As Covid-19 spread across the United States this spring, Jodie Pond, the health director in Teton County, Wyoming, looked forward to deploying a new weapon against the pandemic. Technologists were racing to create apps that would quickly and quietly identify people who had been close to others who were infected. This process of “contact tracing” was traditionally an arduous manual task: Contacting the infected person, retracing their steps over the previous days, and identifying anyone who may have been nearby. That was nothing new for local health officials like Pond. But Covid-19 meant contact tracing on a larger scale larger than ever. So the brightest minds of Silicon Valley proposed to harness the power of the smartphones we carry with us each day to identify potentially exposed people automatically.
To Pond, that plan looked especially useful. Teton County has 25,000 residents spread across an area twice the size of Delaware. But it also sits on the doorstep of Yellowstone National Park and attracts about 40,000 people a day from elsewhere during the summer. A smartphone app would help her small team of contact tracers find people exposed to the virus more quickly.
Soon, though, Pond realized it would not be that easy. She initially chose an app developed by MIT researchers that relied on GPS signals to track people’s whereabouts. But it wasn’t accurate enough to identify who else they may have exposed. So in July, when Covid-19 cases in the county spiked, Pond’s small department traced contacts manually and individually. “We were completely overrun from a contact-tracing capability,” she says. “We didn’t think we would get the lid on.” She hired more tracers and made some calls herself on weekends. The county imposed a local mask order—the only one in the state—and her office worked to arrange more testing at the local hospital. The surge died down.
Pond still wanted an app in time for the winter ski season, when people would be thrown together on lifts and in lines. But in the intervening months, digital contact tracing had changed. It had coalesced around a system developed by Apple and Google that used more precise Bluetooth technology, and her developers planned to use it. So she set up training for her contact tracers, reorganized the office’s record-keeping systems, and waged a local campaign to encourage residents to download the app.
Then, last month, she was jolted again—by a press conference 400 miles away in Cheyenne: The state health department, which handles contact tracing for some Wyoming counties—but not for Teton County—had signed on for a different app, called Care19. She had never heard of it. But under rules set by Apple and Google, it was the only app that would work with Bluetooth on Apple and Android phones in Wyoming. After months of preparation, Pond would have to start from scratch. She wasn’t sure if she would.
“It’s been difficult, and it’s taking a lot longer than we thought,” Pond says. “We’re a tiny little health department, and we’re just trying to do this on our own.”
Pond’s struggles encapsulate how digital technology has so far failed to stem the worst pandemic in a century. In the US, the shortcomings of digital contact tracing are tied to, and reflect, a haphazard response that has left more than 180,000 people dead, and more than 6 million others infected. Contact tracing relies on plentiful, cheap, and rapid testing, so both the infected and those exposed can quickly be identified. Most of the US lacks that capacity. It also depends on public trust in the government and the health system, along with a social safety net for those who are forced to miss work or need childcare when quarantining. All of those are also in short supply.
The fragmented US response is its own stumbling block. In European countries, where the pandemic response is coordinated nationally, releasing an app and tying it to the existing public health system has been fairly straightforward. But here, policies, attitudes, and even software systems differ between and within states. Often, they conflict—think fights over state and local mask policies, or guidelines for who should get a test. Sometimes state governments handle contact tracing. Sometimes counties do. Or a city, or a college campus. Often, as in Wyoming, it’s a hybrid. Without a national strategy, strapped local health departments must develop their own plans.
“It’s not the apps that are the problem. It’s the fractured nature of contact tracing and the political issues. That’s the mess,” says Tim Brookins, the lead developer for the Care19 app, which is used in Wyoming and North Dakota.
When Apple and Google released their technology in April, they planned to be hands-off. The companies, whose operating systems control virtually all US smartphones, would provide the underlying technology and guidance to states. But states would select developers to design and launch apps—and sort out how to integrate them into their local health systems. That was meant to give states flexibility, but instead it turned digital contact tracing into a political football and a headache for public health officials. Only six states—Alabama, Arizona, Nevada, North Dakota, Virginia, and Wyoming—signed on to that vision. Forty-four did not.
Last week, Apple and Google tacitly acknowledged shortcomings with that state-by-state approach. Now, instead of relying on states to develop their own apps, Apple will offer an option built into its operating system, and Google will provide custom apps for each state. States will still need to opt in, but the tech companies will take care of more of what is solidly in their hands: the technology. Still, getting those apps working at a large scale, so that they become useful tools to public health officials, will require overcoming the nation’s patchwork pandemic response. “It only works if we have the public health infrastructure to get out of the pandemic, which we do not have,” says Harper Reed, who was chief technology officer for Barack Obama’s 2012 campaign.
