Noninvasive Ventilation in Acute Respiratory Failure: Who Will Benefit?
Of course, these inconsistent findings could partly been explained by the difference between SOT and HFNC therapy; more importantly, the heterogeneity of included patients may play an important role in it.
ARF is a complex clinical syndrome in clinical practice, which could be caused by many diseases such as pneumonia, acute respiratory distress syndrome (ARDS), cardiac failure, pulmonary embolism, etc. The response to NIV therapy is largely different in these subgroups. For example, NIV has been proved to be effective in cardiac failure (4), whereas inappropriate in severe ARDS (5). In the current meta-analysis, 11 studies were pooled. However, the characteristic of included patients were diverse, and random-effect model used as the heterogeneity was significant (I2 = 69%). We noticed that compared with studies with low PaO2/FIO2 (red circles in Fig. 1: Frat 2015, Lemiale 2015, and Delclaux 2000), studies with high PaO2/FIO2 (> 200 mm Hg) had more significant reduction of intubation rate (smaller and significant RR) (Fig. 1). This finding is consistent with the conclusion by Bellani et al (5) that in patients with ARDS, NIV was associated with higher ICU mortality in the subset with PaO2/FIO2 lower than 150 mm Hg, in which the ventilator-induced lung injury is more likely to happen as NIV is commonly associated with unavoidable large tide volume. On the other hand, we also noticed that three studies with low PaO2/FIO2 still showed a significant reduction of intubation rate, compared with SOT. However, two of these (green circles in Fig. 1: Ferrer 2003 and Antonelli 2000) had a high proportion of patient with cardiac-related ARF (22.5%, 28.5%, respectively). Dozens of studies have reported that NIV support would benefit patients with heart decompensation (4). Thus, the inclusion of these patients may cause bias conclusion.
The authors presented an enlightening study. Yet, we still want to emphasis that inappropriate NIV support is associated with high mortality, and further investigations regarding patient selection before NIV initiation are urgently needed.