Why projections of Covid waves are so complicated

SCIENCE – Starting in early June, we can say this Tuesday, July 26 that the 7th wave of Covid-19 in France has reached its peak. On the side of positive cases such as hospitalizations, the indicators fall, as you can see in the graph below.

Good news (if the trend continues), because this peak linked to the BA.5 variant has recently been equal to that linked to BA.2, far from the first “Omicron wave”, due to BA.1. It is also good news because, although the situation in Portugal and South Africa makes it possible to be less alarmed, this is the first wave in which the Scientific Council (and therefore the government) has no model to rely on.

And at the moment, the Institut Pasteur has no model for the future waves that will undoubtedly come when the summer ends and our resistance slowly decreases.

This did not prevent the Scientific Council, in the final opinion issued on July 19 and detailed in a press conference, from working on three scenarios for the fall. But these trajectories are very generic: a return to existing variants, a descendant variant of Omicron or, worse, a very different and possibly more dangerous variant.

The models became very complex

But why doesn’t the Scientific Council have a projection that would allow it to wait? “This BA.5 wave is the first one where we don’t have a model, because in the last two years we started with simple models, which gradually became more complex to integrate the effect of the epidemic on variants, vaccines, as well as the reduction of immunity,” explains Simon Cauchemez, modeller at the Institut Pasteur and member of the Scientific Council. “These overly complex models are now uncertain”.

To make projections of the Covid-19 curves (more details in this interview with Simon Cauchemez dated from 2020), we make assumptions about the virus (its contagiousness, the duration of the infection, the severity , etc.), and the target population (the number of contaminated people, the number of contacts at risk, age, etc.).

At the beginning of the epidemic, things were (unfortunately) simple: almost anyone could catch Covid-19. Even after the first wave that only affected 5% of the population. There are only two ways to reach the top. Let the virus spread until not enough people are infected (thereby leaving tens or even hundreds of thousands of people to die). Take measures to limit our contacts at risk to break the chains of transmission of the coronavirus.

But since then, things have improved in many ways. First, thanks to vaccines, which give us very important protection against serious forms and, in a lighter and more ephemeral way, against infection. There are also variants to be integrated into the models. Does this new set of mutations make the virus more contagious? Not so cruel? Can the vaccine escape? Previous infection?

“We know very little about cross-immunity between variants”

All these parameters make the projections more complicated, but the modellers still see things clearly. “Until then, you can simplify by putting people in boxes. People vaccinated with one dose, two doses, a booster, those infected with natural immunity”, explained Samuel Alizon, Research Director of the CNRS, specialist in the modeling of infectious diseases. “But those Omicron’s wave blew away the categories.”

With the arrival of the highly infectious variant of Omicron, most Western countries, widely vaccinated, tired of repeated imprisonment and unable to take non-forceful braking measures, chose to abandon the epidemic . By doing this, we accepted a very large wave of cases, hospitalization, but with a lower toll than the previous variants of an unvaccinated population. There is also the vague and cynical hope that this wave will be the last, causing “herd immunity”, which prevents the coronavirus from circulating.

But reality, once again, very quickly caught up with thin hopes. This natural immunity, as we already know, does not last forever. It diminishes and disappears over time (although that against severe forms seems to stabilize after three doses or three infections).

And that’s a big part of the problem. “While vaccine immunity is easy to control and monitor, natural immunity is less well known,” explained Samuel Alizon. Especially in multiplying variants and situations. In what box to put a vaccinated person, contaminated 3 months later, then who has a recall in January? How does his immunity compare to someone infected in 2020, vaccinated 2 times in 2021, then re-infected with BA.1 in January 2022? Or in BA.2 in March?

“We know very little about cross-immunity between the variants, for example we have seen that BA.5 can avoid part of the immunity caused by infection with BA.1”, described Samuel Alizon. All of these boxes become very difficult to handle so that an epidemiological model can provide accurate projections without risking being completely misled.

Simplify without distortion

However, one should always anticipate as much as possible. “The pandemic is not over yet. We are dealing with a virus with a genetic evolution that is difficult to predict”, warned Jean-François Delfraissy, the president of the Scientific Council, as a preamble to the presentation of his latest opinion.

But can we adjust the calculations to this new situation? “Current models are very complex and therefore unstable. A compromise must be found with several parsimonious models, taking into account different resistance profiles. This is a work in progress”, explained Simon Cauchemez.

In a prepublished article on June 15, Samuel Alizon and two colleagues tested a new concept in an attempt to take into account reduced resistance. “The idea is to include in the model how long individuals are in a particular state, for example how long it took their last vaccine dose,” he explained. “One of the results is that even without a new variant, large annual waves can be observed linked to winter and the (limited) reduction of resistance”.

Surprisingly, scenarios in which the entire population is vaccinated at the same time each fall lead to a greater increase than when a booster is offered only to the elderly and vulnerable (although many are vaccinated, more less died). The reason put forward by researchers: by vaccinating everyone at the same time, the level of immunity is synchronized. It is clear that suddenly there are many people who are susceptible to infection. Another lesson from the study, says the researcher: “Furthermore, so-called non-pharmaceutical interventions (improving air quality, wearing a mask, etc.) can be as effective as annual vaccination campaign. Ultimately, the best effectiveness is obtained by combining these interventions and vaccine boosters. “

Obviously, this type of general projection has limitations. “The longer we plan ourselves in the long term, the more qualitative the model”, specified Samuel Alizon. However, what health authorities prefer are “quantitative” projections. In simple words, let’s say that a qualitative model tries to imagine the general flow of the curve of Covid-19 in the long term according to different assumptions. The quantitative model tries to predict the number of people who are infected or hospitalized. “But when we go beyond the moon, these quantitative models become dangerous because of the many unknowns, and we still need to evaluate different scenarios”.

In conclusion, we must clearly remember that we are not powerless to monitor this pandemic. “We must be careful about the next occurrences, because it is very difficult to tell the dates and the magnitude of the peaks. Today, we observe what is happening in our neighbors and it is very instructive”, recalls Arnaud Fontanet, epidemiologist and member of the Scientific Council, during the press conference. However, the future dominant variant should not come out of France. “When we are on the front line, it is difficult, we must remember the possibility of a little you disturbing presence”.

See also at The HuffPost: Why vaccinating people over 60 is not excessive

Leave a Reply

Your email address will not be published.