Predicting Blood Pressure and Blood Pressure Variability in Spontaneous Intracerebral Hemorrhage in the Emergency Department Using Machine Learning

TL;DR

This study uses machine learning to identify factors that predict blood pressure and blood pressure variability in patients with spontaneous intracerebral hemorrhage in emergency departments.

Contribution

The study introduces machine learning to predict blood pressure outcomes and variability in sICH patients, highlighting initial blood pressure as a key factor.

Findings

Initial systolic blood pressure at ED triage was the top predictor of achieving a target SBP ≤160 mmHg at discharge.

Predictors of increased blood pressure variability included mechanical ventilation, elevated serum glucose, and high initial SBP.

142 patients were evaluated, with 60% discharged at or below the target SBP.

Abstract

Introduction: Spontaneous intracerebral hemorrhage (sICH) is a devastating type of stroke. Blood pressure reduction is crucial in its management and is well mentioned in current guidelines; however, the role of blood pressure variability (BPV) in emergency departments (EDs) has not been well studied. This study aimed to identify predictors of lower systolic blood pressure (SBP) (≤160 mmHg) and BPV at ED discharge and course, respectively. Methods: This is a retrospective study of prospectively collected data at a quaternary care center of adult patients diagnosed and treated with sICH between 1 January 2017 and 31 December 2020. The primary outcome of interest was SBP at ED discharge; this was divided into two groups: a control group composed of patients discharged with an SBP ≤ 160 mmHg and a comparison group composed of patients discharged with an SBP > 160 mmHg. Secondary outcomes included measures of BPV, specifically successive variation (SBPSV), and standard deviation (SBPSD) during ED course. Machine learning algorithms were used to identify predictors of SBP at ED discharge: SBPSV and SBPSV. Results: This study evaluated 142 patients, of which 85 (60%) were discharged with SBP ≤ 160 mmHg. The mean SBP at ED discharge was 133 (±16.1) mmHg for the control group and 184 (±21.3) for the comparison group (difference −51; 95% CI −58 to −45; p < 0.001). The top five predictors for the primary outcome identified by machine learning included initial SBP at ED triage, serum sodium, clevidipine administration, serum glucose, and serum creatinine. Predictors for secondary outcome included mechanical ventilation, serum glucose, and initial SBP at ED triage. Conclusion: Initial SBP was the top predictor of achieving a goal SBP ≤160 mmHg at ED discharge in patients with sICH. Predictors of increased BPV included mechanical ventilation, elevated serum glucose, and high initial SBP in the ED. While further studies are necessary to confirm our observations, clinicians should consider these factors when they care for patients with sICH.

Figures