High-flow nasal cannula oxygen therapy versus conventional oxygen therapy in patients with acute respiratory failure: a systematic review and meta-analysis of randomized controlled trials

Acute respiratory failure (ARF) is a common and life-threatening medical emergency in patients admitted to hospitals [1]. It is caused by a variety of diseases, including heart failure, pneumonia, and exacerbations of chronic obstructive pulmonary disease. Many patients with ARF require oxygen therapy. The devices for oxygen therapy include unassisted oxygen delivery devices and assisted ventilation devices [2]. Unassisted oxygen therapy is also called conventional oxygen therapy (COT). It is the main supportive treatment administered to patients with ARF and is usually delivered with nasal prongs or facemasks. Assisted ventilation devices that are commonly used in hospitals include noninvasive ventilation (NIV, e.g., continuous positive airway pressure and biphasic positive airway pressure) and invasive mechanical ventilation (IMV). Previous studies have shown that avoiding IMV significantly decreases the risk of death [3, 4]. Therefore, choosing an optimal oxygen therapy device is very important for reducing the rates of IMV and mortality while also ensuring patients’ safety and comfort.

The effect of COT is limited. The maximal flow rate that these COT devices can deliver is typically only 15 L/min (except for the Venturi mask), which is far lower than the demands of patients with ARF. This discrepancy leads to a significant decrease in the fraction of inspired oxygen (FiO₂) that ultimately reaches a patient’s lungs [5].

ARF patients admitted to the hospital may receive NIV. However, currently, the effects of NIV for these patients with respect to improvements in outcomes are conflicting and the use of NIV in hypoxemic ARF has recently been questioned [6‐8]. Furthermore, NIV is not without limitations. The effect of NIV is highly dependent on a patient’s cooperation, which is also called patient-ventilator synchrony. Additionally, there are many factors that affect the comfort of patients undergoing NIV that may lead to NIV failure, such as the interface, the amount of air leaks, the ventilator settings, pressurization and triggering performances [9]. Moreover, NIV is associated with gastric distension, which may further reduce the functional residual capacity and is poorly tolerated in some patients. [10, 11] Therefore, IMV may still be needed [11].

The high-flow nasal cannula (HFNC) is a recently developed oxygen therapy device in adult patients that can deliver a humidified and heated mixture of air and oxygen at a very high flow rate. It can provide a maximal flow rate of up to 60 l per minute with an FiO₂ of 100% [5]. The use of an HFNC has been demonstrated to generate positive airway pressure at end-expiration, ameliorate oxygenation and dyspnea, reduce the work of breathing and the respiratory rate, and be more comfortable for patients [5, 12‐19]. These benefits are attributed to the mechanisms of HFNCs, including their ability to more adequately meet the peak flow of inspiration, flush the anatomical dead space, and deliver warm and humidified gas, thereby promoting mucociliary function [20, 21].

The use of HFNCs has become increasing popular in the treatment of many diseases and conditions, such as post-extubation, pre-intubation, sleep-related hypoventilation, cardiac surgery, and heart failure, and as an alternative to NIV [20]. However, currently, whether ARF patients benefit from this therapy is unclear and there is a lack of large-scale evidence on the use of HFNCs in patients with ARF. Some studies have shown that HFNC therapy is associated with an improved respiratory state or mortality rate in patients with ARF [2, 22]. However, other studies have not found significant differences between HFNC and COT groups [23, 24].

Recently, some meta-analyses tried to assess the efficiency of HFNC therapy in ARF patients. However, there were controversial results between these studies. [25, 26] After fully reviewing these meta-analyses, we found some studies that evaluated HFNC therapy in post-extubation patients were also involved in these studies. Though some post-extubation patients may suffer reintubation due to ARF, they constitute a different patient population rather than actual ARF patients. In the present systematic review and meta-analysis, we sought to evaluate whether there were differences between HFNC therapy and COT in treating ARF patients rather than post-extubation patients regarding the escalation of respiratory support and other aspects.

We performed this systematic review and meta-analysis according to the guidelines described in the Cochrane Handbook for Systematic Reviews of Interventions [27] and PRISMA statements.

The selection process of the eligible studies is shown in Fig. 1. Initially, 1030 potentially relevant records were identified. By screening the titles and evaluating the abstracts, we removed duplicate studies, reviews, case reports, animal studies, comments, and studies that were not randomized controlled studies, resulting in 8 studies that remained for assessment. Of these, 1 study compared HFNC therapy with NIV [28], 1 study evaluated the effect of HFNC therapy during endotracheal intubation [29], and 2 studies were conducted to prevent ARF after planned extubation [12, 30]. These studies were excluded. Finally, 4 studies were included in the present meta-analysis [2, 22‐24]. The quality of the included studies is shown in Table 1.

A total of 703 patients with ARF were included in this meta-analysis. Of these patients, 371 were randomly assigned to the HFNC group and 332 were assigned to the COT group. Table 2 shows the basic demographic characteristics of all included patients.

Recently, HFNC oxygen therapy has achieved widespread use in adult ARF patients in emergency departments and intensive care units [20, 31, 32]. However, the effect of HFNC therapy in adult ARF patients remains inconclusive. The present meta-analysis included 4 RCTs that studied 703 patients (371 HFNC and 332 COT patients) to examine whether there were differences between HFNC therapy and COT in the treatment of ARF. The overall estimates of this meta-analysis showed that there were no significant differences between the HFNC and COT groups in the rates of escalation of respiratory support (RR, 0.68; 95% CI, 0.37, 1.27; z = 1.20, P = 0.23), intubation (RR, 0.74; 95% CI, 0.55, 1.00; z = 1.95, P = 0.05), mortality (RR, 0.82; 95% CI, 0.36, 1.88; z = 0.47, P = 0.64), or ICU transfer (RR, 1.09; 95% CI, 0.57, 2.09; z = 0.26, P = 0.79) in the treatment of ARF. Our results were similar to a previous study [26].

