Personalized predictive hemodynamic management for major oncologic surgery: effect of progressive implementation of
Gilda Pasta1, Luciano Frassanito2, Mariangela Calabria1
1Department of Anesthesiology, Pain Therapy and Intensive Care INT IRCCS G. Pascale Foundation, Naples, Italy.
Background:
Avoiding intraoperative hypotension and maintaining an adequate cardiac output (CO) during surgery is crucial to ensure tissue oxygen delivery to the tissues and avoid excessive fluid administration. Assisted fluid management (AFM) is a "decision-support" system based on artificial intelligence (AI) that helps the clinician to manage fluids during major surgeries. Hypotension Prediction Index (HPI) is a predictive parameter of intraoperative hypotension. The aim of this study was to assess the relative contribution of different levels of technologic assistance (CO monitoring, CO plus HPI, and CO plus HPI and AFM) to the improvement of hemodynamic management, applied to comparable cohorts of non-cardiac surgical patients.
Methods:
We conducted a retrospective analysis of consecutive patients undergoing major oncologic abdominal surgery, monitored with an arterial radial catheter progressively upgraded with three increasing levels of forecasting event technology. A personalized goal directed fluid therapy (GDT) protocol was applied in all groups. The primary outcome was the time-weighted average (TWA) mean arterial pressure (MAP) <65 mmHg among the three cohorts. Secondary outcomes were the percentage of monitoring time spent with stroke volume variation > 12% and with Cardiac Index (CI) <2 L/min/m2.
Results:
Eighty-two consecutive patients were enrolled: 26 patients in the GDT group, 28 in the HPI group and 28 in the AFM group. TWA-MAP<65 mmHg was 1.13 (0.13-1.83) mmHg in the GDT group, 0.96 (0.26-1.85) mmHg in the HPI group, 0.42 (0.07-0.93) mmHg in the AFM group. Patients with AFM spent roughly 30% of the monitoring time with a CI<2 L/min/m2 compared to less than 10% in the other two groups (Kruskal-Wallis P value 0.013).
Conclusions:
An increasing levels of artificial intelligence-based hemodynamic monitoring and decision-support tools shows a trend towards decreasing IOH, but it did not reach statistical significance.
