Prophylactic high-flow nasal oxygen vs. conventional oxygen for prevention of postoperative respiratory complications: a randomized trial

Abstract

Background

Postoperative pulmonary complications (PPCs) are a major cause of morbidity after lung resection, particularly in older adults. Following extubation, noninvasive respiratory support strategies are increasingly used to prevent respiratory deterioration, but the role of prophylactic high-flow nasal cannula (HFNC) in this setting remains uncertain.

Methods

In this single-center, open-label randomized controlled trial, patients undergoing video-assisted thoracoscopic surgery (VATS) lobectomy or segmentectomy for stage I–IIIa non-small cell lung cancer (NSCLC) were randomized to HFNC or conventional oxygen therapy (COT). The primary outcome was PPC severity during hospitalization, analyzed using ordinal logistic regression; the incidence of clinically relevant PPCs (grade ≥ 2) was also compared between groups. Secondary outcomes included longitudinal respiratory symptoms (mMRC, LCQ-MC, sCSS) and symptom-related functional interference (MDASI-LC), analyzed using mixed-effects models.

Results

A total of 210 patients were included in the modified intention-to-treat analysis (HFNC, n = 104; COT, n = 106). HFNC reduced PPC severity (ordinal OR, 0.50; 95% CI, 0.28–0.91; P = 0.022) and moderate-to-severe PPCs (23.1% vs. 39.6%; absolute risk difference, -15.6%; 95% CI, -28.0% to -3.2%; P = 0.010). HFNC was associated with a lower burden of clinically relevant PPCs. Secondary analyses also suggested improvements in early postoperative respiratory symptoms and symptom-related functional interference. Secondary analyses also suggested improvements in early postoperative respiratory symptoms and symptom-related functional interference. No differences were observed in physiological outcomes or length of stay.

Conclusion

In older adults undergoing lung resection, prophylactic HFNC reduced in-hospital PPC burden and was associated with more favorable early postoperative respiratory recovery. These findings suggest that prophylactic HFNC may provide benefit as a potential preventive respiratory support option in selected older or higher-risk postoperative patients, although routine prophylactic use in all patients undergoing lung resection is not supported by the present findings.

Trial registration

This study was approved by the Chinese Clinical Trial Registry (ChiCTR2200063161) on 01 Sep 2022 and the first patient was enrolled after pre-registration.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s13054-026-06205-4.

Background

Lung cancer remains the leading cause of cancer-related mortality worldwide, and surgical resection offers the best chance of cure for resectable non–small cell lung cancer (NSCLC) [1, 2]. Despite advances in video-assisted thoracoscopic surgery (VATS), postoperative pulmonary complications (PPCs) remain common (19%–59%), prolonging hospitalization and impairing quality of life (QOL) [3–6]. Older adults are particularly susceptible due to age-related reductions in physiological reserve, including sarcopenia and declining pulmonary function [7–9].

High-flow nasal cannula (HFNC) delivers heated humidified oxygen at high flow and is increasingly used perioperatively to improve oxygenation and reduce escalation of respiratory support [10]. In critically ill and high-risk postoperative patients, early noninvasive respiratory support strategies are commonly applied to prevent respiratory failure and avoid escalation to invasive ventilation. However, the role of prophylactic HFNC as a preventive respiratory support strategy in surgical populations, particularly among older adults at high risk of pulmonary complications, remains uncertain. Evidence in lung cancer surgery is limited, as prior randomized trials have included mixed surgical populations or focused primarily on short-term physiological outcomes [11–13].

We therefore conducted a randomized controlled trial to evaluate whether early prophylactic HFNC, compared with conventional oxygen therapy (COT), reduces PPCs in older adults undergoing VATS lobectomy or segmentectomy for NSCLC. Key secondary outcomes included postoperative respiratory symptoms and symptom-related functional interference.

Methods

Study design and participants

This single-center, open-label, randomized controlled trial was conducted in the Department of Thoracic Surgery of Fujian Provincial Hospital between September 2022 and September 2024. The protocol was approved by the local ethics committee and registered with the Chinese Clinical Trial Registry before enrollment. The study followed the Declaration of Helsinki, with written informed consent obtained. The full trial protocol is provided in the Supplementary Material, and patients and the public were not involved in the design, conduct, reporting, or dissemination of this research.

Eligible patients were aged 60–80 years, underwent VATS lobectomy or segmentectomy, and had pathologically confirmed stage I–IIIa NSCLC. Exclusion criteria comprised bilateral pulmonary surgery or pneumonectomy; conversion to thoracotomy or intraoperative blood loss > 1000 mL; planned postoperative intensive care unit (ICU) admission; neoadjuvant chemotherapy, immunotherapy, or targeted therapy; and refusal to participate or dropout.

