Skip to content

How many people need to use exposure notification for it to work?


Q: How many people need to use exposure notification for it work? 

A: Exposure notification doesn’t work if you are the only one using it. Some time after Alice infects Bob, Alice tests positive, shares that diagnosis, and this anonymously notifies Bob that he was exposed. This leads Bob to skip visiting his grandmother - even though she is vaccinated - and this saves her from catching COVID-19. (Vaccines aren’t 100% effective.) Notice that the person who was helped (the grandmother) doesn’t even have to use the app herself, but both Alice and Bob must use the app in order to protect her.

The first models of exposure notification suggested that if 56% of people adopted exposure notification, that would be enough to suppress the pandemic altogether, even as most of us returned to normal life. The same researchers quickly stressed that the press had taken this benchmark number out of context, and that lower levels of adoption can still help

Today we can go beyond those theoretical models and the many assumptions they make, because we have high quality data on the NHS app. With 28% of the U.K. population using the app, and 72% of those users consenting to share their diagnosis after testing positive, the app prevented about half a million infections. That is about 20% of how many people would likely have been infected if the app hadn’t been there at all. The researchers reached this conclusion in part by seeing how much better regions of the country with more app use did, compared to regions with less app use. Even after controlling for wealth, demographics, and everything they could think of, how much app usage there was in a region made a substantial difference. 

The U.K. got such high levels of app adoption because they made the same app useful for other things too. To go to a restaurant in the U.K., you need to write down your name and phone number, in case contact tracers want to reach everyone who was there. But if you have the app, then instead you can check in more privately by scanning a QR code. The same app also lets people order a COVID-19 test, and access their test results through the app. These other features help encourage more people to use the app. WeHealth is in the process of adding similarly synergistic features to our apps. The U.S. doesn’t require venue check-ins like the U.K. and many other nations do, but we can still make our app useful for navigating the world of tests and vaccines.

Adoption rates in the US are less well documented - generally only downloads are counted, not the smaller number of active users that the U.K study counted - but they are clearly lower. But what matters is not how many people in a whole country use the app, but how many people in your own very local community. Do the Bobs in your life have the app? If not, can you persuade them to? Can you get them to persuade the Alices that they interact with? If you get to that point, you are protected. 

Many of us are in a position to act on this. If you run a retail business, you could politely request that your customers please download the app as they enter the building, putting up a QR code to direct them, and put that request right next to your sign about masks. If you do this, then after an infected person spends time in your building, the employees who were close to them will get notified, and that will help you get on top of any outbreak right from the start, before the virus spreads to all of you. The same works for many workplaces, universities, or other communities.

The University of Arizona focused on getting on-campus students to adopt our app last Fall. But we were worried that the sort of student who downloaded the app would be the sort who was unlikely to get infected, and that the students who were out partying would not have the app. To assess adoption, we therefore took into account the fact that exposure notification apps don’t do anything if you aren’t even in contact with anyone. The adoption that matters is among the Alices and Bobs who go on to become infected with COVID-19. Alice needs not just to have the app, but also to share her diagnosis. Bob just needs to have the app. The University of Arizona’s contact tracing team asked infected people about their prior app usage - 46% met the criteria for Bob and 25% for Alice. Put those together and the app is used on both sides of 11% of transmission pairs. That’s not enough to be a magic bullet on its own, but it’s enough to significantly flatten the outbreak curve. It allows a lot more reopening than would otherwise be safely possible.

Starting with clusters of adoption and spreading out from there is also good strategy. It worked for driving widespread adoption of Facebook, Whatsapp, and Uber - it should work for exposure notification too. A key factor in people’s decision about whether to use exposure notification is what they think other people are doing. We are all part of communities, and we all need to do our bit to keep our family, friends, and community safe. If you can get your community to use exposure notification, it doesn’t matter whether usage is high or low in your State as a whole - your community are protecting one another.


Have a question? Ask the professor: