Effect of Flow Rates of High-Flow Nasal Cannula on Extubation Outcomes

Abstract

Background

High-flow nasal cannula (HFNC) has emerged as a promising intervention for postextubation oxygen therapy, with the potential to reduce the need for reintubation. However, it remains unclear whether using a higher flow setting provides better outcomes than the commonly used flow rate of 30 to 50 L/min.

Research Question

Does setting the flow rate of HFNC at 60 L/min vs 40 L/min for postextubation care result in different extubation outcomes?

Study Design and Methods

This randomized controlled trial assigned intubated patients to receive HFNC at either a 60 L/min or 40 L/min flow rate following extubation. The assigned flow rate was maintained for 24 hours. The primary outcome was a composite of reintubation or the use of noninvasive ventilation (NIV) within 48 hours’ after extubation. Key secondary outcomes included ahead-of-schedule changes in HFNC settings and mortality.

Results

A total of 180 patients were randomized; 169 were included in the analysis (86 in the 40 L/min group and 83 in the 60 L/min group). The primary outcome events occurred in 19 patients (22.1%) in the 40 L/min group and in 14 patients (16.9%) in the 60 L/min group (risk difference, 5.2%; 95% CI, –6.7% to 17.1%; P = .39). For secondary outcomes, the 40 L/min group was associated with a higher risk of escalation in respiratory support, defined as using NIV or up-titration of HFNC settings (24 [27.9%] vs 8 [9.6%]; P = .002).

Interpretation

In unselected extubated patients, setting the HFNC flow rate at 60 L/min did not reduce the risk of reintubation or NIV use compared with a flow rate of 40 L/min. Using a flow rate of 40 L/min with as-needed up-titration may be a reasonable alternative to setting the flow at 60 L/min for postextubation care. However, this trial may not have been sufficiently powered to exclude a small between-group difference.

Clinical Trial Registration

ClinicalTrials.gov; No.: NCT04934163; URL: www.clinicaltrials.gov.

FOR EDITORIAL COMMENT, SEE PAGE 1257

Take-Home Points.

Study Question: Does setting the flow rate of high-flow nasal cannula at 60 L/min vs 40 L/min for postextubation care result in different extubation outcomes?

Results: This randomized trial comparing 2 high-flow nasal cannula flow rates (60 L/min vs 40 L/min) for postextubation care found no significant difference in the composite outcome of reintubation or rescue use of noninvasive ventilation, although the 40 L/min group had a higher incidence of respiratory support escalation.

Interpretation: A flow rate of 40 L/min with as-needed up-titration may be a reasonable approach compared with routinely setting the flow at 60 L/min.

Postextubation respiratory care is a challenge for health care providers working in the ICU. Extubated patients often require oxygen therapy to correct residual oxygenation impairment resulting from the initial respiratory failure event.^1,2 With conventional oxygen therapy, extubated patients still face a 10% to 20% risk of extubation failure even after passing a successful spontaneous breathing trial (SBT).^2,3 The pathophysiology of extubation failure typically involves airway and cardiopulmonary issues, such as changes in airway patency, alveolar edema, and lung derecruitment when removing the artificial airway and transitioning from positive pressure ventilation to unassisted breathing.^1,4,5 Noninvasive ventilation (NIV) and high-flow nasal cannula (HFNC) oxygen therapy are increasingly used to improve extubation outcomes due to their advantages over conventional oxygen therapy for postextubation respiratory support.6, 7, 8

Clinical trial data have shown that HFNC can reduce the risk of reintubation compared with conventional oxygen therapy.^9,10 HFNC delivers heated and humidified gases with airflow rates as high as 30 to 60 L/min, which can provide consistent oxygen concentration, a low level of positive end-expiratory pressure (PEEP), and CO₂ washout in the upper airways.11, 12, 13 Physiologic studies suggest that these beneficial effects are correlated with flow rates.^12,14,15 However, it remains uncertain whether these flow-dependent physiologic benefits can translate into improved clinical outcomes when a higher flow rate is applied to patients after extubation.

We therefore conducted a randomized controlled trial to compare a flow rate of 60 L/min vs 40 L/min for HFNC on extubation outcomes among patients undergoing planned extubation. We hypothesized that HFNC with a flow rate of 60 L/min would provide a more reliable Fio₂, higher PEEP, and greater CO₂ washout compared with a flow rate of 40 L/min,^12,15,16 leading to better extubation outcomes in terms of the risk of reintubation and the need for rescue use of NIV.

