High-flow nasal cannula versus noninvasive ventilation in patients with hypoxemic respiratory failure: a prospective cohort study

Abstract

High-flow nasal cannula (HFNC) and non-invasive ventilation(NIV) are commonly used for hypoxemic respiratory failure, but their comparative efficacy remains unclear. This prospective cohort study enrolled 259 non-hypercapnic hypoxemic respiratory failure patients (PaO₂ <60 mmHg, PaCO₂ <50 mmHg, pH ≥ 7.30) in a Chinese ICU. Patients were allocated to HFNC (n = 128) or NIV (n = 131) based on physician judgment. Primary outcome was 28-day intubation rate; secondary outcomes included 28-day mortality, treatment withdrawal, and ICU/hospital stay duration. Baseline characteristics were similar except for higher respiratory/heart rates and disease severity in the NIV group. Before adjustment, the HFNC group had a lower intubation rate (P < 0.05) and fewer events of death or treatment withdrawal(P<0.01). After propensity score matching, all group differences became non-significant: intubation(aOR 2.61,95%CI 0.58–11.68,P = 0.21), hospital mortality (aOR 0.91,95%CI 0.31–2.62,P = 0.86),death or withdrawing treatment༈aOR 0.54,95%CI 0.26–1.13,P = 0.10),and the composite endpoint (aOR = 1.14,95%CI 0.42–3.08, P = 0.80).The durations of ICU stay (8 vs. 8 days, P = 0.87) and hospital stay (13 vs. 12 days, P = 0.10) showed no significant differences both before and after matching.Patients transitioning from HFNC to NIV had longer ICU stays than NIV-only patients (12 vs. 8 days, P < 0.05). Outcomes did not differ between HFNC-failure patients transitioning to NIV or intubation (P > 0.05). HFNC and NIV show similar efficacy in preventing intubation and reducing mortality. NIV may shorten ICU stay in severe cases, but escalation to NIV or intubation after HFNC failure does not improve outcomes. Treatment should be individualized based on disease severity and patient response. However, given the non-randomized design and potential for residual confounding despite multivariate adjustment, these findings should be interpreted with caution and require validation in randomized controlled trials.

Introduction

Acute hypoxemic respiratory failure (AHRF) represents a significant proportion of intensive care unit (ICU) admissions globally¹. The provision of respiratory support is paramount for these patients, with the objectives of enhancing alveolar ventilation, ensuring sufficient oxygenation, alleviating respiratory muscle fatigue, and ultimately augmenting survival rates. Non-invasive oxygenation strategies have demonstrated efficacy in averting endotracheal intubation in AHRF patients when compared to conventional oxygen therapy².According to the 2023 Global Initiative for Chronic Obstructive Pulmonary Disease (GOLD) report³, non-invasive ventilation (NIV) is preferred over invasive ventilation as the initial modality for managing acute respiratory failure in hospitalized patients with acute exacerbations of chronic obstructive pulmonary disease (AECOPD)⁴. Nonetheless, NIV is associated with a considerable incidence of adverse effects, including claustrophobia, nasofacial skin breakdown, ocular irritation, and gastrointestinal distension.High-flow nasal cannula (HFNC) therapy has been shown to offer superior comfort and oxygenation compared to standard oxygen therapy administered via face mask^5–7. In recent years, HFNC has gained prominence in ICUs for the management of acute respiratory failure, effectively bridging the therapeutic gap between mask oxygen delivery and non-invasive or advanced invasive mechanical ventilation⁸.

The utility of non-invasive oxygenation strategies in patients with moderate to severe hypoxemia remains ambiguous⁹. The FLORALI trial, which investigated subjects with acute hypoxemic respiratory failure, found that neither NIV nor HFNC reduced the intubation rate. However, HFNC was associated with a reduction in 90-day ICU mortality compared to both standard oxygen therapy and NIV¹⁰. Current clinical practice guidelines lack definitive recommendations regarding the role of non-invasive respiratory support strategies in AHRF¹¹.

Consequently, we intend to investigate the clinical efficacy of HFNC and NIV in patients with nonhypercapnic acute hypoxemic respiratory failure. This study aims to furnish additional evidence to assist clinicians in making individualized decisions between HFNC and NIV.

