Influence of Noninvasive Ventilation Protocol on Intubation Rates in Subjects With De Novo Respiratory Failure.

Coudroy, Rémi; Hoppe, Marie-Anne; Robert, René; Frat, Jean-Pierre; Thille, Arnaud W

BACKGROUND The use of noninvasive ventilation (NIV) is debated in de novo respiratory failure. Prolonged sessions, using a dedicated NIV ventilator, with high PEEP levels could be associated with better outcomes than shorter sessions using an ICU ventilator, with low PEEP levels. We performed a systematic review of randomized controlled trials to test whether the incidence of intubation was influenced by the NIV protocol in subjects admitted to the ICU for de novo respiratory failure. METHODS We selected randomized trials on NIV indexed in medical literature databases from their inception to April 2018. Pediatric studies, those performed outside of the ICU, trials with subjects on NIV for a reason other than de novo respiratory failure, and studies in which NIV protocol was not specified were excluded. Two authors independently extracted intubation rates and the NIV protocol (prolonged or short sessions, type of ventilator, and PEEP levels). RESULTS Fourteen studies, which included 750 subjects treated with NIV for de novo respiratory failure in ICU, were analyzed. Overall intubation rate was 38%, 95% CI 31-45% and was not influenced by prolonged NIV sessions or the type of ventilator. The 154 subjects treated with PEEP greater than the median overall PEEP (6 cm H2O) had a PEEP level of 8 ± 2 cm H2O and a pressure support level of 7 ± 2 cm H2O. Their intubation rate was lower than the 293 subjects treated with lower PEEP levels (25%, 95% CI 15-37% vs 43%, 95% CI 33-54%, respectively, P = .03). Inclusion criteria were heterogeneous, and critical information on NIV application were frequently lacking. CONCLUSIONS Except for high PEEP levels that might be associated with lower intubation rates, the protocol for carrying out NIV does not seem to influence intubation rate in patients with de novo respiratory failure.