Abstract
Using smartphone location data and a novel application of publicly available employment data, we map how California communities are connected to nearby prisons through the movement of prison staff, and we measure the role these connections play in spreading infectious diseases. Leveraging an exogenous prisoner transfer-induced COVID-19 outbreak at San Quentin state prison in June 2020 as a quasiexperiment, we examine the unidirectional spread of the disease from the prison to surrounding communities. This outbreak was unique: its origin from outside Northern California was clearly documented and nonstaff entry and exit was severely limited during this time. Our identification strategy compares zip codes connected and unconnected to the prison via staff movement. Compared to unconnected zip codes with similar pretransfer COVID-19 rates and demographic characteristics (race/ethnicity, education, household income, age, and population), zip codes connected to San Quentin had 13% more new COVID-19 cases in July and 30% more in August. Our results suggest that a hypothetical novel infectious disease that emerged in California prisons could lead to almost 15,000 community infections within 1 month from staff movements alone. These findings identify the degree to which "closed institutions" are-even during lockdowns-epidemiologically porous, highlighting the need for public health interventions to reduce the unintended consequences of such connections on the spread of infectious disease.