Methods

Study design and participants

This investigation was structured as a single-center, prospective cohort study, executed within the Respiratory Intensive Care Unit (RICU) of The First Affiliated Hospital of Chongqing Medical University, a prominent tertiary academic medical center located in Chongqing, China. The institution boasts a comprehensive capacity of 3,200 inpatient beds, inclusive of 20 specialized RICU beds. The study protocol received ethical endorsement from the Institutional Review Board (IRB) of The First Affiliated Hospital of Chongqing Medical University. All methods were performed in accordance with the relevant guidelines and regulations. Written informed consent was procured from all participating patients, or alternatively, from their nearest kin or an authorized surrogate decision-maker, in accordance with ethical standards and regulatory requirements.

A total of 259 patients presenting with respiratory failure were enrolled in the study between April 2021 and August 2023. Consecutive patients aged 18 years or older were included if they fulfilled all five of the following inclusion criteria: a respiratory rate exceeding 24 breaths per minute, a ratio of the partial pressure of arterial oxygen (PaO₂) to the fraction of inspired oxygen (FiO₂) of 300 mmHg or less while receiving oxygen at a flow rate of 8 L per minute or higher for a minimum duration of 15 min, a partial pressure of arterial carbon dioxide (PaCO₂) not exceeding 45 mmHg, an arterial blood gas pH of 7.3 or greater, and a confirmed absence of COVID-19 diagnosis. Patients who initiated nasal high-flow oxygen therapy (HFNC) or non-invasive ventilation (NIV) following the cessation of invasive mechanical ventilation were excluded from the study. Additionally, patients who had already received HFNC or NIV prior to admission to the intensive care unit (ICU) were also excluded.

Interventions

At our institution, the selection of the respiratory support modality and the initial treatment parameters were determined at the discretion of the attending clinician. Participants were stratified into two cohorts: the “HFNC group,” comprising patients administered nasal high-flow oxygen therapy, and the “NIV group,” consisting of patients receiving non-invasive ventilation.

In the HFNC group, high-flow oxygen therapy was administered via an Optiflow™ nasal cannula interface utilizing an AIRVO™ 2 humidification system (Fisher & Paykel Healthcare, Auckland, New Zealand). The initial flow rate was titrated to 40–60 L per minute (L/min), with the fraction of inspired oxygen (FiO₂) set at 100% and the humidification temperature maintained at 37 °C. The FiO₂ was subsequently adjusted to ensure a pulse oximetry reading (SpO₂) of ≥ 92%. Continuous monitoring of the patient’s clinical status, including vital signs and arterial blood gas analysis, was conducted within the first 1–2 h following treatment initiation.Early termination criteria for HFNC included: intolerance to the high-flow nasal cannula, deterioration of respiratory failure (defined as a respiratory rate > 35 breaths per minute, SpO₂ <88%, or a ROX index < 2.85), paradoxical thoracoabdominal movement, arterial pH < 7.35, or a partial pressure of arterial carbon dioxide (PaCO₂) > 45 mmHg. If any of these criteria were met, the respiratory support was escalated to non-invasive ventilation (NIV) or directly transitioned to invasive mechanical ventilation, as clinically indicated.

In the NIV group, non-invasive ventilation was administered via a full-face mask interface connected to a dedicated ventilator (Philips V60 or BiPAP Vision), operating in a pressure-support ventilation mode. The pressure-support level was meticulously titrated to achieve an expired tidal volume (Vt) within the range of 7 to 10 milliliters per kilogram of predicted body weight. The initial positive end-expiratory pressure (PEEP) was set between 5 and 10 centimeters of water (cm H₂O). Subsequent adjustments to the fraction of inspired oxygen (FiO₂) and/or PEEP were made to maintain a peripheral oxygen saturation (SpO₂) of 92% or higher, ensuring optimal oxygenation and patient comfort.

In addition to the study intervention, all participants received standardized dyspnea management protocols, which included nebulized bronchodilator therapy, endotracheal or oropharyngeal suctioning for secretion clearance, and comprehensive antimicrobial therapy as indicated.

