High-Flow Nasal Cannula Compared With Noninvasive Positive Pressure Ventilation in Acute Hypoxic Respiratory Failure: A Systematic Review and Meta-Analysis

OBJECTIVES:

To evaluate the efficacy and cost-effectiveness of high-flow nasal cannula (HFNC) when compared with noninvasive positive pressure ventilation (NIPPV) in patients with acute hypoxic respiratory failure (AHRF).

DATA SOURCES:

We performed a comprehensive search of MEDLINE, Embase, CINAHL, the Cochrane library, and the international Health Technology Assessment database from inception to September 14, 2022.

STUDY SELECTION:

We included randomized control studies that compared HFNC to NIPPV in adult patients with AHRF. For clinical outcomes, we included only parallel group and crossover randomized control trials (RCTs). For economic outcomes, we included any study design that evaluated cost-effectiveness, cost-utility, or cost benefit analyses.

DATA EXTRACTION:

Clinical outcomes of interest included intubation, mortality, ICU and hospital length of stay (LOS), and patient-reported dyspnea. Economic outcomes of interest included costs, cost-effectiveness, and cost-utility.

DATA SYNTHESIS:

We included nine RCTs (n = 1,539 patients) and one cost-effectiveness study. Compared with NIPPV, HFNC may have no effect on the need for intubation (relative risk [RR], 0.93; 95% CI, 0.69–1.27; low certainty) and an uncertain effect on mortality (RR, 0.84; 95% CI, 0.59–1.21; very low certainty). In subgroup analysis, NIPPV delivered through the helmet interface—as opposed to the facemask interface—may reduce intubation compared with HFNC (p = 0.006; moderate credibility of subgroup effect). There was no difference in ICU or hospital LOS (both low certainty) and an uncertain effect on patient-reported dyspnea (very low certainty). We could make no conclusions regarding the cost-effectiveness of HFNC compared with NIPPV.

CONCLUSIONS:

HFNC and NIPPV may be similarly effective at reducing the need for intubation with an uncertain effect on mortality in hospitalized patients with hypoxemic respiratory failure. More research evaluating different interfaces in varying clinical contexts is needed to improve generalizability and precision of findings.

KEY POINTS

Question: How does high-flow nasal cannula (HFNC) compare to noninvasive positive pressure ventilation (NIPPV) in terms of efficacy and cost-effectiveness in patients with acute hypoxic respiratory failure?

Findings: HFNC and NIPPV may be similarly effective at reducing the need for intubation with an uncertain effect on mortality in hospitalized patients with hypoxemic respiratory failure. Subgroup analysis suggests NIPPV delivered via the helmet interface may reduce intubation compared with HFNC.

Meaning: HFNC can likely be used as an alternative to NIPPV in patients with hypoxic respiratory failure, however, more research is needed evaluating different modalities of NIPPV in varying clinical contexts to determine how these technologies can be used.

Oxygen therapy for patients with acute hypoxic respiratory failure (AHRF) is delivered through a variety of modalities. For those that are less hypoxemic, low-flow oxygen therapy, delivered through nasal cannula or Venturi mask, may be suitable. If respiratory failure is more severe, high-flow nasal cannula (HFNC) or noninvasive positive pressure ventilation (NIPPV) may be necessary. HFNC enables delivery of heated and humidified oxygen through hard plastic nasal cannula at flow rates of 40–60 L/min, which more closely match the inspiratory needs of dyspneic patients (1). HFNC can also provide a modest amount of positive end-expiratory pressure (depending on mouth opening; ~ 7 cm H₂O) and decrease both pharyngeal dead space and nasopharyngeal resistance (2, 3). Furthermore, HFNC may be more comfortable and less obtrusive than other forms of oxygen delivery for patients (3).

NIPPV, another modality used in those with more severe respiratory failure, provides positive pressure ventilation through a face mask or helmet interface (4). This allows for titratable delivery of positive end-expiratory pressure (PEEP) and augmentation of tidal volumes in response to patient-triggered inspiration. NIPPV requires more monitoring compared with HFNC and is often less well tolerated (5). While certain conditions such as hypercapnic respiratory failure may be better treated with NIPPV, the optimal oxygen-delivery method for those with severe AHRF remains uncertain (6).

Guidelines evaluating HFNC and NIPPV are available but most often do not address comparison between these two modalities, instead focusing on comparison to standard low-flow oxygen therapy (7, 8). Considering the judicious use of healthcare resources, there are potential advantages to HFNC given it is less resource intensive and requires less setup, maintenance and monitoring compared with NIPPV (6). The primary question for this review addresses the comparison between HFNC and NIPPV, evaluating both efficacy and cost effectiveness in adult patients with AHRF.