Study Design and Methods

Trial Design

This investigator-initiated, open-label, parallel-group, randomized controlled trial¹⁷ was conducted in 5 medical ICUs at the National Taiwan University Hospital in Taiwan from September 2021 to April 2023. The goal was to compare 2 flow rates of HFNC (60 L/min vs 40 L/min) used for postextubation oxygen therapy and their effects on subsequent extubation outcomes. The study flow is summarized in e-Figure 1. The trial protocol (202103054RINB) was approved by the research ethics committee of the National Taiwan University Hospital. Written informed consent was obtained from each patient or their next of kin prior to enrollment in the trial. The study was conducted in accordance with the Declaration of Helsinki and local regulations.

Patients

Adult patients were eligible for enrollment in the trial if they were originally intubated for acute hypoxemic respiratory failure with a ratio of Pao₂ to Fio₂ (P/F) < 300 mm Hg, had been receiving mechanical ventilation for > 48 hours, and were ready for planned extubation after passing an SBT. Exclusion criteria included do-not-reintubate orders, tracheostomies, terminal cancer, pregnant women, infeasibility of using HFNC, and plans to use preventive NIV following extubation by the primary care team.

Patients were weaned according to the recommendations of the clinical practice guidelines regarding liberation from mechanical ventilation.^2,6 The weaning protocol included daily screening for weaning readiness and a once-daily 1-hour SBT if the patient met the criteria for weaning readiness. The criteria for SBT success were based on previous guidelines.²

Randomization

Eligible patients were enrolled prior to the planned extubation. Randomization was performed via a centralized web-based management system (REDCap system, Vanderbilt University). Using a random-assignment sequence based on permuted blocks of 4 participants, patients were randomly assigned in a 1:1 ratio to receive HFNC with a flow rate of 40 L/min or 60 L/min following extubation. The investigators were unaware of the block size.

Intervention: Setting Flow Rates and Fio₂ of HFNC

HFNC oxygen therapy was delivered by the Optiflow nasal cannula with Airvo 2 system (Fisher & Paykel Healthcare). The intervention and study flow are summarized in e-Figure 1. The nasal cannula of HFNC was properly positioned prior to removing the endotracheal tube from the patients. The flow rates of HFNC were directly set at 40 L/min and 60 L/min without a flow titration process in the 2 trial groups. The temperature was set at 34 °C for better patient tolerance,¹⁸ and Fio₂ was titrated to keep arterial oxygen saturation > 92%. The flow rate setting was kept unchanged for 24 ± 6 hours unless clinical deterioration or intolerance was observed. The decision to adjust HFNC settings in response to clinical deterioration was at the discretion of the primary care team. However, any ahead-of-schedule change in flow rates and Fio₂ needed to be reported with reasons.

Following 24 hours of intervention, the flow rate could be tapered by 10 L/min every 4 to 8 hours if the patient was clinically stable. If the patient was stable on a flow of 30 L/min, a switch from HFNC to conventional low-flow oxygen therapy was encouraged. All patients were followed up until day 28 following extubation or discharge from the hospital. If patients were discharged prior to day 28, the vital status at discharge was treated as the vital status on day 28.

Outcomes

The primary outcome was a composite of reintubation or rescue use of NIV within 48 hours following extubation.¹⁹ Death was not included in the composite outcome and was not considered a competing risk for the primary outcome because any event leading to death would trigger reintubation prior to death, as patients with do-not-resuscitate orders were excluded from this trial. The choice of 48 hours as the timeframe for the primary outcome measure was based on previous studies suggesting that a 48-hour observation period is sufficient to assess extubation outcomes^2,19 as well as practical considerations at the study institute.

Secondary outcomes included the following: the proportion of intolerance to HFNC settings (up-titration or down-titration of flow rates earlier than planned); the average respiratory rates and heart rates over 24 hours (calculated by the area under the curve of respiratory rates and heart rates divided by time); the change in P/F ratios and Paco₂ levels within 24 hours after extubation; the proportion of respiratory acidosis (elevated Paco₂ with a pH < 7.35) in 24 hours; and death in the ICU and in the hospital.

To explore possible heterogeneity in the treatment effect, prespecified subgroup analyses of the primary composite outcome were performed. Subgroup analyses included sex (male or female), age (≥ 65 years or < 65 years), heart failure or no heart failure, BMI (≥ 25 kg/m² or < 25 kg/m²), duration of mechanical ventilation prior to extubation (> 7 days or ≤ 7 days), and P/F ratios (≥ 300 mm Hg or < 300 mm Hg). Logistic regression models with interaction terms were used to test for significant differences between subgroups.

