As the world descended into COVID-19 chaos in spring 2020, Apple and Google announced a rare partnership.
The two rivals were collaborating on a plan to transform smartphones into powerful weapons to stop the coronavirus from spreading. Using
wireless signals, a person’s phone would keep track of all the other phones it crossed paths with — if anyone in the chain turned out to be infected, there would be an easy way to find and notify those at risk.
“Contact tracing can help slow the spread of COVID-19 and can be done without compromising user privacy,” Apple CEO Tim Cook wrote in a tweet on April 10, 2020, trumpeting the partnership with Sundar Pichai, his counterpart at the helm of Google.
The two tech titans rolled out their “exposure-notification” tools in record time. But more than a year since the first apps using Apple and Google’s technology were released, and with a new coronavirus variant on the march, smartphone contact tracing has struggled to live up to its promises and prove its worth in the US.
Only slightly more than half of US states have rolled out a contact-tracing app. And an Insider investigation of those apps found critical shortcomings.
About one in four people activated the tech on their phones, on average, in states where the apps were available. And in those states, an average of just 2% of the people who tested positive for COVID-19 logged the information into the app — significantly limiting the value of the tool.
“Just having the app isn’t going to do anything, if people don’t input results,” said Isobel Braithwaite, a public-health official in the UK who has studied the efficacy of smartphone-based contact tracing.
The data contrasts with relatively high usage in certain other countries and highlights an area in which America has stumbled in its response to the pandemic. As policymakers, healthcare experts, and the public grapple with the latest coronavirus surge, the apps’ poor track record in the US raises important questions about how to effectively implement technology in the fight against disease and the risks of relying too much on it.
In an emailed statement shared on the condition it could not be quoted from directly, a public-relations representative speaking for both Google and Apple said that the technology was intended to bolster authorities’ ability to combat COVID-19 and that research showed it has saved thousands of lives.
Using a smartphone to fight COVID-19
Apple and Google’s tools have a simple premise: What if your smartphone could tell you when you’ve been exposed to someone diagnosed with COVID-19?
To do this, the apps use phones’ Bluetooth sensors to detect occasions when users have been in close proximity to other people who also have the apps installed. If a user is diagnosed with the virus, they enter their positive diagnosis into the service. The service then checks the list of who they’ve been in close contact within the past week or so and sends them all notifications informing them about their exposure so they can quarantine or get tested.
In an effort to assuage privacy concerns, the notified users are never told who exposed them, who else is diagnosed with COVID-19, or where the incident occurred.
Apple and Google developed technology to let governments build contact-tracing tools as downloadable apps for their countries’ residents, and Apple later built the system directly into its mobile operating system.
It’s designed to supplement traditional contact tracing — humans manually calling the contacts of infected people to inform them of their exposure. (Because the tools don’t share location, they’re sometimes referred to as “exposure notifications” rather than “contact tracing.”)
In the US, apps were built at the state level. Insider reached out to the health departments of the 25 states, plus Washington, DC, that developed and released the tools statewide to ask for data on user numbers, number of positive COVID-19 cases logged in the systems, and number of notifications sent. Eight states didn’t provide any information — Alabama, Connecticut, Hawaii, Louisiana, Missouri, New Mexico, North Dakota, and Utah — while the other 17 and DC provided at least partial data.
A lot of people are installing the app — but less log infections
By at least one metric, Maryland is first in the nation. It has the highest proportion of any state’s COVID-19 diagnoses entered into Google and Apple’s tools — 9.7% — while there have been 2.4 million installations of the service, equivalent to roughly 40% of the state’s residents.
California’s system has been activated on phones 12 million times — effectively about 30% of the state’s population — but of the almost 2.5 million COVID-19 cases reported in California since the app was launched, only about 3% were logged in its system.
And in Nevada, while the equivalent of more than 50% of the state’s population has adopted the smartphone apps, only 470 positive COVID-19 diagnoses have been entered into the service in the past year — just 0.157% of cases reported in the state during the same time frame.
Some states struggled to get residents to even consider the service. Arizona closed its program in July after a number equal to 1.3% of its population installed it. In Michigan, 6.3% of residents have activated it. And in Wyoming, 0.69% of residents — about 4,000 people — have installed the service.
A little under half of US states opted not to not build a service in the first place, limiting the tools’ reach. The 28 states and territories that use the service represent roughly 186.8 million people — meaning another 141.5 million Americans (43.1% of the overall population) aren’t covered in any capacity.
The low usage comes as the US has struggled to convince much of the population to get vaccinated, despite some of the easiest access to vaccines in the world, and there has been revolts against public-health measures like mask-wearing and social distancing.
It’s unclear how the lack of stats from the 10 states that didn’t provide Insider with data on the number of cases logged (which collectively represent 34.6 million people) might affect overall averages and totals.
Why so low?
In interviews, public-health experts and technologists identified multiple possible causes for the relatively low usage of the tools.
The people who install the app may not be entirely representative of the state overall, and Braithwaite, the UK public-health official, hypothesized that app users might be wealthier and “structurally less likely to get COVID.” As such, the percentage of COVID-19 cases in a given state that get logged in its app is not necessarily going to be as high as the percentage of people in that state who have installed the app.