Although the present meta-analysis found no significant differences between HFNC therapy and COT for the treatment of adult ARF patients, it should be noted that there was significant heterogeneity between the RCTs included in the present study, which may have affected our conclusions. A series of factors may have led to this significant heterogeneity. First, as shown in Table 4, the criteria for ARF differed between the 4 studies (Table 4). The different criteria for inclusion led to different degrees of ARF among the patients in the four studies. In the study by Frat and colleagues, there was no significant difference in the intubation rate between the two groups, but when they conducted a post hoc analysis according to the ratio of PaO₂/FiO₂ at enrollment (≤ 200 mmHg versus >200 mmHg), they found that for the subgroup of patients with a PaO₂/FiO₂ ≤ 200 mmHg, the intubation rate was significantly lower in the HFNC group than in the COT group [22]. Therefore, we consider the degree of ARF to be an important factor influencing the effectiveness of HFNC therapy. In addition, the differences in the degree of ARF experienced by the patients in the present meta-analysis may have eliminated the potential differences between the two groups. In fact, some therapies may only be useful in more critically ill patients.

Second, the starting flow of HFNC therapy may also affect results. The starting flows of HFNC therapy differed between the four studies (Table 2). In a study by Parke et al., researchers measured patients’ nasopharyngeal pressure when HFNC therapy was performed with gas flows of 30, 40, and 50 L/min [14]. They found that during HFNC therapy, the mean nasopharyngeal airway pressures were 1.5 ± 0.6, 2.2 ± 0.8, and 3.1 ± 1.2 mmHg at 30, 40, and 50 L/min, respectively. They demonstrated that the level of PEEP as a benefit of HFNC therapy was flow-dependent. The different starting flows may have led to different levels of PEEP and could have affected the results. Consistently, in two studies by Bell et al. and Frat et al., the starting flows were all 50 L/min and these studies showed more benefits in the HFNC group than in the COT group. [2, 22]

Third, as shown in Table 3, the criteria for the escalation of respiratory support differed between the 4 studies. In addition, the strategies for escalation also differed. The different criteria and strategies used for the escalation of respiratory support may have led to a bias. Furthermore, three of the studies included the option of escalating to NIV as a strategy for the escalation of respiratory support in the HFNC group. However, whether the patients who failed to improve with HFNC therapy could be recovered by escalating to NIV is currently unclear [8].

The subgroup analysis showed that HFNC therapy may decrease the rates of escalation of respiratory support (RR, 0.71; 95% CI, 0.53, 0.97; z = 2.15, P = 0.03) and intubation (RR, 0.71; 95% CI, 0.53, 0.97; z = 2.15, P = 0.03) when ARF patients are treated with HFNC therapy for ≥24 h compared with COT, which is not surprising. As we know, some therapies may only be useful with a sufficient duration. In the other two RCTs [2, 23], the durations of HFNC therapy were only 2 h, which may be too short to show its benefit. HFNCs have shown greater benefits with longer durations in post-extubation patients. A previous randomized controlled study by Maggiore et al. compared HFNC therapy with COT in 105 patients after extubation [12]. The duration of HFNC therapy was at least 48 h. Notably, the results showed that the reintubation rate or any form of escalation of respiratory support was significantly lower in the HFNC group than in the COT group. The duration of HFNC therapy in their study was similar to that used in the study by Frat et al., which was included in this meta-analysis. In addition, in the study by Frat et al., HFNC oxygen therapy also showed a positive effect in decreasing 90-day mortality [22]. Conversely, when the duration of HFNC oxygen therapy was shorter, studies comparing HFNC therapy and COT often showed negative results, including the study by Lemiale et al. that was included in the present meta-analysis [23]. Therefore, it seems that the duration of HFNC therapy is associated with its efficacy, with longer durations of HFNC therapy potentially leading to better results. The optimal duration of HFNC therapy in patients with ARF is still unclear and the durations of HFNC therapy in the relevant studies varied greatly. Our present meta-analysis could provide useful information in this regard.

There are several limitations of our meta-analysis. First, there were few studies that compared HFNC therapy and COT in patients with ARF and the number of patients included in our meta-analysis was limited. According to the study by Frat et al., the intubation rate was 38% in the HFNC group and 47% in the COT group. Assuming an intubation rate of 45% in the COT group, to detect a 5-percentage point reduction in the intubation rate in the HFNC group with an α of 0.05 and a β of 0.20 would require each group to enroll approximately 1220 subjects. Therefore, further large-scale studies are needed to confirm our results. Second, the effects of HFNC therapy in hypoxic ARF may be different from those in hypercapnic ARF, however, we could not perform a subgroup analysis relative to this aspect due to a lack of raw data. Third, since only 4 studies were included in the present meta-analysis, a funnel plot could not provide sufficient power to reveal a publication bias. Fourth, it should be noted that the durations of HFNC therapy were different among the 4 studies, especially in the studies by Bell and Lemiale [2, 23] in which the durations of HFNC therapy were only 2 h. Our subgroup analysis showed that a longer duration of HFNC therapy (≥24 h) may benefit ARF patients, so the inclusion of studies with different durations of HFNC therapy might produce biases.

Our meta-analysis demonstrated that HFNC therapy was similar to COT in ARF patients. The subgroup analysis showed that HFNC therapy may decrease the rates of the escalation of respiratory support and intubation when ARF patients were treated with HFNC therapy for ≥24 h compared with COT. Further high-quality, large-scale studies are needed to confirm our results.