In DepthCOVID-19

Can phone apps slow the spread of the coronavirus?

See allHide authors and affiliations

Science  19 Jun 2020:
Vol. 368, Issue 6497, pp. 1296-1297
DOI: 10.1126/science.368.6497.1296

Science's COVID-19 coverage is supported by the Pulitzer Center.

Health departments around the world are betting on technology to help stem the stealthy spread of the coronavirus: cellphone apps that aim to identify and alert those who recently came into contact with an infected person. By encouraging those potentially exposed to COVID-19 to self-isolate, the thinking goes, a phone app could swiftly cut off chains of transmission. Dozens of local and national governments have launched official apps or are developing them.

“It's very appealing—that you have an app that does all this work,” says Hannah Clapham, an epidemiologist at the National University of Singapore. But she and others warn that an app can't replace human contact tracers. “I worry that we think it's going to save us.”

So far, only epidemiological models suggest apps can change a pandemic's course. Ensuring that an app detects risky contacts without overwhelming users with false alarms is one challenge; getting enough people to download an app is another. As health officials weigh competing apps and prepare pitches to privacy-conscious citizens, epidemiologists, engineers, and behavioral scientists are considering how to put an app to the test.

People can apparently transmit the coronavirus for days before they develop symptoms, so by the time health departments learn of a case, they have precious little time before infected contacts start to spread the virus. “You have a couple days to chase people down,” says C. Jason Wang, a health policy researcher at Stanford University who works with health departments on COVID-19. And traditional contact tracing—interviewing the infected person, tracking down the recent contacts they can recall, and telling those people to self-isolate—is time consuming.

In contrast, phones could detect when two users are close enough to share the virus, and an app could alert one when the other gets sick—even if those people are strangers who happened to sit in adjacent subway seats. “The technology response is absolutely necessary,” Wang says, “and it needs to be fast.”


Embedded Image

Millions of people have downloaded apps designed to alert them to coronavirus exposure.

PHOTO: MINZAYAR OO/PANOS PICTURES/PANOS PICTURES/REDUX

Some Asian countries that faced early outbreaks have already put smartphone tracking to use. South Korea collects GPS data to publicize recent paths of newly infected people, and last week announced it would require facilities including bars and night clubs to collect identifying information from visitors' phones to share with health authorities. The Chinese government has combined location data with other surveillance to restrict citizens' movement if they may be infected. GPS location tracking is also a component of apps in India, Iceland, and U.S. states including North Dakota and Utah. But GPS technology isn't precise enough to gauge short distances between two phones. And widespread location tracking raises privacy concerns.

Many governments are instead developing apps that rely on short-range Bluetooth radio signals, which can detect close encounters but don't track movements. Each phone broadcasts an ID number to nearby phones that record such “handshakes.” If a user has symptoms or tests positive, the app can trigger alerts to those recent contacts.

To be effective, these apps will first have to reliably estimate distance, based on the strength of the received signals. In a test of Singapore's TraceTogether app, computer scientists Douglas Leith and Stephen Farrell at Trinity College Dublin found flaws in those estimates. The study, posted online but not peer reviewed, showed that when people sat across a table, the received signals were much weaker if their phones were in their pockets versus on the table. Sometimes the signal increased as people moved farther apart—perhaps because of reflection off metal surfaces such as supermarket shelves. The imprecision could lead to both missed encounters and false alarms, Leith warns.

Imperfect Bluetooth readouts can still help if interpreted conservatively, says Marcel Salathé, an epidemiologist at the Swiss Federal Institute of Technology Lausanne. He advises the Swiss government on an app now in a pilot phase that aims to detect when someone comes within 2 meters of an infected person for at least 15 minutes. He thinks his team can tune the system so that “if somebody gets an exposure notification, we will feel damn sure it's actually been a contact.”

An app can only catch interactions between people who have installed it. Singapore, which launched TraceTogether on 20 March, says roughly one-third of the population has signed on. But uptake of an app “needs to be almost improbably high to really capture all of the contacts that might be relevant,” says Allison Black, a genetic epidemiologist at the University of Washington, Seattle, and the Fred Hutchinson Cancer Research Center.

Recent modeling by infectious disease epidemiologist Christophe Fraser and colleagues at the University of Oxford predicted that if about 56% of a population used an app, it alone could reduce the virus' reproduction number—how many people catch the virus from each infected person—enough to control the outbreak. But Salathé says that, even at relatively low levels of uptake, an app could still prevent infection and save lives: “As soon as you have double digits, I think the effect is already quite substantial.”

An app's impact depends on whether people stay home when it tells them to. Human contact tracers connect a potentially infected person to social supports that help them self-quarantine—for example by arranging grocery delivery or even a hotel stay. But because most apps keep users anonymous, health officials won't automatically know who gets an alert or what they need to stay home, Wang says. And users might not dutifully check in with health officials when they get an alert. “That's too optimistic,” he says. “We tell people to stay at home; they go to the beach.”

To test whether an app accurately flags risky encounters, health officials want to know the proportion of identified contacts who end up sick—the “secondary attack rate.” As a rule of thumb, if an app's attack rate matches or exceeds that of traditional contact tracing, “we know the app is doing a really good job,” Salathé says.

Some app designs allow health officials access to anonymized ID codes of infected users and all their contacts. That means officials can calculate the secondary attack rate and fine-tune the app by checking how many notified users later report symptoms or a positive test through the app. In Norway, which launched such a centralized app in April, municipalities will compare how many, and how quickly, contacts are identified by the app versus through traditional contact tracing, says Emily MacDonald, an epidemiologist at the Norwegian Institute of Public Health.

Other apps, including Switzerland's and one released this week in Germany, are decentralized, meaning data about interactions stay on a phone. Privacy advocates favor this design, and Google and Apple encourage it. Last month, they released technology to support Bluetooth tracking, but only for apps with a decentralized design. With these apps, health departments will know about potentially infected users only if they report getting an alert. Critics say that design could make it harder to evaluate the app's performance.

The ultimate test, some researchers say, is a randomized trial to gauge whether using an app brings down rates of infection. But such trials “would be very costly and difficult,” says Oxford behavioral economist Johannes Abeler. Because COVID-19 is relatively rare, such trials would have to be large. And to gauge effectiveness, researchers would have to factor in the proportion of a participant's contacts who had downloaded the app—which might be nearly impossible to know.

Another possibility is to compare changes in infection rates between geographic areas or demographic groups with different levels of app use, suggests Rosalind Eggo, an epidemiologist at the London School of Hygiene & Tropical Medicine who hopes to study the impact of apps as data accumulate. “We have a lot of technology that can help us here,” she says. “There's quite a lot of people saying, ‘Oh, it won't work.’ I think we need to try.”

View Abstract

Stay Connected to Science

Navigate This Article