Modelling changing patterns in the COVID‐19 geographical distribution: Madrid’s case

We analyse the transmission factors shaping the spatial distribution of COVID‐19 infections during the distinct phases of the pandemic’s first wave in Madrid, Spain, by fitting a spatial regression model capturing neighbourhood effects between municipalities. Our findings highlight that factors such as population, mobility, and tourism were instrumental in the days before the national lockdown. As a result, already in the early part of the lockdown phase, a geographical pattern emerged in the spread of the disease, along with the positive (negative) impact of age (wealth) on virus transmission. Thereafter, spatial links between municipalities weakened, as the influences of mobility and tourism were eroded by mass quarantine. However, in the de‐escalation phase, mobility reappeared, reinforcing the geographical pattern, an issue that policymakers must pay heed to. Indeed, a counterfactual analysis shows that the number of infections without the lockdown would have been around 170% higher.
Explanatory factors of the spread of the pandemic in Madrid changed according to the phase of national lockdown. As for the effectiveness of social distance measures, a counterfactual analysis shows that the number of infections without the lockdown would have been around 170% higher.