One thing Apple and Google did not change last week was their approach to privacy and security. Their system was designed to make it impossible to identify individuals or to track their location. Devices share randomly generated codes via Bluetooth with compatible devices nearby. When a user receives a positive Covid-19 test result, and has it verified by a local health authority, they are given a passkey to send to a central server. This causes their unique codes to be broadcast to other phones, so that those who may have been exposed get alerts. In theory, no one should be able to tell who exposed you to Covid-19, or where and when.
In late April, in an effort to soften the idea, Apple and Google rebranded their effort from “contact tracing” to “exposure notification.” To avoid confusion, the companies said governments could give only one developer permission to use the protocol in each state or country.
The system isn’t perfect. Studies at Trinity College Dublin found that apps using the Apple-Google method performed inconsistently on buses and trams due to interference. They suggested changing the threshold for what qualifies as “exposure” to being within 6 feet of an infected person for 10 minutes, as opposed to 15 minutes.
Still, some countries rushed to develop apps using the system. Germany and the UK abandoned apps they had built in favor of new ones based on the Apple-Google scheme. In Ireland, around 40 percent of smartphone users have installed the country’s Covid Tracker app. Gar Mac Criosta, product manager at Ireland’s Health Service Executive, says the system has sent out more than 800 exposure notifications since it launched on July 7, just before a recent spike in cases.
Mac Criosta says the Irish effort has faced pushback over privacy and security, but the government has responded with a coordinated national plan and consistent messaging. He says those responsible for designing and developing national Covid apps meet regularly on Zoom to discuss ideas, problems, and successes. “It’s beers on a Friday night with guys from all over the world,” he says. No one from the US has attended.
In Switzerland, 35 percent of smartphone users have downloaded the national contract-tracing app. Carmela Troncoso, an assistant professor at the Swiss Federal Institute of Technology who is leading the country’s digital contact-tracing effort, says 26 people recently reported testing positive, and went into quarantine, after receiving an alert via the app. Others may have received alerts and chosen to self-quarantine, she says.
Switzerland is reporting several hundred new positive cases a day on average. Troncoso and her colleagues believe the app is cutting off some crucial new lines of infection. “Every single positively tested person in quarantine makes a difference,” says Salathé Marcel, an epidemiologist involved in the country’s Covid-19 response. “Because every single such case could potentially have kicked off the next super-spreader event if they had not been in quarantine.”
It’s unclear how many people need to use an app for it to be effective. In April, an epidemiological model from researchers at Oxford University made headlines for suggesting that 60 percent of a population would need to use a contact-tracing app in order to stop an outbreak. But the same study showed that even lower levels of adoption, like those seen in Ireland and Switzerland, could help quell new infections.
Troncoso says her country’s experience shows government support for contact-tracing is vital. For instance, after some employers told staff not to use the app for fear that they would miss work, the government introduced a scheme to cover absentee workers. “I wouldn’t say that Switzerland is a success story yet,” Troncoso says. “But we are headed towards success.”
In the US, efforts at contact tracing have become mired in the same politics afflicting other aspects of Covid-19. Consider South Carolina, which announced plans to develop an app using the Apple-Google framework in May—one of only three states to do so at the time. Just as quickly, it was over. In June, just before the legislature’s summer recess, lawmakers added language to a Covid-19 spending bill that banned state agencies from using digital contact-tracing apps. “There was a fear by some that this contact-tracing platform would not be able to be turned off, so you could be followed around with your phone,” says Mark Sweatman, a lobbyist for the Medical University of South Carolina, which was slated to develop the state’s exposure notification app. By mid-July, the state’s outbreak of Covid-19 cases had surged beyond public health officials’ ability to manually trace cases.
In part, such suspicion reflected bad experiences elsewhere. Utah, for example, spent millions on a GPS-based tracing app in April. Few people downloaded it, county public health departments declined to use it, and the state later turned off its location-tracking features. Other location-based apps either proved insufficiently accurate or were found to send location data to third parties. The opponents in South Carolina were not convinced that the Apple-Google system would be more privacy-preserving than GPS-dependent apps.
The patchwork US response has led to strategy conflicts. In Teton County, where much of the pandemic response has been local, officials took their own initiative with an app, only to have a state effort usurp them. On the other hand, at the University of Illinois, researchers designed a Bluetooth-based exposure notification app for the campus, but found it couldn’t use the Apple-Google system because the state has not selected a developer to make one. Sweatman says the Medical University of South Carolina has fielded inquiries about using its app on college campuses, but it can’t offer the tool because the state health department has not blessed it. (The university says it is in discussions with the state for permission to begin a research trial of the app.)
