A method for monitoring and improving patient: ventilator interaction

Abstract

Objective: To evaluate a new approach for monitoring and improving patient-ventilator interaction that utilizes a signal generated by the equation of motion, using improvised values for resistance and elastance obtained noninvasively.

Design and setting: Observational study in intensive care units in five European centers.

Patients: We studied 21 stable patients instrumented with esophageal/gastric catheters for a previous study and ventilated alternately with pressure support (PSV) and proportional assist (PAV) ventilation with a Tyco 840 ventilator.

Measurements and results: Previously recorded digital files were analyzed in real-time by a prototype incorporating the new technology (PVI monitor, YRT, Winnipeg, Canada). Actual onsets (P(DI)-T(ONSET)) and ends (P(DI)-T(END)) of inspiratory efforts, ineffective efforts, and patient respiratory rate were identified visually from transdiaphragmatic or calculated respiratory muscle pressure. Monitor-identified T(ONSET) occurred 0.107 +/- 0.074 s after P(DI)-T(ONSET), substantially less than trigger delay observed with conventional triggering (0.326 +/- 0.086 s). End of effort was identified 0.097 +/- 0.096 s after P(DI)-T(END), significantly less than actual cycling-off delay during PSV (0.486 +/- 0.307 s) or PAV (0.277 +/- 0.084 s). The monitor detected 80% of ineffective efforts. There was excellent agreement between monitor-estimated respiratory rate and actual patient rate over a wide range (17-59/min) of patient rates (mean (+/- SD) of difference -0.2 +/- 1.9/min for pressure support and 0.2 +/- 0.9/min for proportional assist) even when large discrepancies existed (> 35/min) between patient and ventilator rates.

Conclusions: The proposed approach should make it possible to improve patient-ventilator interaction and to obtain accurate estimates of true patient respiratory rate when there is nonsynchrony.