Methods of measurement and outcome measures

Vital signs and arterial blood gas analysis results were meticulously documented at 1 h, 12 h, 24 h, 48 h, and 72 h following the initiation of either HFNC or NIV therapy. The variables evaluated encompassed demographic data such as age and sex, comorbid conditions, and comprehensive laboratory test results. The primary outcome was 28-day intubation rate, defined as the proportion of patients who required invasive mechanical ventilation (endotracheal intubation) within 28 days after initiation of high-flow nasal cannula (HFNC) or non-invasive ventilation (NIV) (the index intervention). To standardize intubation indications across study sites and mitigate the risk of delayed intubation, predefined criteria for endotracheal intubation were employed. These criteria included hemodynamic instability, deterioration in neurologic status, or evidence of persistent or worsening respiratory failure, as indicated by at least two of the following: a respiratory rate exceeding 40 breaths per minute, lack of improvement in signs of elevated respiratory-muscle workload, development of copious tracheal secretions, acidosis with a pH below 7.35, or an SpO₂ of less than 90% for more than 5 min in the absence of technical dysfunction.Secondary outcomes included: 28-day mortality: Defined as death from any cause within 28 days after the index intervention, confirmed by EMRs (death date, death certificate) or telephone follow-up with family members (for patients discharged before 28 days). Treatment withdrawal rate: Defined as the proportion of patients whose HFNC/NIV was discontinued due to “active treatment withdrawal” (decision by patients/family members to forgo further respiratory support, excluding discontinuation due to treatment success or intubation) within 28 days.Duration of ICU stay : Calculated as the number of days from ICU admission to ICU discharge (for survivors) or to death (for non-survivors). Duration of hospital stay: Calculated as the number of days from hospital admission to hospital discharge (for survivors) or to death (for non-survivors).To avoid misclassification, mortality was analyzed as a separate secondary outcome and not considered part of the primary outcome (28-day intubation rate). Patients who were intubated and later died were counted in both outcomes. Those who died without intubation were only included in the mortality analysis, with their data for intubation censored at the time of death.

Statistical analysis

All statistical analyses were performed using SPSS software (version 22.0). Continuous variables with a normal distribution are presented as mean ± standard deviation and were compared using independent-samples t-tests. Continuous variables with a non-normal distribution are presented as median (interquartile range) and were analyzed using the Mann-Whitney U test. Categorical variables are expressed as percentages and compared using the chi-square test.

To evaluate the associations between risk factors and the outcome, both univariate and adjusted analyses were conducted. For the adjusted analysis, a separate model was constructed for each risk factor, controlling for a pre-defined set of confounders (including age, gender, underlying diseases, vital signs, white blood cell count, hemoglobin, albumin, and arterial blood gas parameters). Results are reported as crude odds ratios (ORs) and adjusted odds ratios (aORs) with corresponding 95% confidence intervals (CIs).

Propensity scores were estimated using multivariable logistic regression, adjusting for age, gender, underlying diseases, vital signs, white blood cell count, hemoglobin, albumin, and arterial blood gas parameters. Patients in the HFNC and NIV groups were then matched 1:1 using the nearest-neighbor method without replacement, with a caliper width of 0.1.

A time-to-event analysis was performed to compare the risk of intubation between patients initially treated with high-flow nasal cannula (HFNC) and those initially treated with non-invasive ventilation (NIV). Kaplan-Meier curves were plotted, and group differences were assessed using the log-rank test. The time to intubation was defined as the duration from the initiation of initial respiratory support (HFNC or NIV) until the event of intubation. Patients who were not intubated by the end of the follow-up period were right-censored.

A two-sided p-value < 0.05 was considered statistically significant for all tests.

Results

During the study period spanning from April 2021 to August 2023, a total of 809 subjects diagnosed with hypoxemic respiratory failure were admitted to the Respiratory Intensive Care Unit (RICU). Of these, 503 subjects did not meet the predefined inclusion criteria, including 69 cases due to facial deformity/trauma causing mask leakage, 50 due to upper airway obstruction, 100 due to high aspiration risk, 93 due to impaired consciousness (Glasgow Coma Score < 10), 102 cases diagnosed with COVID-19 and 89 due to intolerance to non-invasive ventilation. and an additional 47 cases of intubated patients or patients requiring emergency intubation were excluded based on the exclusion criteria. Consequently, 259 subjects were enrolled in the study, with 128 assigned to the High-Flow Nasal Cannula (HFNC) group and 131 assigned to the Non-Invasive Ventilation (NIV) group. The flow of subjects through the study is illustrated in Fig. 1.(Fig. 1).