States such as Massachusetts and New Jersey have left their app plans in limbo or abandoned them to focus on the bigger picture. “States are struggling to stand up a public health response in general, including plain old vanilla contact tracing,” says Margaret Bourdeaux, a researcher at Harvard Kennedy School. Inadequate testing in some areas has undermined even traditional contact tracing. States that launched large-scale contact-tracing efforts have run into problems ramping up and managing these efforts. Researchers at George Washington University estimate that many states have fewer contact tracers than they need. California, for example, needs more than 30,000 contact tracers to keep up with the virus’s spread, according to the researchers. An NPR survey last month found the state had less than a third of that number.
Contact tracers report difficulties getting people to respond or follow advice to quarantine and seek testing, especially when they can’t also offer resources such as childcare to help. Mistrust of the government, especially among immigrants and lower-income residents, has been compounded by online misinformation about contact tracing.
“It gets to a fundamental thing where some people have started to get disenchanted with the test, trace, isolate programs to control the virus,” says Justin Lessler, a professor of public health at Johns Hopkins University who studies the effectiveness of contact-tracing programs. Rather than give in to that disenchantment, he recommends investing more to make the programs effective. “The challenges here are fast follow-up, trust, and also having quarantine and isolation procedures to help them protect others in their household.”
Digital contact tracing can bolster those efforts, he says, making it easier to reach more people quickly, especially contacts an infected person may not know—like strangers at the store, or colleagues in a different department. But if there’s little support and concerns about digital tracking deter vulnerable groups from adopting the apps, the effort can widen the gaps. “Public trust is everything during a health crisis, so anything that might risk it—especially if the benefit is unclear—is viewed as not worth the effort,” Bourdeaux says.
Early on, it had become clear to some that the US needed a more national response. “We are not Canada,” which has a national health care system and a national contact-tracing app, says Scott Becker, executive director of the Association of Public Health Laboratories (APHL). “It would be a lot simpler if there were a national approach, but that’s not how public health works in the United States.”
In May, a group of officials, including state health officers and technologists at the Centers for Disease Control and Prevention, formed a working group to tackle the problem. One obvious issue was allowing state apps to talk to one another, says Becker, who later joined the discussions. Although Google and Apple offered common technology for exchanging keys over Bluetooth, states initially built their own servers for sending notifications. That posed a problem for interstate tourists and commuters, who might be using apps hooked into different systems. (Plus, having a separate server in every state wasn’t great for security.) So the APHL, which frequently acts as a conduit for public health data between localities and federal agencies, set up a common server for state apps to talk to each other.
A bigger technical issue for states was deciding what they needed in a contact-tracing app and finding someone to build it. Apple and Google’s announcement potentially relieves states of this burden. Instead, they’ll be able to provide some basic information they wish to include in the apps, plus their desired statewide branding, and Apple and Google will generate offerings from a template.
It’s still not clear whether these new apps, named Exposure Notification Express, will prove more successful at slowing the spread of the disease. One big issue: connecting with local health officials, who often oversee traditional contact tracing. “The real friction is the way the contact tracing happens,” says Brookins, the developer of the Care19 app in North Dakota and Wyoming. “It’s the haphazard, heterogenous jurisdictions that’s difficult.”
Google and Apple’s original plan had seemed to recognize the challenges of this “last-mile problem,” as Brookins puts it, by relying on developers to build custom state apps. Brookins, for example, had set up Care19 with those local distinctions in mind. In North Dakota, where universities are back in session, each campus has its own set of guidelines and resources. So the app allows people to sign up with a school affiliation and receive local instructions. Similarly, in Wyoming, he says, with its complex overlay of contact-tracing jurisdictions, places like Teton County could customize the app’s features for local residents. That also allows public health officials to see more data on app usage and how many alerts are sent.
Pond says she’s open to using Care19 but remains unsure. Her small office doesn’t have an IT department, and she must attend to many other issues before contemplating another app. In the meantime, the original GPS app is still live on phones across the county. Recently, she trained her contact tracers to use that app as a kind of memory tool—a way to jog recollections of where infected people had been, so that the tracers could reach their contacts faster. It wasn’t the exposure alert system she had hoped for, and she still fears ski season. But in her local response against the virus, every tool she has could make a difference.
- 📩 Want the latest on tech, science, and more? Sign up for our newsletters!
- San Francisco was uniquely prepared for Covid-19
- The pandemic could derail a generation of young scientists
- Tips to treat and prevent face mascne
- Parenting in the age of the pandemic pod
- Bill Gates on Covid: Most US tests are “completely garbage”
- Read all of our coronavirus coverage here
