Subarachnoid hemorrhage (SAH) is a critical cerebrovascular event caused by bleeding into the subarachnoid space, frequently resulting from ruptured intracranial aneurysms [10,15,17]. Mortality caused by SAH has been reported in a range of 20%-50% from prior studies [3,9]. SAH not only provides immediate life-threatening risks, but it also causes long-term functional deficits in survivors. Greebe et al. examined long-term functional outcomes using the modified Ranking Scale (mRS) and reported that 55.7% of SAH patients improved their mRS after a 4-month follow-up [5]. Therefore, A prognostic evaluation in SAH patients is required to develop appropriate therapeutic strategies.

Intracranial pressure (ICP) is one of the key factors that determines a patient’s outcome after the accumulation of blood between the arachnoid and pia mater. However, the optimal ICP thresholds for SAH patients have been debated from the review of the literature. Marbacher et al. investigated the pathogenesis of brain injury following SAH in animal experiments and found that SAH induced a rapid increase in ICP, which corresponded to decreased cerebral perfusion pressure (CPP) and regional cerebral blood flow [12]. Consequently, sustained high ICP correlates with poor neurological outcomes [1]. Zoerle et al. found that SAH patients with ICP greater than 20 mmHg had a significantly higher mortality rate than patients with ICP 20 mmHg or less [28]. According to Heuer et al., the following is a negative correlation between an increase in ICP and a favorable outcome. Favorable 6-month outcomes were observed in 71.9% of patients with ICP less than 20 mmHg, 63.5% of patients with ICP between 20 and 50 mmHg, and 33.3% of patients with ICP greater than 50 mmHg [6].

Though neurocritical care has advanced, the optimal ICP criteria for thresholds for guiding personalized treatment will be better neurological outcomes and prognosis. The review of the literature has not clearly defined the optimal ICP cutoffs for prognostic prediction [26]. The present study aimed to determine the prognostic value of different ICP thresholds in SAH patients. By identifying critical ICP cutoffs, our findings may help improve clinical decision-making and patient management strategies.

As a result, Table 1 presents the clinical features of the 110 patients with SAH. There was a female predominance at 66.4%, with a mean age of 60.59 (±16.42) years. Hypertension, dyslipidemia, diabetes mellitus, and ischemic heart disease were identified in 46.4%, 23.6%, 10.9%, and 6.4% of the patients, respectively. Headache and motor weakness were common symptoms in 57.3% and 57.3% of cases, respectively. Furthermore, seizure was found in 15.5% of the present cohort. In terms of clinical severity grading scales, 47.3% of cases were Hunt and Hess grade IV, with 18.2% being grade V. In WFNS grade, 76.3% of the present cohort were grade IV-V.

Modified Fisher grade IV is the most common imaging severity in the present study, accounting for 70.0% of cases, whereas obliterated basal cistern was observed in 18.2% of cases. According to the cerebral angiogram, the ruptured aneurysm of the anterior cerebral circulation was 51.8%, while 48.2% of SAH cases had the ruptured aneurysm from the posterior circulation. Based on treatment modalities, 65.5% of patients underwent endovascular coiling, and surgical clipping was performed in 34.5% of cases. Furthermore, all patients underwent ventriculostomy and the average intraoperative ICP was 17.90 (+8.47) mmHg.

The mean follow-up time was 15.81 (+24.01) months and the maximum follow-up period was 106 months. As a result, 1-year, 2-year, and 5-year survival probabilities were 0.62 (95% CI 0.53-0.73), 0.59 (95% CI 0.49-0.71), and 0.54 (95% CI 0.42-0.70), respectively. Additionally, the median overall survival time was determined; however, the estimated median survival was not achieved from the entire dataset and the overall survival curve is demonstrated in Fig. 1.

From univariate analysis, the candidate variables with a p-value of 0.1 or less included Hunt and Hess grade, WFNS grade, pupillary light reflex, and ICP value. Hence, the final prediction model that had the lowest AIC value comprised pupillary light reflex, and ICP value, as shown in Table 2. The p-value of the global Schoenfeld test was 0.73 and the final model did not violate the proportional hazards assumption. Furthermore, Harrell’s c-index was 0.728 for estimating the predictive performance.

As a result, HR with 95% confidence intervals was computed across various ICP levels, and a dose-response curve was constructed, as illustrated in Fig. 2. The ICP value of 24.34 mmHg was the potential risk transition point from the HR of one. The ICP at the potential risk transition point is positively related to mortality. In addition, dose-response curves were created according to the pupillary light reflex group for subgroup analysis. As shown in Fig. 3, only the fixed-both-eyes group showed the potential risk transition point at the ICP of 25.75 mmHg, while other groups did not. Therefore, SAH patients were dichotomized into two groups: the group with an ICP value of less than 24 mmHg and the group with an ICP value of 24 mmHg or higher. The group of an ICP value of 24 mmHg or higher was significantly associated with poor prognosis (p-value of log-rank test=0.01), as demonstrated in Fig. 4.

