Patients
A total of 252 patients were included in the analyses. Flowchart of the study inclusion is shown in additional figure 1. Twelve patients (4.8%) were re-intubated within 72 hours after extubation. APACHE II score, Simplified Acute Physiology Score (SAPS) II, duration of mechanical ventilation, length of ICU-stay and in-hospital mortality were significantly higher in the extubation failure group than in the extubation success group (Table 1). Table 2 shows indications for intubation.
Associations between PCED and CPF and between diaphragm peak velocity and CPF
The Pearson coefficient was 0.496 (p<0.001) for the correlation between PCED and CPF and 0.347 (p<0.001) for the correlation between diaphragm peak velocity and CPF. A simple regression model with CPF as the dependent variable in relation to PCED showed significant associations between PCED and CPF (P < 0.001, beta coefficient 11.9, 95% CI 9.28–14.5, adjusted R2 = 0.243, Figure. 3) and between diaphragm velocity and CPF (P < 0.001, beta coefficient 1.97, 95% CI 1.44–2.49, adjusted R2 = 0.175, Figure. 4). A multiple regression model adjusted for age and sex showed modestly stronger associations between PCED and CPF (P < 0.001, adjusted beta coefficient 11.4, 95% CI 8.88–14.0, adjusted R2 = 0.287) and between diaphragm velocity and CPF (P < 0.001, adjusted beta coefficient 1.71, 95% CI 1.91–2.24, adjusted R2 = 0.235). Height was not used in the regression models because it was not significantly associated with CPF in relation to PCED or diaphragm peak velocity.
The equation for predicting CPF with PCED, age, and sex in mechanically ventilated patients was
Predicted CPF = PCED (cm) × 11.4 − age (years) × 0.197 + male sex × 10.2 + 50.5.
The equation for predicting CPF with diaphragm velocity, age, and sex in mechanically ventilated patients was
Predicted CPF = diaphragm peak velocity (cm/sec) × 1.71 − age (years) × 0.009 + male sex × 14.9 + 43.2,
where male sex = 1 and female sex = 0.
Bland–Altman plots were used to assess agreement between measured CPF and CPF predicted by PCED, age, and sex and between measured CPF and CPF predicted by diaphragm peak velocity, age, and sex (Figure 5 and 6). The differences between predicted CPF and measured CPF were larger at higher values in both PCED and diaphragm peak velocity.
Ultrasonographic indices, CPF, and extubation outcome
PCED and CPF were significantly lower in the extubation failure group than in the extubation success group (mean PCED: 1.22 ± 0.67 cm vs 2.32±1.13 cm, p=0.001; mean CPF: 47.1 ± 21.3 L/min vs 71.1 ± 26.9 L/min, p=0.003), while diaphragm peak velocity did not significantly differ between the two groups (Table 1). A PCED less than 1.6 cm and CPF less than 50 L/min were significantly associated with extubation failure, after adjusting for APACHE II score (adjusted OR for PCED: 7.28; 95% CI, 1.88–28.3, p < 0.001; adjusted OR for CPF: 6.1; 95% CI, 1.81–20.6, p < 0.001). Figure 7 shows the ROC curves of PCED, diaphragm peak velocity, and CPF to predict extubation failure. The AUCs of PCED, diaphragm peak velocity, and CPF for extubation failure were 0.791 (95% Cl 0.668–0.914), 0.587 (95% Cl 0.426–0.748), and 0.765 (95% Cl 0.609–0.922), respectively. The specificity and sensitivity for extubation failure with a PCED of ≤1.6 cm H2O were 0.708 and 0.750, respectively. The specificity and sensitivity for extubation failure with a CPF of ≤50 L/min were 0.741 and 0.666, respectively. There was no significant difference in predictive accuracy between PCED and CPF (p=0.61).