Margaret Bourdeaux, Mary Gray, Barbara Grosz
Given the ongoing sense of urgency, some, particularly in technology circles, wonder if waiting for a healthcare worker to track down the virus makes sense. In a fight that depends on quickly finding and containing the coronavirus, couldn't we do more with our mobile phones to put our country in a better position? Could the digital technologies that already connect us—sensing and tracing our every move—stop infected people in their tracks and help skip the step of large armies of human contact tracers having to call people? Unfortunately, this line of thinking has dominated much of the debate around how best to apply technology to contact tracing in the U.S.
In many ways, the phrase contact tracing elides the harder and most important part of a contact tracer's work. Identifying and tracing potentially infected people are only the first steps in the process. Frontline human contact tracers need to persuade people, particularly those at our community's margins—whether undocumented workers or the elderly living alone—to disclose a Covid-19 status that could stigmatize or further isolate them. Much of the work in stopping a wildly infectious disease involves communities and healthcare workers quickly mustering the resources to monitor and manage patients' health. Yet these healthcare heroes have never had technologies built to specifically make the challenges of their jobs less daunting. That is why we need the HCI research community, well versed in human-centered design, to take the lead on building the technological innovations to trace and treat Covid-19.
Contact tracing is a core element of all plans for safely easing the social distancing and shutdowns resulting from the pandemic. For more than a century, this epidemiologists' tool has proved the key to saving lives threatened by infectious diseases. By systematically mapping those exposed to contagious infections, it enables the containment of disease transmission. The success of contact tracing depends not only on meticulous data collection but also on the effective counseling of the people exposed and at risk. These two critical components must roll out as a coordinated effort. The people cornered by the pandemic are vulnerable and fearful for themselves and their loved ones. That is why successful contact tracing requires the counseling skills of trained healthcare workers connected to the communities that they serve. Perhaps more surprisingly, the data collection does too.
The magnitude of the labor force needed for contact tracing is daunting, leading some to propose that technology should take over. On Friday, April 10, Apple and Google announced joining forces to reduce the need for human contact tracers and allow people to take contact tracing into their own hands. These two tech companies, and many others, aim to empower anyone with a cellphone to anonymously text their Covid-19 infection status and to receive text alerts if they have come close to someone infected with the virus. Other computer scientists  and technologists  are developing methods for ensuring privacy and anonymity. But there are growing concerns that states and consumers might end up with poorly built apps that either leak information  or are simply ineffective . We—a computer scientist, a social scientist, and a physician who is involved in standing up the contact tracing program in Massachusetts—have deep reservations about the effectiveness and equity of this particular approach. We do, however, believe that human contact tracer teams would benefit enormously from a different approach to deploying technology to meet the challenges of contact tracing. Medical communities and public health depend on bringing a human-centered design approach to the challenge of large-scale contact tracing. The HCI research community could develop the technological support systems that healthcare workers and their patients will need to weather the threats of Covid-19.
Let's review why digital contact tracing apps are, at best, a partial solution.
Cellphone data will miss, for example, data from an infected person who leaves their cellphone behind when they go grocery shopping. It will also miss the millions of Americans who don't have their own cellphones or live in rural parts of the country with limited cellular or Internet access. More crucially, cellphone tracking alone cannot accurately report the nature of contacts. High-risk contacts range from being within six feet of an infected person who is not wearing a mask for longer than 10 minutes to touching a contaminated surface. Irrelevant contact data collected by technology on autopilot will needlessly consume precious human contact tracers' time. The data collection needed for successful contact tracing relies on trained healthcare workers who are able to assess the potential risk associated with each interaction. This in turn requires the ability to patiently encourage someone who is sick and anxious to remember who they interacted with under relevant conditions.
