Racial Disparities in ICU Ventilation for COPD: A Retrospective Study Using MIMIC-IV Data

Introduction: Prior studies suggest Black ICU patients may receive invasive mechanical ventilation (IMV) less often than White patients, but the role of clinical versus structural factors remains unclear. This study examines racial disparities in IMV use among ICU patients with chronic obstructive pulmonary disease (COPD) and evaluates whether adjusting for clinical, demographic, and illness severity variables accounts for observed differences.


Methods: We conducted a retrospective cohort study using the MIMIC-IV database to assess racial disparities in IMV among ICU patients with COPD. Patients were identified by ICD-9/10 codes and included if admitted to the ICU and self-identified as Black or White, yielding a final cohort of 1,961. The primary outcome was receipt of IV during the ICU stay. We applied stepwise adjustment across five covariate domains (demographics, socioeconomic status, comorbidities, severity of illness, and anthropometrics) using propensity score models, logistic regression, and targeted maximum likelihood estimation (TMLE). Isolated domain analyses assessed each factor’s independent impact on disparity attenuation.


Results: The cohort was predominantly White (82.9%) and male (55.1%), with a median age of 69 years. Black patients comprised 17.1%. Race was linked to differences in baseline clinical, socioeconomic, and illness severity characteristics (baseline pseudo-R² increased from 0.01 to 0.09 with full adjustment). In logistic models, Black patients initially had lower odds of receiving IMV than White patients (adjusted OR = 0.49; 95% CI: 0.10–2.30; p < 0.001), but this disparity became non-significant after adjusting for socioeconomic, comorbidity, and severity variables (fully adjusted OR = 0.86; 95% CI: 0.67–1.09; p = 0.224). TMLE showed a lower, but not significant, unadjusted risk for Black patients (RD = –0.17; 95% CI: –0.51 to 0.17), which attenuated to the null in fully adjusted models (RD = –0.03; 95% CI: –1.81 to 1.74).


Conclusions: Black COPD patients had lower IMV rates than White patients, but this disparity was not significant after covariate adjustment. Propensity score models suggested modest structural confounding, while TMLE showed no residual racial effect post-adjustment. Differences were largely mediated by illness severity at presentation.