ABSTRACT
To better understand the drivers of spread of an infectious disease such as COVID-19, it is crucial to identify the most likely individuals to become infected. The continued circulation of SARS-CoV-2 remains a global concern, however with falls in testing only a small fraction of infections are now being recorded. Comprehensive data from earlier periods may then prove a unique resource for probing finer patterns of infection. To this end we analyse the fine spatio-temporal distribution of over 450,000 cases of COVID-19 in Scotland in waves of the B.1.1.529 Omicron and B.1.617.2 Delta lineages, from May 2021 to January 2022. We use random forest regression on case numbers, informed by measures of geography, sociodemographics, testing and vaccination. We then identify indicators for higher case numbers, showing that despite marked differences in the velocity of the outbreaks, the risk factors are remarkably similar. We show how finer variation is only adequately explained through the use of multiple explanatory variables, implying that case heterogeneity resulted from a complex interplay of individual behaviour, immunity, and willingness to test. This analysis also provides evidence that the case distribution may be biased relative to that of all infections, particularly with respect to local deprivation.
