NHS Covid-19 app alerts 1.7 million contacts

After almost three months’ silence regarding the product’s development and the assistance that it has given the UK’s National Health Service (NHS), the developer of the UK’s Covid-19 contact-tracing app has revealed that approximately 600,000 cases have been prevented by the app since September 2020, possibly preventing 6,000 deaths.

After a catalogue of delays, missteps and a complete technological volte-face, as well as arriving over four months later than originally proposed, the UK’s Covid-19 contact-tracing app was officially launched on 24 September 2020.

Despite earlier hyperbole from a number of politicians, the app’s role is essentially to act as a tool working alongside the UK’s currently struggling traditional contact-tracing apparatus.

It forms a central part of the NHS Test and Trace service in England and the NHS Wales Test, Trace, Protect programme, identifying contacts of those who have tested positive for Covid-19 and helping to prevent further spread of the virus.

Available to smartphone users aged 16 and over in multiple languages, the app includes proximity tracing using Bluetooth Low Energy (BLE), risk alerts based on postcode district, QR check-in at venues, a symptom checker and test booking.

The contact-tracing element of the app works by logging the amount of time users spend near other app users, and the distance between them, so it can alert users if they have been close to a person who later tests positive for the virus. The app notifies contacts as quickly as 15 minutes after a user inputs a positive test result.

In building the app, the digital technology arm of the NHS, NHSX, worked closely with major tech companies, including Google and Apple, scientists within the Alan Turing Institute and Oxford University, Zuhlke Engineering, medical experts, privacy groups, at-risk communities, and teams in countries across the world using similar apps – such as those behind the very popular and successful German app.

The NHS said that almost six months into its full operational life, the app has been breaking chains of transmission to protect users and their communities since its launch in September.

The NHS has now released the results of a research study showing that the app has now been downloaded 21.63 million times, representing 56% of the eligible population aged over 16 with a smartphone, and was the second most downloaded free iPhone app on the Apple App Store in 2020.

Furthermore, more than 1.7 million app users across England and Wales have been advised to isolate by the NHS Covid-19 app following a close contact with someone who later tested positive.

As well as contact tracing and booking a test, the app allows users to check their symptoms via the symptom checker. Coronavirus symptoms have been reported into the app over 1.4 million times in England and Wales since 24 September. More than 3.1 million test results have been entered into the app across England and Wales, of which 825,388 were positive.

This is a combination of both tests booked through the app and test results manually entered into the app. However, the NHS stressed that as users are more likely to manually enter a positive test result to trigger contact tracing, rather than input a negative result, these figures cannot be used to calculate a positivity rate for the NHS Covid-19 app comparable to manual contact tracing.

Figures on the app’s wider features show users have checked into a venue more than 103 million times. A total of 253 venues were identified as “at risk” as a result of an outbreak since 10 December, triggering “warn and inform” alerts to app users who had checked into those venues, protecting app users by letting them know as quickly as possible when they may have been at risk so they can monitor symptoms. A “warn and inform” notification will only be sent if an outbreak is identified at a venue and is not equivalent to an instruction to isolate.

In revealing the new statistics, the NHS also published research conducted by scientists at The Alan Turing Institute and Oxford University showing for every 1% increase in app users, the number of coronavirus cases in the population can be reduced by 2.3%.

Looking at the epidemiological impact of the app, the Turing/Oxford researchers also pointed to a causal link between app use and reduced case numbers, with their analysis suggesting the NHS Covid-19 app has prevented 600,000 cases since it was launched.

The app operates on a principle of continual improvement and has had a series of updates since its launch in September 2020, including an update to its risk-scoring algorithm to utilise GAEN API Mode 2, which better estimates distance based on Bluetooth signal strength, increasing its accuracy and making this the first contact-tracing app worldwide to harness this technology.

Talking to Computer Weekly about the development path of the app to date since November 2020, CEO of Zuhlke UK Wolfgang Emmerich, who has been at the forefront of the app’s development since the earliest days in its development, revealed that the firm has evolved the app continuously, releasing an update almost once a week.

The most significant of these were the switch to version two of the exposure notification application programming interfaces (APIs) from Apple and Google to work with the leading available smartphones. The result is an app that has greater accuracy and delivers fewer false positive results.

“The Turing Institute devised quite a few fairly sophisticated algorithms for the first app and we have now implemented in this in this app that are currently in use,” said Emmerich.

“The way to do that was to persuade Apple and Google to provide the primitives that are required to get the right set of data out of the [product’s] Bluetooth modules. So, we switched to an API version in November following quite extensive field testing to persuade ourselves that this was a genuine improvement.

“We’ve implemented [filters] and that has improved the accuracy very considerably of that contact tracing, and so we have a lower false positive rate .We have, altogether, a better contact-tracing algorithm.”