key: cord-0689552-v102m39s
authors: Biondi, Breanne E.; Leifheit, Kathryn M.; Mitchell, Carmen R.; Skinner, Alexandra; Brinkley-Rubinstein, Lauren; Raifman, Julia
title: Association of State COVID-19 Vaccination Prioritization With Vaccination Rates Among Incarcerated Persons
date: 2022-04-12
journal: JAMA Netw Open
DOI: 10.1001/jamanetworkopen.2022.6960
sha: cbe7a7bbf9ab1e4c3dc7beb7379013f21aa1198d
doc_id: 689552
cord_uid: v102m39s

This cross-sectional study examines the association between prioritization of COVID-19 vaccination for state prison systems and the rate of vaccination among incarcerated persons.

State data on weekly COVID-19 vaccination counts among IP incarcerated persons and monthly prison population counts were from The Marshall Project and Associated Press. 1 Monthly prison population counts were applied to each week.

State policies regarding vaccine allocation, from the COVID-19 US state policy database. 2

States that were excluded from the analysis did not report full vaccination rates (i.e. cumulative percents of incarcerated persons IP receiving two doses of Pfizer or Moderna vaccine or one dose of Johnson&Johnson), only partial (i.e. cumulative percent receiving at least one dose of vaccine). Nebraska did not report any vaccinations. For states that had missing cumulative vaccination data for a given week (N=57 of 1,294 state-weeks), we imputed the average of the previous and following week.

The event study is used to analyze changes in outcomes in the pre-and post-periods by using binary indicators for weeks pre-and post-incarcerated persons IP vaccine eligibility date. Binary indicators for states that did not have a policy to prioritize incarcerated personsIP for vaccination were all set to zero. The event study model allows us to analyze changes pre and post-policy change similar to a difference-indifferences analysis, except that the predicted outcome can vary across leads and lags. The estimated effect (the percentage point difference in the cumulative percent of incarcerated personsIP fully vaccinated) from the event study model are the effect of policies prioritizing incarcerated personsIP to be vaccinated on the cumulative percent of incarcerated personsIP vaccinated.

Analyses were performed using Stata/MP, v17.0 (StataCorp LLC, College Station, Texas). states that never lifted) (x weeks post)st = binary indicator of whether outcome was observed x weeks after the week incarcerated personsIP became prioritized for vaccination (among states that ever lifted; always 0 for states that never lifted) (week)t = fixed effects for calendar week (state)s = fixed effects for state

The Marshall Project

COVID-19 US state policy database

COVID-19 Vaccinations in the United States,Jurisdiction | Data | Centers for Disease Control and Prevention. Accessed

Difference-in-Differences with Variation in Treatment Timing