Three tests for the COVID-19 contact tracing app

The outbreak of coronavirus (COVID-19) has been accompanied by an explosion in the use of digital technology. The health service has rolled out virtual consultations at astonishing speed and NHSX has been developing a contact tracing app. The app may soon be using data from our phones to determine if we have come into contact with someone who has tested positive for the virus, or may have been exposed to it.

The government is now prioritising the rollout of a telephone-based NHS Test and Trace service ahead of the app. This ordering has some advantages because it means that the human element – critical for any contact tracing approach –  is put in place sooner. The app, once it is available, will automate some of the steps. It will also mean that more contacts can be traced, since the app will record contacts that people may have forgotten about as well as contacts with strangers.

Experts analysing the spread of the virus have argued that ‘viral spread is too fast to be contained by manual contact tracing, but could be controlled if this process was faster, more efficient and happened at scale.’ The instant tracing offered by the app is clearly attractive – the reason why at least 29 countries are developing one. Yet the app also raises some big questions, and its effectiveness is still unknown.

With the app now nearing the end of its initial pilot on the Isle of Wight, this blog sets out three tests that must be met to ensure the app meets the goal of reducing transmission of the virus.  

1. Has the risk of unequal impact been understood and addressed?

The challenge facing any digital team is to ensure that the solutions they are developing meet the need. We know that COVID-19 is having a disproportionate impact on poorer areas, older people and some minority ethnic people, so the app needs to be designed with these communities in mind.

App users will potentially benefit from a better understanding of their risk of infection, access to tests, quicker diagnosis, and potentially greater individual freedom to travel and work. However, the ethics board advising on the development of the app have warned that up to 21% of the UK population do not own a smartphone, and the Lloyds UK Consumer Digital Index shows that 13% lack the skills to open an app. These figures raise questions about whether the benefits of the app will go to those who need them most.

Meanwhile, the app will generate false alerts and these may have unintended consequences. Some groups may be hit by repeated instructions to self-isolate, with a greater impact on their ability to work. Will this disproportionately affect people more likely to already find themselves in perilous financial situations?

With the impact of COVID-19 already being felt unequally across society, the risk that the app might have unequal impact needs to be examined and addressed. The government is supporting the DevicesDotNow campaign to target the UK’s digitally excluded during COVID-19, but NHSX needs to start reporting on the impact the app is having on different population groups (including rates of uptake, false negatives and positives, and overall effectiveness). The wider NHS Test and Trace system also needs a more comprehensive strategy to tackle health inequalities.

2. Have the evaluation’s approach and findings been shared transparently?

To some extent, the public is being asked to trust their safety, and perhaps elements of their personal freedom, to an app. So, the public should have access to information about how the success of the app is being evaluated, and what those findings are.

There should also be clarity about the underlying logic of the intervention – what are the intended outcomes, and how will the intervention deliver these? What assumptions is this based on? How will each of these things be measured, and how will an evaluation inform decisions about wider implementation and ongoing use of the app? And what decisions need to be made, when, by whom and how, and what evidence will inform these decisions?

NHSX has already made a commitment to open sourcing the underlying code for the app and has recognised the benefits this approach would bring. We now need the same level of transparency about the evaluation.

3. Are patients and the public being involved at every stage of the app’s development and implementation?

The public is very interested in the app, as is evident from the intense media coverage, but the challenge is to channel this interest into meaningful engagement throughout the entire process of developing and implementing the technology.

The data collected through the app are very valuable, not only for contact tracing but also research into how the virus is spreading and how effective (and necessary) the various containment measures are. We know very little about how comfortable the public is about using the smartphone data for this research, so public deliberation is urgently needed.

Anybody with a history of working with health data knows the critical importance of keeping the public onside. But public engagement needs to go beyond questions about how the data is used, to help to shape the impacts that the app is having. The app raises many questions, such as whether it is reasonable to expect people to self-isolate based on the results of the algorithm, or in the public interest that we carry our phones with us. Constant engagement is necessary to surface and begin to answer questions.

If the app works it will be an amazing achievement. And the chances of success will be higher if there is constant public engagement, transparent evaluation and if the risks to inequalities are understood and addressed. 

Adam Steventon (@ASteventonTHF) is Director in the Data Analytics team at the Health Foundation.