In Italy and many other countries, digital contact tracing (DCT) through smartphone apps has played a patchy role in the management of the COVID-19 pandemic so far. But, it deserves another chance, according to a study1 published in Nature Communications by an international team led by Bruno Lepri from the Fondazione Bruno Kessler in Trento. The authors show that tracing apps can be effective at containing new outbreaks even when scarcely adopted, and without quarantining large fractions of the population — provided the right physical distancing measures are kept in place.
The scientists used data derived from the Copenhagen Networks Study (CNS), a large-scale reconstruction of social interactions and proximity in a sample population, based on data exchanged and collected by smartphones. “The study includes different settings such as a university campus, a high school and an office building,” Lepri explains. “This allowed us to estimate the tracing efficiency of different policies, measured as the fraction of the individuals infected by a single person who are warned by the application and quarantined,” he continues.
In the study, the least restrictive policy warns and quarantines only those who had contacts with infected individuals for more than 30 minutes at a distance around 1 meter, explains Giulia Cencetti, a researcher at the Fondazione Kessler who co-led the study. The most restrictive one quarantines also the contacts below 5 minutes, and at distances above 1 meter. Various intermediate policies were also tested, and for each of them the authors estimated the efficiency and the rate of false positives (notifications reaching people who do not then test positive) and false negatives (infected people who do not get notified).
The researchers then fed the estimated tracing efficiency into a mathematical model of the epidemic. For an adoption rate of 20%, they found that DCT leads to a decrease in daily new cases even with the least restrictive policy, provided that the app’s warning results in effective isolation, and that masks and physical distancing measures keep the reproduction number (the average number of people infected by each positive case) at 1.2. If the adoption rate gets to 40%, it becomes possible to relax physical distancing (allowing for a reproduction number of 1.5) but a more stringent tracing policy is needed, leading to quarantine up to 5% of the population.
The potential role of DCT has been highlighted since the early days of the pandemic. In May 2020, scientists at the University of Oxford showed2 that any contact tracing strategy would be ineffective if notifications were delayed more than three days, as people infected with SARS-CoV-2 are able to transmit the virus several days before they develop symptoms.Manual contact tracing suffers from such notification delays, whereas DCT can warn those at risk instantaneously.
In Italy, the Immuni application has been downloaded nearly 10 million times but only sent 94,000 risk notifications, and the French app TousAntiCovid has similar numbers, with 14 million downloads, but only 199,000 risk notifications. The UK has done better with nearly 22 million downloads and more than 1,880,000 risk warnings.
“Until now the impact of DCT was not quantified” says Luca Ferretti, a researcher at the Oxford Big Data Institute and co-author of the Oxford study, who was not involved in this research. “The work by Lepri and co-authors fills this gap and offers a cost-benefit analysis”.