Catherine Hill is an epidemiologist. She has worked at the Inserm, at Harvard and at the Institute Gustave-Roussy in Villejuif. She was also a member of the scientific council of the Agency of the drug, where she was one of the few supporters of dr Irène Frachon, who had just reported the case to the Mediator. Its work focused, in particular on the causes and evaluation of cancer screening. She has also conducted the first studies quantifying the mortality related to the Mediator.

On this topic, read also our folder Pick, anatomy of a scandal health

The Point : While France is déconfine, new outbreaks of contamination Covid-19 are repeated, especially in France and South Korea. Should we be worried ?

Catherine Hill : Yes, the situation in France remains a concern. In South Korea, the virus has not circulated virtually. In France, its spread has been slowed thanks to the containment, but it has never ceased to circulate. The situations are very different. In South Korea, there has been a total of 260 death of the Covid-19 on the date of may 14, for a population of 51 million inhabitants, and the number of people infected increases only very slowly for several weeks ; overall, the epidemic is controlled. The south Korean population has been extensively tested : we know that the date of April 20, for 100 tests performed, only two were positive. On the contrary, in France, the epidemic is not really controlled : the containment has stopped the runaway exponential, which was double the number of cases and the number of the dead every three days, but the virus continues to circulate. On the same date of April 20, there were 33 positive tests in France for 100 tests. This is the sign that we can reserve our tests for the most likely, and this strategy still prevails in the party today. The French population has not been extensively tested, and has been observed to more than 27 000 deaths for a population of 67 million inhabitants. Where in Korea, we can speak of recovery, in France, it is the continuity.

The strategy selected by Olivier Véran is to test only the cases already presenting symptoms, on medical prescription, then the people with whom they were in contact in the 48 hours prior to these symptoms. According to him, screening wider there would be " no sense from the medical and scientific. "You are not of this opinion...

This strategy is insufficient to control the epidemic, at least for three reasons. The first is the frequency of asymptomatic carriers, these people who spread the virus without anyone being aware of this. Researchers at the Scripps Research Institute have compiled studies to estimate the frequency of asymptomatic carriage in populations tested more or less systematically. One of the studies the most informative, because it focuses on 1 046-positive individuals, is that of the sailors of the aircraft carrier Charles de Gaulle : in this study, 48 % of those tested were asymptomatic. Another study to be informative is that the people of the Diamond Princess, which had 46 % of asymptomatic on 712 positive. All of the data collected by the researchers of the Scripps Institute provides a global estimate of the asymptomatic carriage of 52 %, based on the observation of the nearly 3,000 positive cases. To be clear : half of the people infected by the Covid-19 don't know, and will never know since they will not develop symptoms.

Or (and this is the second reason that shows the inadequacy of this strategy), several studies (such as Ferretti et al., Science 2020, or Cheng Jama Internal Medicine May 2020, ED) estimates that half of the contamination comes from direct contact with an asymptomatic individual.

The third reason, finally, is that the bulk of the contamination is done in the 5 days preceding the onset of symptoms and within 5 days after the first symptoms. Limit the search to the contacts in the 48 hours preceding the onset of symptoms ignores a portion of the contacts potentially contaminated.

in Short, it is necessary to do this as recommended by the WHO since the 16 march : "test, test, Test..."

of Course. It is necessary to test massively the population. But for this it is necessary to change the strategy, because obviously, we cannot carry 67 million tests per week. A solution, used in other countries, is to group test : samples are collected, for example, among 20 people, we split each sample into two, and then you gather the samples n° 1 to each person in a single tube, and it is in this tube that you are in search of the virus. If the result is negative, each of the 20 persons is presumed to be negative ; if the result is positive, we can test one by one samples n° 2. If the virus is not very common, this system significantly reduces the number of tests required, and the cost of operations. We can play on the size of the groups and on procedures a lot more sophisticated to optimize these group tests.

another solution, which has the merit of simplicity, is to search for the virus in wastewater : if there is no virus in the sewers to the exit of a small town, it is probably no inhabitant is positive. Conversely, if the virus is found in the sewers, you can make a collection upstream in the circuit of the sewage to find the inhabitants to be tested in priority.

We are in our first weekend of " déconfinement ". Are you concerned about that ?

The current situation is still critical. On 15 may, there was still 2 to 162 persons in the intensive Covid-19, and by adding the number of 3 219 people in the icu for another cause observed on the 30th of April, it happens to nearly 5 400 people in the icu. Small outbreaks reported in recent days are only the signal of the movement of the virus, movement more visible than before to the extent that the symptomatic cases and their contacts are tested a bit more systematically.

A recent study from the Pasteur Institute shows that the 11 may, to 4.4% of the French population, or 2.8 million people, had been infected by the new coronavirus. What can we learn from this study ?

These are estimates obtained from the available data, using a classical model. The first information that we can draw, is that there is still more than 64 million people at risk of catching the Covid-19, which is not really good news. The second information is only 0.7 % of infected people die, as 3.6% of persons infected are hospitalized, and 18.1 % of hospitalized die. The mortality is under 1 %, which is not too bad news.

Few studies have been conducted in France to evaluate the progress of the epidemic. Is this a wrong ? What could have been done ?

We have no estimate of the prevalence of the virus in the population, neither globally or by region. It would have been necessary to test representative samples of the population by RT-PCR at regular intervals to get an idea of the spread of the virus. In the absence of representative samples, one could test, for example, all women who have recently given birth. There are 2,000 per day, this is not a bad sample. Today, while the population is starting to flow, who can say what is the proportion of infectious people in the paris metro, in Britain, New Aquitaine ?

Is there still time to straighten the bar ? With what measures, for example ?

It must be tested heavily, and it is a serious mistake to have banned companies from offering tests to their staff via occupational health. When the situation is back to normal, we will not fail to question the strategy that has led to confining the major part of the population, leaving the virus to circulate to the outside instead of actively seeking individuals with the virus and their contacts in order to isolate them.