The use of the hypotension prediction index (HPI) software in the operating room: A narrative review

Abstract

Hypotension is a daily problem in the operating room and a monitoring platform allowing prediction of arterial hypotension and providing assistance to diagnose its mechanisms is clinically needed. The use of HPI (hypotension prediction index) as an alarm system based on a proprietary algorithm derived using machine learning from multiple components of the arterial waveform to predict shortly hypotension has been validated in studies using invasive and non-invasive continuous arterial pressure monitoring in cardiac and non-cardiac surgery. It also has been compared to standard of care in managing intraoperative hypotension with conflicting results. As HPI is very strongly correlated with mean arterial pressure (MAP), it could simply mirror concurrent MAP and have equal predictive performance. Even if HPI and MAP are not interchangeable, the overwhelming influence of MAP in the model could thus lead to a minimal diagnostic advantage of HPI in clinical practice. The additional value of HPI associated with guidance treatment protocols to improve patients’ outcomes should probably be further assessed in large-scale well-designed studies to justify its extra-cost before widen the range of patients that might benefit.