Fig. 1.

Flow chart. HFNC= high-flow nasal cannula, NIV= noninvasive ventilation.

Characteristics of study subjects

Prior to propensity score matching, patients in the NIV group had more severe clinical conditions—such as higher respiratory and heart rates (P < 0.05, Table 1)—than those in the HFNC group. To minimize baseline confounding, a 1:1 propensity score matching was performed using the nearest-neighbor method with a caliper width of 0.1. After matching, the two groups were well-balanced in baseline characteristics. The main causes of respiratory failure in the matched cohort were severe pneumonia and chronic obstructive pulmonary disease.

Table 1.

Baseline data between patients who initiated with HFNC and NIV.

	Overall cohort			Propensity-matched cohort
	HFNC N = 128	NIV N = 131	p	HFNC N = 100	NIV N = 100	p
Age, years	68 ± 15	67 ± 16	0.79	68 ± 13	68 ± 15	0.74
Male	100 (78%)	98 (75%)	0.56	75 (75%)	76 (76%)	> 0.99
Respiratory rate, breaths/min	27 ± 6	31 ± 10	< 0.01*	28 ± 6	29 ± 7	0.22
Heart rate, beats/min	102 ± 22	109 ± 23	< 0.01*	102 ± 22	106 ± 22	0.15
Systolic blood pressure, mmHg	134 ± 22	137 ± 24	0.26	133 ± 22	136 ± 24	0.43
Diastolic blood pressure, mmHg	78 ± 15	80 ± 15	0.27	78 ± 16	79 ± 15	0.59
pH	7.45 ± 0.06	7.44 ± 0.09	0.22	7.46 ± 0.06	7.44 ± 0.09	0.78
PaCO2, mmHg	36 ± 7	35 ± 8	0.55	36 ± 7	36 ± 8	0.83
PaO2/FiO2, mmHg	211 ± 77	191 ± 86	0.06	203 ± 74	195 ± 87	0.43
Underlying disease
Hypertension	54 (42%)	62 (47%)	0.45	43 (43%)	45 (45%)	0.89
Diabetes mellitus	34 (27%)	41 (31%)	0.41	29 (29%)	34 (34%)	0.54
Presence of tumor	23 (18%)	24 (18%)	> 0.99	18 (18%)	15 (15%)	0.70
Presence of immunosuppression	7 (5%)	13 (10%)	0.25	5 (5%)	6 (6%)	> 0.99
Chronic kidney disease	17 (13%)	17 (13%)	> 0.99	15 (15%)	12 (12%)	0.68
Chronic heart disease	36 (28%)	43 (33%)	0.42	31 (31%)	33 (33%)	0.88
Chronic lung disease	38 (30%)	52 (40%)	0.12	34 (34%)	42 (42%)	0.31
WBC,×109/L	11.0 ± 5.5	11.0 ± 5.8	0.97	10.7 ± 5.1	11.1 ± 6.0	0.61
Hemoglobin, g/dL	11.8 ± 2.6	11.7 ± 2.8	0.87	11.6 ± 2.6	11.8 ± 2.8	0.55
Albumin, g/L	33 ± 6	32 ± 6	0.88	33 ± 6	33 ± 6	0.90
Outcomes
Escalation to NIV	21 (16%)	-	-	15 (15%)	-	-
NIV intolerance and use of HFNC	-	10 (13%)	-	-	8 (8%)	-
Intubation	10 (8%)	24 (18%)	0.02*	8 (8%)	5 (5%)	0.57
Intubation or death	16 (13%)	34 (26%)	< 0.01	12 (12%)	13 (13%)	> 0.99
Hospital mortality	12 (9%)	22 (17%)	0.10	9 (9%)	12 (12%)	0.65
Death or withdrawing treatment	27 (21%)	52 (40%)	< 0.01*	20 (20%)	32 (32%)	0.08
Duration of ICU stay, days	8 (5–12)	8 (4–12)	0.87	8 (5–12)	9 (5–12)	0.74
Duration of hospital stay, days	13 (8–19)	12 (7–20)	0.33	13 (8–19)	13 (7–22)	0.89

*p < 0.05.

