Study of video electroencephalogram features for predicting 28-day survival in critical hematologic patients with disorders of consciousness

Abstract Background: Critical hematologic patients frequently develop disorders of consciousness due to infection, metabolic derangements, or malignant infiltration, markedly worsening their survival. Video electroencephalography (vEEG) is an objective measure of brain function and has demonstrated prognostic value in disorders of consciousness secondary to space-occupying lesions or traumatic brain injury. However, its predictive utility in critically ill hematology patients remains underexplored. This study evaluates the independent prognostic value of vEEG features and clinical factors for 28-day survival in these patients using Cox multivariate regression. Methods: We retrospectively enrolled 66 patients with critical hematologic conditions and disorders of consciousness admitted to the Hematological Care Unit (HCU) of the Institute of Hematology, Chinese Academy of Medical Sciences, from May 2024 to December 2024. Patients were categorized by 28-day outcome (Alive vs. Deceased) and all completed a 28-day follow-up. We collected demographics, clinical diagnoses, vital signs, laboratory data, and vEEG recordings. Univariate analyses (χ² and independent-samples t-tests) identified variables with P < 0.05, which were entered into a multivariate Cox proportional-hazards model. We report corrected hazard ratios (csHR) with 95% confidence intervals (95% CI). Results: 1. The 28-day mortality rate of patients with critical blood disorders and disorders of consciousness admitted to the HCU was 57.6% (38/66);Sluggish pupillary light reflex (csHR = 3.185, 95% CI = 1.263-8.034), absence of pupillary light reflex (csHR = 12.276, 95% CI = 3.329–53.378), intracerebral hemorrhage (csHR = 8.334, 95% CI = 2.895–23.992), subarachnoid hemorrhage (csHR=11.184, 95% CI=4.418–31.733), heart failure (csHR=3.193, 95% CI=1.498–6.807), and endotracheal intubation (csHR=3.048, 95% CI=1.330–6.986) were independent clinical risk factors for 28-day mortality;EEG nonreactivity was an independent risk factor for 28-day mortality (csHR = 2.87, 95% CI = 1.42–5.81, P = 0.003); sleep elements were an independent protective factor (csHR = 0.45, 95% CI = 0.22–0.91, P = 0.027). Conclusion: In critical hematology patients with disorders of consciousness admitted to the HCU, EEG non-reactivity independently predicts higher short-term (28-day) mortality, while the presence of sleep elements confers an independent protective factor. Enhanced monitoring of clinical and vEEG-related indicator is conducive to optimizing the management of critical hematologic patients in the HCU and improving prognosis.