[Early mean arterial pressure variability and in-hospital mortality from severe subarachnoid hemorrhage: a U-shaped association based on MIMIC-IV and eICU-CRD data].

OBJECTIVES

To explore the association between mean arterial pressure variability (MBPV) and in-hospital mortality in patients with non-traumatic subarachnoid hemorrhage (NSAH).

METHODS

This study was conducted using the data of NSAH patients from the Medical Information Mart for Intensive Care IV (MIMIC-IV) database and the eICU-CRD, with the final sample size ranging from 1345 to 1820 subjects across different MBPV indicators. Mean arterial pressure (MAP) data within 24, 48, and 72 h after admission were extracted, and the standard deviation (SD), average real variability (ARV), range, and successive variability (SV) of MBPV were calculated. Restricted cubic spline (RCS) combined with Cox regression analysis was adopted to evaluate the nonlinear correlation and determine the risk inflection point using piecewise regression. Time-dependent Cox regression was performed only for variables violating the proportional hazards assumption. Two sensitivity analyses were conducted (excluding patients with surgery within 72 h of admission and those with missing data of medication) to validate the robustness of the results. The primary outcome was all-cause in-hospital mortality.

RESULTS

MBPV at all time windows showed a U-shaped nonlinear association with in-hospital mortality of the patients (all P<0.05), and the strongest association was found at 24 h. The inflection point of MBP SD was 8.9, and either excessively low or high fluctuation increased the mortality risk. Time-dependent analysis showed that ARV and SV of MBPV became more hazardous after 4-7 days, while its SD and range were risky mainly within the first 14 days. Both sensitivity analyses confirmed the robustness of the main results.

CONCLUSIONS

Early MBPV presents a U-shaped association with in-hospital mortality from NSAH, and moderate fluctuation predicts better prognosis. MBP SDs within 24 and 72 h are the most robust indicators, and the former can serve as the preferred indicator for early blood pressure monitoring.