HFNC = high-flow nasal cannula, NIV = noninvasive ventilation, ICU = intensive care unit.

Primary and secondary outcomes

The unadjusted analysis initially suggested that the intubation rate was significantly lower in the HFNC group at 8%, compared to 18% in the NIV group (P = 0.02). The case of death or withdrawing treatment was significant lower in HFNC group at 21%,compared to 40% in the NIV group(P<0.01). And a significant difference was observed between the two groups in the composite endpoint of intubation or death(P<0.01).While there were no statistically significant differences between the two groups in terms of hospital mortality, duration of ICU stay, or duration of hospital stay (P > 0.05)(Table 1). However, After adjustment for baseline characteristics and propensity score, none of the outcomes showed statistically significant differences between the two groups. Specifically, the adjusted odds ratios were 2.61 (95% CI, 0.58–11.68; P = 0.21) for intubation; 0.91 (95% CI, 0.31–2.62; P = 0.86) for in-hospital death; 0.54 (95% CI, 0.26–1.13; P = 0.10) for death or treatment withdrawal; and 1.14 (95% CI, 0.42–3.08; P = 0.80) for the composite endpoint of intubation or death (Table 2).

Table 2.

Odds ratio of death, withdrawing treatment, or intubation in hospital for patients who initiated as HFNC versus NIV.

	Crude OR	P	Adjusted OR#	p
Overall cohort
Death in hospital	0.51 (0.24–1.09)	0.08	0.68 (0.27–1.76)	0.43
Death or withdrawing treatment	0.41 (0.23–0.71)	< 0.01	0.42 (0.21–0.83)	0.01
Intubation	0.38 (0.17–0.83)	0.02	0.48 (0.19–1.27)	0.14
Intubation or death	0.41 (0.21–0.78)	< 0.01	0.51 (0.22–1.16)	0.11
Propensity-matched cohort
Death in hospital	0.73 (0.29–1.81)	0.49	0.91 (0.31–2.62)	0.86
Death or withdrawing treatment	0.53 (0.28–1.01)	0.06	0.54 (0.26–1.13)	0.10
Intubation	1.65 (0.52–5.24)	0.39	2.61 (0.58–11.68)	0.21
Intubation or death	0.91 (0.39–2.11)	0.83	1.14 (0.42–3.08)	0.80

#It was adjusted by age, gender, underlying disease, vital signs, white blood cell counts, hemoglobin, albumin, and arterial blood gas tests.

HFNC = high-flow nasal cannula, NIV = noninvasive ventilation, OR = odds ratio.

The duration of ICU stay for patients who transitioned to NIV following HFNC failure (mean 12 days) was significantly longer compared to that of patients who exclusively received NIV (mean 8 days), with a statistically significant difference (P < 0.05). However, no significant differences were observed in the intubation rate, mortality, or duration of hospital stay between these patient subgroups (P > 0.05, Table 3).

Table 3.

Clinical features of the patients who escalated from HFNC to NIV and those who directly received NIV.

	Escalation from HFNC to NIV N = 21	NIV N = 131	p
Age, years	70 ± 16	67 ± 16	0.41
Male	15 (71%)	98 (75%)	0.79
Respiratory rate, breaths/min	29 ± 6	31 ± 10	0.45
Heart rate, beats/min	110 ± 21	109 ± 23	0.88
Systolic blood pressure, mmHg	133 ± 26	137 ± 24	0.46
Diastolic blood pressure, mmHg	79 ± 13	80 ± 15	0.70
pH	7.47 ± 0.05	7.44 ± 0.09	0.15
PaCO2, mmHg	34 ± 8	35 ± 8	0.60
PaO2/FiO2, mmHg	190 ± 71	191 ± 86	0.98
Underlying disease
Hypertension	13 (62%)	62 (47%)	0.25
Diabetes mellitus	7 (33%)	41 (31%)	> 0.99
Presence of tumor	4 (19%)	24 (18%)	> 0.99
Presence of immunosuppression	1 (5%)	13 (10%)	0.69
Chronic kidney disease	3 (14%)	17 (13%)	> 0.99
Chronic heart disease	9 (43%)	43 (33%)	0.46
Chronic lung disease	10 (48%)	52 (40%)	0.63
WBC,×109/L	12.0 ± 7.1	11.0 ± 5.8	0.46
Hemoglobin, g/dL	10.9 ± 1.9	11.7 ± 2.8	0.22
Albumin, g/L	33 ± 5	32 ± 6	0.51
Outcomes
Intubation	2 (10%)	24 (18%)	0.53
Hospital mortality	5 (24%)	22 (17%)	0.54
Death or withdrawing treatment	11 (52%)	52 (40%)	0.34
Duration of ICU stay, days	12 (7–15)	8 (4–12)	0.02*
Duration of hospital stay, days	14 (8–24)	12 (7–20)	0.27

