UK Contact Tracing App Switches Approach

Italy and Germany have both released apps using this model recently whilst Australia continues with the centralised model (with varying degrees of success), but it is unsure when it will be available in the UK.

This change follows the NHS running testing on both the centralised and decentralised models, with both displaying advantages and disadvantages. Our article in April herediscussed the NHS's initial approach in applying a centralised data-matching approach for its contract tracing app.

While the centralised NHSX model that was trialled on the Isle of Wight was able to accurately measure the distance between users, and registered 75% of nearby Android phones, the software could not effectively register Apple phones, registering just 4% of close-by handsets. This seemingly stems from Apple devices being ineffective at running software that uses anything other than Apple's own native technology, restricting the Bluetooth usage of third-party apps, which is crucial to the efficacy of this contact-tracing model.

While the decentralised Apple/Google model was able to register 99% of both Android and Apple devices nearby, its downfall was in its accuracy in calculating the distance between users. Testing in both the UK and Ireland has shown that this software can sometimes fail to differentiate between devices that are 1 metre apart, and those 3 metres apart, despite one being inside and one being outside the current 2 metre radius for social distancing (reducing to 1 metre plus from 4 July 2020 where 2 metres is not possible) meaning it cannot be relied on yet to make decisions about who should self-isolate.

Apple and Google's design has however been held up as being more friendly to data privacy concerns, not tracking the location of the user, and with device-matching taking place on the devices themselves as opposed to in a centralised server, only storing a user's Bluetooth ID centrally. However, the NHS has warned that the trade-off is that in this approach, the NHS will have severely limited visibility of health data, hampering wider data analysis (e.g. spotting localised outbreaks of Covid-19 before they happen).

As a result of both approaches being found wanting, where does the UK go from here? For now, the Test and Trace team have confirmed that they while continue to work on the new decentralised app, they will focus on manual contact tracing, with individuals testing positive for COVID-19 being asked about places they have recently visited, and details of people they have been in close contact with. This is only effective up to a point as it relies on individuals first responding to these questions, and being able to give a full account of this information. Figures released on 18 June showed that of 14,045 people who tested positive and were referred to the manual scheme, 10,192 provided details of recent contacts, a response rate of 72.6%.  That said both versions of the app still involved voluntary disclosures by individuals as opposed to any automatic contact tracing or mandatory reporting.

How can this be improved? Quite ironically, it appears only through the use of more intensive surveillance of the public. Singapore, South Korea and China have each been equally lauded and criticised re their privacy intrusive approaches, which utilise significant and wide-ranging levels of surveillance, with CCTV, bank cards and mobile phone usage each being tracked to identify who has come into close proximity with individuals who have tested positive, and therefore who to test. Whilst this has shown to be effective, the resistance felt against using a "mere" centralised tracing approach in the UK indicates that these even more "invasive" techniques would not be introduced by the NHS without considerable pushback from data privacy advocates.