Coronavirus: Ministers consider NHS contact-tracing app rethink, or maybe not

This article from the BBC critical about the efficacy of the NHX contact tracing app appears on the same day as a much more bullish article from the Times.

Is No 10 briefing in two different directions? Is this one of Dominic Cummings’ “do the unexpected” cunning plans? Or is the Government in disarray?

Owl is posting both for comparison and contrast – first the BBC: 

Coronavirus: Ministers consider NHS contact-tracing app rethink

Concerns about the risks of deploying a go-it-alone UK coronavirus contact-tracing app are causing further delays.

A second version of the smartphone software was due to have begun testing on the Isle of Wight on Tuesday, but the government decided to postpone the trial.

Ministers are considering switching the app over to tech developed by Apple and Google.

But countries testing that model are experiencing issues of their own.

Health Secretary Matt Hancock had originally said the NHS Covid-19 app was to be launched across England – and possibly other parts of the UK – by 1 June.

But he subsequently said the government had decided it would be better to establish a network of human contract tracers first.

However, the BBC has discovered that one of the main reasons the initiative is running behind schedule is that developers are having problems using Bluetooth as a means to estimate distance.

Even so, they still believe they are better placed to tackle the challenge than counterparts overseas who are working under constraints imposed by the two US tech firms.

Bluetooth handshakes

Contact-tracing apps are designed to prevent a second wave of infections by keeping a log of when two people are in close proximity to each other and for how long.

If one of the users later tests positive for the disease, the records are used to determine how likely it is they infected the other. If required, an alert is triggered to help prevent the further spread of the virus.

The UK has adopted what is known as a “centralised” approach, meaning that the contact-matching process is carried out on a remote computer server. One benefit is it offers epidemiologists more data to tackle the pandemic. France and India are other countries to have adopted this model.

By contrast, Apple and Google’s “decentralised” approach carries out the matches on the handsets themselves, on the grounds this better protects users’ privacy.

Poland switched its app from a centralised to decentralised approach on Tuesday. Switzerland, Ireland, Germany, Italy and Latvia are among others to have adopted the tech giants’ design.

Both systems rely on Bluetooth “handshakes” to work.

Number 10 is concerned that iPhones will not always detect each other because of a restriction Apple has imposed on apps that do not adopt its model.

But the UK team has devised a workaround and is more concerned about other limitations of using Bluetooth.

Train trouble

Some of these issues were outlined in a study published by Trinity College Dublin last month.

It highlighted problems with using received Bluetooth signal strength as a means to estimate distance.

Researchers warned signal strength “can vary substantially” depending on:

  • how deeply a handset is placed in a bag
  • whether the signal has to pass through a human body to reach the other phone
  • if the two people are walking side-by-side or one behind the other
  • if the devices are indoors rather than outdoors
  • whether the smartphone is surrounded by metal objects

The report highlighted troubling results when Singapore’s TraceTogether app was tested.

Image copyright PA Media Image caption Ireland’s police force is among organisations trialling a contact-tracing app ahead of a planned national rollout

An experiment within a stationary train carriage found that when users moved from a distance of 3.5m (11.5ft) to 4m, signals became stronger rather than weaker because of the way metal objects were reflecting the radio waves.

A trial in a supermarket also found the received signal strength was the same whether two people were walking close together or 2m apart.

Follow-up tests using Apple-Google’s tech are currently under way.

“The work is ongoing, but preliminary results are broadly consistent with previous observations,” said Dr Brendan Jennings, who has been tasked with assessing the effectiveness of Ireland’s Covid-19 app.

Hidden data

The team behind Switzerland’s SwissCovid app is carrying out tests of its own.

Its Bluetooth measurement chief believes the issue can be partly addressed by taking a range of readings over a period of five minutes or more.

But he added that Apple and Google had placed curbs on what could be achieved.

“The Google and Apple API [application programming interface] limits the amount of raw information that is actually exposed to the app,” Prof Mathias Payer told the BBC.

“For maximum utility, we would get all the different measurements, but this has privacy implications.”