Statistical Analysis

We determined that enrollment of 180 patients would provide the trial with 80% power to detect a 20% between-group difference in the incidence of the primary outcome, at a 2-sided alpha level of 0.05. We assumed that the incidence of composite outcome events, specifically NIV used or reintubation, would be 20% in the group with a flow rate of 60 L/min.^20,21 All analyses were performed on an intention-to-treat basis.

For the primary outcome, the proportion of rescue use of NIV and reintubation within 48 hours was compared between the 2 trial groups by using the χ² test. A post hoc analysis using a multivariable logistic regression model was conducted to account for imbalances in baseline characteristics. For the secondary outcomes reported as proportions (the proportion of changing HFNC settings in 24 hours, ICU mortality, and in-hospital mortality), the χ² test was used for between-group comparisons. For continuous secondary outcomes, including the average heart rates and respiratory rates over 24 hours, and the change in P/F ratios and Paco₂ with 24 hours following extubation, skewness and kurtosis tests were used to assess data normality, and Levene tests were used to evaluate homogeneity of variance. Depending on the data, either two-sample t tests or Wilcoxon rank-sum tests were used for between-group comparisons. In addition, linear mixed-effects models were applied in a post hoc analysis to evaluate the longitudinal data. We did not adjust for multiple comparisons in any secondary outcome or subgroup analyses, as the secondary outcomes were based on preplanned hypotheses and were intended for exploratory purposes.

STATA version 18 (StataCorp) was used for all analyses.

Results

Study Population

During the study period, 674 patients eligible for planned extubation were screened (Fig 1). Of these, 180 were randomized to treatment, and 169 (86 in the 40 L/min group and 83 in the 60 L/min group) were included in the analysis. The Consolidated Standards of Reporting Trials flow diagram for the trial is presented in e-Figure 2. The baseline characteristics of the patients were similar between the 2 groups, except for the numerically lower percentage of male participants and those with pneumonia and a higher percentage of heart failure and renal failure in the 60 L/min group (Table 1).

Figure 1.

Enrollment, randomization, and follow-up of the study population.

Table 1.

Baseline Patient Characteristics

Characteristic	Flow Rate
	40 L/min (n = 86)	60 L/min (n = 83)
Age, y	70 ± 14	71 ± 14
Male sex	55 (64)	47 (57)
BMI, kg/m2	24 ± 5	24 ± 4
APACHE II score	23 ± 7	23 ± 8
Length of MV prior to extubation, d	8 (5-12)	8 (5-11)
Causes of acute respiratory failure
Pneumonia	43 (50)	33 (40)
Lung edema	17 (20)	26 (31)
Pleural effusion	6 (7)	3 (4)
Aspiration or airway obstruction	5 (6)	4 (5)
Other	15 (17)	17 (20)
Comorbidities
BMI > 25 kg/m2	31 (36)	31 (37)
Cerebrovascular diseases	11 (13)	11 (13)
Heart failure	19 (22)	25 (30)
COPD	4 (5)	10 (12)
Renal failure requiring dialysis	14 (16)	21 (25)
Cancer	35 (41)	39 (47)
Arterial blood gas prior to extubation
Pao2/Fio2 ratio, mm Hg	330 ± 77	311 ± 78
Fio2	0.3 ± 0.06	0.31 ± 0.05
Paco2, mm Hg	37 ± 5	38 ± 7
pH	7.46 ± 0.05	7.45 ± 0.05
Arterial blood gas at the initiation of MV
Pao2/Fio2 ratio, mm Hg	233 ± 109	230 ± 104
Fio2	0.54 ± 0.18	0.53 ± 0.15
Paco2, mm Hg	37 ± 11	36 ± 9
pH	7.41 ± 0.09	7.42 ± 0.09

Data are presented as mean ± SD, No. (%), or median (interquartile range). APACHE II = Acute Physiology and Chronic Health Evaluation II; MV = mechanical ventilation.

Flow Rate Intervention

Following extubation, the difference in flow rates between the groups was apparent from the beginning and persisted throughout the remainder of the 24-hour intervention period (Fig 2). The set Fio₂ values were comparable in both groups.

Figure 2.

Flow rate and Fio₂ settings of high-flow nasal cannula during the intervention period.

