UK contact-tracing app developers hit back at effectiveness and privacy doubts

With the official launch of the second phase of the UK’s Covid-19 contact-tracing app now under way on the Isle of Wight, the UK government has come out fighting against a legion of critics from legal and scientific communities who are casting doubt on whether the app will be of use in the fight against coronavirus.

Yesterday (4 May), the government, backed by leading experts in the UK’s epidemiology, IT and communications sectors, released more information on how the app – the first details of which were announced on 24 April – would work in its first scale trial on the Isle of Wight.

Developed by NHSX, the NHS’s digital healthcare innovation unit, with support from supplier VMware Pivotal Labs, the contract-tracing app works by using Bluetooth to automate the “laborious” process of contact tracing and has the goal of reducing transmission of the virus by alerting people who may have been exposed to it, so they can take action to protect themselves.

Once installed, the app will start logging the distance between a user’s smartphone and other phones nearby that also have the app installed using Bluetooth Low Energy. The anonymous log of how close users are to others will be stored securely on each user’s phone.

If a user becomes unwell with symptoms of Covid-19, they can use the app to inform the NHS, which, subject to sophisticated risk analysis, will trigger an anonymous alert to other app users with whom the user came into significant contact over the previous few days.

Yet almost as soon as the first details of the app’s capability were announced, critics weighed in with concerns over what the app could achieve and whether the UK public could, or would, make representative use of it. In particular, the main dispute is whether the app’s centralised nature will lead to privacy breaches, and also whether it will be of any use at all if there is a lack of user uptake.

Reacting to the doubters, John Newton, director of health improvement for Public Health England and national Covid-19 testing coordinator, said the time was now right to roll out the app in the Isle of Wight, given that UK scientists were learning all the time about how to integrate what they have learned with facts on the ground. He said the app project was all about defining risk and that the centralised approach was the right one.

“With the benefit of these big datasets and all the sophisticated technology of machine learning and artificial intelligence [AI], we can draw in all this information and use it to refine the risk assessment, and therefore the advice is given to people who use the app,” said Newton.

“People will know about the symptom tracker, which has been used by millions of people, and that tells a lot about what people describe. We can then link that with the test results and then the context test results, so we get really concrete evidence to link to the recorded symptoms, and all that will be incorporated into the development of the app.

“This is a really strong reason for having a centralised approach, so we can draw all this evidence together and then put it out again so that people get the best possible risk assessment, and therefore the best possible advice on what to do, both for contacts and for the cases.”

Jonathan Van-Tam, deputy chief medical officer for England and professor of health protection at the University of Nottingham, was very supportive of the centralised approach, noting that the anonymised dataset it can generate could be subjected to AI techniques and machine learning to gain knowledge.

“We will not only understand the proximity, and the duration of proximity, of individual contacts to a case, but we will be able to deselect that out, according to the day of illness of the case,” he said. “And therefore, we will be able to learn, for example, that 30 minutes’ exposure at close quarters to a case at day one of symptoms is equivalent to two hours’ exposure at one-and-a-half days before their symptoms began or three days after their symptoms began when the virus load is starting to decline.

“So, these are the kinds of granularity of contact transmission that, in a science sense, we are going to learn a huge amount from, and not just learning for learning’s sake, and we can go back into refining the algorithms. So, over time, the whole thing becomes much more efficient and a better tool to control the disease burden.”

The scientific experts stressed that uptake of the app would be vital and that by regularly using the app over a sustained period of time, people will engage with something that is clearly about protecting the NHS.

“There is obviously a mathematical function around the fact that not only do we need people to download the app, but the effectiveness of the app is essentially a squared function of the number of people that download it because it relies on the context of cases yet to emerge,” said Newton. “The effectiveness in terms of being able to drop to R Zero [the virus transmission coefficient] will be key to uptake and usage.”

Uptake of the app was one of the key issues that NHSX chief executive Matthew Gould was questioned about while giving evidence to the House of Commons Science and Technology Committee on 28 April, as part of an inquiry into UK science, research and technology capability during the coronavirus outbreak.

Updating the assurances that he gave to MPs, Gould said the NHS was going to mount a really serious campaign to make sure “people know that if they do want to carry on saving lives, protecting the NHS and get the country back on at speed, then downloading the app is one way that they can do that”. But Gould conceded that uptake at the hoped-for levels would be “a stretch”.

Giving evidence to the parliamentary committee, Christophe Fraser, senior group leader in pathogen dynamics at the University of Oxford’s Big Data Institute, said about 60% of the UK population would need to use the app to effectively combat the spread of Covid-19.

Recognising the challenge of this, particularly given the much lower adoption rates shown by similar apps in other countries, Gould insisted it was worth saying that even if the app’s usage didn’t reach “very high levels” of downloads, there was still huge value to be gained.

“Even having the levels that we’ve seen in some other countries, the app will be fantastically valuable,” he said. “It will give us an insight of the speed of contact tracing, and it will anyway be supplemented by the more traditional contact tracing, so it’s not all dependent on the app and the level of downloads anyway.”

Gould also remarked on a parallel development by the Scottish government, which announced just as the Isle of Wight trial was unveiled that it had developed a web-based tool for the NHS in Scotland, accessible on smartphones or computers, which would allow citizens to input details of people that they have been in close contact with, and for these to be sent directly and securely to contact-tracing teams.

The Scottish government added that it was seeking to ensure greater involvement in the development of the NHSX app and that it needed to understand how data from the app would interface with its own approach to contact tracing. The Scottish government did accept that not everyone in Scotland would want, or be able, to use a web-based tool.

Gould’s analysis was that there was not a great deal of space between what Scotland’s first minister has said and NHSX’s approach.

“What we are placing is not a single solution or a magic bullet,” he said, “but as part of a strategy where we have testing, and we have contact tracing, and all the elements of the strategy reinforce each other. So actually [the Scottish alternative] sounds very compatible.”