Randomization and sample size calculation

Randomization used a computer-generated sequence (SPSS 22.0; IBM Corp; fixed seed 2022901) prepared by an independent researcher not involved in enrollment or outcome assessment. Allocation concealment was ensured via a pre-generated list. Blinding was infeasible due to the intervention’s nature.

Sample size was estimated on the basis of the expected incidence of PPCs from preliminary retrospective data, which showed a PPC incidence of 45.2% in the COT group and 23.8% in the HFNC group. Assuming a two-sided α of 0.05 and 90% power, approximately 101 patients per group were required. To account for a potential 25% loss to follow-up during the planned 6-month postoperative follow-up period, the recruitment target was increased to 125 patients per group. Actual attrition during the study was substantially lower than anticipated.

The preliminary retrospective dataset available during study planning recorded PPCs as binary incidence outcomes rather than according to the modified Kroenke severity grading system used in the present trial. These estimates were derived from a retrospective cohort of 343 older adults undergoing VATS lung resection for NSCLC at our institution (Supplementary Material). Therefore, reliable assumptions for ordinal-outcome sample size estimation were not available at the design stage.

Interventions

Both interventions were applied as post-extubation respiratory support strategies aimed at preventing respiratory deterioration. Patients received allocated oxygen therapy. In the COT group, oxygen was via nasal cannula starting at 3 L/min, titrated to peripheral oxygen saturation (SpO2) targets (≥ 95% or ≥ 93% in chronic obstructive pulmonary disease [COPD] or obesity).

In the HFNC group, therapy was delivered using the Optiflow™ system (Fisher & Paykel Healthcare). Gas temperature was initially set at 31 °C and adjusted for comfort. Initial flow rate was 40 L/min (range 20–50), and fraction of inspired oxygen (FiO2) 40%. Flow and FiO2 were titrated to maintain SpO₂ ≥95% (≥ 93% in COPD or obesity). These target ranges were prespecified in the study protocol and reflected routine clinical practice to avoid excessive oxygen administration in patients at risk of hypercapnic respiratory failure.

HFNC settings were adjusted according to oxygenation targets and patient comfort during routine postoperative care. HFNC exposure duration was calculated according to recorded treatment initiation and discontinuation orders. Delivered flow rates and FiO2 values were extracted from routine clinical records. Temporary bedside interruptions during mobilization or routine nursing care were not prospectively recorded.

All patients underwent two-port VATS lobectomy or segmentectomy with systematic lymph node sampling or dissection according to institutional standards [14, 15]. Surgical and postoperative management followed institutional protocols. Apart from the assigned oxygen therapy strategy, perioperative management was otherwise standardized between groups, including postoperative analgesia, early mobilization, chest physiotherapy, and routine postoperative nursing care.

Primary outcome measures

The primary outcome was PPCs status during hospitalization, graded on an ordinal scale from 0 to 5 using a modified Kroenke et al. definition (e-Appendix 1) [16]. For primary analysis, grades were grouped as 0–1, 2, and ≥ 3. Grade 0 indicated no symptoms; grade 4, reintubation or invasive ventilation ≥ 48 h; grade 5, death before discharge. PPC severity was assessed daily until discharge using predefined criteria, and the highest PPC grade recorded during hospitalization was used for analysis. No separate prespecified intervention-related harms were defined because both groups received routine non-invasive oxygen therapy. Clinically relevant unintended events were monitored during routine clinical care.

Secondary outcome measures

Secondary outcomes included physiological measures, respiratory symptoms, and symptom-related functional interference. Physiological outcomes comprised arterial blood gas on postoperative day (POD) 2 and the 6-minute walk test (6MWT), which was performed before discharge after chest tube removal and stopping intravenous therapy. POD 2 was selected prospectively to provide a more comparable duration of exposure to the assigned oxygen therapy before arterial blood gas assessment, while avoiding potential selection bias from later assessment after oxygen discontinuation or early discharge.

Dyspnea was assessed with the modified Medical Research Council dyspnea scale (mMRC-DS), grading breathlessness-related limitation from 0 (strenuous exercise only) to 4 (too breathless to leave the house or dress/undress) (e-Appendix 2) [17]. Cough-related QOL was evaluated using the Leicester Cough Questionnaire Mandarin Chinese version (LCQ-MC), a 19-item instrument with physical (8), psychological (7), and social (4) domains. Each item is scored 1–7, yielding total scores from 3 to 21 (higher=better) (e-Appendix 3) [18]. Cough severity was measured using the simplified cough symptom score (sCSS), assessing daytime and nighttime cough separately (each 0–3; total 0–6, with higher scores indicating greater cough severity) (e-Appendix 4) [19]. Prespecified longitudinal respiratory symptom outcomes included the LCQ-MC total score and domain scores, sCSS daytime, nighttime, and total scores, and mMRC-DS.

