Digital contact tracing data for COVID-19 unveils detailed epidemic dynamics

Data from digital contact tracing, using examples from the National Health Service (NHS) COVID-19 app for England and Wales, has revealed epidemic dynamics in unprecedented detail. The findings show that digital contact tracing can provide insights with unprecedented time resolution, in addition to its primary purpose of reducing transmission. “Digital tools are likely to play an even larger role in the next pandemic, given their unparalleled scalability and the insight they can provide for precision public health,” write the study’s authors. Epidemics are driven by complex and dynamic transmission patterns, making it challenging to measure all aspects of the process. This is especially true for rapidly spreading respiratory pathogens. Detailed characterization of these systems can inform public health decisions and allow for targeted interventions. Digital contact tracing, proposed early in the COVID-19 pandemic, has been evaluated for its effectiveness in reducing transmission, but its potential for improving epidemic monitoring has remained largely underexplored. Here, Michelle Kendall and colleagues report an analysis of anonymized data from the NHS COVID-19 app for England and Wales, which was in use between 2021 and early 2023. Kendall et al. found that data from the app captured significant changes in contact rates and transmissibility reflecting events such as easing lockdowns, Christmas holidays, and the Euro 2020 championship. The introduction of vaccines and the emergence of Delta and Omicron variants also influenced these dynamics. The analysis revealed a shift from household infections to fleeting contacts, particularly on weekends and during holidays. Notably, fleeting infections peaked on Saturdays and household infections on Sundays. The app detected 57,000 infection events during Euro 2020, highlighting the impact of nationwide gatherings on transmission. Seasonal holidays showed a pattern of increased mixing followed by a quiet period in January. Moreover, the data allowed Kendall et al. to develop a novel indicator – R_app(t) – which breaks down the reproduction number R(t) into contact rates and transmission rate, which were accurate and available at least 5 days before other available estimates.