Association of documented obstructive sleep apnea with mortality in ICU patients with hypertension: a cohort study using the MIMIC-IV database

Obstructive sleep apnea (OSA) is highly prevalent among patients with hypertension, yet its influence on mortality in critically ill hypertensive individuals admitted to the intensive care unit (ICU) remains unclear. This study aimed to investigate the association between OSA and short-term and long-term mortality in this specific population. We performed a retrospective cohort analysis using the MIMIC-IV database (version 3.1). Adult hypertensive patients with their first ICU admission between 2008 and 2022 were included. OSA was identified using International Classification of Diseases (ICD) codes recorded at discharge for the index hospitalization. The primary outcomes were all-cause in-hospital, 30-day, 90-day, and 365-day mortality. The association was examined using multivariate Cox, multiple imputation for missing data, Kaplan-Meier survival analysis, and landmark analysis. The proportional hazards assumption was assessed using Schoenfeld residual tests. A total of 41,493 patients were enrolled, of whom 4,550 (11.0%) had a diagnosis of OSA. Patients with OSA were younger and had a higher body mass index. In univariate analysis, OSA was associated with lower mortality at all time points. After full adjustment for demographics, vital signs, laboratory values, severity scores, and comorbidities, OSA remained significantly associated with lower mortality: in-hospital hazard ratio (HR) 0.75 ( 95% CI 0.67 ~ 0.85), 30-day HR 0.75 ( 95% CI 0.68 ~ 0.84), 90-day HR 0.78 ( 95% CI 0.71 ~ 0.85), and 365-day HR 0.82 ( 95% CI 0.76 ~ 0.88); all P < 0.001. Landmark analysis revealed a consistent inverse association between OSA and mortality across different time windows after ICU admission (e.g., for in-hospital mortality among patients alive at day 3: HR 0.67, 95% CI 0.59 ~ 0.76), although the magnitude of association was numerically smaller at later landmark time points. Sensitivity analyses with complete-case data yielded similar but attenuated results. Kaplan-Meier curves consistently demonstrated higher survival probabilities for the OSA group across all follow-up periods. In this large cohort of critically ill patients with hypertension, a documented diagnosis of OSA was associated with lower short- and long-term mortality in multivariable-adjusted analyses. However, this association is most plausibly explained by survivor bias, detection bias, immortal-time bias due to the timing of OSA diagnosis, and residual confounding, rather than a true protective biological effect. These findings should be interpreted as a prognostic association of a documented diagnosis, not as evidence that OSA confers any clinical benefit.