Theres is a widespread agreement that technology must be part of the solution to the COVID-19 crisis, but what are some of the difficulties and risks associated with digital contract tracing apps, and what problems cannot be addressed by these means?
As the crisis around Covid-19 unfolds around the globe, many countries are moving between two different types of intervention; lock-down, a massive and massively expensive position of last resort where an epidemic is potentially or actually raging out of control and what has come to be called the ‘dance’ where societies open back up, albeit with restrictions and mitigation strategies to help manage and remove the threat.
The complex social, political and economic ‘dance’ as we seek to manage a deadly threat to lives and our way of life is now playing out and can be expected to dominate for many months to come. The core idea here is that each country will, for political, economic and social/historical reasons interpret and act on the scientific evidence and technical options in different ways. Contact tracing applications are no different from any other aspect of this situation; the when, what, how, why and with what effect their application will have varies hugely between countries and indeed within some.
In order to restart something akin to normal daily life and restore the economy in a society experiencing a dangerous epidemic that is out of control, or at least was, much needs to be done including building a capacity to conduct testing and contact tracing at scale and speedily.
The country needs to be able to test, at a minimum, all people with symptoms and locate their contacts. A system and method needs to be in place to deal with the implications of this scale of testing. Something of the order of 3% of tests might turn out positive. These people need to be identified along with as high a proportion of their contacts as possible (say 70% to 90% of them) and isolation and quarantine result.
Having identified a clinical case by testing any of their recent and significant previous contacts need to be identified within 72 hours or less as further delay defeats the purpose of this approach. A national system staffed by thousands of trained tracers is required to do that; and also technology. Testing and contact tracing provide the basis for the mitigating actions; isolation for identified cases and quarantine for their immediate contacts. It is these actions in combination that will reduce the number of cases and open the country.
To attempt to open up a large, densely populated country before these resources are in place is to invite disaster so the rest of this post is written as though these preliminary conditions have been met. At the time of writing however, in early May 2020, this has not yet been achieved in the UK. The country does not have the trained tracers, widely available rapid testing nor indeed (yet) a robust tracking application all in place and ready to roll.
There is widespread agreement that technology must be part of the solution here and this quick note is an attempt to look at some of the problems and issues that have and are arising around these technologies. To summarise the issue we are concerned with here it is item 3 on the list below; contact tracing, remembering that all of these measures need to be undertaken together for this approach to mitigation to work:
- Testing, to find out who is infected
- Isolation, to prevent the infected from infecting others
- Contact tracing, to identify the people with whom they’ve been in contact
- Quarantine, to prevent these contacts from infecting others
To put it mildly there is a great deal of disagreement about the role and utility of proposed technologies and the trade-off involved in their use, especially around efficacy, privacy and individual rights. It would be unhelpful to add to this clamour, a level of noise that its already obscuring – as far as the public is concerned – what to do for the best.
Under the present circumstances clarity of exposition and of purpose are supremely important. So this blog post is offered with all humility and a health warning. The science and social/technology aspects of contact tracing is evolving quickly. This is a real world problem and any science here is applied science, all practitioners are having to answer complex questions without definitive answers at this point in time. Certainty, however is not in short supply, numerous commentators and indeed players in this situation are very certain as to what should be done and also certain that other less honest or well-informed actors are entirely wrong in their opposing views.
These controversies range from the ethics and politics of contact tracing right down to the question of how to manage the process of opening up the country given that this is a political decision. However much it may suit some some actors to claim that the way forward is being drawn only with reference to “the science”.
“Most of the bluetooth contact tracing apps built today are amazing pieces of technology that will be useless unless they get some fundamental changes.”
“In the age of the coronavirus, more AI-powered surveillance and tracking is inevitable, but the trick is how to do it constructively.”
Digital Contact Tracing Apps
Globally, the work of the two US technology companies with 99% of the market for mobile telephone operating systems; Apple and Google, has dominated the discussion. They have been lauded, vilified, rejected, embraced and criticised by health researchers, politicians, ethicists and others for the technologies they have proposed.
It is worth quickly making it clear that these two companies have not themselves set out to make public health contact tracing applications as such. They have not sought to provide public health actors grappling with the impact of Coronavirus with all of the data they might require in order to conduct the dance, the opening up of society safely and quickly. Rather they have begun by creating the API’s (Application Programming Interfaces) that will allow apps built by other parties to be created and distributed via the app stores maintained by the two companies.
