Why are there so many mediocre restaurants

Corona and gastronomyWhy it is difficult to assess the risk of infection in restaurants

What do we know about the role of restaurants in the pandemic?

There are reliable reports of infections that have taken place in the catering sector. At the beginning of the epidemic in Germany, there was a major outbreak at an auto supplier near Munich. That was the first 16 cases in Germany. A team of researchers tracked the chains of infection at the time. An employee had infected a colleague in the canteen. The two had only turned briefly to hand the salt shaker. Otherwise they had sat with their backs to each other. The research team described this case in the journal "The Lancet: Infectious Diseases" in mid-May. Another case occurred at the opposite end of the republic: in Leer in East Frisia. There had been a party in a restaurant at which about thirty guests had become infected. So a typical superspreading event. Similar incidents were also reported, for example from Berlin.

(dpa-Pool / Michael Kappeler) The federal government wants to implement these corona restrictions
With drastic restrictions from November 2, the federal government wants to curb the massively increasing number of corona infections again. An overview.

How big is the risk that people will be infected in the restaurant?

The data situation is rather thin. There are only a few studies. One was done by the US CDC in Atlanta. The experts there have come to the conclusion that going to a restaurant is a certain risk factor for getting infected. This was a case-control study in which infected cases are compared with a non-infected group, i.e. the control group. In this case, these were people who had reported to be tested - because they had symptoms typical of COVID-19. But only about half of the tests were positive, so only those were infected.

The subjects in both groups were asked: What did they do in the two weeks before the onset of symptoms? Have you met a large group of other people? Were you shopping? Did you go to a restaurant? Or a church service? Have you been to the gym or a bar? In the office or at the hairdresser's? The answers of both groups were then compared. And there was a clear difference in behavior between the infected group and the non-infected control group only when visiting restaurants and a subgroup when visiting bars and cafes. The infected reported about twice as likely to have visited a restaurant in the time before the onset of symptoms than the non-infected. The CDC team concludes that eating out is a risk factor. Because you cannot eat or drink with a mask, they recommend further measures, such as good ventilation.

How reliable are the results of this US study?

The experts from the CDC themselves list a few points that limit the informative value. On the one hand, there is the number of test subjects: there were a total of just under 320 test subjects, half infected, the other half not infected. And in terms of their composition, the groups were not identical in terms of age and gender, for example. That is, there may be random fluctuations in the data that skew the result. The result does not necessarily have to be transferable to the entire country. The groups were also not asked whether they had eaten inside or outside. That would also be important information for assessing the risk. And: At the time of their survey, the test subjects knew whether they were infected or not. It's possible that this influenced their responses. So: there are some uncertainties that need to be taken into account.

Overview on the subject of coronavirus (imago / Rob Engelaar / Hollandse Hoogte)

What is it like in Germany?

For Germany, the Robert Koch Institute published a study with such data in Epidemiological Bulletin 38/2020. The data for this come from the health authorities, some of which also transmit to the RKI where someone is infected. And here the residential properties are way ahead, i.e. private households, but also residential homes such as old people's and nursing homes or refugee homes. Around 30,000 cases, individuals. In restaurants, on the other hand, just 300 cases were recorded, one hundredth of the infections in homes. So that seems to paint a different picture than the study from the USA.

Are the numbers from Germany more reliable?

These figures should also be treated with caution. On the one hand: They do not depict everything that is happening. As I said: The health authorities tell the RKI where an infection has taken place. However, an outbreak event can only be assigned to a quarter of the cases. This information exists only for 50,000 cases out of approximately 200,000. For the rest, for three quarters of the cases, we simply do not know, and this unreported figure could of course hide contagions in the restaurant. And then there is also a statistical bias due to the fact that these chains of infection can be traced back more easily in the household than, for example, in public spaces, such as in local public transport or in restaurants.

What is the conclusion that can be drawn?

The British scientific advisory body SAGE put together an overview in September of how various non-pharmaceutical measures are likely to affect the spread of the pandemic. The closure of restaurants therefore has a mediocre, moderate effect. But even this statement is only moderately reliable, with a medium confidence. The effective reproduction number could thus probably be reduced by 0.1 to 0.2.

What we knew relatively well are the boundary conditions under which the virus spreads: that is, close contact between people, no mouth and nose protection, poor ventilation and loud singing and speaking. The fact that these circumstances are not the same everywhere also makes the evaluation in the restaurants difficult. You cannot lump all businesses together with the same brush. It makes a difference whether I sit in a restaurant with good air conditioning and enough space between the tables. Or if I celebrate in a pub as if it were Ischgl 2019. But if I really want that, then of course I can do it at home.