Hard decisions in digital contact tracing

– OPINION –

The concept of digital contract tracing raises tough questions that require hard decisions, writes Koi Tū Research Fellow Dr Andrew Chen.

 

Since we discussed contact tracing in the media over the past week, our Centre has had many people get in contact who are developing digital contact tracing solutions and trying to get them adopted by the government. Meanwhile Dr Ayesha Verrall from the University of Otago reports that she’s “getting 20 messages a day from people who can design apps.”

This probably shouldn’t be that surprising – the technology that underpins these apps is new, but not that new, and is readily accessible to the world’s developers. Technologists tend to be doers, and for those who can’t work on developing a vaccine or manufacturing more face masks, it makes sense that they might pour their efforts into making apps.

But having so many options and apps can be overwhelming for a government already under severe pressure. Every solution has its strengths and weaknesses, but time is also of the essence because we need to have something in place before we leave Level 4. We need to remember that among the options being presented by the private sector, the government would also be considering internal development, or just focusing on improving manual contact tracing (although this may become overwhelmed if numbers increase). An options analysis has to be conducted to assess which might be the best solution in the New Zealand context, and as other commentators Craig Squire and Tom Barraclough have emphasised, it needs an inter-disciplinary approach that draws from a variety of perspectives because there are a lot of factors to consider. What might some of those factors be?

One of the most important questions is whether government contact tracing needs to cover the entire population or not. There is no doubt that effective pandemic control requires significant contact tracing; as people are infectious before symptoms appear, it is critical that as soon as a carrier is identified, that person is isolated and all their contacts followed. Speed is critical as the longer we wait, the more potential contacts they have, and the increased likelihood that they may have already passed the virus onto others. There is increasing epidemiological data and comparative observation that suggests digital technologies can assist with this process, while allowing the movement of people.

If we use cell phone location data from the telcos, we might get 90-95% coverage of the population, but that may come at a huge privacy cost. If we tell people to download an app and we get a participation rate of 20-30%, is that enough? Modelling from the University of Oxford last week suggests that speed is more important than coverage – that if we can get the time for contact tracing down from three days to minutes or hours then it drastically improves the likelihood of containing the epidemic even with partial participation. If we can use the digital contact tracing data to augment manual processes, then we can achieve a high level of protection.

In Singapore, the TraceTogether app has been downloaded by approximately one million people – roughly 16% of the population – but the government says they need closer to 75% of the population using the app in order for it to be effective. There’s a small subtlety to note in the statistic though – it counts the number of people who have downloaded the app, which is different to the number of active users who have Bluetooth on and are logging contacts (which is likely less). People may also have more than one device with the app downloaded, further dropping the actual proportion of the population using it.

We don’t have much formal evidence beyond anecdata for how many New Zealanders would use and download an app for digital contact tracing purposes. The Google Consumer Barometer estimates that 88% of the population uses smartphones. That’s reasonably high, but there’s still 12% that would miss out if we used an app-based approach, and we need to understand which segments of the population they belong to or we might further exacerbate existing inequalities. There are also certainly people who would oppose the use of surveillance to improve contact tracing, and while an opt-in tool might appease them, it could also reduce the effectiveness of any solution.

However, it is important to remember that if we use digital contact tracing to augment rather than replace manual contact tracing, then we don’t need full coverage of the population. Being able to take some of the load off manual contact tracers by finding contacts before conducting a formal interview, replacing some phone calls with push notifications, or speeding things up by sending an initial notification to people to self-isolate and to then be followed up a phone call later, could make a big difference in the overall effectiveness of contact tracing.

Another question is whether we need to know the locations of people, or if we would prefer to know about the interactions and contacts people have, or if we need both. Both the Singaporean and European Union apps are based on contacts, using Bluetooth to pick up who you have been near, rather than where you have been. But Russia is using QR codes to make people register when they leave their house, and South Korea is using a combination of security camera footage, credit card transaction records, and GPS data to pinpoint exactly where a person has been. On the face of it, a contact-based system would be better for protecting privacy than a location-based system; but which is most acceptable and suitable in the New Zealand context?

Then there are more tough factors to be considered. Are we just using this for contact tracing, or would we also want to track and enforce social/physical distancing like in the US or UK? Which solution would best complement the existing manual contact tracing processes? How quickly can we get a system deployed that works without too many errors and before the lockdown ends? Is the data stored on the device or in the cloud, and does the government have full visibility?

Who will have access to the data produced by this system, and when will we shut it down? Is data deleted automatically after a certain time? Who will provide oversight for how the system is being used? What option will give us flexibility to respond and make changes as our understanding of SARS-CoV-2/COVID-19 improves over time? What option provides an appropriate trade-off between utility and privacy? What digital security measures need to be included to protect people’s information? And what option helps us keep up morale and trust in the government in order to encourage compliance and participation? It’s not easy.

We should also talk about incentives – a lot of the discussion assumes that the government releases an app and then tries to use advertising and marketing methods to encourage people to download and use the app. The key message to push for downloading an app is to protect not only yourself, but also the broader community around you. But there have been a couple of other suggestions for how to better encourage uptake. For example, supermarkets could require you to have the app before they allow you to enter the store, as part of keeping their staff safe. Alternatively, the government could offer a $50 voucher to each person who downloads the app and ties it to their NHI number or RealMe account, which would have a small economic stimulus effect as a bonus. There are ways for us to nudge behaviours beyond advertising on TV and the internet, and it is a worthwhile investment to make.

There are a lot of people out there who want to help, and I want to recognise those who have already developed their own solutions and offered them to the government. But whatever the solution deployed by government, one thing that is important is that we all need to get behind it and support that solution. One of the greatest dangers of having many options is if multiple apps get deployed, and those apps don’t talk to each other or share information. This would lead to fragmentation of the population, with people using different apps reporting data to different people, and it would significantly reduce the effectiveness of all of the solutions. If we are going to deploy digital contact tracing, we need everyone on the same platform quickly, working together to combat the spread of this disease.

 

This article has been peer-reviewed by Peter Gluckman, Anne Bardsley, and Jill Rolston.

Andrew is a Research Fellow with Koi Tū: The Centre for Informed Futures. He comes from a technical background with a PhD in Computer Systems Engineering from the University of Auckland. His PhD research focused on the use of camera-based person tracking, as well as how might use technology to help protect the privacy of people. Andrew also has a background in civics education and policy analysis, and is now using that to explore how we can bring technical expertise with other forms of evidence to form pragmatic and practical policy suggestions that can help our society find a path forward through the digital revolution.