MATERIALS AND METHODS

We registered the protocol for this systematic review and meta-analysis on PROSPERO (No. 42022328066) and used the Preferred Reporting Items for Systematic Reviews and Meta-Analyses checklist to report the findings (Supplementary Table 3, http://links.lww.com/CCX/B163).

Literature Search Strategy

We performed a comprehensive search of MEDLINE, Embase, CINAHL, the Cochrane library, and the international Health Technology Assessment database from inception until September 14, 2022. We included both controlled vocabulary and keywords in the search and used the main concepts of acute hypoxemic respiratory failure, HFNC, and NIPPV. A copy of the search strategy is included in the Supplementary Materials (http://links.lww.com/CCX/B163). We did not exclude trials based on language or quality. We also searched for unpublished studies using the Canadian Agency for Drugs and Technology in Health grey literature checklist, the references of included articles and prior meta-analyses addressing the research question. Finally, we consulted experts in the field to identify any unpublished studies.

Selection Criteria and Method

We screened citations independently and in duplicate (D.C., V.T.) in two stages; first examining the titles and abstracts and then the full text of selected citations. Any disagreements at the full text stage were adjudicated using a third reviewer (B.R.). We captured reasons for study exclusion after reviewing the full texts of identified trials.

We included studies that compared HFNC to NIPPV in adult patients (> 18 yr) hospitalized with AHRF that reported on at least one of the following outcomes of interest: mortality, need for intubation, ICU length of stay, hospital length of stay, or patient comfort (using any validated scale). We defined NIPPV as any device that delivers mechanical respiratory support without the need for endotracheal intubation through an interface that delivers continuous positive airway support (CPAP) or bilevel positive airway support. For studies of heterogeneous patient populations (including hypercapnic patients), we included the study if greater than 50% of those randomized had AHRF. For clinical outcomes, we included only parallel group and crossover randomized control trials (RCTs). For the cost-effectiveness outcomes, we included any study design that evaluated cost effectiveness, cost utility or cost benefit analyses and reported on at least one of the following outcomes of interest: costs, quality-adjusted life years (QALYs), incremental costs/effectiveness or incremental cost/QALY.

Data Extraction and Quality Assessment

We abstracted data independently and in duplicate (D.C., V.T.) using a standardized data abstraction form. For the clinical outcomes, we collected data on trial characteristics, demographic data, intervention and control arm management, and outcomes of interest. For the economic outcomes, we collected data on analytic technique, study design, time horizon, study perspective, and study characteristics.

We assessed risk of bias (ROB) for clinical studies using the modified Cochrane ROB tool 2 for RCTs (9). We assessed each RCT in the following domains: randomization sequence generation, allocation concealment, blinding, incomplete data, selective reporting, and other bias. For each domain, we rated ROB as “high,” “probably high,” “probably low,” or “low.” We assigned the overall ROB for each trial as the highest risk attributed to any domain except for blinding. Given the challenges with blinding the caregiver and patients in these trials, as long as the outcome assessors were blinded, we did not rate down these trials for ROB related to blinding. We assessed the overall certainty of evidence for each outcome using the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) framework (10).

For economic studies, quality was assessed using the Consolidated Health Economic Evaluation Reporting Standards checklist (11). We assessed each study using the standardized 24-point checklist and rated studies as high, moderate, or low quality.

Data Analysis Methods

For clinical outcomes, we used the DerSimonian-Laird random effects model with inverse-variance weighting to generate pooled treatment effects across studies. We assessed heterogeneity between trials using the chi-square test, the I² statistic, and visual inspection of the forest plots (12). We present results of dichotomous outcomes using relative risk (RR) and continuous outcomes as mean difference (MD), both with 95% CIs. We tabulated absolute differences with 95% CIs, using the NIPPV group to set baseline risk. We performed all statistical analysis using RevMan 5.3 (Cochrane Collaboration, Oxford, United Kingdom) software. For economic outcomes, we summarized study results using tables, along with quality assessments.

We had planned for two a priori subgroup analyses: 1) helmet NIPPV versus facemask NIPPV and 2) high ROB studies versus low ROB studies. We hypothesized that HFNC would show a more beneficial effect versus facemask NIPPV compared with helmet NIPPV and that HFNC would be more effective in high ROB studies. For all subgroup analysis, if the p value for the test of interaction was less than 0.05, we used the Instrument for assessing the Credibility of Effect Modification Analysis tool (13) to assess for credibility of subgroup effects. This tool considers factors such as: whether effect modification is based on comparisons within rather than between trials, whether the effect modification was correctly hypothesized a priori, whether the effect modification was supported by prior evidence, how many subgroups were investigated, and whether random or fixed effects model were used. In all other cases, we assumed that the subgroup differences are not credible enough in influence conclusions regarding the effect size or are due to chance alone.

