected during surveillance of seasonal epidemics, or of re-evaluating patients whose pneumonia diagnosis remained unexplained but who had suggestive imaging findings, are certainly promising avenues for targeting these reanalyses. A particular benefit of working with virological samples would lie in the possibility of sequencing the genome of isolated viruses (where applicable) and comparing it to existing viruses that have been circulating since 2019.