Image copyright Getty Images Image caption Swiss MPs have given permission for the country’s contact-tracing app to be launched nationwide

Apps using Google and Apple’s tech do not get to see the actual signal strength but rather one of three values, based on calculations used to normalise the different ways Bluetooth behaves on different handsets,

By contrast, the UK team can currently obtain the measurements directly.

Those responsible believe a further advantage of their centralised approach is that the data can be processed on the server involved, since it would be too taxing a task to be done on smartphones.

But part of their challenge is communicating this to Baroness Dido Harding – who heads up the wider Test and Trace programme – and 10 Downing Street itself.

A spokesman for the prime minister declined to comment.

Now the Times

Tracing app cuts spread of coronavirus on Isle of Wight

Oliver Wright, Policy Editor

Contact tracing and the NHS’s covid smartphone app have identified infections and reduced the spread of the virus, official figures suggest.

An analysis of the integrated system, that is being used on the Isle of Wight, appears to show that it has succeeded in containing the infection during a month long trial.

The findings are likely to be highlighted by an official evaluation of the project to be published next week and will increase pressure on the government to launch the app across the country.

There have been reports that Downing Street is cooling on the new NHS Covid 19 app amid delays to the nationwide roll out.

However this was denied yesterday by senior government sources who insisted that the app still had an important role to play in containing the virus while easing lockdown restrictions.

Official government data of confirmed cases of coronavirus, broken down by local authority areas, show that at the start of the trial on the Isle of Wight on May 7 there had been 154 cases on the island.

Over the following month, during which time the app and tracing system was in place, a further 47 cases were identified – a rise of 30.5 per cent.

Across England, during the same period, the number of confirmed cases of Covid 19 only rose by 12 per cent.

But the data also reveals that the vast majority of the new infections on the island were picked up in the first few weeks of the trial.

How the NHS Covid-19 contact tracing app works

Between May 7 when the system was put in place and May 26, 45 new cases were identified. However since then only two new infections have been identified.

This, experts said, could be evidence that the integrated system was successful at picking up more cases early and eventually halting the spread of the virus on the island.

The figures could be influenced by the greater use of tests on the island, which were only made available nationally to everyone with symptoms nationwide on May 18.

Nevertheless the data does give hope that the integrated system can work to control infection rates as the lockdown begins to ease.

Paul Hunter, professor of medicine at the University of East Anglia said the figures suggested that the rate of infection on the island had declined “more rapidly” than might have been expected.

“It is very hard to draw a definitive conclusion because the overall case numbers on the island are small which makes it hard to prove in a statistically significant way,” he said.

“But it does look promising.”

The island’s MP Bob Seely described the results as “fascinating”.

“Clearly these results need to be studied carefully, but one interpretation of this data may be that the app helped to identify cases – as it is designed to do – and by identifying more cases in the population more quickly it has helped people to take action earlier which has prevented the further spread of the virus, which has resulted in a steeper decline.

“One should be cautious, but it may show that the Island has not only provided a place to test the app effectively, but has also shown the effectiveness of the app too.”

The data comes amid reports that Downing Street is cooling on the concept of the app after initial enthusiasm.

Senior government figures admitted that they had perhaps been “too keen” on the app to begin at the expense of more traditional contact tracing.

“If you look around the world you see that the emphasis everywhere is on more traditional techniques to trace contacts. That doesn’t mean the app doesn’t have a role to play but it is not a single solution.”

Supporters of the programme point out that as lockdown restrictions ease so more people will come into contact with people that they do not know making the app more valuable as a way of controlling the virus.

They also denied suggestions that NHSX, which developed the app, was preparing to scrap it in favour of a system designed by Apple and Google.

“To change tac at this stage would lead to significant delays and there are no other advantages in doing so.”

Lord Bethell of Romford, the minister for innovation, said: “We will publish an evaluation soon explaining the lessons we have learnt and the changes to the app we have made as a result.

“The app will complement the NHS test and trace programme, helping control the spread of the virus by rapidly and anonymously alerting contacts who users may not know, and who would be missed through standard contact tracing methods.”