At the request of reviewers, we also performed post hoc sensitivity analyses excluding trials presented in abstract form for the outcomes of intubation and mortality and excluding trials that did not include only hypoxic respiratory failure patients.

RESULTS

Clinical Review

Search Strategy and Study Characteristics

We reviewed 499 citations and included nine RCTs (n = 1,539) (6, 14–21) (Fig. 1). The characteristics of the included RCTs are depicted in Supplementary Table 1 (http://links.lww.com/CCX/B163). RCTs included between 15 and 791 patients. Of the nine RCTs, only one was a crossover study (16) and two were published in abstract form (14, 18). Overall, three studies compared HFNC to helmet NIPPV (6, 14, 16), five compared it to facemask NIPPV (15, 17–20) and one compared it to both facemask and helmet NIPPV (21).

Figure 1.

Preferred Reporting Items for Systematic Reviews and Meta-Analyses study flow diagram for clinical review.

Three trials included patients with mixed hypoxic respiratory failure (16, 18, 19), one included patients with mixed respiratory failure (20), one included patients with cardiogenic pulmonary edema (14), another included only immunocompromised patients (15), three included patients with COVID-19 hypoxic respiratory failure (6, 17, 21). Supplementary Tables 4 and 5 (http://links.lww.com/CCX/B163) summarizes the ROB for included RCTs. Three trials were adjudicated to have high ROB (14, 15, 18), while the remainder had low ROB or some concerns.

Outcomes

Supplementary Table 2 (http://links.lww.com/CCX/B163) summarizes the GRADE certainties and pooled estimates. Compared with NIPPV, HFNC may have no effect on intubation (eight trials; 1,189 patients; RR, 0.93; 95% CI, 0.69–1.27; low certainty; Fig. 2) and has an uncertain effect on mortality (six trials; 1,161 patients; RR, 0.84; 95% CI, 0.59–1.21; very low certainty; Fig. 3). Patients who received helmet NIPPV were less likely to be intubated when compared with HFNC, an effect which was not seen when compared with facemask NIPPV (p value for subgroup interaction = 0.006; moderate credibility; Supplementary Materials, http://links.lww.com/CCX/B163). There was no credible subgroup effect when comparing high versus low or intermediate ROB trials (Supplementary Fig. 4, http://links.lww.com/CCX/B163) or for the outcome of mortality (Fig. 3 and Supplementary Fig. 5, http://links.lww.com/CCX/B163). Sensitivity analyses excluding trials that were presented in abstract form only did not affect conclusions regarding mortality and intubation (Supplementary Figs. 6 and 7, http://links.lww.com/CCX/B163).

Figure 2.

Effect of high-flow nasal cannula (HFNC) compared with noninvasive positive pressure ventilation (NIPPV) on intubation. Studies subdivided by type of NIPPV. df = degrees of freedom, M-H = Mantel-Haenszel, NIV = noninvasive ventilation.

Figure 3.

Effect of high-flow nasal cannula (HFNC) compared with noninvasive positive pressure ventilation (NIPPV) on mortality. Studies subdivided by type of NIPPV. df = degrees of freedom, M-H = Mantel-Haenszel, NIV = noninvasive ventilation.

HFNC likely has no effect on ICU length of stay (four trials; 1,071 patients; MD, 0.00 d; 95% CI, 0.48 d less to 0.49 d more; moderate certainty; Supplementary Fig. 1, http://links.lww.com/CCX/B163) or hospital length of stay (five trials; 1,289 patients; MD, 0.34 d higher; 95% CI, 0.35 d less to 1.03 d more; low certainty; Supplementary Fig. 2, http://links.lww.com/CCX/B163). Finally, HFNC has an uncertain effect on patient dyspnea, measured using a Visual Analog Scale (four trials; 543 patients; MD, 1.14 higher; 95% CI, 1.28 lower to 3.56 higher; very low certainty evidence; Supplementary Fig. 3, http://links.lww.com/CCX/B1633).

Sensitivity analysis removing the one study that included mixed respiratory failure patients (20) did not alter any of our conclusions (Supplementary Figs. 9–11, http://links.lww.com/CCX/B163).

Economic Review

Quantity of Studies Available

Of the eight citations identified, we included one cost effectiveness study (22) (Supplementary Fig. 8, http://links.lww.com/CCX/B163). The clinical input for this study was a previously published RCT (the FLORALI study) comparing HFNC to facemask NIPPV in patients with hypoxic respiratory failure of any cause (19). Given that our clinical meta-analyses included eight other RCTs, which when pooled resulted in a substantially different effect estimate of intubation, we did not think that the cost-effectiveness study accurately represented the effectiveness of HFNC when compared with NIPPV. As a result, we opted to exclude this study and, thus, not perform a cost-effectiveness analysis.

