Figure 1.

Flow diagram depicting the phases of enrollment, exclusion, and data analysis for Model 1 and Model 2. ^a Patients with implausible data points such as a COVID-19 diagnosis in 2017 or individuals in whom the date of death predated their date of hospital admission were excluded. ^b If a patient was missing a specific covariate for the Cox model, they were excluded for complete case analysis. These specific covariates included age, body mass index, modified Charlson comorbidity index, and modified Sequential Organ Failure Assessment score (central nervous system and respiratory component removed). ^c Compared to Model 1, additional Model 2 covariates included PaO₂/FiO₂ and modified Sequential Organ Failure Assessment score with respiratory component added for complete cases analyses. ^d Unmatched patients between the dexmedetomidine group and no dexmedetomidine group on propensity score calculation were removed.