[Development and validation of an extremely-low frequency pressure reactivity index for traumatic brain injury...

Objective: To develop an ultra-low-frequency pressure reactivity index (PRx) (EL-PRx) based on hourly data (0.000 28 Hz) as an alternative to the conventional PRx, which requires high-frequency sampling, and provide a prognostic tool for traumatic brain injury (TBI) in resource-limited settings. Methods: This multicenter retrospective cohort study included 473 participants, including patients with TBI who were identified from the MIMIC-Ⅳ and eICU-CRD databases and those admitted to Fujian Provincial Hospital affiliated with Fuzhou University between April 2018 and April 2025. They were divided into survival (n=360) and non-survival (n=113) groups based on in-hospital all-cause mortality data. EL-PRx was calculated using 9-23 h moving windows. Propensity score matching (1∶1, caliper width 0.1×logit standard deviation) was performed to adjust for confounders, including demographics, vital signs, laboratory findings, comorbidities, supportive therapies, and IMPACT model variables. Restricted cubic spline analysis, univariate logistic regression, and receiver operating characteristic curve analysis were used to evaluate the association between EL-PRx and outcomes and its predictive performance. The optimal clinical threshold was determined using sequential Chi-squared testing. Results: A total of 473 patients were included (360 survivors and 113 non-survivors). EL-PRx was significantly higher for non-survivors than for survivors. The values were 0.14 (0.00, 0.30) and 0.07 (-0.10, 0.16) (P=0.003), respectively, within the 9-h window, and consistent differences were observed across other time windows (all P<0.05). Restricted cubic spline analysis demonstrated a non-linear positive association between EL-PRx and mortality risk. Univariate logistic regression showed that EL-PRx was significantly associated with mortality across different time windows, with odds ratios ranging from 3.825 to 8.073 (all P<0.05); the strongest effect was observed within the 17-h window (OR=8.073, 95%CI 2.053-35.697). Receiver operating characteristic curve analysis indicated that predictive performance was optimal on days 5-6 with a maximum AUC of 0.667. Sequential Chi-squared testing identified 0.15 as the optimal clinical threshold for EL-PRx, which became stable when monitoring duration was≥3-4 days. Conclusion: EL-PRx, which is derived from routinely collected hourly data, can effectively predict in-hospital mortality for patients with TBI. Its threshold is comparable to that of conventional PRx, making it suitable for intensive care settings with limited monitoring resources.