DISCUSSION

When compared with NIPPV, HFNC likely had no effect on the need for intubation and an uncertain effect on mortality. In subgroup analysis, as opposed to the facemask interface, NIPPV delivered through the helmet interface may reduce intubation compared with HFNC. In terms of other outcomes, there was likely no difference in ICU or hospital length of stay and an uncertain effect on patient-reported dyspnea. No conclusion could be made regarding the cost effectiveness of HFNC compared with NIPPV.

While previous systematic reviews have been conducted evaluating this research question (23), this is the first review to be performed following the publication of two large RCTs in 2021 (6, 17). The HENIVOT trial (6) compared helmet NIPPV to HFNC and while it showed no difference in the primary outcome of respiratory support free days, it did show a significant reduction in intubation rates with helmet NIPPV. It is worth noting, HENIVOT was underpowered for intubation, and thus, results are imprecise impacting strength of conclusions. The other recent RCT, RECOVERY-RS, was a three-arm trial that compared HFNC and NIPPV to conventional oxygen therapy (COT) (17). While the primary composite outcome of tracheal intubation and mortality showed no difference between NIPPV and HFNC when compared against COT, it is important to note, that this study had significant numbers of crossover. Additionally, NIPPV in RECOVERY was strictly CPAP, rather than the bilevel ventilation with pressure support. This may have reduced the generalizability of the results as most clinicians use bilevel ventilation through the helmet or facemask interfaces. While both of these newer RCTs seem to show a benefit of NIPPV over HFNC, when pooled together with other RCTs, this effect is tempered. Thus, any conclusion regarding the effectiveness of one mode over another remains challenging.

Considering the different modalities of NIPPV in this review, a credible subgroup effect was seen when comparing helmet NIPPV to HFNC as opposed to facemask NIPPV for the outcome of intubation, favoring helmet. Prior RCTs (24) and systematic reviews (4) examining this topic seem to support this finding, with low certainty evidence indicating that the helmet NIPPV may reduce mortality and intubation when compared with facemask NIPPV. The physiology may be related to improved PEEP and increased tolerability and comfort offered by the helmet (16, 25). However, only three studies evaluated this comparison (6, 14, 16), of which only two examined intubation. Thus, while certainly hypothesis generating, the sample sizes and effect sizes used to make this comparison are too small, to allow for definitive conclusions.

Beyond hypoxemic respiratory failure, many questions regarding the use of HFNC versus NIPPV remain. For example, the optimal oxygenation strategy in hypercapnic respiratory failure and in patients with post-extubation respiratory failure remains unclear. While current clinical guidelines recommend the use of HFNC over COT in many of these selected populations (hypoxemic respiratory failure, post extubation failure, and prophylactic use in high-risk postoperative patients) (8), it is unclear whether there could be incremental benefit to NIPPV beyond what is provided with HFNC. Weighed against the potential incremental clinical benefit of NIPPV, is the higher degree of patient comfort reported with HFNC, especially compared with facemask interface (26). Unfortunately, in this review, patient comfort was not sufficiently reported among the included studies to allow for pooling. While pooled dyspnea scores do not favor one technology over the another, dyspnea is only one aspect of comfort, and other important aspects such as tolerability, pain or agitation were not collected or reported. Thus, whether HFNC is truly more comfortable and thus tolerable when compared with NIPPV remains uncertain and could be interface dependent.

This is the largest and most comprehensive systematic review and meta-analysis to specifically compare HFNC and NIPPV in AHRF. Strengths of this study include a comprehensive search, a high degree of methodologic rigor, a pre-registered protocol, assessment of GRADE certainty allowing for appropriate contextualization of results, and inclusion of the most recent large recent RCTs compared with previous reviews. This review also has limitations. First, the total number of included patients and the number of events are small contributing to ongoing imprecision and low or very low certainty in pooled outcomes. Second, by including all studies that compared NIPPV to HFNC, there may be important clinical and statistical heterogeneity across trials. To help address this, we performed subgroup analysis evaluating HFNC compared with facemask NIPPV versus helmet NIPPV. Despite planned analysis, sparsity of data did not allow for evaluation of effect based on etiology of hypoxemic respiratory failure or NIPPV technique (comparing CPAP vs bilevel ventilation). Further, no included studies evaluated combination therapy of various modalities of respiratory support—which may be the optimal approach to noninvasive respiratory support. This highlights the need for further study on how specific causes of acute respiratory failure respond to HFNC and NIPPV.

CONCLUSIONS

HFNC and NIPPV may be similarly effective at reducing the need for intubation in hospitalized patients with hypoxemic respiratory failure. More research is needed evaluating different interfaces, their cost-effectiveness and in varying clinical contexts to improve generalizability and precision of findings.

Supplementary Material

cc9-5-e0892-s001.pdf ^{(1.9MB, pdf)}