Study population and setting
This retrospective cohort included all 706 consecutive patients admitted to two Michigan hospitals between March 10 and May 10, 2020 with confirmed SARS-CoV-2 infection via polymerase chain-reaction. The authors collected data on participant incarceration status (the exposure), age, sex, race, chronic medical conditions, smoking, status, body mass index, reported symptoms, vital signs, medications, laboratory findings, complications, and treatment outcomes — specifically in-hospital mortality and 30-day mortality. They compared baseline covariates among incarcerated persons (n = 108) and non-incarcerated persons (n = 598), and modelled inpatient mortality using logistic regression and 30-day mortality using Cox proportional hazards models. For each outcome, they presented three models: 1) only incarceration status; 2) Model 1 plus age, sex, and race; and 3) Model 2 plus obesity and Charlson Comorbitiy Index (a medical comorbidity index).
Summary of Main Findings
Of the 796 participants included, 108 (15.3%) were incarcerated. Incarcerated adults were more likely to be male (98.2% versus 49.0%), younger (mean age 61.2 versus 67.7 years), Black (50% versus 43.2%), former or active smokers (80.2% versus 37.6%), and have COPD (37% versus 10.5%). Incarcerated participants were less likely to have chronic kidney disease (10.2% versus 23.9%) and obstructive sleep apnea (5.6% versus 12.2%). Incarcerated participants were also more likely to have rapid breathing (44.4% versus 27.9%), a fever (26.9% versus 14.9%), and a low oxygen saturation (48.2% versus 37.3%) on presentation. They were also more likely to need more intensive respiratory support (high flow nasal cannula or mechanical ventilation) than non-incarcerated participants. Incarcerated participants were also more likely to die in the hospital (29.6% versus 20.1%) and die within 30 days of admission (34.3% versus 24.6%) than non-incarcerated participants. After correction for covariates (Model 3), incarcerated participants were at a 2.32 (95% CI: 1.33, 4.05) times increased odds of dying in the hospital and 2 (95% CI: 1.33, 3) times increased hazard of dying 30 days from admission compared to non-incarcerated participants.
This study attempted to control for confounding variables and used appropriate modeling strategies to assess COVID-19-related in-patient mortality (logistic regression) and 30-day mortality (Cox proportional hazards) in incarcerated individuals. They used multiple imputation for missing baseline variables and checked the proportional hazards assumption graphically and with Schoenfeld residuals.
To be transferred to the hospital for admission, incarcerated adults were required to meet Michigan Department of Correction criteria (based on age, vital signs, oxygen requirements, and mental status), whereas all non-incarcerated participants who needed supplemental oxygen were admitted. Incarcerated patients were therefore more likely to be sicker at admission than non-incarcerated patients, and at higher risk of death. This difference is likely to at least partially explain the worse outcomes among incarcerated participants, however, the analyses did not adjust for COVID-19 severity at admission. Additionally, this study only included individuals in one region of the United States, and the findings may not be generalizable to other incarcerated individuals in the United States or elsewhere. Although they used the Charlson Comorbidity Index to adjust for pre-existing comorbidities, this index weights most comorbidities similarly, though each have different risks associated with COVID-19 severity. Therefore, residual confounding may bias the estimates.
This study shed light on COVID-19 outcomes among incarcerated adults, a marginalized population who have been disproportionately impacted by the COVID-19 pandemic.
This review was posted on: 23 April 2021