Previous Abstract | Next Abstract
Printable Version
A3028
October 05, 2020
10/5/2020 10:00:00 AM - 10/5/2020 11:00:00 AM
Room Virtual
Closed-loop Vasopressor Infusion Using Phenylephrine Compared To Norepinephrine: An In-silico Exploratory Study
Joseph Rinehart, M.D., Sashini Weeraman, M.B.,Ch.B., Michael Ma, B.S., Michael-David Calderon, B.S., Paulette Mensah, B.S., Hailey Maxwell, B.S., Alexandre P. Joosten, M.D.,Ph.D.
University of California Irvine, Orange, California, United States
Disclosures: J. Rinehart: Consulting Fees; Self; Edwards Lifesciences. S. Weeraman: None. M. Ma: None. M. Calderon: None. P. Mensah: None. H. Maxwell: None. A.P. Joosten: None.
Background: We have previously developed and tested the feasibility of closed-loop norepinephrine infusions in-silico, in animal models, and most recently in human cohorts, and shown consistently its ability to minimize the incidence of hypotension. Phenylephrine is another vasopressor agent commonly used as an infusion to maintain a target arterial pressure — with a longer physiologic half-life — requiring a change in control characteristics compared to NE. In this study, we explored the ability of our existing controller to manage phenylephrine infusions with minor adjustments to control characteristics. We hypothesized that in this simple resuscitation challenge mean time-in-target would be within 5% of our previous NE performance. Methods: The proportional-integral-derivative (PID) core of the controller was reweighted for use of phenylephrine through a series of exploratory simulation experiments. Ultimately, the P and I values were left stable while the D value was adjusted to include a longer forward-looking window given the longer half-life of the drug. We tested the controller in-silico using our validated physiologic simulator and an updated phenylephrine model 2. The simulator was configured to exhibit a variable degree of septic vasodilation and hypotension, and the controller allowed to correct blood pressure in a 30-minute window without external disturbances to measure overshoot and oscillation. The controller was also run with NE as a baseline comparison group to the PE group. The primary outcome for testing was percentage of case time in target range when configured to achieve a mean arterial pressure (MAP) of 70 ± 5 mmHg. Results: 100 distinct cases for each simulation condition (NE and PE) were run. The NE group spent 96.2% (28.8 of the 30 minutes) of the time in target, including the time required initially to restore the starting hypotension in the total time. Once inside the target range the arterial pressure stayed inside with minimal overshoot (<5mmHg) and no observable rebound undershoot. The PE group spent 93% time (27.9 min) in target, including the initial run-up period. Again, all patients stayed with 5mmHg of target on initial overshoot and rebound undershoot was <1 mm Hg. Spaghetti plots of both conditions are shown in Figure 1. Conclusion: This in silico study demonstrated that the controller for vasopressor administration is able to manage phenylephrine infusions with minor adjustments. Robustness studies will be needed to test the pharmacologic assumptions.

Figure 1

Copyright © 2020 American Society of Anesthesiologists