PURPOSE: We recently proposed a real-time, non-invasive, automatic method to describe patient-ventilator asynchrony (PVA) using time-frequency analysis of airway flow signals1. The aim of the present study is to determine the relationship of patient outcome to PVA, characterized by spectral analysis.
METHODS: Prospective, longitudinal, observational study of adult critically ill patients enrolled within 24 hours of endotracheal intubation. A computerized data acquisition system recorded the airway flow signal at 30 Hz and calculated a frequency spectrum at 2.5 min intervals from enrollment to weaning or to ICU discharge. The amplitude ratio of the first harmonic to the zero frequency component (H1/DC) was used to determine of asynchrony. We averaged all H1/DC values computed during the full period of observation and used a value of < 45% as the cutoff for asynchrony. Other ventilator and hemodynamic data were also acquired on a continuous basis. Complications, e.g. barotrauma as well as 28-day and 180-day mortality were recorded. Figures are shown as median [IQR].
RESULTS: We enrolled 18 patients aged 61 [52-72] years of which 8 were males; SOFA and SAPS II scores upon enrollment were 5.0 [5.0-7.8] and 43 [35-46], respectively. For the group, mean H1/DC during the observation period was 44.2 [41.8-48.9] %. Ten patients had mean H1/DC during the observation period < 45%. Time on mechanical ventilation for this group was 6.5 [2.5-9.0] days whereas those with H1/DC > 45% was 2.0 [1.8-3.0] days (p = 0.077 by Mann Whitney test). There was a trend towards higher mortality in the asynchronous group (4/10 vs 1/8). SAPS, SOFA scores and other recorded ventilator and hemodynamic variables were not associated with outcome.
CONCLUSIONS: In this preliminary study, we noted a significant trend towards longer time on mechanical ventilation for patients with PVA as determined by H1/DC < 45% calculated by spectral analysis.
CLINICAL IMPLICATIONS: Spectral analysis of airway signals is an automatic, non-invasive method that may prove useful in detecting and correcting PVA and may eventually have a role in improving patient outcome.
DISCLOSURE: The following authors have nothing to disclose: Binu George, Aparna Das, Eliana Gonzalez, Connor Edsall, Guillermo Ballarino, Guillermo Gutierrez
No Product/Research Disclosure Information