Most problematic, however, is assuming that the nuances of contract tracing can be reduced to simply tracing networks of contacts from location data. To succeed, contact tracing programs require that people trust the entity to whom they are reporting. Trust is built on empathy , patience, and the ability to help someone who has just been exposed to a life-threatening disease. The delivery of troubling or frightening health news takes more than the ping of a text message. Human contact tracers are able to guide a rattled parent to think through who their child might have played with at a neighborhood potluck two weeks ago or help an undocumented immigrant find support and care should they fall ill. They also convey understanding and help people marshal the resources they will need to sustain a 14-day quarantine after they have been exposed. Thus, contact tracing hinges on deeply human exchanges. There is no app for that.
Contact tracing hinges on deeply human exchanges. There is no app for that.
Too narrow a focus on cellphone location technologies could distract policymakers from building out the ranks of workers critical to implementing a comprehensive contact tracing strategy. And a focus on this use of technology alone could distract computer scientists and systems engineers from developing the technologies that could make contact tracing more efficient and effective. These monumental contact tracing efforts require technologies built to assist rather than replace the legions of healthcare workers necessary for the months that there is neither a treatment nor a vaccine for Covid-19. Technologies, properly designed to support healthcare personnel in the tasks their work comprises, will be essential.
The Massachusetts Covid-19 Community Tracing Collaborative (CTC) has deployed the first comprehensive contact tracing program in the U.S. The CTC combines the leadership and state resources of the Baker administration's Covid-19 Response Command Center, the state Department of Public Health, and the Massachusetts Health Connector with the deep domain expertise of Partners in Health, a Boston-based healthcare nonprofit known for tackling outbreaks of Ebola, HIV, and other deadly infections around the world. Partners in Health is training thousands of healthcare workers to fan out, virtually, across the state to begin the hard work of contact tracing. To determine the ways in which technology can help requires first understanding the four pillars that make contact tracing effective. These four key elements are:
- Identifying who has tested positive for Covid-19
- Helping those who have tested positive identify people they may have exposed to Covid-19
- Contacting those exposed and at risk of contracting Covid-19, connecting them with testing facilities, and counseling them through effective self-quarantining measures
- Finding a safe place to shelter those who cannot quarantine or isolate at home, equipping them with the resources they need, and monitoring their health and well-being.
The fourth pillar is especially challenging, because of the diversity of people needing such help—including frontline medical personnel, people fleeing domestic violence, and the homeless. We will need technologies that connect people to ongoing community-based support and guidance that's relevant to their needs and worthy of their trust.
Computing technologies have great potential to assist with the work entailed by each pillar. For them to be useful and effective, though, requires that technologists work with epidemiologists and expert contract tracers to understand their work and their workflows, and from this, to identify their information needs and determine the best ways to meet them. The development of these technologies must follow established methods for human-computer interaction design, including the critical steps of iterative design. Otherwise they may make contact tracers' work more difficult and time consuming in ways that electronic health records have done for doctors and gig-work platforms have done for gig workers .
To be helpful, contact tracing technologies must be designed to work well for the people using them—namely, both human healthcare tracers and ordinary citizens whose contacts must be traced. It involves more than dropping a virtual pin on the map that logs the location of an infected person. Technical specs, like Bluetooth ecosystems or applications of what's called differential privacy prioritize anonymity. But those technologies cannot get at the nature and quality of contact with any accuracy.
Contact tracing involves talking with people, bringing empathy and care to a call and evaluating and assessing the type of contact someone might have had with another potentially infected person. Until the world's population has herd immunity (hopefully through effective and equitable vaccination programs), we will need contact tracing. This contact tracing will involve building rapport and keeping in touch with people so that when an inevitable outbreak throws us into a panic, there's a calm, well-trained professional to guide us through the best next steps to protect ourselves and each other. In these cases, there's no technological substitute for the distinctly human capacity of a healthcare worker.