But Braithwaite added she “definitely wouldn’t think that’s sufficient to see that reduction in percentage” seen in the data.
Other factors may include everything from forgetfulness to tech issues and failures by state health departments to communicate with residents about the tools.
“It’s certainly not as high as anyone would hope,” said Adriane Casalotti, the chief of public and government affairs for the National Association of County and City Health Officials. “But given the challenges with getting folks to even sign up for these services, and the lack of community conversation around them, especially in the last couple of months … I’m not as surprised.”
Problems may include American cultural skepticism toward sharing health data, she said, and the difficulty of creating tools and generating trust in them midcrisis.
“It’s really hard to build these things up in the middle of the problem,” she said, “without having that more thoughtful partnership before the problem strikes.”
A privacy snafu in February, when The Markup reported that certain contact-tracing information stored on Android phones could be accessed by other apps, may have also reinforced wariness to use the apps among some Americans.
The fact that individual states were responsible for launching the apps, rather than the federal government — unlike the UK and other countries — may have also hampered efforts to raise nationwide awareness and encourage users to input their test results, said Ashkan Soltani, a former chief technologist for the Federal Trade Commission who has expressed skepticism about contact-tracing apps.
“The numbers you’re looking at would probably have looked very different if the federal government had got behind this system and done a wider, coordinated rollout of the app, of these sorts of apps, across the states,” he said.
Nonresidents are also able to activate the service in a particular state, meaning activation rates in a state may be higher than the true number of residents who have installed the service. (Washington, DC, was excluded from Insider’s calculations on total installations because it has a higher number than the area’s total population.)
A notable limitation to the data is that states didn’t provide information on the number of cases inputted on a monthly (or weekly) basis. Some states may have started slowly before improving adoption rates or had a strong start before usage declined, and the lack of granular data obscures this.
Bryant Karras, the chief informatics officer at Washington State Department of Health, said usage had grown in recent months, particularly after Apple built tools directly into iOS and once the state started sending automated texts to residents diagnosed with COVID-19 encouraging them to use the service. If the same data was prepared for a shorter timeframe, he said, “I think you’d be surprised at what we’re seeing.”
Low usage may hamper efficacy
Pichai, Google’s CEO, has set public expectations about usage of the contract-tracing system. In an interview with Wired in May 2020, he said that “even if only 10 to 20 percent of users opt in, this will have a real, meaningful impact. The more, the better.”
Installation numbers indicate at least 10% of residents in 12 states have activated Apple and Google’s exposure-notification tools, out of the 17 that responded to Insider. The overall figure across all 17 is 24.8%.
But none of the states have had even 10% of the total COVID-19 cases in their area entered into the system since its launch, with a combined percentage of 2.14%.
As a proportion of the total COVID-19 cases in the entire US, the percentage of cases being logged into the apps is likely significantly lower.
Meanwhile, more than 40% of weekly cases in the UK are being logged into its app.
It’s unclear how low levels of usage will affect the efficacy of the system.
A 2020 research paper partly written by Google employees estimated that even at “relatively low levels of adoption” — which it pegged at 15% — smartphone-based contact tracing could reduce infections of COVID-19 by 8% and deaths by 6%. But the study appears to assume that the 15% of people who install the service then go on to routinely enter any positive COVID-19 diagnoses they receive. The data suggests that in practice, this is not happening.
Still, health officials from several states pointed to the study as evidence that their programs were effective, despite low levels of usage — including Michigan, where about 0.17% of cases have been logged in the system, and Delaware, where 0.39% of cases have been logged.
Apple and Google do not provide states with the identities of the users who are notified, and multiple public-health departments said that these privacy protections made it difficult to empirically measure the tools’ effectiveness. Some said they viewed them as a supplement to traditional contact tracing and pandemic-suppression efforts, rather than their primary tool.
“Even though we don’t really know the efficacy, we are delighted to have this as another tool for public-health agencies,” Katherine Feldman, the director of the contact-tracing unit at the Maryland Department of Health, said.
The perils of tech solutionism
Soltani, the former FTC chief technologist, said he was concerned that the focus on smartphone-based contact tracing early in the pandemic may have detracted from other efforts.
“The amount of attention and focus that so many policymakers, so many legislators, tech companies, experts, all paid to this system — had that interest, effort, experience been focused on things like getting vaccinated or disinformation, misinformation around COVID and contact tracing,” that would’ve likely had a much bigger influence, he said.
Braithwaite disagreed, saying that while there was an “overfocus on these types of technological solutions,” they’re relatively cheap to develop. The effect of the apps in other countries, like the up to 8,000 lives it is estimated to have saved in the UK, makes it worth it, she said. Multiple state health officials were similarly effusive about the technology.
Still, Casalotti cautioned against reflexive technological solutionism in the face of hard problems. “When we started looking at some of these tools,” she said, “the idea was … ‘We just need a technological fix to this,’ and it really shows that it’s much more nuanced than that.”