Figure 2

Changes in PaCO2 based on survival status, response to HFOV and diseaseseverity. A: Differences in PaCO₂ in survivors (opencircles) and non-survivors (filled circles) over 72 hours of HFOV. X-axisindicates time (hours) on HFOV. Time 0 is the time on CMV immediately precedingthe change to HFOV. 'Last' is the last measurement on HFOV prior to returningto CMV. There was no difference in the trend of change in PaCO₂ between the two outcome groups. Data are displayed as mean and error barsrepresent SEM at each time-point. B: Patients were subdivided into twogroups (responders and non-responders) based on the analysis of the ROC curve.They were considered responders if their PaO₂/FiO₂ ratioimproved >38% from baseline. Responders show a trend towards better CO₂ clearance (P = 0.07). Data are displayed as mean and error barsrepresent SEM. C: Patients were subdivided into two groups based ondisease severity. Patients with more severe disease (open circles) have agreater clearance in PaCO₂. The number of patients (Survivors - S;and non-survivors -NS) at different time- points after initiating HFOV were: atbaseline (S, n = 45; NS, n = 57); 12 hours (S, n =44; NS, n = 38); 24 hours (S, n = 39; NS, n = 28);48 hours (S, n = 31; NS, n = 24); 72 hours (S, n =25; NS, n = 19). CMV, conventional mechanical ventilation; FiO_2, fraction of inspired oxygen; HFOV, high-frequency oscillatoryventilation; PaCO₂, partial pressure of carbon dioxide in arterialblood; ROC, receiver-operating curve; SEM, standard error of the mean.