*p < 0.05.

HFNC = high-flow nasal cannula, NIV = noninvasive ventilation, ICU = intensive care unit.

Patients who escalated from HFNC to NIV showed significant improvement in oxygenation.

index 24 h after getting on the machine(P < 0.05)(Fig. 2).

Fig. 2.

Comparison of indicators between the observation group after HFNC failure and the control group. H0 = At the initiation of getting on the machine, H1 = 1 h after getting on the machine,. H12 = 12 h after getting on the machine, H24 = 24 h after getting on the machine.

The mortality rate for the cohort that escalated from HFNC to NIV (5 of 21 subjects,24%) did not differ significantly to the patients in NIV group(22 of 131 subjects,17%)(aOR 0.90,95%CI 0.19–4.27,P = 0.89 Table 4).

Table 4.

Odds ratio of death or withdrawing treatment in hospital for patients who escalated from HFNC to NIV versus those who directively received NIV.

	Crude OR	P	Adjusted OR#	p
Death in hospital	1.55 (0.51–4.67)	0.44	0.90 (0.19–4.27)	0.89
Death or withdrawing treatment	1.67 (0.66–4.22)	0.28	1.40 (0.45–4.34)	0.56

#It was adjusted by age, gender, underlying disease, vital signs, white blood cell counts, hemoglobin, albumin, and arterial blood gas tests.

HFNC = high-flow nasal cannula, NIV = noninvasive ventilation, OR = odds ratio.

There were no statistically significant differences in mortality rates, duration of ICU stay, or total hospital length of stay between the two cohorts: one transitioning to non-invasive ventilation following high-flow therapy failure and the other undergoing direct intubation after high-flow therapy failure (P > 0.05, Table 5).

Table 5.

Clinical features of the patients who escalated from HFNC to NIV and those who escalated from HFNC to Intubation.

	Escalation from HFNC to NIV N = 21	Escalation from HFNC to Intubation N = 10	p
Age	76 ± 6	68 ± 4	0.76
Male	13(61.9%)	7(70%)	0.66
Respiratory rate, breaths/min	29 ± 6	31 ± 9	0.502
Heart rate, beats/min	109 ± 20	111 ± 20	0.807
Systolic blood pressure, mmHg	131 ± 25	151 ± 17	0.056
Diastolic blood pressure, mmHg	78 ± 12	81 ± 13	0.539
pH	7.36 ± 0.49	7.00 ± 0.00	0.002**
PaCO2, mmHg	36 ± 10	36 ± 8	0.990
FiO2	41(33–43)	33(33–42)	0.29
Underlying disease
Hypertension	6(28%)	4(40%)	0.26
Diabetes mellitus	5(24%)	0(0%)	0.59
Presence of tumor	3(14%)	0(0%)	0.63
Presence of immunosuppression	0(0%)	0(0%)	> 0.99
Chronic kidney disease	1(5%)	2(20%)	0.69
Chronic heart disease	0(0%)	0(0%)	> 0.99
Chronic lung disease	2(9%)	1(10%)	> 0.99
WBC,×109/L	12 ± 7	12 ± 3	0.96
Hemoglobin, g/dL	76 ± 22	87 ± 3	0.030*
Albumin, g/L	108 ± 19	115 ± 29	0.5
Outcomes
Hospital mortality	5(23.8%)	4(40%)	0.353
Duration of ICU stay, days	13(9–19)	8(5–14)	0.286
Duration of hospital stay, days	16(9–25)	10(6–15)	0.412

*p < 0.05.