Primary Outcome

There was no significant difference in the primary outcome between the 2 groups. The composite outcome of reintubation or use of NIV within the first 48 hours following extubation occurred in 19 patients (22.1%) in the 40 L/min group and in 14 patients (16.9%) in the 60 L/min group (Table 2), with a risk ratio of 1.31 (95% CI, 0.7 to 2.44; P = .39) and a risk difference of 5.2% (95% CI, –6.7% to 17.1%). A post hoc analysis using logistic regression produced results consistent with those of the χ² test (OR, 1.40; P = .39). Multivariable logistic regression, adjusted for sex, causes of acute respiratory failure, and comorbidities, revealed no significant association between the intervention assignment and the primary outcome (OR, 1.56; P = .29).

Table 2.

Results of Primary and Secondary Outcomes

Outcome	Flow Rate		Risk Ratio (95% CI)	P Value
	40 L/min (n = 86)	60 L/min (n = 83)
Primary outcomes
Composite of reintubation or use of NIV in 48 h	19 (22.1)	14 (16.9)	1.31 (0.7 to 2.44)	.39
Reintubation in 48 h	12 (14)	11 (13.3)	1.05 (0.49 to 2.25)	.89
Use of NIV in 48 h	9 (10.5)	7 (8.4)	1.24 (0.48 to 3.18)	.65
Secondary outcomes
Ahead-of-schedule change in flow rates in 24 h	22 (25.6)	5 (6.0)	4.24 (1.69 to 10.69)	< .001
Up-titration in flow rates	15 (17.4)	0 (0)	NA	< .001
Down-titration in flow rates	7 (8.1)	5 (6.0)	1.35 (0.45 to 4.09)	.59
Up-titration in Fio2 in 24 h	15 (17.4)	5 (6.0)	2.9 (1.1 to 7.61)	.02
Use of NIV in 24 h	4 (4.7)	4 (4.8)	0.97 (0.25 to 3.73)	.96
Escalation of respiratory support in 24 ha	24 (27.9)	8 (9.6)	2.9 (1.38 to 6.08)	.002
Average heart rate in 24 h, beats/min, mean (95% CI)	88.4 (84.6 to 92.2)	87.6 (84.1 to 91.1)	Difference 0.8 (–4.4 to 5.9)	.65
Average respiratory rate in 24 h, breaths/min, mean (95% CI)	19.3 (18.4 to 20.2)	18.9 (18.1 to 19.8)	Difference 0.4 (–0.8 to 1.6)	.53
Change in Pao2/Fio2 ratios after 24 h, mm Hg (95% CI)	–41 (–57 to –25)	–37 (–56 to –18)	Difference –4 (–29 to 21)	.75
Change in Paco2 after 24 h, mm Hg (95% CI)	1 (0 to 2.1)	–0.4 (–1.5 to 0.7)	Difference 1.4 (–0.1 to 2.9)	.07
Respiratory acidosis	0 (0)	0 (0)	NA	NA
ICU mortality	7 (8.1)	3 (3.6)	2.25 (0.6 to 8.42)	.21
In-hospital mortality	7 (8.1)	10 (12.1)	0.68 (0.27 to 1.69)	.4

Data are presented as No. (%) unless otherwise indicated. AUC = area under the curve; IQR = interquartile range; NA = not applicable; NIV = noninvasive ventilation.

^aEscalation of respiratory support indicated using NIV or up-titration in either flow rates or Fio₂ of high-flow nasal cannula.

There was no significant difference in the proportion of reintubation (12 [14%] vs 11 [13.3%]; P = .89) or the use of NIV (9 [10.5%] vs 7 [8.4%]; P = .65) between the 40 L/min and 60 L/min groups (Table 2).

Secondary Outcomes

The proportion of ahead-of-schedule change in flow rates during the 24-hour intervention period, including both up-titration and down-titration, was significantly higher in the 40 L/min group than in the 60 L/min group (22 [25.6%] vs 5 [6%]; P < .001) (Table 2). The proportion of flow up-titration was significantly higher in the 40 L/min group (15 [17.4%] vs 0 [0%]; P < .001), but there was no significant difference in the proportion of flow down-titration (7 [8.1%] vs 5 [6%]; P = .59). Twelve patients underwent down-titration earlier than planned; in 11 patients (7 in the 40 L/min group and 4 in the 60 L/min group), this was attributed to attempts for early transfer out of the ICU due to ICU bed allocation issue, and 1 patient in the 60 L/min group experienced intolerance to the high flow due to discomfort. The proportion of up-titration in Fio₂ during the first 24 hours was also higher in the 40 L/min group than in the 60 L/min group (15 [17.4%] vs 5 [6%]; P = .02). Overall, the 40 L/min group was associated with a higher risk of escalation of respiratory support, defined as the use of NIV or up-titration of flow rates or Fio₂ with HFNC, than the 60 L/min group (24 [27.9%] vs 8 [9.6%]; P = .002). Other secondary outcomes were similar between the 2 groups (e-Fig 3, Table 2).