Functional interference was assessed using the MD Anderson Symptom Inventory for Lung Cancer (MDASI-LC), six items measuring symptom interference with daily life over the past 24 h on an 11-point scale (0 = did not interfere, 10 = interfered completely) (e-Appendix 5) [20].

Patients completed the mMRC-DS, LCQ-MC, sCSS and MDASI-LC preoperatively, at discharge, and at 7 days, 1 month, 2 months, 3 months, and 6 months after discharge. Follow-up was planned for 6 months after discharge. However, if patients reported complete resolution of respiratory symptoms and no further symptom-related functional interference, follow-up assessments could be discontinued early according to a prespecified study stopping rule defined in the trial protocol. All available data obtained before discontinuation were included in the longitudinal analyses.

Statistical analysis

All values are presented as mean (standard deviation, SD), median (interquartile range, IQR), or number (percentage), as appropriate. Baseline characteristics were summarized descriptively. Group comparisons for study outcomes were performed using the independent-samples t test or Mann-Whitney U test for continuous variables and the χ² test or Fisher exact test for categorical variables.

The primary outcome was analyzed in two complementary ways. First, PPC grade (0–1, 2, and ≥ 3) was analyzed using ordinal logistic regression to estimate the odds of greater complication severity in the HFNC group relative to the COT group. Second, because the grading system also identified the occurrence of clinically relevant PPCs, the incidence of moderate-to-severe PPCs (grade ≥ 2) was compared between groups using the χ² test, and the absolute risk difference with 95% confidence interval (CI) was reported. This binary outcome also corresponded conceptually to the original event-based framework used in the sample size estimation. Sensitivity analyses for the primary outcome were additionally performed using multivariable ordinal logistic regression adjusted for COPD, smoking history, and TNM stage.

Longitudinal changes in continuous symptom scores, including MDASI-LC interference and LCQ-MC, were analyzed using linear mixed-effects regression models. Time was modeled as a categorical variable with seven levels: preoperative baseline, discharge, and 7 days, 1, 2, 3, and 6 months after discharge. Fixed effects included treatment group, time, and their interaction, with a participant-level random intercept to account for repeated measurements. Pairwise comparisons among the six postoperative time points were performed with Bonferroni adjustment for multiple testing, with the preoperative measurement as reference.

Ordinal outcomes, including mMRC-DS and sCSS, were analyzed using mixed-effects ordinal logistic regression models, with treatment group, time, and their interaction as fixed effects and a participant-level random intercept. Clinically meaningful dyspnea was defined as mMRC-DS ≥ 2, and adjusted predicted probabilities and risk differences were estimated from the fitted models.

All mixed-effects models incorporated all available observations and accounted for within-patient correlation under the missing-at-random assumption. Because follow-up assessments could be discontinued after complete symptom resolution according to the prespecified protocol-defined stopping rule, missingness in longitudinal symptom outcomes may have been related to recovery status. All analyses were performed according to the modified intention-to-treat principle. The modified intention-to-treat population included all randomized patients who underwent the intended VATS lobectomy or segmentectomy, had pathologically confirmed NSCLC, and had at least one postintervention outcome assessment.

Statistical tests were two-sided, and P values < 0.05 were considered statistically significant. Analyses were conducted using Stata/MP version 18.0 (StataCorp). No important changes to the trial methods or outcomes were made after trial commencement, and no interim analyses or stopping guidelines were planned.

Results

Baseline characteristics

A total of 238 patients were assessed for eligibility; 15 were excluded (13 did not meet inclusion criteria, 2 declined), and the remaining 223 were randomized to HFNC (n = 110) or COT (n = 113) (Fig. 1).

Fig. 1.

Study flow diagram. HFNC, high-flow nasal cannula; COT, conventional oxygen therapy

In the HFNC group, 107 received the allocated intervention; three did not (one conversion to thoracotomy, one conversion to wedge resection, one intraoperative frozen section showing benign disease). One was lost to follow-up, and two declined further follow-up after postoperative readmission, leaving 104 for analysis.

In the COT group, 111 received the allocated intervention; two did not (one conversion to thoracotomy, one intraoperative frozen section showing metastatic disease). One was lost to follow-up, one died from stroke, and three declined further follow-up after postoperative readmission, leaving 106 patients for analysis.

Baseline demographic and clinical characteristics are presented in Table 1.

Table 1.

