Aspirin Therapy and 28-Day Mortality in ICU Patients: A Retrospective Observational Study From Two Large Databases

Abstract

PurposeAspirin is widely used in patients in the intensive care unit (ICU); nonetheless, its effects on these patients remain controversial. This retrospective analysis of data from clinical practice investigated the effects of aspirin on 28-day mortality in ICU patients. MethodsThis retrospective study included data from patients in the Medical Information Mart for Intensive Care (MIMIC)-III database and the eICU–Collaborative Research Database (CRD). Patients aged 18 to 90 years and admitted to the ICU were eligible and were assigned to one of two groups according to whether they were given aspirin during their ICU stay. Multiple imputation was used for patients with >10% missing data. Multivariate Cox models and propensity score analysis were used to estimate the association of aspirin treatment with 28-day mortality among patients admitted to the ICU. FindingsIn total, 146,191 patients were enrolled in this study, and 27,424 (18.8%) used aspirin. Aspirin treatment in ICU patients, especially in nonseptic patients, was associated with a lower 28-day all-cause mortality on multivariate Cox analysis (eICU-CRD, hazard ratio [HR] = 0.81, [95% CI, 0.75–0.87]; MIMIC-III, HR = 0.72 [95% CI, 0.68–0.76]). Aspirin treatment was associated with lower 28-day all-cause mortality after propensity score matching (eICU-CRD, HR = 0.80 [95% CI, 0.72–0.88]; MIMIC-III, HR = 0.80 [95% CI, 0.76–0.85]). However, on subgroup analysis, aspirin therapy was not associated with a lower 28-day mortality in patients without systemic inflammatory response syndrome (SIRS) symptoms or with sepsis in either database. ImplicationsAspirin treatment during the ICU stay was associated with a significantly reduced 28-day all-cause mortality, particularly in patients with SIRS symptoms but without sepsis. In patients with sepsis and with/without SIRS symptoms, beneficial effects were not clear, or more careful patient selection is required.