Isle of Wight trial shows positive effect of contact-tracing app, but data privacy body slams UK

The UK government’s Covid-19 contact-tracing app programme has to date been a catalogue of missed schedules, delays, technology setbacks and has generated much, mostly deserved, criticism, but details have emerged of how the official trial of the yet-to-be-launched product on the Isle of Wight has generated positive results in the fight against the Covid-19 coronavirus.

However, this rare good news associated with the project came just as citizens’ rights and legal bodies concerned with data protection and privacy amplified their concerns regarding the fundamental nature of the programme and who is running it.

The contact-tracing app’s whole development has been plagued by missteps since details of its construction were first aired in April 2020, and it has fallen behind on every launch target.

In June, the app took a sea-change in its form and is now being developed using a decentralised data collection app model based on Google and Apple application programming interface (API) technology, rather than the previous, much-criticised, centralised database structure that is still in test phase on the Isle of Wight.

The latest phase of the project has seen Dido Harding, executive chair of NHS Test and Trace, which is leading the development of the app, promise that it would be launched “shortly”. This phase brings together the work done so far using the Google and Apple-based framework, for which initial programming carried out by VMware was recently handed over to the London office of Swiss software firm Zühlke Engineering, and that of the original programme under test on the Isle of Wight, which since the end of May has somewhat fallen under the radar after a string of negative publicity surrounding technical mishaps that had been revealed in the trial.

App has potential to control coronavirus

Yet at a Westminster Health Forum policy conference looking at the development and roll-out of Test and Trace, along with the future of the contact-tracing app, Christophe Fraser, senior group leader in pathogen dynamics at the Big Data Institute and a professor in the Nuffield department of medicine at the University of Oxford, said the results of the trial were encouraging.

As a co-lead in the app’s development, Fraser revealed that one of the findings of the Isle of Wight trial was that simulations had shown that by combining manual contact tracing with social distancing, the app could contribute to stopping the epidemic.

In June, development of the UK’s coronavirus contract-tracing app switched from a centralised database structure to a decentralised data collection model based on technology developed by Google and Apple

With uptake of the app a critical factor in its success, he noted that many European countries with similar approaches saw varied app uptake of between 10% and 30% of the population. But, ultimately, he said that with a combination of interventions, app-based contact tracing could add to control.

Another key point highlighted by Fraser was that the app developed using Google and Apple technology wasn’t a panacea, and like the original app, there were still pros and cons. An example was in generating pre-warning. “If somebody shows symptoms where they have tested positive, now you can pre-warn some of their contacts that they’ve been in contact with somebody who may have Covid,” he said.

“The problem is this generates more false notifications. But then, on the other hand, some of those false notifications are true. So, you could have that as an amber notification saying, ‘If you receive this notification, you don’t need to quarantine, but now’s not a good time to go and visit an elderly relative or somebody who’s vulnerable’. You can confirm that after the index person has had a test, you get a red notification and a request to quarantine.”

Fraser revealed that NHSX, the digital unit of the NHS responsible for developing the app, has been working with colleagues in New Zealand on more personalised messaging. He pointed out that there was a huge local effect in the app’s implementation. “If you have a cluster of people who use the app and contact tracing, it doesn’t really matter that much what happens nationally – you have a local effect that you can stop outbreaks, a bit of a Neighbourhood Watch effect,” he remarked.

Referring specifically to the Isle of Wight trial, Fraser noted the adverse publicity at the start, but said a number of points should be taken in mitigation, in particular the fact that when the trial started in May, the app was at a very early stage.

“[In terms of] what actually happened, there was roughly a month where the app was being used on the island, and where there was no sort of national testing programme and no national testing and tracing. During that time, 160 cases were referred [from] manual contact-tracing, [subsequently] reporting 163 contacts. [In all] 1,524 people during that period reported symptoms [on] the app, resulting in 1,188 exposure notifications. It could be thought that many of these were false positives, and this highlighted, maybe, the problem of notifying based on symptoms, and people didn’t quite know what to do with the exposure notification that came back, loud and clear,” he observed.

