Last spring, Apple and Google launched an ambitious effort to harness technology in the fight against Covid-19, building a powerful smartphone tool that would alert people of possible exposure to the coronavirus.
The software could play an important role in helping curb outbreaks, new data shows, but as the pandemic’s winter wave engulfs the United States, the technology remains largely unused. It is available only in about a third of states, stymied by privacy concerns, a lack of public awareness and interest, poor access to fast testing, and a patchwork system of government health authorities.
“It turns out that it is very, very challenging to get people to use a Covid app,” said James Larus, a computer scientist and dean of the School of Computer and Communication Sciences at the Swiss university EPFL, who has worked on the effort with Apple, Google and public health officials. “We went into it thinking that of course people would want to use this, and we have been very surprised.”
Engineers at the companies built a system that respected people’s privacy, reasoning that without such safeguards, no one would sign on. The technology, called “exposure notifications,” doesn’t track users’ locations, instead relying on Bluetooth to detect which phones have been within several feet of one another for more than a few minutes. When a user receives a positive test result, the local health system provides a code by email, text message or phone call to enter into the app. That will alert anyone who was in proximity while the person was infectious.
After some states balked at the effort required to make an app, the tech companies this fall tried to make the process easier, allowing states to roll out the technology without creating a stand-alone application.
This week California will launch its version of the tool, called Exposure Notifications Express, joining four other states and the District of Columbia in using the simplified program. It is an important test for the country’s most populous state, as Gov. Gavin Newsom orders parts of it to shut down amid surging cases. Because of its size and prominence, the tech giants’ home state could provide momentum for the technology, but it remains to be seen how many people will sign up and whether California has enough fast testing capacity for the tool to be helpful.
Areas using Express have an advantage in marketing because Apple and Google send push alerts to people’s phones when it becomes available in their state. The District of Columbia has seen the highest participation rate: more than 60 percent of its population. About 20 percent of residents in Colorado, Connecticut and Maryland — which are all using Express — have joined; Washington State has had about 13 percent participation since introducing the tool earlier this month.
In most of the other 12 states using exposure notifications, rates are in the single digits.
Early on, epidemiologists at Oxford suggested in a paper that if 60 percent of people in an area used a digital contact-tracing app, the pandemic could be brought under control without a lockdown. Later epidemiological models indicated that apps could help reduce viral spread even if just 15 percent of a population used them.
In Switzerland, about 22 percent of the population is using the technology. In a study of contact tracing in Zurich, researchers calculated that for every 100 people who tested positive, the app correctly notified 24 contacts who had caught the virus — a success rate similar to what is seen in human contact tracing.
Public officials have cited the importance of contact tracing in battling the pandemic, but U.S. health workers have had trouble keeping up with high infection rates and persuading people to cooperate.
A pilot program at the University of Arizona provided what may be the first example of an app slowing the viral spread in the United States. During an outbreak there this fall, it sent alerts for as many as 12 percent of transmissions, researchers estimated.
“We believe that the outbreak on campus had a flatter curve because of the app,” said Joanna Masel, a mathematical biology professor who has helped supervise the rollout of the program, called Covid Watch.
Anna Giudici, a junior at the university, said the alert she received in mid-September led her to stay away from other people, even though her only symptom was a mild headache.
“If I hadn’t gotten a notification, I wouldn’t have thought anything of it,” she said, adding that she had not been alerted by a traditional contact tracer. She got a rapid test and went into isolation when it came back positive. To her knowledge, she did not infect anyone else; even her roommates tested negative.
Several other students and staff members said they appreciated having the app, even as they acknowledged that it didn’t work perfectly, especially at first. In an early version, iOS users received notifications of “possible exposures” when they were unlikely to have been infected but had briefly been in range of someone with the virus — messaging that some found jarring and confusing.
In at least some cases, the early app appeared to notify users about someone who was nearby — but on the other side of a wall. Nathalie Riddell, a senior, said she had received several alerts of “significant exposure” when her roommate had the virus and was isolating in another part of their apartment.
