Impact of concordant ICU triage on hospital mortality: a nationwide retrospective multilevel analysis

Background Specialty intensive care units (ICUs) provide diagnosis-specific patient care. Nonetheless, the advantage or disadvantage of ICU specialization remains debatable. We conducted a multicenter retrospective cohort study to evaluate the clinical outcomes of specialty ICUs in critical care. Methods This retrospective study reviewed adult ICU patients in the eICU Collaborative Research Database (eICU-CRD), a United States (US)-based multicentered ICU cohort database. Patients over 16 years old with documented Acute Physiology and Chronic Health Evaluation (APACHE) IV scores were enrolled and grouped by their admitted ICUs. Multilevel logistic regression models were performed to assess the relationships between hospital mortality and ICU type/ICU triage conditions in patients admitted with a diagnosis of cerebrovascular accident (CVA), acute coronary syndrome (ACS), coronary artery bypass graft (CABG), pneumonia (PNA) and other respiratory diseases, and trauma. A parallel multiple mediation analysis was performed to examine whether interventions in the ICU mediated the association between ICU triage and hospital mortality. Results A total of 136,236 patients admitted to 271 ICUs from 186 hospitals were enrolled. In multilevel analysis, admission to cardiac ICU (OR: 0.81, p < 0.001) and surgical ICU (OR: 0.88, p = 0.041) were associated with lower mortality risks, and admission to Neuro-ICU (OR: 1.32, p < 0.001) was linked to a higher risk of mortality compared to general ICU in total cohort. In patients without trauma, general ICU or discordant triage was not associated with a higher in-hospital mortality risk than concordant ICU triage. In trauma patients, discordant ICU triage was associated with higher hospital mortality (OR: 1.87, p = 0.005), partially mediated by mechanical ventilation (OR = 1.007, p < 0.001). Conclusion Specialized concordant ICUs were not associated with lower hospital mortality risk compared to general ICUs after adjusting for confounders. Under resource-constrained settings, discordant ICU triage may be a viable option.