Showing just how much this had an impact on the epidemic, Fraser presented data showing people who test in hospital and people testing in the community, compared with happened on the Isle of Wight and nationally, and whether infections declined more quickly on the Isle of Wight. The data showed only one confirmed case on the Isle of Wight in the past month.

“The interesting thing is, the pilot corresponds with this decline in the infection rate on the Isle of Wight. So, if you compare the reduction in the Isle of Wight to all 149 other areas of England, you can say the Isle of Wight declined much more quickly than the other areas of England. And if you calculate the so-called R number, on the Isle of Wight you get a reduction that really corresponds with the introduction of the test and trace, in terms of time, and down to very low numbers that really haven’t been seen in many other places during the critical weeks of the trial down, down to 0.3-0.5.”

Fraser concluded that the tests showed evidence that stopping infection was possible by using the app together with other measures, including ongoing social distancing. However, he stressed that the Isle of Wight results represented causal proof, and that a randomised control trial was also necessary. But he believed the work had resulted in a very encouraging analysis.

Getting communities on board

At the vanguard of making sure the test actually happened was the Isle of Wight Council, which serves a closed community of about 140,000 residents, 40% of whom are categorised as retired.

Council leader Dave Stewart said the initial message to the community was that they would be leading the nation as they protected themselves. He noted that it was important to understand and communicate exactly how the app worked, and the confidential aspects of the app, tapping into what he said was the islanders’ culture of commitment and community spirit. He also quashed a few received wisdoms about specific technology hitches regarding the app, in particular with iPhones.

“We encouraged people to download the app because that is one of the big challenges we started with. In fact, over 54,000 people downloaded the app onto their phone,” he recalled. “[Picking] up that point about iPhones, I have an iPhone 6, which is not the latest model, and I had no problem downloading [the app]. My wife has a Samsung and she had a problem. That’s an Android, so there was some problem with the downloading. But in the main, most people told me they had no problem at all.

“By participating in the pilot, residents were also being more careful about social distancing and hand washing, and infection rates fell even more”
Dave Stewart, Isle of Wight Council

“It was an achievement to get to those numbers, and a few naysayers very quickly turned around and realised that it was the right thing to do. It was simple to download, and we had a position where a lot of people were using their iPhones because of lockdown to communicate with each other, anyway. Indeed, some of the elderly residents of the island got an iPhone so they could communicate with family.

“We actually gained more traction on it than we expected. And I remember asking somebody why they downloaded the app, and they said, ‘Well, it was the right thing to do’.

“A second message for me is that people like things to do. We often sit and watch TV, and we feel bit helpless, so downloading the app and creating face coverings were things people could do. And again, they could then get aligned with it.”

Interestingly, Stewart pointed out that key to addressing community danger was effective messaging. Many people were interested in the download numbers and that, he said, became a bit of an interest on a regular basis, and ultimately became a common cause that motivated the community.

“People became more aware of the risk that Covid posed, which was a benefit because it meant that if people understood the risk, they would therefore manage that risk better. What we didn’t know was that by participating in the pilot, residents were also being more careful about social distancing and hand washing, and infection rates fell even more than we thought,” he said.

“I think it shows that people not only had a practical tool to use, but it was like a reminder in their pocket – you know, ‘keep your distance’, ‘two metres if you can’, ‘wash your hands’ – because they really didn’t want to get ‘the ping’. You didn’t want your app to go off, because if it didn’t, it meant you were keeping yourself safe in mind of the elders. The community will accept if you engage with them. The community was a common cause for people who got behind it. And people could do something. We refer to [the app] frequently as an extra tool in the keepsake toolbox.”

