High-flow nasal cannula in adults with acute respiratory failure and after extubation: a systematic review and meta-analysis

Acute respiratory failure (ARF) is one of the most common causes of intensive care unit (ICU) mortality [1‐3]. Oxygen therapy is a main stay of treatment for patients with hypoxemic respiratory failure. Several devices can be used to administer conventional oxygen treatments (COT), including nasal cannula, simple face masks, Venturi masks, and high-concentration reservoir masks [4]. The maximal flow rate that can be achieved with COT is 15 L/min which is lower than the inspiratory flow of most patients with ARF. Room air is often added to increase flow but at the expense of reducing the final concentration of oxygen delivered to patients at the alveolar level [5, 6]. Additionally, insufficient moisture and a lack of warm air during COT can induce discomfort for patients who require supplemental oxygen [7, 8].

High-flow nasal cannula (HFNC) delivers heated and humidified oxygen gas through the nasal or tracheal route with flow rates as high as 60 L/min in adults [6]. Several clinical trials have reported that HFNC improves oxygenation prior to intubation and reduces episodes of severe hypoxemia during intubation [9], post-cardiothoracic surgery [10], during bronchoscopy [11] and after extubation from invasive mechanical ventilation (IMV) in patients with ARF [12, 13]. Despite encouraging results from preliminary randomized controlled trials (RCTs), clarity is lacking regarding specific patient populations who may benefit from HFNC use [14‐17]. To address this deficiency in the literature, we performed the current meta-analysis to compare the effect of HFNC, COT and noninvasive ventilation (NIV) on clinical outcomes of patients receiving either initial ARF treatment or respiratory support after extubation.

We found that HFNC was superior to COT in reducing treatment failure when used as an initial support strategy and reduced rates of extubation failure, and reintubation when used after extubation. In few trials, HFNC reduced intubation rate compared to NIV when used as initial support strategy but did not impact rates of treatment failure and reintubation when used after extubation.

To date meta-analyses have shown different effects of HFNC on intubation in patients with ARF [35‐41]. The meta-analysis by Maitra et al. included 7 trials (n = 1699) found no benefit of HFNC compared to COT or NIV [35]. Subsequently, Monro-Somerville et al. combined data from 9 trials (n = 2507) comparing HFNC to other forms of respiratory support, including COT and NIV (as usual care), found no significant differences between treatment strategies in intubation and mortality rates [36]. Similarly, the review of Nedel and colleagues included 9 trials (n = 1552) of critically ill patients with or at risk of ARF found that HFNC therapy was not superior to COT or NIV [37]. In 2 trials (n = 495), a meta-analysis comparing HFNC in cardiac surgery patients, found that HFNC reduced escalation of respiratory support compared to COT [38]. Conversely, a recent meta-analysis by Ni et al. pooled 8 studies (n = 1084) including RCTs and retrospective studies and identified that HFNC reduced intubation rate compared to COT and NIV [39]. Huang et al. had found that HFNC may be benefit to avoid reintubation in critically ill patients with ARF by pooling data of 7 trials (n = 2781) [40]. The most recent review by Zhao et al. included 11 trials (n = 3459) compared HFNC to COT or NIV [41] and found that HFNC reduced intubation rate compared to COT but not to NIV. Our meta-analysis differs from previous meta-analyses in the inclusion criteria utilized, the number of trials and patients included, the outcomes reported, and in summary estimates of treatment effect. We focused on clinical indications for use of HFNC and compared its use to alternative treatments (COT and NIV). Our review represents the largest meta-analysis conducted to date including 18 RCTs and 4251 patients. We found that HFNC (vs. COT) reduced treatment failure when used as an initial support strategy in patients with ARF. Contrary to the findings of Zhao et al. [41], we found that compared to COT, HFNC reduced the rate of treatment failure (low quality) but not intubation rate (low quality evidence). Additionally, we found that HFNC (vs. COT) significantly reduced rates of both extubation failure (moderate quality evidence) and reintubation (moderate quality evidence) when used after extubation. Similar to studies conducted preterm infants, these findings suggests a potential clinical role for HFNC in the post extubation period [42]. Finally, compared to NIV, we found promising preliminary data in 2 trials that HFNC may reduce the rate of intubation when used as an initial support strategy. Taken together these findings support the use of HFNC versus COT as an initial support strategy and after extubation. Notwithstanding, several questions regarding HFNC application remain to be addressed. Further trials are needed to clarify the role for HFNC in different etiologies of ARF and compared to NIV after extubation. Several trials are currently underway to evaluate the effect of HFNC in moderate and severe ARF and in AECOPD (ClinicalTrials.gov: NCT02687074, NCT02439333).

Although HFNC therapy was initially developed for neonatal patients, indications for its use have recently been expanded to include adult patients [6, 43]. Several mechanisms have been postulated to improve oxygenation in patients who are treated with HFNC. First, the high flow rates with HFNC ‘washout’ carbon dioxide in upper airways and reduce dead space [44]. Second, the peak inspiratory flow of dyspneic patients can be met, and even exceeded, by the administration of high flow gas with HFNC thus reducing the dilution effects of the administered gas with room air [5]. Third, the ability to heated and humidified gas with HFNC facilitates tolerance [45]. Fourth, HFNC may create a small amount of positive pressure in the nasopharynx [46], which may help prevent atelectasis and recruit collapsed alveoli [47]. Patients, especially those with hypoxemic ARF, may benefit from some or all of the purported mechanisms of action associated with HFNC.

Our meta-analysis has several strengths. It is the largest meta-analyses conducted to date and to evaluate HFNC use by clinical indication. It is strengthened by an extensive search, duplicate citation screening and data abstraction, and conduct of a prespecified subgroup analysis. Our meta-analysis also has limitations. First, despite an extensive literature search, we identified only 4 trials [10, 24, 26, 34] comparing HFNC and NIV including 2 initial support strategy trials and 2 trials post-extubation trials. Second, by necessity all trials were deemed to be at high risk of performance bias as the nature of the interventions being applied precluded blinding after treatment allocation. Third, we did not construct funnel plots as fewer than 10 trials were identified for each comparison. Finally, we were not able to pool all data reported for outcomes including ICU and hospital stay, respiratory rate, and PaO₂/FiO₂ ratio due to variability in measuring and reporting of these outcomes.

We found that compared to COT, HFNC significantly reduced treatment failure when used as an initial support strategy and when used after extubation reduced both extubation failure and reintubation rates. In few trials, HFNC reduced intubation rate compared to NIV when used as initial support strategy but did not impact rates of treatment failure and reintubation when used after extubation.