Dominic Cummings’ lockdown travels and the exams fiasco could have contributed to dooming the government’s Covid contact-tracing app before it even launches, a technology expert has warned.
Evidence of low uptake overseas also suggests the app may not live up to ministers’ early hopes of a panacea. In late May, the health secretary, Matt Hancock, admitted it was “the cherry on the cake but [not] the cake”; in recent weeks it has barely been mentioned.
The app, which is due to launch in England and Wales on Thursday 24 September, will use the bluetooth signal in mobile phones to track close and sustained contact between users and then warn those who may have been exposed to an infectious person that they should self-isolate.
But to succeed at that goal, the app will need to be installed by a large proportion of the public. That could be hard to achieve, warned Imogen Parker, the head of policy at the tech thinktank Ada Lovelace Institute, because of a series of trust-diminishing scandals over the summer.
“In the original modelling, which made the case for a contact tracing app, the magic number to suppress transmission effectively would be 60% – or 80% of smart phones users – downloading, running and adhering to the app. But internationally, the best case scenario we’ve seen has been about 40% uptake, and that’s in countries where the majority of the population lives in a single urban area, and have had apps for longer, like Iceland and Singapore. Examples from larger countries like Germany, and geographically and culturally closer countries like Ireland and Scotland, suggest we’re looking more like 18-30% a few weeks after launch,” she said.
“In the UK, uptake is going to be related to trust in government. While we were doing some public work on trust over May, you had the Barnard Castle incident; after that you had the A-level algorithm. But the flip side is that the NHS brand itself is incredibly trusted.”
Parker also raised alarm at the prospect of large numbers of people being advised to self-isolate based on “false positive” results. “The best data I’ve seen suggests 45% false positives and 33% false negatives,” she said, “but phone proximity isn’t everything. The growing body of evidence about things like the substantially limited risk outside versus inside really matters. We need to make sure the app can identify risk, not just identify phones.”
The latest version of the contact-tracing app is substantially rebuilt from an earlier version. It was pulled from public release at the last minute after tests in the Isle of Wight revealed several flaws with the iPhone version. Some of those changes should help increase uptake and efficacy, said the University of Oxford’s Prof Christophe Fraser, a scientific adviser to the national test-and-trace programme.
“We and others have shown through simulations, where we show the integration of the app with manual tracing, social distancing, and so on, even 10 to 15% uptake can have an effect,” he says.
The newest version of the app is built with a framework created by Apple and Google, which means it can begin working even before it is installed on devices. It also includes a QR code-led “check-in” function, which lets users record that they have been to public locations and receive alerts for any outbreak linked to that venue.
Those features, Fraser says, should help people see that the app isn’t just important for public health but for individual outcomes too. That means that even “false positive” warnings can be useful.
“Localised contact tracing provides information, even if you’re not infectious,” he said. “It’s not really a ‘false positive’, because it’s very important to know that Covid-19 is spreading in your local area. We’re faced with a difficult winter, a grave winter, and every little behavioural nudge matters. A little bit of ventilation, mask wearing and hand hygiene does make a difference.”
The Department of Health and Social Care has been approached for comment.
o This article was amended on 21 and 22 September 2020. Point number seven in the graphic was amended because it is the app user’s personal bluetooth key that is uploaded to a central server, not the data gathered by their phone as an earlier version said. This article was further amended to correctly quote Imogen Parker.