The median time from HFNC initiation to escalation to NIV was 3.0 days (IQR: 1.7–5.9), and to direct intubation was 3.8 days (IQR: 0.9–10.0). The median time from HFNC initiation to intubation (including via NIV) was 4.0 days (IQR: 1.1–10.5). For patients who failed NIV, the median time from NIV initiation to subsequent intubation was 2.2 days (IQR: 0.7–5.1).The time to intubation (from either HFNC or NIV initiation) showed no significant association with key clinical outcomes(P = 0.22)(Fig. 3).

Fig. 3.

Comparison of time from HFNC or NIV initiation to intubation.

Discussion

A growing body of evidence supports the efficacy and safety of high-flow nasal cannula (HFNC) in patients with acute hypoxemic respiratory failure (AHRF), particularly in non-hypercapnic cases. Studies have consistently associated HFNC with improved patient comfort, reduced dyspnea severity, and decreased respiratory rate^12–15. A meta-analysis of randomized controlled trials further confirms that HFNC represents a viable initial treatment strategy for AHRF¹⁶.A growing body of research has corroborated the efficacy of HFNC in managing non-hypercapnic respiratory failure¹⁷. However, there remains a paucity of studies directly comparing HFNC to non-invasive ventilation (NIV) as the initial treatment modality for AHRF.

This study evaluated the comparative efficacy of high-flow nasal cannula (HFNC) and non-invasive ventilation (NIV) in a specific patient cohort.In our study, while unadjusted analysis suggested a potential reduction in intubation risk with HFNC this benefit did not persist after adjusting for baseline severity and propensity scores. After controlling for confounders through multivariate adjustment and propensity score matching, we observed no significant differences between HFNC and NIV in either intubation rates or in-hospital mortality.

The initially observed disparity likely stemmed from clinical confounding, particularly the preferential use of NIV in patients with greater disease severity and higher intrinsic intubation risk. Our findings align with the FLORALI trial¹⁰, which also reported comparable outcomes between HFNC and NIV after balancing baseline characteristics. Notably, although unadjusted analysis indicated higher treatment withdrawal rates in the NIV group—a decision often influenced by disease severity, prognosis, and patient/family preferences—this difference did not translate into a statistically significant mortality gap after rigorous adjustment.

These results underscore the critical importance of accounting for baseline imbalances in observational comparisons of respiratory support strategies. The similar risks of intubation and mortality after adjustment suggest that underlying patient factors may exert a greater influence on clinical outcomes than the choice of initial respiratory support modality.

In cases where HFNC fails and treatment is escalated to NIV, patients typically demonstrate significant improvement in oxygenation indices. This enhancement can be attributed to the physiological advantages of NIV, including higher levels of ventilatory support, more efficient oxygen delivery, reduced respiratory muscle fatigue, and improved ventilation-perfusion matching. However, this transition is associated with a prolonged ICU stay, despite no statistically significant differences in intubation rates, mortality, or total hospital length of stay.Patients commencing treatment directly with NIV experienced a median ICU stay of approximately 8 days. In contrast, those failing HFNC and transitioning to NIV had a median ICU stay of around 12 days—a 4-day extension that consequently increases ICU-related costs and amplifies the financial burden on patients. This prolonged stay may be explained by the fact that HFNC failures often present with more severe clinical conditions, heightened dyspnea, and lower oxygenation indices, all contributing to a more protracted recovery.

For patients with respiratory failure and severe dyspnea, HFNC as initial therapy may not always be the most appropriate choice. Although it can alleviate dyspnea and hypoxemia in selected cases, its efficacy may be limited in severe respiratory failure. While HFNC improves oxygenation and respiratory comfort through high-flow oxygen delivery and humidification, it may not significantly enhance oxygen partial pressure in cases with significant ventilatory insufficiency or increased pulmonary shunting, thereby constraining its therapeutic impact. These observations are consistent with another study¹⁸ suggesting that NIV may be superior to HFNC as initial therapy in certain clinical scenarios.

