O032 Virtual Ventilation: an evaluation of non-invasive ventilation polysomnography and utility of ventilator recorded data

INTRODUCTION: Non-invasive ventilation (NIV) is a treatment for respiratory failure for a wide range of medical conditions. Following technological evolution of ventilator software, ventilator data is available for review remotely. There has been little scientific evaluation of whether this remotely recorded and stored data should be used to optimise ventilator settings. The aim of this study was to compare the agreement of recommended ventilator settings based on three methods: laboratory NIV polysomnography (PSG), ventilator data only and ventilator data with augmented signals. METHODS: Fifty NIV PSG studies were retrospectively selected. Each study was duplicated using three different signal types: laboratory NIV PSG (A), ventilator only signals (B) and ventilator signals with respiratory bands, transcutaneous carbon dioxide and oximetry (C). Two respiratory and sleep physicians were provided deidentified clinical vignettes, asked to review 150 ventilation studies, and prescribe their recommended end-of-night ventilator settings. RESULTS: Comparative analysis of pressure support using Bland and Altman plots between study type A and B demonstrated mean difference (MD) 0.66 (limits of agreement (LOA) -1.92 to 3.25)cmH2O, study type A and C MD 0.08 (LOA -2.72 to 2.88)cmH2O, study type B and C MD 0.58 (LOA -3.12 to 1.96)cmH2O. CONCLUSION: Comparison of full-signal NIV PSG with ventilator only and augmented ventilator signals did not result in a significant difference in mean pressure support prescribed by the physician. However, there were wide limits of agreement, suggesting that in some circumstances, a reliance upon ventilator only signals may result in significantly different pressure support prescription for the patient.