Association between trajectory of intracranial pressure and all-cause mortality in patients with non-traumatic subarachnoid hemorrhage: A retrospective cohort study from 2008 to 2022.

Our study aimed to assess the association between intracranial pressure (ICP) trajectories and 30-day, 90-day, 180-day, and 365-day all-cause mortality in patients with non-traumatic subarachnoid hemorrhage (SAH). We identified patients diagnosed with non-traumatic SAH from the MIMIC-IV 3.1 database. Group-based trajectory modeling was applied to identify ICP trajectories with selection based on Log-likelihood, Akaike information criterion (AIC), Bayesian information criterion (BIC), Hannan-Quinn information criterion (HQIC), Odds of correct classification (Occ), and Average posterior probability (Avepp) to determine the optimal number of classes. Logistic regression with a multi-model approach was used to compare all-cause mortality among patients with different ICP trajectories. Subgroup analyses were performed to assess the interactions. To demonstrate robustness, the relationship between traditional ICP metrics and all-cause mortality was evaluated using restricted cubic splines (RCS) and receiver operating characteristic (ROC) curves. Of the 1052 patients with non-traumatic SAH, 312 were included. Four distinct ICP trajectories were identified: (1) Class 1 (sustained decline and stabilization at a low level), (2) Class 2 (stabilization at the lowest level), (3) Class 3 (initial decline followed by elevation with fluctuations at moderate-to-high levels), and (4) Class 4 (initial highest level followed by decline with high fluctuations). Class 4 had significantly higher mortality rates at all time points. Additionally, multi-model logistic regression analyses showed that, compared to Class 1, Class 4 was associated with a higher risk of death at 30 days (adjusted OR: 5.36, 95%CI: 1.13-25.50, P = 0.035), 90 days (adjusted OR: 5.54, 95%CI: 1.21-25.28, P = 0.027), 180 days (adjusted OR: 7.66, 95%CI: 1.75-33.55, P = 0.007), and 365 days (adjusted OR: 6.74, 95%CI: 1.52-29.89, P = 0.012). RCS analysis revealed a non-linear association between average ICP and ICU mortality, with no linear association observed for other mortality outcomes. Threshold effect analysis identified the optimal threshold of the average ICP as 11.08mmHg. The AUCs of %ICP>20mmHg, average ICP, and baseline ICP for predicting 30-day mortality were 0.567, 0.567, and 0.542, respectively. Furthermore, subgroup analyses showed no interactions between ICP trajectory classes and subgroups of age, sex, race, hypertension, external ventricular drainage, or mannitol use. In patients with non-traumatic SAH, early ICP trajectories were significantly associated with 30-day, 90-day, 180-day, and 365-day all-cause mortalities. This indicates that ICP trajectories can serve as effective indicators for prognostic assessments.