Abstract
BACKGROUND: Data on criminal justice system involvement can support public health efforts in ways that have been recognized for decades, but data protections and jurisdictional boundaries can make data sharing difficult. In many jurisdictions, carceral facilities are required to publish lists of currently incarcerated individuals. Automated collection of these lists may be one way for public health to access this information. The purpose of this project was to evaluate the availability, completeness, and utility of carceral data collected from public-facing registers in Washington State. METHODS: Program staff at the Washington State Department of Health catalogued the websites of all carceral facilities in Washington State and identified what information was available about currently incarcerated individuals. This information was downloaded daily from 1/1/2023 to 12/31/2023 using R software. The completeness of this data was compared relative to the Washington State Jail Booking and Release System (JBRS) during the same time frame. To evaluate the utility of the scraped data (which may contain only partial identifiers) for record linkage, we performed a set of simulated linkages between two external datasets with a known relationship (King County Jail bookings and a surveillance list of people living with HIV who may be out of care). We applied a simple match algorithm to copies of these datasets that had been reduced to match the different combinations of identifier variables available in the public data (full names and ages, partial names, etc.) We compared the sensitivity and positive predictive value (PPV) of the algorithm applied to the reduced datasets and calculated an estimate for the entire dataset weighted by incarcerated population size. RESULTS: At the time of the project, 61 of 71 facilities in Washington State published information about current inmates. 100% of these 61 published names of inmates, 33% age or date of birth, and 13% other identifiers. We collected data from 58 facilities over the project span. 89% of individuals in JBRS were present in the daily scraped data and 95% of individuals in JBRS who were incarcerated for more than 24 h. We estimated that the collected data had 87.7% sensitivity of and 88.8% PPV in linkages with HIV registries. CONCLUSIONS: Public facing carceral data in Washington State constitute a data source with high completeness and adequate information for record linkage.