Did Eat Out to Help Out cause Covid to spread?

We’ve been seeing a number of claims debating whether the government’s Eat Out to Help Out (EOHO) scheme led to an increase in Covid-19 infections. 

Eat Out to Help Out (EOHO) was a government scheme that encouraged people to visit restaurants by offering discounted meals last August.

The Chancellor of the Exchequer, Rishi Sunak has recently denied the scheme was to blame for the rise in infections.

We can see that Covid-19 daily case numbers did increase between mid-August and mid-September (during and following the scheme) but we have seen claims in the media and elsewhere that EOHO contributed significantly to this rise. 

Much of this has been fuelled by a paper published in October which suggested the scheme was responsible for 8-17% of new infection clusters (i.e. infections that shared a common location).

This appears to have been misunderstood by several media outlets, who conflated it with a rise in actual infection rates. 

So what does this paper say? And how certain can we be about the relationship between Eat Out to Help Out and Covid-19 infections?

What did the paper say?

The paper, which was published by the University of Warwick and has not been peer reviewed, assessed whether there was a new Covid-19 infection cluster made up of at least three cases within each calendar week where EOHO was active.

It found that “the program is accountable for between 8 to 17 percent of all new local infection clusters during that time period”. This would mean it caused more Covid infections to happen, but not that it caused a rise of total infections by 8 to 17%, as some outlets suggested.

The research used data including the restaurants registered with the Eat Out to Help Out scheme, the number of bookings made via table booking service OpenTable and even levels of rainfall which may have put people off dining out. 

This study found that areas with a higher rate of uptake in the scheme were more likely to experience a new Covid-19 infection cluster. 

A University of Warwick press release at the time said in its headline “‘Eat Out to Help Out’ scheme drove new Covid-19 infections up by between 8 and 17%, new research finds.” As we have said, this is not quite right.

How reliable are these findings?

It is plausible that EOHO increased contact between people (the possibility of transmission via restaurants was acknowledged recently by Professor Jonathan Van-Tam) and this would therefore raise the likelihood of them spreading Covid. It is hard to estimate the size of this specific effect, however.     

The study itself refers to the 8-17% figure as a “back of the envelope calculation”, suggesting it is a rough estimate. The 17% figure is the upper limit of this estimate, but it could be as low as 8%.

Other experts have also warned against relying too heavily on the numbers in the paper’s conclusions.