Social distancing, population density and spread of COVID-19 in England: longitudinal study

Background: The UK government introduced social distancing measures between March 16-22, 2020, aiming to slow down transmission of COVID-19. Aim: To explore the spreading of COVID-19 in relation to population density after the introduction of social distancing measures. Design & setting: Longitudinal design with 5 weekly COVID-19 incidence rates per 100,000 for 149 English Upper Tier Local Authorities (UTLAs), March 16-April 19. Method: Multivariable multilevel model to analyse weekly incidence rates per 100,000; time was level-1 unit and UTLA level-2 unit. Population density was divided into quartiles. The model included an interaction between week and population density. Potential confounders: percentage aged 65+, percentage non-white British, percentage in two highest classes of the National Statistics Socio-economic Classification. Co-variates: male life expectancy at birth, COVID-19 prevalence rate per 100,000 on March 15. Confounders and co-variates were standardized around the mean. Results: Incidence rates per 100,000 peaked in the week of March 30-April 5, showing higher adjusted incidence rate per 100,000 (46.2; 95% confidence interval (95%CI) 40.6-51.8) in most densely populated ULTAs (quartile 4) than in less densely populated ULTAs (quartile 1: 33.3, 95%CI 27.4-37.2; quartile 2: 35.9, 95%CI 31.6-40.1). Thereafter, incidence rate dropped in the most densely populated ULTAs resulting in rate of 22.4 (95%CI 16.9-28.0) in the week of April 13-19; this was lower than in quartiles 1, 2, and 3, respectively 31.4 (95%CI 26.5-36.3), 34.2 (95%CI 29.9-38.5), and 43.2 (39.0-47.4). Conclusion: After the introduction of social distancing measures, the incidence rates per 100,000 dropped strongest in most densely populated ULTAs.