COVID-19 exposure app helped prevent thousands of cases in Washington state, new study shows
WA Notify, the COVID-19 exposure notification tool for Washington state, saved an estimated 40 to 115 lives and prevented between 2,800 and 8,200 cases during four months of use this winter, according to a recent study by researchers at the University of Washington and Washington State Department of Health.
The new study is the first to assess the public health value of smartphone-enabled exposure notification systems in the U.S., according to a statement from the DoH.
About one third of smartphone users in Washington state have enabled the tool, which alerts people anonymously if they may have been exposed to COVID-19. And as Washington prepares to lift pandemic restrictions, health officials encourage more uptake of WA Notify, released by the state on Nov. 30.
“This is an incredible and important tool as we reopen,” said Bryant Karras, chief informatics officer and senior epidemiologist at the Department of Health, and a co-author on the study.
Seattle Mariners games are returning to full capacity, the Amazon banana stand is back, and people are getting closer to each other and to people they don’t know. The state will open before the June 30 target date if 70% of people over 16 have snagged their first COVID-19 shot, and the state is edging close.
“The combination of relaxation of masking and distancing recommendations, opening of large events, increased travel, new variants of concern, and uneven vaccination coverage makes for a lot of uncertainty,” said Janet Baseman, lead author of the study and associate dean for Public Health Practice at the University of Washington. “The pandemic is not over.”
WA Notify uses a system developed collaboratively by Apple and Google, now widely used for exposure notification tools worldwide. About half of states have launched such tools, and almost all use the Apple-Google technology.
iPhone users enable the system by tapping a button in their settings, “Turn on Exposure Notifications”, and Android users download an app. The tool works by exchanging random anonymous codes via Bluetooth between phones near each other. The temporary codes contain no location information and no personal information.
“It is just quietly working in the background. No news is good news,” said Karras. Users who receive a notice of a positive COVID test can alert others by tapping a link, releasing their codes to be matched with other WA Notify users who may have been in close proximity for 15 minutes or more. “It understands how far away you were from the person, but it never keeps track of where that proximity event occurred,” said Karras.
The tool is meant to augment conventional contact tracing, and is potentially better at identifying people unknown to the infected individual — such as that nearby fan at the Mariners game. As more people get vaccinated and cases decrease, “exposure notification now can help us find that needle in a haystack,” said Scott Becker, executive director of the nonprofit Association of Public Health Laboratories, which operates a server that helps match the released codes.
The new study, published as a preprint and not yet peer reviewed, adapted an approach published in the international science journal Nature by researchers from Oxford University and their colleagues. That study concluded that a similar tool in the United Kingdom averted hundreds of thousands of infections.
“The methodology is sound,” said Becker of the new research.
The Washington researchers pulled data from a range of sources, including anonymous, aggregated datasets of the number of notifications generated and opened. The study period, from November 30, 2020 to March 31, 2021, coincided with nearly 200,000 reported cases and almost 2,500 deaths, during the most severe phase of the pandemic.
They found that 10,084 users anonymously reported their COVID-19 diagnosis, generating an estimated 3.4 exposure notifications per infected user, outpacing the 2.7 notifications generated on average through the state’s conventional contact tracing program.
To estimate the number of infections and deaths averted, the researchers assessed parameters such as the probability a notified contact became infected, which they set at 12.085% for their model.
Exposure notification tools “need to be understood as part of a tool of public health,” said Susan Landau, a professor of computer science at Tufts University who directs a program in cybersecurity and public policy. Landau has written about such tools, praising the privacy of the Apple-Google system, but also raising questions about their uptake in communities highly affected by COVID-19, often lower income and with a high percentage of Black and Hispanic individuals.
Such communities may receive more frequent notifications, but people may be less able to act in response, for instance by taking time away from their workplace. And some may work in proximity to others but wear a protective mask, she has noted.
“Who is using the apps?” asked Landau in an email to GeekWire. “Is the result of their use an increase of testing and medical resources to an already well-served community?”
Such data are difficult to gather, in part because of the anonymity of the system, notes a separate report examining the development and implementation of WA Notify, published in the Online Journal of Public Health Informatics. The rollout of the tool was pinned to an ongoing public relations campaign after extensive pilot testing, which may have contributed to its uptake at one of the highest rates in any state. The more people who use such tools, the more effective they are.
“Now that we have an evidence base that it’s making a difference, I think now public health can’t ignore this as a tool,” said Karras. “Most importantly, the public needs to keep it on their phones so that it can work.”
Editors Note: The researchers have updated their study since they first uploaded it onto the preprint server medRxiv, and expect the server to soon post a paper with updated data, estimating 40 to 115 lives saved. The original results estimated lives saved at between 30 and 120.