Cerebral Autoregulation, Mannitol Response, and Outcomes in Traumatic Brain Injury: A Structural Causal Model Approach

BACKGROUND AND OBJECTIVE: Intravenous mannitol serves as an intermediate strategy for reducing intracranial pressure (ICP). Favorable mannitol responsiveness conditions and its impact on prognosis remain poorly understood. This study aimed to develop a structural causal model (SCMMannitol) to estimate heterogeneous treatment effects and reveal causal relationships between patient status, mannitol response, and prognosis. METHODS: Arterial blood pressure and ICP data were collected from 53 patients with traumatic brain injury. To maintain adequate ICP, arterial blood pressure, and cerebral perfusion pressure, 20% mannitol was administered. Cerebrovascular pressure reactivity (PRx) and resistance of mannitol (RMannitol) were derived. The data set included sex, age, Glasgow Coma Scale, pupil reactivity, major extracranial injury, and Glasgow Outcome Scale-Extended (GOSE). IMPACT and CRASH prognostic models were retrained to include RMannitol. SCM analyzed causal relationships between PRx, RMannitol, and GOSE. RESULTS: Mannitol significantly reduced ICP at pretreatment PRx <0.2 (P < .001). PRx <0.2 yielded lower RMannitol than PRx ≥0.2 (P = .017). RMannitol and GOSE were negatively correlated (Pearson r = -0.33, P = .016). The RMannitol predictive capacity for mortality was 0.81 (cutoff, 0.68; 95% CI, 0.68-0.94; AUROC, P = .001). In the outcome prediction model, PRx before mannitol and RMannitol inclusion improved outcome prediction (P < .05). SCM identified a causal pathway linking PRx before mannitol, RMannitol, and outcomes. SCMMannitol estimated heterogeneous treatment effects of mannitol in an individual patient. CONCLUSION: PRx before mannitol and RMannitol causality suggests that compensatory cerebral vasoconstriction drives mannitol response. RMannitol and GOSE inverse relationship underscores the importance of ICP monitoring and timely treatment escalation. The PRx before mannitol, RMannitol, and GOSE causal relationship suggests that considering both RMannitol and PRx may improve patient outcomes. SCMMannitol may estimate heterogeneous treatment effects, advancing precision medicine for patients with traumatic brain injury.