So, what kinds of technologies could make a difference? They must be intentionally built to assist rather than attempt to replace the most vital piece of contact tracing: the caregiver in the healthcare loop. Tech designed to help healthcare workers coordinate care for families and friends in the midst and aftermath of the virus will win the day. We list a few suggestions below, drawn from computer science and engineering research, that point toward the kinds of computer-human collaborative work technologies from which a newly recruited army of human contact tracers (HCTs) would benefit as they attack Covid-19 head on:
- Dynamic electronic reference tools that provide well-indexed access to answers to likely challenging questions, thus helping contact tracers offer consistent responses when asked something they aren't sure how to answer
- For long-term monitoring, shared HCT records could assist teams in collaborating as they muster resources for someone who cannot quarantine at home and in routing care where needed
- Secure, shared databases and local area networks to support networked HCT teams doing intakes. HCT teams will need to coordinate initial calls, particularly to vulnerable, hard-to-reach groups, so that more than one person with the necessary language or cultural background can reach out, multiple times, for data collection
- Last, health workers and their agencies, across private-public partnerships, will need secure, centralized data storage that they can, collectively, trust to share information at the local, state, and federal levels.
Glamorizing the promise of tech and assuming it can substitute for the healthcare worker at the core of contact tracing's effectiveness could waste more than time—it could cost lives. Who among the millions of Americans with either poor Internet access or no smartphone at all will we miss along the way? And, more important, what is in place to help people comply with quarantine and isolation directives?
As is well documented in epidemiology, those who have the most to risk when asked to share information about themselves can be the hardest to reach in a pandemic. Think of the risks an undocumented delivery person carries right now. They are in the thick of the threat that this virus poses because they are bringing food and comfort to anyone who can afford to order home deliveries. These workers need more than an app to navigate the risks that they face if they fall ill with Covid-19 and must come forward. Data collection, even the most privacy attuned, is only the first step. Tech innovation could be a game changer in the hands of health-work professionals. Teams of healthcare workers, assisted by technologies built to support their workflows and teamwork, will be critical in the coming months. Ultimately, the best technological interventions will be the ones that ask epidemiologists how they do their best work. It is worth investing in the contact tracing technologies we need not just for Covid-19 this year but also for the inevitable next pandemic.
It is not too late to make a coordinated offensive strike. We could combine the power of human-centered tech design with the irreplaceable kindness of frontline health workers to help us keep Covid-19 at bay until we have a vaccine. In fact, doing anything less misses the opportunity of our lifetimes to unleash the deeply social potential of technologies.
3. Coronavirus: Security flaws found in NHS contact-tracing app. BBC News Technology. May 19. 2020; https://www.bbc.com/news/technology-52725810
4. Landau, S. Looking beyond contact tracing to stop the spread. Lawfare. Apr. 10, 2020; https://www.lawfareblog.com/looking-beyond-contact-tracing-stop-spread
Margaret Bourdeaux is the policy liaison for Partners in Health's Covid-19 contact tracing program. She holds appointments at Harvard Medical School, Brigham and Women's Hospital, and the Belfer Center for Science and International Affairs at the Harvard Kennedy School of Government. Bourdeaux conducts research and fieldwork focused on health systems and institutions in conflict-affected states. [email protected]
Mary L. Gray is a senior principal researcher at Microsoft Research, a Berkman Klein Faculty Affiliate at Harvard University, and a faculty member in the School of Informatics, Computing, and Engineering, with affiliations in anthropology and gender studies, at Indiana University. Her research focuses on how everyday uses of technologies transform people's lives. [email protected]
Barbara Grosz is Higgins Research Professor of Natural Sciences in the Paulson School of Engineering and Applied Sciences at Harvard University and external faculty at the Santa Fe Institute. Her pioneering research in artificial intelligence (AI) has developed foundational theories of multi-agent collaboration and applied them to the design of healthcare coordination systems. [email protected]
Copyright held by authors. Publication rights licensed to ACM.
The Digital Library is published by the Association for Computing Machinery. Copyright © 2020 ACM, Inc.