Do Different Data Processing Methods Distort Minute Ventilation - Carbon Dioxide Production Relationship (VE/VCO2 slope)?

Abstract

The minute ventilation - carbon dioxide production relationship (VE/VCO2 slope) and ventilatory equivalent for carbon dioxide at ventilatory threshold (VE/VCO2 at VT) has been demonstrated to be associated with reduced short- and medium-term survival in many circumstances such as heart failure and post-major surgery. However, there exist a broad variety of data processing methods for expired gas analysis, including time-averaged and breath-averaged. PURPOSE: To determine the level of agreement in VE/VCO2 slope between different data processing methods for cardiopulmonary exercise testing (CPET) and to explore their ability to discriminate between healthy subjects and patients with corrected Tetralogy of Fallot (TOF). METHODS: We retrospectively analyzed the CPET data of 51 healthy subjects and 51 adult patients with corrected TOF. The expired gas data were processed through different methods (raw; time-averaging: 10s, 15s, 30s and 60s; breath-averaging: moving 5 breath, moving 11 breath and middle 5 of 7 breath). We recorded the VE/VCO2 slope in each data processing method and VE/VCO2 at VT. A Bland-Altman analysis was used to assess the level of agreement in VE/VCO2 slope between the different data processing methods. RESULTS: The Bland-Altman analysis indicates that the 95% limits of agreement in VE/VCO2 slope between 30s and 10s time-averaged methods range from -0.74 to 0.44. The two methods consistently provide similar findings. The 95% limits of agreement in VE/VCO2 slope between 5 breath-averaged and 10s time-averaged methods ranged from -1.03 to 2.76, which might include clinically important discrepancies. As for between group comparison, there was a statistically significant difference in VE/VCO2 at VT between healthy subjects and patients with corrected TOF (25.8 vs 27.2, p = 0.025), while no significant difference was observed in VE/VCO2 slope (p > 0.05) with different data processing methods. CONCLUSIONS: There was no clinical significant difference in VE/VCO2 slope within different time-averaged methods or within different breath-averaged methods, while clinically important discrepancies might exist between time-averaged and breath-averaged methods. The VE/VCO2 at VT is the only analyzed parameter that could discriminate between healthy and TOF subjects.