In the effort of fighting the COVID-19 pandemic, several governments world-wide have imposed unprecedented mobility restrictions and social distancing policies, as these - combined with contact tracing and isolation of cases - represent the most effective strategy to slow down the spread of SARS-CoV-2. Italy has been the first EU country to adopt such interventions, by imposing a national lockdown on March 12 that was lifted about 2 months later, once the epidemic curve had been suppressed. Here, we extensively investigate the socioeconomic determinants of the responses to mobility restrictions imposed in Italy during the full course of the first wave of the COVID-19 epidemic, from February until June 2020, through the analysis of human mobility patterns derived from anonymized and aggregated mobile phone data. To this aim, we monitored the timelines of mobility responses using several mobility metrics during 3 phases of the COVID-19 outbreak in Italy both at the province level and at the district level in 3 major cities. Through statistical modeling, we identified the association between mobility responses and several demographic, economic, and epidemiological covariates, across different spatial scales. The results show that behavioral responses were associated to varying socioeconomic factors in the different phases of the pandemic management cycle, and across different geographic scales, highlighting the complex landscape of the determinants of behavioral responses to non-pharmaceutical interventions.
