Understanding the mechanisms behind rotational and localized lockdowns: How mobility and spatial disparities modulate COVID-19 transmission

City-level curfews and other non-pharmaceutical interventions (NPIs) targeting specific communities or finer within-city aggregation, such as neighborhoods or localities, were used to slow the transmission of SARS-CoV-2 upon the announcement by the World Health Organization of the emergence of a virus with a potential pandemic threat. Community transmission was reduced with coarser NPIs, but the effectiveness of NPIs restricting finer spatial scales remains to be refined, especially in big urban centers with heterogeneous spatial distribution in their mobility and socioeconomic variables. Communities are segregated spatially in large urban centers based on socioeconomic status ^1–4 . In developing countries, the lowest socioeconomic statuses must commute to far locations the most ⁵ and sustain their economies based on informal activities ⁶ . Consequently, it is unclear what spatial aggregation is appropriate to intervene while minimizing economic disruption. This work evaluates an NPI implemented in Bogotá, a megacity in Latin America in Colombia. A set of rotational lockdowns at the locality scale (an administrative aggregation of neighborhoods) was implemented in the city. First, we use mobility data to investigate how commuting changed at the scale of the restrictions. Second, using population epidemiological surveillance data of COVID-19, we quantify community transmission at the scale of the restrictions by estimating the effective reproductive number R _eff and investigating if changes in mobility correlate with those in transmission. Third, we use an epidemiological transmission model to simulate counterfactual scenarios in the absence of NPIs. We compared the counterfactual projections with the reported incident infections, mortality, and community transmission at the city scale. Thus, we will determine whether the finer NPIs reduced the level of community transmission in the city. Finally, we use socioeconomic data at the scale of the restrictions (localities) to investigate how these and mobility predicted transmission. We incorporate random variability across space to simulate uncontrolled sources of variability. Our results show how finer NPIs change mobility at different spatial scales. Although the intervention reduces mobility between units, the finer-scale mobility is not perturbed. We have evidence that the first sets of interventions were the most effective in reducing transmission. Finally, we found substantial spatial heterogeneity in the combined effect of socioeconomic variables and mobility in predicting transmission.