One of the tried and tested strategies to combat a pandemic is to identify those who have the disease, trace all their contacts, and isolate them to prevent spread. It used to be the task of those with good personal skills, persistence and (in the more distant past) good walking shoes. Now many governments are turning to ‘smart’ phones in an effort to control the spread of Covid-19. This article is the first in a series about one such attempt: the new app currently being trialled by England’s National Health and Public Health Services.

This app has already been hugely controversial, even before it has been released. As I’m one of the first members of the general public to use it, I’ll be examining and reporting about it here. From what I’ve seen so far, it’s not the app that we’ve been expecting.

Locking the population down in small, stable units is effective at stopping the spread of infection, but comes at great cost. Most governments are eager to limit the social and economic damage as soon as they can by easing those restrictions on movement and contact. But unless they can introduce new measures to control spread, it’s almost certain that their populations will suffer second waves of infection and deaths. In small and relatively static groups, traditional methods of contact tracing can be highly effective. But in urban areas with many using public transport, and denser groups, so many contacts are anonymous that tracing them using those methods is ineffective. Easing restrictions in those areas becomes a high risk strategy.

The app

The app is named NHS COVID-19, and is described by the NHS as “an automated system for rapid symptom reporting, ordering of swab tests, and sending targeted alerts to app users”. It’s one of three parts of the trial which has just started here, the other two being:

• “web-based Contact Tracing and Advisory Service (CTAS) and increased capacity to provide tailored alerts to all contacts by phone.”
• “widespread availability of rapid swab testing kits to make sure confirmed cases remain in isolation, and support rapid detection and isolation of higher-risk contacts.”

This is quite different from what has been generally reported: the NHS sees the app primarily for rapid symptom reporting and the ordering of swab test kits.

Installing and configuring the app is simple, provided that you have an iPhone running iOS 11 or later, or a compatible Android phone, about which I will say no more.

The only item of personal information which the app requires is the first part of the user’s postal code. According to the information given about the app’s security and privacy protection, this is combined with information about the phone, such as its model, to generate a ‘hash’ with which its data is associated. At this stage, the app doesn’t request any other details such as the user’s name or email address.

You need to give your consent to two protected features for the app to operate: background Bluetooth LE, and Bluetooth needs to be left on at all times to enable the phone to recognise other app users in the vicinity, and push notifications. The latter enable the app to warn you if any recent contact has developed Covid-19.

These are followed by Apple’s standard user consent dialogs.

Once those are complete, the app performs a self-test and displays confirmation that it’s working properly, and you’re asked how you’re feeling today. If you’re feeling well, you close the app; if you’re feeling unwell, you start its structured questions designed to elicit symptoms of Covid-19 infection.

I’m currently using version 1.0.2 of the app, which is 3.6 MB in size, and stores around 330 KB of data on the phone. It appears a light user of data connections, and a light user of the battery too.

What is strange for a contact tracing tool is that the user gets no feedback at all unless and until a close contact develops symptoms. The app doesn’t tell the user how many other users it has come into contact with, nor whether the user’s phone came dangerously close to any other user’s phone for a significant period. Indeed, the definition of such potentially hazardous encounters isn’t given, and changes with improvements in the central model being used.

The app’s self-test also doesn’t include any assessment of whether the app can ‘ping’ a nearby phone with another copy of the app. You could wander around all day with your phone shielded from Bluetooth signals, and be unaware that the app had been unable to contact any other device. It also takes no interest in whether you live with someone else with the app installed on their phone: it takes no steps to establish whether that person is in your household, therefore doesn’t need to be treated as any other contact.

The trial

Even more of a puzzle is why this app is being trialled on the Isle of Wight.

The Island has a population of around 140,000, 28% of whom are 65 and over. With so many older people, ‘smart’ phones are relatively unusual compared with areas in large cities such as London and Birmingham. This also means that around a quarter of the population are in groups which are vulnerable to Covid-19, thus have been advised to shelter and to avoid leaving their homes at all. The Island is one of the few relatively physically isolated parts of England: its only links to the mainland are ferries, whose services have been cut to a minimum to transport supplies to the Island, and move a small number of key workers.

So far, Covid-19 appears to have been relatively uncommon on the Island. According to official figures, around 0.1% of the population have tested positive to date. However, with so many older people, the death rate has been disturbingly high at around 20-25% of positive cases. Unlike much of the UK, its care homes are generally small, and few deaths have occurred in them so far.

Some of this is recognised in the information given about the trial. For instance, “One of the reasons we’ve chosen the Isle of Wight is that, compared to the UK as a whole, there have been fewer COVID-19 cases.” To carry out meaningful trials on an app whose primary purpose is contact tracing requires an abundant supply of fresh cases, with contacts to trace. Yet the trial area was selected for its low rate of fresh cases, among other factors.

Primary purpose

As far as the user is concerned, the app is about capturing their earliest symptoms of what could be Covid-19 infection. Unless they receive notifications of earlier contacts developing the disease, its sole feature is to provide them with a way of reporting their symptoms. There’s even a reward at the end of it: if they do, they’re going to be sent a test kit so that a lab can confirm whether or not they do have the infection.

I know of seven people on the Island who have possibly had Covid-19, but none was ever tested. That wasn’t because they didn’t want tests, but because they weren’t on offer to them unless they were sick enough to be admitted to hospital. I’m sure that, if they had been offered an app which recorded their symptoms and rewarded them with a test kit, each would have been only too delighted to use it and report their details.

For a government intending to ease lockdown restrictions, such early reporting and testing is crucial. It enables them to know within a couple of days where outbreaks are occurring, and whether changing rules and behaviours are leading to a second wave. Without this app, it has been taking two weeks and longer for cases to be reported.

This explains why the app’s developers weren’t interested in the frameworks offered by Apple and Google. Being decentralised, they didn’t meet the primary requirements for gathering early reports of Covid-19 centrally, however good they may have been at tracing contacts. In the UK model, contact tracing is also performed centrally, using a combination of techniques rather than just ‘smart’ phones.

Impressions

Early indications here are that this app protects the user’s privacy, doesn’t track users at all, doesn’t flatten batteries, and is unobtrusive to the point where you can’t even tell whether it is detecting contacts. It doesn’t appear to be the contact tracing app which was expected, though: it’s not ‘track and trace’ so much as ‘diagnose and test’, and may explain where Boris Johnson intends sending his promised 200,000 test kits a day.