Diverse and geographic reach

Following on from the Isle of Wight story and drilling deeper into the app’s general efficacy, Adam Steventon, director of data analytics at the Health Foundation, revealed that the latest figures from the NHS Test and Trace programme showed that 80% of people who are identified call and give their contacts – about five or six, on average – and about 75% of those contacts are then reached. The app’s job, he said, was about increasing those figures so the Test and Trace programme becomes more effective and people stay safe as we head into autumn and winter.

“Society will only be safe from coronavirus if all parts of society are safe, so the app needs to reach different groups of people to help us identify more contacts”
Adam Steventon, The Health Foundation

Yet he stressed one central key point: “Society will only be safe from coronavirus if all parts of society are safe. And so the app needs to be reaching, not the same people as the Test and Trace programme, but different groups of people to help us identify more contacts. And that’s really going to be the bottom line for when it comes to judging the success of the app going forward.

“We know from the careful research over the past four months that Covid-19 is having a disproportionate impact on people in poor areas of the UK, on older people, on people with disabilities, as well as people with specific clinical conditions, as well as many black, Asian and minority ethnic communities. The challenge is to make sure the app works with these communities, so that they can stay safe, and so that we can all stay safe. We shouldn’t assume that the engagement of people on the Isle of Wight is going to work everywhere, and we need to be able to adapt the roll-out to meet the needs of different communities across the UK.”

The central message, said Steventon, was that the programme needed to be understanding and address the risk that the app might have different impacts among people. The risk was that the app might not benefit those people who are at greater risk of coronavirus, and would therefore not have a significant impact on top of what is already in place with the Test and Trace programme.”

The issue of trust

Risk of a different kind was indicated by Silkie Carlo, director of civil liberties and privacy campaigning organisation Big Brother Watch. From the outset in her presentation at the Westminster Health Forum policy conference, Carlo said it may be helpful to talk about the app as an exposure notification app, separating out the role of contact tracing and clarifying precisely what it is the app can do and the role it can play.

She suggested that earlier conversations may have glossed over slightly what she called the “cacophony of errors” there have been in attempting to develop the British app, and the “huge waste of public money and critical time” in pursuing a “particularly invasive” version of the app.

“I wish the government would make it easy by making something simple, in line with the rest of Europe, that does what it says on the tin, with a strict assurance and safeguards that it’s not going to do anything else”
Silkie Carlo, Big Brother Watch

One thing that’s been clear right from the start, she said, was that the key to the app having any kind of successful impact would be uptake, and that to have huge uptake you needed to have “enormous” trust, and that trying to build trust in a government IT project was no easy task.

“I really wish that the government would make it easy for themselves by making something simple, something that’s in line with the rest of Europe, something that does what it says on the tin, with a strict assurance and safeguards that it’s not going to do anything else,” she said.

“We have to bear in mind that a fifth to a quarter of the population do not have a smartphone, particularly poorer and more vulnerable people, and critically 45% of over-55s don’t have a smartphone – and these are the people who are particularly at risk. This is not going to be the ultimate solution, and one of the problems with the government’s messaging was that initially people were told they had a duty to download [the app] because it was going to have this kind of overwhelming impact. Bear in mind the key here is to build trust, to build public understanding, to make the messages as simple and trustworthy as possible.”

In this regard, Carlo reserved particular ire for Dido Harding, about whom she said had led a company, TalkTalk, that was “responsible for one of the worst hacks in the UK in recent years”.

“This doesn’t inspire trust,” she said. “Sometimes it seems like the government is not even really trying very hard to build trust, and I think that is going to prove a fatal mistake. I really use that word in both senses, and not lightly. It’s very rare in the field of privacy and civil liberties that the stakes are so high, and unfortunately in this case the government’s undervaluing of privacy and of trust, and of the public’s appetite for privacy insecurity in digital technologies, will prove to be really quite self-defeating.

“So there’s a set of principles that government will need to follow with the new app, whenever that surfaces, to foster that kind of public trust. Also, to make sure it does truly respect rights, that we’re going beyond rhetoric and that we build something that is watertight, in terms of safeguards for data protection and public rights.”