Quantifying meaningful adoption of a SARS-CoV-2 exposure notification app on the campus of the University of Arizona

Digital exposure notification requires both the primary and secondary cases to have previously installed a smartphone application, and the primary case to rapidly report their positive diagnosis. These conditions were met for an estimated 12% of transmission pairs during a SARS-CoV-2 outbreak on the campus of the University of Arizona.

were able to anonymously trigger notifications, including testing and quarantine 23 recommendations, for other Covid Watch users with whom they had been in contact. 24 Here we quantify meaningful uptake. App download numbers are readily accessible, but 25 overstate active app usage (3). Effective uptake requires not just app installation in both a 26 primary and a secondary case, but also that the primary case report a positive diagnosis by 27 obtaining and entering a secure verification code. 28 The effectiveness of exposure notification depends not on total population usage, but on 29 usage among individuals who go on to be infected by SARS-CoV-2. One concern is that 30 individuals who tend to comply with public health guidelines might be both more likely to 31 download an exposure notification app and less likely to contract SARS-CoV-2, leading to the 32 overestimation of effective usage. With the help of a third party, the University of Arizona used 33 social marketing tools to promote app usage. Identified influencers proved successful with quick 34 download rates among students.

35
Evaluation of GAEN apps is challenging because of their strict privacy protections. We 36 therefore partnered with case investigation of primarily university students to ask three 37 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted February 3, 2021.  Combining these numbers, 26% of all infected individuals interviewed by manual contact 58 tracers reported having previously entered a verification code into their app to anonymously 59 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted February 3, 2021. ; https://doi.org/10.1101/2021.02.02.21251022 doi: medRxiv preprint notify their contacts. One caveat is that the cases that contact tracers succeed in interviewing 60 (response rate is ~70% of assigned cases) may have higher app use than the general campus 61 population. However, we get the same estimate of 26% by dividing the total number of 62 verification codes issued by the total number of positive tests from our campus testing facilities. 63 We count verification codes (613 in total) at point of issue rather than upon usage, but because 64 obtaining a code requires some action from an infected individual (either a phone call or clicking 65 on a request link), we expect this to be only a slight overestimate. There are also factors that 66 might cause this number to be an underestimate; specifically, we count positive tests not cases, 67 and some individuals tested positive on multiple test platforms (PCR and antigen) for the same 68 illness. These two independent approaches both yield the same estimate that 26% of cases 69 notified their contacts using the Covid Watch app.

70
Because the three questions regarding app usage were asked during the case investigation 71 interview, our results demonstrate that notification via the Covid Watch app was more rapid not 72 only than traditional contact tracing at the University of Arizona, but also more rapid than other 73 digital exposure notification workflows where traditional contact tracers provide verification 74 codes over the phone. Our automated code delivery via an end-user test results portal is now 75 adopted by commercial test providers in Arizona. 76 We propose and estimate a metric of meaningful usage among cases. Because the app's 77 purpose is to isolate the infected, focusing on cases is more epidemiologically meaningful than 78 usage among the general population. We consider the scenario where a primary case infects a 79 secondary case within a tightly interconnected community such as a college campus, and 80 estimate the probability that both cases have used the app to the minimum necessary level to 81 potentially impact transmission. The estimated probability for the primary case is 26% (where 82 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted February 3, 2021. ; https://doi.org/10.1101/2021.02.02.21251022 doi: medRxiv preprint verification code entry is required) and 47% for the secondary case (where only app activation is 83 required). Combining these by assuming a well-mixed population, and neglecting transmission 84 from outside campus given low community prevalence at the time this pilot study was 85 conducted, app usage is estimated to affect 12% of transmission pairs. In a structured population 86 where individuals in the same transmission pair have more similar app usage rates, this value 87 will be higher than 12%.  However, R(t) is also indirectly reduced when the far larger number of exposure events that do 96 not lead to transmission also lead to behavior change, which either prevents infection in the 97 notification recipient, or elicits first quarantine and then isolation after they have been infected 98 by a different exposure within the same social network (8).

99
Here we have proposed a new metric for assessing app usage within a tightly 100 interconnected community that does its own testing and tracing. We show that usage on the 101 University of Arizona campus is high enough to make it a useful tool that complements and 102 augments traditional contact tracing.

159
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted February 3, 2021. ; https://doi.org/10.1101/2021.02.02.21251022 doi: medRxiv preprint