However the two companies have made very serious design choices which run counter to the expectations and demands of some health workers, researchers and politicians. Put simply, they have chosen not to acquire or store centrally a continuous record of their customers location data. Any decisions to release the contact data stored on the customers phone is a decision for the individual not the company. What they are providing is a notification system not a public health surveillance tool. Apple and Google’s project attempts to balance privacy with the usefulness of tracking potential exposure. The name of the project: “Privacy-Protecting Contact Tracing”, highlights the significance of the design decisions they have made.
“Privacy, transparency, and consent are of utmost importance in this effort, and we look forward to building this functionality in consultation with interested stakeholders.” (Apple & Google, April 2020)
One strand of the criticism directed at Apple and Google seems especially unfair in that they are being accused in some quarters (Washington Post, May 2020, paywalled) of denying health researchers access to the data that they themselves have, or will acquire and store. This is expressly not happening and the two companies have gone to great lengths to explain that they will not be doing so. The system they have designed is encrypted with data stores locally on a users phone and not central servers. Data can be shared between users and their doctors or whomsoever might operate an application built upon the API’s but that would be at the discretion of the person who owns the telephone.
Apple and Google have elected to design a system which can be used to notify individuals if they have been in contact with people or persons who have themselves tested positive for the virus. The system is not designed to provide continuous tracking data on all private individuals who download and use apps built using the framework they have designed.
Two points from this debate are relevant to what follows below; first the two companies have made it clear that they needed to build an application that would not seriously impact battery life and thereby reduce the level at which applications are downloaded and used. Arguing against continuous monitoring and data transfers that would reduce the utility of the phone for other purposes.
Second that they believed themselves to be limited by the requirement to meet their customers expectations with regard to privacy and the law of the land in countries in which they operate.
It is obvious that considerations affecting the rate of uptake and usage of these applications is very important because of the need to provide sufficient coverage in order to impact the spread of the disease. If people’s phones do not work because of the way in which location tracking is implemented, then the system would not be of much value to anyone not least because too few people would actually use it.
Similarly if the company violates its customers expectations of privacy in an arbitrary way that might influence trust both now and in the future it will influence uptake. It has been suggested that people will not mind a decision being made for them as to who should know where they have been and for how long under the present circumstances. But the fact remains we do not know the answer to this question because no one has asked, which speaks loudly as to the status of the rights of individuals in some quarters.
Finally there is the question of authority of the state to implement more powerful tracking if they so chose and indeed that looks to be the route taken by some countries. As has been suggested here in the UK the government needs to pass primary enabling legislation if they wish to do this. It is not feasible for governments, especially within the EU, to pass the problem on to Apple and Google. In any event it seems entirely reasonable for Apple and Google to have drawn the inference that they should obey existing EU law in drawing up their proposals. The preamble to the GDPR regulations make clear the expectations that their initiative must meet: “Natural persons should have control of their own personal data.” That’s the very core of the design choices that led to Apple and Google’s system.
The question as to what contribution can digital contact tracing apps make in the present circumstances, in managing the dance as it were, is still up for debate and hence this post. First because it is not clear as to what it is that they should actually be doing as we have discussed above. But also because we need to answer some pretty basic questions as to the parameters that will influence their efficiency and effectiveness.
Setting aside the question as to the extent and degree to which public testing and in particular rapid, targeted testing and subsequent action can take place. Tracking, whether involving modern apps or proven methods using conventional tools, tested in other campaigns such as those against Ebola and SARS needs to meet some basic requirements. Essentially any effort to test and trace requires that the available methods and resources can acquire a useful minimum number of contacts within a useful period of time after a positive test.
The Public Health Imperative
in order to understand the impact of technology upon tracing in a pandemic such innovations have to be considered in context. In practice tracking apps have not formed the cornerstone of any successful strategy but they have been employed as part of a stringent lock-down monitoring in countries such as China, S.Korea and Taiwan. A failure by someone quarantined at home to keep the phone on and charged could lead to a visit form the police in Taiwan back in March.
The use of the apps for more general tracking and as part of tracing efforts has evolved rather differently. Here the aim has been to supplement the core strategy of tracing contacts based on established methods essentially human workers largely using telephones, field interviews and variants of customer management software systems to find and notify relevant contacts of the newly infected. The fact is that this work on the ground is a very human activity and indeed in some settings it cannot be otherwise. If the contact tracers cannot be drawn from the relevant community and they should be, then they will need knowledge of the settings and the circumstances of people they engage with.
To give a few obvious examples of how and why the technology cannot be the only answer, in any sense and indeed will form only a limited part of an effective response we need to look first at people and geography.
The first law of geography states that “everything is related to everything else, but near things are more related than distant things”. In order to navigate the geography of a rapidly spreading and entirely invisible illness an understanding of this basic law is required. In San Francisco the homeless people of the city proved to be particular vector for the disease and the task of entering the community and finding contacts a sensitive task that required skilled staff and trust. Certain sections of the Jewish Community in New York have behaved in ways that deeply threaten the health and indeed the lives, particularly of elderly members of the community.
Tracking contacts here is anything but straightforward as funerals and other services are being held not in private but in secret and in defiance of the law. That these occasions create contacts that spread the disease is beyond doubt. In fact many of the what are coming to be called super spreader events; such as a choir practice in the Massachusetts that left 68 infected and 2 dead or a funeral service in Jerusalem similarly point to the importance of an understanding of people and geography.
The problem is that the in the absence of the ‘magic bullet’ tracking app which does not yet exist and in all likelihood will not arrive in time, human understanding and contact are and will be required, if civil society is to reopen. It is reasonable to point out that in the UK and the USA it is just those sections of the population most at risk of severe symptoms and death resulting from exposure to the disease that do not in fact use or own smartphones. Only 49% of adults aged 75 and over had used the Internet in the last month at the most recent survey here in Wales (ONS, 2019), this rises to 79% in those aged 65 to 74. Mobile phone ownership follows the same trend line. In fact other communities such as the homeless betray a similar divergence from the trend towards universal mobile phone use. However the iron law of distance also benefits those charged with tracing, frequently the most important targets; people with whom the new patient has been in extended contact with are workmates and family. This works in favour of human based tracking systems; being able to establish trust and recruit patients and their contacts is perhaps more important than the limited, but not useless bluetooth tracking technologies.
If we have left any doubt concerning the position of digital tracking technologies in the ‘mix’ of solutions, let one of the lead engineers on the first widely deployed tracking Apps, Tracetogether used in Singapore put the matter succinctly;
“If you ask me whether any Bluetooth contact-tracing system deployed or under development anywhere in the world is ready to replace manual contact tracing, I will say, without qualification, that the answer is, ‘No,’” – Jason Bay TraceTogether app.
“Any attempt to believe otherwise is an exercise in hubris, and technology triumphalism. There are lives at stake. False positives and false negatives have real-life (and death) consequences. We use TraceTogether to supplement contact tracing—not replace it.” (MIT Technology Review, 2020).
What digital tracking applications can do is carry some of the burden, not least because tracking is effective when it covers as many people as possible. The painstaking and human centred approach can be bolstered by the speed and scale of digital apps under some circumstances. But even there the fundamental limits of the technology, such as the inability of Bluetooth beacons to distinguish between vertically separated contacts, to readily and accurately monitor people within an effective radius which varies dynamically from setting to setting must raise problems. This is an Apple practice and experience will count for a good deal in developing an effective system. In the USA practitioners and managers who served on programmes to manage the Ebola outbreak have been recruited to train and supervise new staff.
The coronavirus pandemic in its consequences in our society is not a discrete event but a combination of systemic, infrastructural degradation over a period of years. What Etic Lab in another recent report referred to as the desertification of civil society institutions (Etic Lab, 2020).
The outcomes we’re currently coping with may appear sudden but have developed over many years. Decisions made in the design, funding and structure of our social/health-care, political, economic, and civil society systems are now leading to unintended consequences at scale and quickly.
In essence the stability and indeed the sustainability of our society in the face of extreme external stressors is being tested. What we are finding is that our systems are not resilient, indeed they are prone to failure and cascading failures at that. It is essential however that as we enter the phase of the ‘dance’ of reopening society within a pandemic, that the public health task of testing, tracing and isolating be successful and quickly. If there is one thing that the horrific loss of life in the care home sector has shown, is that time is a key asset. Delays in implementing a comprehensive and nuanced test, track and isolate system will prove expensive indeed.
It is deeply worrying that at the time writing, mid-May 2020, the UK does not have such a system in place now two months after they have been deployed in other countries across the world. Technology alone cannot solve the problem but good public health care practice has been shown to be capable of addressing such issues, with access to all of the knowledge, skill and resources required.