Treatment selection should be individualized based on illness severity. For patients with mild to moderate respiratory failure, HFNC may serve as the preferred initial therapy—a conclusion consistent with previous studies including those on infants with bronchiolitis¹⁹. Conversely, for those presenting with more severe respiratory distress, NIV should be prioritized to effectively reduce ICU stay.

Our analysis of post-HFNC failure pathways yields two key clinical insights. First, escalation to either NIV or invasive mechanical ventilation (IMV) showed no significant impact on patient prognosis, with no statistically significant differences in mortality, ICU stay, or hospital duration. This finding addresses an important evidence gap and offers new perspectives for managing HFNC failure.Second, and more instructive, is the relationship between treatment timing and outcomes. Our data revealed that the time interval from treatment initiation to escalation—whether from HFNC to direct intubation (median 3.8 days), or subsequent intubation after NIV failure (median 2.2 days)—exhibited no independent correlation with mortality or hospital stay. This challenges the conventional wisdom that “delayed intubation invariably worsens outcomes.”

We propose the following mechanistic explanation: the dynamic nature of clinical decision-making decouples time from prognosis. Clinicians likely implemented aggressive escalation for patients showing early rapid deterioration—intervening before organ damage occurred—while justifiably delaying invasive procedures in those with indolent disease progression to avoid overtreatment. This individualized approach transforms intervention timing into a response to evolving clinical conditions rather than an independent prognostic factor. Moreover, HFNC failure itself defines a high-risk population in which the prognostic significance of the “treatment failure” event may outweigh the time interval to subsequent intervention.

This study possesses several strengths, including being the first systematic comparison between NIV and IMV following HFNC failure. Its conclusions may be generalizable to other ICU patients with non-hypercapnic AHRF. However, several limitations warrant consideration. The non-randomized treatment allocation based on physician judgment constitutes a fundamental constraint. Although we employed propensity score matching and multivariate adjustment, unmeasured confounders—such as subtle differences in disease severity or implicit factors in clinical decision-making—may still influence the results.In analyzing the primary outcome of intubation rate, we explicitly defined death as a competing event but employed traditional survival analysis methods. Although supplementary composite endpoint analysis yielded consistent conclusions, and the number of deaths without prior intubation was very low, future larger-scale studies utilizing competing risk models would provide more precise risk estimates.For patients experiencing HFNC failure, focusing solely on the choice between NIV and IMV appears to offer limited clinical value. Instead, developing predictive models for early and accurate identification of failure risk holds greater promise. Future research should validate these findings through multicenter, large-sample randomized controlled trials with standardized protocols to provide more robust evidence for optimizing respiratory support strategies.

Conclusion

In summary, while HFNC and NIV show comparable efficacy for mild to moderate hypoxemic respiratory failure, NIV reduces ICU stay in severe cases. Crucially, once HFNC fails, neither the choice of escalation therapy (NIV or IMV) nor the time to escalation affects outcomes. This suggests the priority should be early identification of HFNC failure and timely intubation rather than intermediate NIV trials. These findings should be validated in randomized controlled trials to establish optimal escalation pathways.

Author contributions

All the authors participated in the article preparation. The authors read and approved the final manuscript.1.Literature search: Zhao Qianru, Jiang Heyue.2.Data collection: Zhao Qianru, Liu Qiao.3.Study design: Hong Yueling，PanLongfan.4.Data analysis: Duan Jun.5.Manuscript preparation: Hong Yueling, Zhao Qianru, Jiang Heyue.6.Manuscript review: Hong Yueling, Pan Longfang.

Funding

This study was supported by the Chongqing Joint Medical Research Program of Science & Health(Project No. 2020FYYX138). The funder had no role in the study design, data collection, statistical analysis, or manuscript writing.

Data availability

All the data analysed during this study are included in this published article.

Declarations

Competing interests

The authors declare no competing interests.

Ethics approval and consent to participate

The study was approved by the Ethics Committee of the First Affiliated Hospital of Chongqing Medical University. Written informed consent was obtained from all participants prior to their inclusion in the study.Ethics Number:2020-1.

Consent for publication

The authors confirm that no identifiable personal data are presented in this manuscript. Therefore, consent for publication was not required.