An illustration showing a functional example of household self-isolation has been temporarily removed as it is being updated in line with the latest changes to the guidelines. Our analysis period means that we address policy compliance issues, both in times of more difficult lockdown (including severity in terms of household visits, see Additional Information 1) and in terms of the impact of vaccine introduction on behaviour. On these points, the issues of budgetary mix and compliance with policy have so far been addressed only indirectly. Fierce debates about the causes and extent of population-wide non-compliance in the UK continue.5 While some survey data suggests that more than 90% of Britons have always adhered to social distancing most of the time26, other survey data suggests that non-compliance with “stay at home” orders is much higher. Hills and Eraso27 note that “contrary to a perceived sense of conformity with people”, 92.8% of Londoners surveyed did not comply with all social distancing rules, with 48.6% involved in “deliberate non-compliance”. Other studies have attempted to identify the causes of non-compliance,28,29,30,31,32,33,34,35, although their collective results are very inconsistent and often contradictory, perhaps given that causal factors can change over time.34 We join various public health researchers and suggest that by using passively collected anonymous mobility data, we can overcome the level of bias in social desirability that typically conveys the self-reporting methods that have always been used to indicate compliance27,28,30,36,37,38,39, and meet the requirement that “future research focus on this.” should assess compliance with objective measures to minimize the likelihood of biased reporting39. In doing so, we hope this will further address the `current` lack of empirical evidence to support the fact that there is significant `fatigue` in the UK in terms of compliance with COVID-19 restrictions, which helps to accurately inform the health policy makers developing the English response. Updated guidelines for households where grandparents, parents and children live together, with information on the spread of COVID-19, as well as financial support and the application of self-isolation. All translations have been removed pending the update.
Added illustration showing a functional example of household self-isolation, updated translations, and easy-to-read instructions. Clusters of visits were removed when they fell into a 30 m buffer zone of an established point of interest (POI). Points of interest were drawn from the ORDNANCE Survey`s POI layer, which provides a comprehensive collection of potential places to visit across the UK, from churches to restaurants to sports stadiums. One of the limitations of this step is that potential household visits to mixed-use areas are more likely to be suppressed. Households that are near or on the same floor area of the building (e.g., inside the same tower at different levels) as a point of interest would be ignored in this study. This is to be expected to have a stronger effect in denser urban areas. After the removal of individual “places of residence and work” as well as visits to POIs and green spaces, a remaining group of visit points were considered unclassified places of visit. To allow for final validation of the estimated household visits, unclassified visits from 2020 were extracted and compared to the Ordnance Survey AddressBase dataset, which records the location of all residential addresses in the UK. This revealed that 89% of unclassified visits took place within a 50 m radius of a residential building. As a result, the remaining 11% of visits (i.e., those located more than 50 m from a residential building) were removed from the remaining analysis. The final register of visits is classified as household visits. Home advice for households likely to be infected with coronavirus (COVID-19).
This study proposes a new framework for identifying visits by anonymous people to non-residential households. This metric is derived from cell phone trajectory data, and by extracting mobility behavior of this type, we are able to derive an indicator of “household mix” by place and time.