Introduction: Non-invasive ventilation (NIV) is a treatment that improves mortality and quality of life in chronic respiratory failure.  The monitoring used to titrate NIV varies substantially; however, treatment is directed towards improving pulmonary gas exchange and sleep quality.  The aim was to evaluate whether different montages of signals, influenced identification of respiratory and patient ventilator asynchrony (PVA) events and ventilation titration.   

Method: 50 laboratory NIV polysomnography (PSG) studies were reviewed under 3 different signal conditions: NIV PSG, ventilator derived signals (V) and polygraphy (PG): ventilator signals, respiratory bands, transcutaneous carbon dioxide and oximetry. Two respiratory and sleep physicians reviewed and scored respiratory and PVA events for each individual study with agreement quantified with Cohen?s kappa statistic.  Physicians? confidence and satisfaction of ventilator settings and titration recommendations were ranked on a 5-point Likert scale and compared with ordinal logistic regression. 

Results: There were marked differences in the identification of respiratory events including upper airway obstruction with reduced respiratory drive; PSG vs V kappa 0.1(CI -0.12, 0.32) and PSG vs V kappa 0.43(CI 0.2, 0.67). Ineffective effort, a measure of PVA, PSG vs V studies demonstrated kappa 0.02(CI 0, 0.05) compared with PSG vs PG kappa 0.34(CI 0.16, 0.53). Physician confidence for ventilation prescription was lower in PG and V studies at -1.3 and -3.5 points respectively when compared with PSG. 

Conclusion: Ventilator only derived signals resulted in key differences in respiratory event, PVA identification and physician confidence.  This may help inform future models of NIV care.