Since the start of the year, around one in five French people has been recognized as positive for SARS-CoV-2, according to data collected on SI-Dep – the database which has been listing, since May 2020, the results of virological tests. How many people have contracted Covid-19 since the start of the epidemic in France more than two years ago? “There will never be raw data” in this regard, warns epidemiologist and modeler Mircea Sofonea, from the University of Montpellier.
If we add the number of people declared positive in the indicators from the SI-Dep database, we reach nearly 24 million cases. We can add nearly 140,000 infections identified by the health authorities before the production of this file (when the patients of Covid-19 were identified in a fairly traditional way).
But this calculation is very imperfect, for several reasons:
- Massive late screening. During the first months of the epidemic, virological tests were reserved for the most serious patients. According to estimates from Public Health France (SPF), only one in twenty-four infections was recorded from March to May 2020. It was not until the summer following the first wave that screening was expanded.
- A changing screening policy and offer. The detection rate of the virus varies greatly over time. The reasons can be personal: thus, the French have been tested more during the holidays, when reuniting with their families. But it is above all the screening policy that has changed many times during these two years. The tests were first PCR exclusively, then antigenic, accessible by prescription then freely, with appointments more or less easy to obtain depending on the period, etc. From October 15, 2021, the cessation of reimbursement for so-called screenings “Comfort” for unvaccinated adults has, for example, led to a drop in screening in unvaccinated and asymptomatic adults, and infections have gone undetected.
- Contaminated people never tested. “Not all infected people get tested, either because they are not symptomatic and do not think about it or because they do not have access to the test for various reasons”summed up Clara Blondel, scientific project manager, during the epidemiological update organized by SPF on March 11.
- Self-tests not counted. Since the end of 2021, the use of self-tests has become widespread, especially among children, but these screenings are not counted in SI-Dep. In theory, a positive result must be confirmed by an antigen test or RT-PCR, but there is no guarantee that this rule is respected. Not to mention that the reliability of the results is not without limits.
- Reinfections counted multiple times. People who have had the misfortune to contract Covid-19 on several occasions have been counted several times in the indicators provided by the health authorities. SPF identified 417,000 people who had at least two positive tests more than two months apart between the 1is January 2021 and January 27, 2022. But due to the absence of a specific information system before May 2020 and changes in the data pseudonymization stage of it in early 2021, this count does not include “possible reinfection events with a first episode in 2020”, underlines SPF in a recent publication. There are no data available on reinfections that have occurred since this publication. For comparison, English health authorities have detected 800,000 possible reinfections out of 17.1 million confirmed cases since the end of January 2020.
Other methods to complete the official balance sheet
From the start of the epidemic, several teams of researchers wanted to estimate the number of people with Covid-19 with methods other than virological tests. This was in particular about “to answer a question: have we achieved collective immunity? »explains Mircea Sofonea.
- First estimates based on hospital data
In France, from the start of the epidemic, researchers from the Institut Pasteur, Public Health France and Inserm combined French hospital data and the survey carried out on board the Diamond-Princess, a cruise ship placed in quarantine in early 2020, to determine the hospitalization and fatality rates of Covid-19. According to this extrapolation, approximately 3.5 million people (between 2.1 million and 6 million) would have been affected as of May 11, 2020, according to their work published in the journal Science le 13 mai 2020.
- The limits of serological surveys
The arrival of seroprevalence surveys, which search blood samples for the presence of antibodies directed against SARS-CoV-2, have given another point of view. However, the relationship between the presence of antibodies and infection with SARS-CoV-2 has been gradually modified by vaccination – antibodies directed against the spike protein (S) can be distinguished, which can also be the result of an infection than vaccination and those directed against the nucleocapsid protein (N), resulting solely from an infection but which would be detectable for a shorter time than the former.
In any case, the results of these studies must be analyzed with caution: “Antibodies are known to decline over time, so it is difficult to interpret seroprevalence data”, underlines Simon Cauchemez, head of the mathematical modeling unit of infectious diseases at the Institut Pasteur. Furthermore, not all cases “of seroconversion, that is to say the fact of producing antibodies”abounds Mr. Sofonea.
However, serological surveys have enabled researchers at the Institut Pasteur to develop a new method: this involves determining the probability of being hospitalized when infected. “by comparing the number of hospitalizations that there were during the first wave with the number of infections estimated by the major Sapris seroprevalence survey during this wave.
Knowing the number of people hospitalized, it thus becomes possible “to rebuild the number of infected” over time, details Mr. Cauchemez. But this method had to stop being used “with the emergence of new, more severe variants and with the vaccination of the population, because these two factors modify the probability (…) hospitalization we need”explains Mr. Cauchemez.
- “Between 65% and 70%” of French people contaminated at least once
Several modeling works are in progress. At the Institut Pasteur, Mr. Cauchemez explains that the teams are in the process of “to develop a more complex approach which will make it possible both to estimate the number of French people who have been infected but also to quantify the different immunity profiles [par exemple vacciné et infecté, etc.] ». But this work is “quite complicated”.
Mr. Sofonea’s approach is based on hospital mortality, “the most reliable data that exists”. Then you have to “going back to the number of infected cases: it’s more complicated, because you have to use the probability of dying while being infected, which varies” over time, due to several factors, such as the variant concerned, the improvement of treatment techniques or the rise in importance of vaccination. While the methodology is still being improved, his team believes that “between 65% and 70%” the number of French people infected at least once since the start of the epidemic.
Our selection of articles on Covid-19
Find all our articles on SARS-CoV-2 and Covid-19 in our section
On the epidemic :
And also :