But the project worked in part because it targeted a specific population that included many young people who were tech-savvy and trusted the app provided by their university. The school also had easily accessible rapid tests, Dr. Masel said. The lack of a fast testing system off campus meant the pilot study could not be extended to the entire state earlier this fall, she added.
“The app doesn’t stand on its own,” she said. “It requires available rapid testing to work.”
As with much of the U.S. coronavirus strategy, decisions about the implementation of exposure-notification apps have been left to the states. North Dakota and Virginia embraced the technology quickly, for instance, but officials elsewhere said they had concerns about efficacy and privacy. Some preferred to focus resources on human contact tracing.
“We had evaluated the risk of being early adopters of untested technology and felt like that would be more problematic than beneficial,” said Sarah Tuneberg, senior adviser to Colorado’s governor for Covid-19 testing containment and technology. But by this fall, she said, state health officials decided that “waiting any longer puts public safety in jeopardy.”
The state introduced the technology in October, using Exposure Notifications Express. Under that option, Colorado didn’t have to create its own app; Google had made one for Android users, and iPhone owners could turn on technology built into the operating system. Ms. Tuneberg said the push notifications from the companies helped the state reach its 20 percent adoption rate.
In states that have their own apps, without the benefit of push notifications, the numbers are far lower: about 5 percent in New York, less than 3 percent in Alabama and about 1 percent in Wyoming. Virginia has had the most success, at nearly 10 percent, having devoted about $1.5 million to public awareness campaigns.
Jeff Stover, the executive adviser to the state’s health commissioner, said that public health departments have for months been encouraging testing and mask-wearing, and that marketing coronavirus exposure apps is also essential. Virginia has “done a good job of continually increasing the proportion of the population who is buying into this,” he said. “We have had to market to different segments of society that might have different reasons not to trust the government.”
A pilot study in California suggested that traditional advertising might not be the most effective way to get people to use the technology. “Far and away the most effective messaging was a text to your phone,” said Dr. Christopher Longhurst, chief information officer at the University of California San Diego Health. The best text message, he said, told people that the app could help them protect their family and friends.
From the beginning, one of the main concerns among the public has been privacy. After years of surveillance scandals, people are reasonably skeptical of technology companies and the government, said Elissa Redmiles, a computer scientist who has studied attitudes toward Covid apps.
“They have this sense that everyone is taking their data constantly, and they don’t want to give up any more data,” she said, or they worry about authoritarianism and think, “I don’t want to be surveilled by the government.”
The focus on privacy has led to something of a Catch-22. Dr. Redmiles’s research shows that people want assurances not only of privacy but also of the technology’s effectiveness before agreeing to use the apps in large numbers. But privacy protections make it harder to collect the very data that can show how well the apps work.
“If you can’t see whether it’s effective, it’s not very compelling,” said Marc Zissman, a computer security researcher at M.I.T.’s Lincoln Laboratory. The Centers for Disease Control and Prevention tapped Dr. Zissman’s lab this fall to help figure out how effective the exposure-notification system is.
The C.D.C. hopes the new research will answer some key questions: “How many more contacts can be notified of their exposure via these apps than traditional contact tracing alone? How much faster are individuals notified? And is the added speed and comprehensiveness worth the investment of states’ resources?” a spokeswoman for the agency said.
But even with assurances of privacy and evidence that the apps can work, many Americans still show little interest in using the technology.
One study indicated that less than half the population supported using exposure-notification apps, even with privacy-protecting features.
In another survey, conducted by Dr. Redmiles, nearly a quarter of people said they would never install such a tool, even if it were completely private and 100 percent effective. More than half would agree to use an app only if it reduced viral spread by 50 percent.
Those who may need the notifications most, because they are engaging in risky behavior, may be less likely to download the apps, said Dr. Masel, the Arizona biologist.
On the university campus, fewer than half of people who tested positive for the virus were using the app. And of the users who tested positive, only about half shared their codes through the app to alert others. Drop-offs have also occurred in the Swiss study and in other states.
“People load the app to know if they were around someone else who tested positive, but don’t want to notify others if they are positive,” either because they are concerned about their privacy or because they have a “selfish mind-set,” said Tim Brookins, one of the developers of the app used in North Dakota.