In preplanned subgroup analyses (Fig 3), no significant interactions were found across the subgroups. However, a trend in favor of the 60 L/min flow was observed in the subgroup with worse oxygenation prior to extubation (P/F ratio < 300 mm Hg).

Figure 3.

Subgroup analyses of the primary composite outcome.

Discussion

In this randomized controlled trial of 169 patients undergoing planned extubation, setting the HFNC flow rate at 60 L/min did not improve the primary outcome of reintubation or rescue NIV compared with 40 L/min. However, the lower flow group experienced a higher incidence of escalation of respiratory support. These findings suggest that a flow rate of 40 L/min with as-needed up-titration is a reasonable strategy compared with routinely setting the flow at 60 L/min for postextubation care. Subgroup analysis, however, indicates a trend that a higher flow setting may benefit certain patients, such as those with worse pre-extubation oxygenation.

Previous studies on HFNC flow rates typically focused on short-term physiologic changes in highly selected patients or simulation models.^11,12,16,22 Although extensive research has been done on the physiologic effects of increasing HFNC flow rates, randomized trials evaluating their impact on extubation outcomes remain scarce. The 3-arm trial of Theologou et al²³ found similar extubation outcomes between HFNC at 60 L/min and 40 L/min, with modest improvements in oxygen saturation and respiratory rates with 60 L/min. For clinically relevant outcomes, our trial, along with that of Theologou et al, suggests that a universal high-flow setting of 60 L/min for patients after extubation does not outperform a lower flow setting of 40 L/min. In addition, starting with a flow rate of 40 L/min may offer advantages in reducing oxygen consumption and improving patient comfort compared with a universally higher flow approach. For example, to provide oxygen therapy with an Fio₂ of 0.4, setting the flow at 60 L/min would consume about 5 L/min more oxygen than setting the flow at 40 L/min.

We initially hypothesized that higher HFNC flow rates would improve extubation outcomes by providing a more reliable Fio₂, reducing alveolar derecruitment, and enhancing carbon dioxide washout.^12,15,16 However, our randomized controlled trial did not confirm this hypothesis. Several factors may explain this discrepancy. First, the lower-than-expected incidence of the primary outcome may have limited the statistical power of the study. A physiologic study evaluating the effects of increasing HFNC flow rates showed that some beneficial effects of HFNC were already obtained at a flow rate of 30 L/min.¹² We speculated that the true effect size in extubation outcomes was small between flow rates of 60 L/min and 40 L/min, and a large-scale trial is needed to prove the small difference. Second, participants in this trial might not be able to maintain continuous mouth closure while breathing on HFNC because the intervention period lasted for 24 hours. Open-mouth breathing may reduce PEEP and other beneficial effects of increasing flow rates.^22,24 which diminish the real-world effectiveness of increasing HFNC flow rates. Third, the study population may have included a relatively low-risk group of patients, as evidenced by the overall reintubation rate of 13.6%. Subgroup analysis indicates a trend that higher flow rates may be beneficial for patients with worse pre-extubation oxygenation. Further research is needed to confirm this finding and to identify the optimal HFNC flow rate for different patient subgroups.

The current study has several limitations. First, this trial was a single-center study with an open-label nature. The impact of the practice pattern in the study hospital on the study results is unknown. Second, the statistical power may be inadequate in this trial for the assessment of the primary outcome and subgroup analyses because the outcome event rate was lower than expected, and the true effect size of the trial intervention might be smaller than estimated. Third, there was some imbalance in baseline characteristics between the 2 study groups. The higher cardiopulmonary comorbidity burden in the 60 L/min group might introduce a bias toward the null. However, the analysis of multivariable models by adjusting for sex and comorbidity did not alter the study findings.

Interpretation

In postextubation care using HFNC, setting flow rates at 60 L/min and 40 L/min did not result in a significant difference in the composite outcome of reintubation and rescue use of NIV. However, the 40 L/min group was associated with a greater proportion of respiratory support escalation within the first 24 hours, and this trial may not have been sufficiently powered to exclude a small between-group difference. The study findings suggest that a flow setting of 40 L/min that included escalation of HFNC settings as rescue therapy is as good as a universal flow setting of 60 L/min for patients after extubation.

Funding/Support

This work was supported, in part, by Taiwan Health Foundation (to S.-Y. R.)

Financial/Nonfinancial Disclosures

None declared.