This present study identified ICP as the potential prognostic factor in patients with SAH, emphasizing its independent association with mortality and poor prognosis. Our findings are consistent with previous studies underscoring the detrimental impact of elevated ICP in patients with hemorrhagic stroke. Zhang et al. [26] revealed that the crucial ICP threshold for patients with spontaneous intracerebral hemorrhage was 16.5 mmHg, whereas we identified a threshold ICP value of 24 mmHg as a potential transition point beyond which the hazard of death increases significantly. The dose-response analysis supports the hypothesis that higher ICP levels are associated with worse outcomes. Kaplan-Meier analysis and multivariate Cox regression confirmed that patients with ICP 24 mmHg or more had notably lower survival probabilities.

Interestingly, the subgroup analysis based on pupillary light reflex revealed a distinct risk transition point (25.75 mmHg) only in patients with fixed pupils, implying that the prognostic value of ICP may be modified by neurological examination findings. From prior studies, several independent prognostic factors have been reported, such as increasing age, high WFNS grade, intraventricular hemorrhage, intracerebral hematoma, and vasospasm [13,16,27]. Additionally, pupillary light reflex has been reported as a predictor in prior studies. Mader et al. [11] created and validated the prognostic score systems based on pupil status and found areas under the receiver operating characteristic curves (AUROC) of the scoring systems ranged from 0.775 to 0.841, while Dengler et al. reported AUROC ranging from 0.75 to 0.82 using a clinical-radiographic scoring system with pupil status for prognostication [4]. The predictive performance of our work is consistent with that of previous studies. While various clinical-radiographic factors remain pivotal in outcome prediction, ICP presents a dynamic and quantifiable parameter that can be continuously monitored and actively managed. Therefore, external validation should be performed in the future for generalizability [18,20,23].

To the authors’ knowledge, although prior studies have explored the prognostic implications of elevated ICP in SAH, dose-response analysis of ICP thresholds remains relatively uncommon [24-26]. The present study contributes to the growing body of evidence by applying a hazard-based dose-response approach to identify a specific ICP cutoff associated with poor outcomes. However, the authors recognized that there are some limitations to the present study.

In the present study, ICP value was collected and analyzed by intraoperative ICP measurement at the time of ventriculostomy. While the precise interval from symptom onset to ICP measurement was not recorded for all patients due to the retrospective nature of the study, we recognized its potential importance. Therefore, average ICP should be used in future prospective investigations.

As a single-center retrospective study, our results were susceptible to selection bias and might not be generalizable to broader populations [7,8]. To reduce bias, multicenter research should be performed in the future to increase the size of the study population and provide new data for updating the prediction model [13,14,21,22]. Furthermore, future research should investigate patient-specific ICP targets, and combining ICP with multimodal monitoring (e.g., CPP, brain tissue oxygenation) could enhance individualized care strategies.

Currently, various prediction tools have been employed for prognostication in SAH patients, including traditional clinical nomograms and more advanced machine learning-based models. These tools aim to improve the accuracy of outcome prediction by integrating multiple clinical and radiological parameters [27-29]. However, a comprehensive comparison of their predictive performance, usability, and real-world applicability remains limited [30]. Future research should focus on systematically evaluating these methodologies to determine their relative strengths, limitations, and potential for integration into clinical workflows, ultimately enhancing individualized patient care and decision-making [2,31].

In clinical implication, these findings may serve as a foundation for developing and refining prediction tools through the Plan-Do-Check-Act (PDCA) cycle [32,33]. Specifically, the Plan phase involves defining an ICP threshold (e.g., 24 mmHg) for closer monitoring and timely intervention. The Do phase integrates the threshold into healthcare workflows, such as alarm systems or escalation protocols. During the Check phase, patient outcomes and intervention effectiveness are assessed using audits or quality indicators. Finally, the Act phase applies the findings to refine prognostic models and improve clinical guidelines. In this context, the PDCA cycle provides continual quality improvement, ensuring that ICP monitoring techniques are evidence-based and adaptive to future data [34].

Our study demonstrates that elevated ICP, particularly beyond the threshold of 24 mmHg, is strongly associated with poor outcomes in SAH patients. These findings support the inclusion of ICP thresholds in prognostic assessment and underscore the need for vigilant ICP monitoring and early intervention in the neurocritical care setting.