Methodology
In order to estimate the projected impact of COVID-19 on incarcerated populations, the Recidiviz COVID-19 Model for Incarceration uses incarceration-specific measures of how COVID-19 would spread within a prison or jail, along with state data on incarcerated populations and community resources to generate a likely estimate of total future COVID-19 cases in the incarcerated population, and related hospitalization use and deaths. This model is similar to models being used to estimate total hospitalizations and deaths across the state and country, but recognizes that social distancing is not possible within prisons and jails, and that many individuals in prisons and jails are at particular risk from this disease.
FWD.us used the April 16, 2020 release of the model and entered inputs for Arizona based on current publicly available data, including the current number of confirmed cases in the state, within the state prison population, and among facility staff, as well as the age breakdown of the state prison population, housing density (90 percent of individuals in Arizona prisons are housed in dorm-style accommodations which greatly increases the rate of spread of the virus), and percentage of capacity used in state prison facilities. FWD.us also included updated data on hospital capacity in the state based on current reported capacity.
FWD.us also updated the hospitalization rate used in the model. While Recidiviz included the mid-range estimate of hospitalizations per confirmed case in the United States, given current limitations in testing capacity, we felt the denominator (total number of cases) might be underestimated and the total rate therefore overestimated. Instead, we used an estimated hospitalization rate of 5%, based on an Imperial College of London paper that included an age-adjusted estimate of hospitalizations per total (including unobserved) hospitalizations. This should be seen as a lower bound for hospital bed use by people in the state prison population.