Baseline characteristics of the study participants by group

Characteristics	HFNC (n = 104)	COT (n = 106)
Age, years, mean (SD)	68.03 (4.83)	66.90 (4.54)
Sex, female, n (%)	56 (53.8)	68 (64.2)
BMI, kg/m2, mean (SD)	24.20 (3.31)	23.85 (3.13)
Comorbidities
Hypertension, n (%)	45 (43.3)	48 (45.3)
Diabetes, n (%)	11 (10.6)	17 (16.0)
Coronary disease, n (%)	9 (8.7)	4 (3.8)
COPD, n (%)	20 (19.2)	17 (16.0)
Asthma, n (%)	1 (1.0)	2 (1.9)
History of smoking, n (%)	33 (31.7)	26 (24.5)
Preoperative spirometry
FEV1, % predicted, mean (SD)	101.13 (21.68)	105.03 (18.44)
FVC, % predicted, mean (SD)	104.82 (19.31)	109.21 (17.39)
6MWT, m, mean (SD)	436.80 (62.05)	445.75 (70.28)
PaO2/FiO2 ratio, mmHg, mean (SD)	373.54 (53.20)	385.56 (43.93)
Extent of resection
VATS lobectomy, n (%)	62 (59.6)	73 (68.9)
VATS segmentectomy, n (%)	42 (40.4)	33 (31.1)
Length of anaesthesia, min, median [IQR]	180 [150, 220]	172 [150, 200]
TNM stage
Stage Ia, n (%)	87 (83.7)	78 (73.6)
Stage Ib–IIIa, n (%)	17 (16.3)	28 (26.4)

Data are presented as mean (SD), n (%), or median [IQR], as appropriate

HFNC, high-flow nasal cannula; COT, conventional oxygen therapy; SD, standard deviation; BMI, body mass index; COPD, chronic obstructive pulmonary disease; FEV1, forced expiratory volume in 1 s; FVC, forced vital capacity; 6MWT, 6-minute walk test; PaO2/FiO2, ratio of arterial oxygen partial pressure to fractional inspired oxygen; VATS, video-assisted thoracic surgery; IQR, interquartile range

HFNC treatment characteristics

Detailed HFNC treatment characteristics are summarized in e-Table 1. Median total HFNC duration was 67.0 h (IQR, 46.5–88.0 h), with a mean duration of 73.4 ± 33.0 h. All patients received HFNC on POD 1, whereas 80 (76.9%) and 44 (42.3%) patients remained on HFNC on POD 2 and POD 3, respectively. Delivered flow rates and FiO2 values remained generally stable during the early postoperative period, with mean flow rates and FiO2 values of approximately 40 L/min and 40%, respectively, across POD 1–3.

Primary outcomes

During hospitalization, the severity distribution of PPCs differed between groups (Table 2; Fig. 2A). None/mild PPCs (grade 0–1) occurred in 79 (76.0%) patients in the HFNC group and 64 (60.4%) in the COT group; moderate PPCs (grade 2) in 17 (16.3%) and 31 (29.2%); and severe PPCs (grade ≥ 3) in 8 (7.7%) and 11 (10.4%). HFNC was associated with lower odds of greater PPC severity (ordinal logistic odds ratio [OR] 0.504, 95% CI 0.280 to 0.907; P = 0.022).

Table 2.

Postoperative outcomes of the study participants by group

Primary outcome	HFNC (n = 104)	COT (n = 106)	Effect estimate (95% CI)a	P value
PPCs grade, n (%)
0–1 (none or mild)	79 (76.0)	64 (60.4)	0.504 (0.280 to 0.907)	0.022b
2 (moderate)	17 (16.3)	31 (29.2)
≥ 3 (severe)	8 (7.7)	11 (10.4)
Moderate-to-severe PPCs (≥ 2), n (%)	25 (23.1)	42 (39.6)	-15.6% (95% CI -28.0% to -3.2%)	0.010c
Secondary outcomes
PaO2/FiO2 ratio on POD 2, mmHg, mean (SD)	297.38 (82.55)	289.28 (58.36)	8.10 (-11.32 to 27.52)	0.413d
6MWT at discharge, m, mean (SD)	320.82 (58.82)	334.15 (61.61)	-13.33 (-29.73 to 3.06)	0.110d
Additional outcomes
Length of postoperative hospital stay, days, median [IQR]	5 [4, 6]	5 [4, 6]	—	0.103e
Δ6MWT from baseline to discharge, m, mean (SD)	-115.98 (51.61)	-112.07 (61.84)	3.91 (-11.60 to 19.43)	0.619d

Data are presented as mean (SD), n (%), or median [IQR], as appropriate

HFNC, high-flow nasal cannula; COT, conventional oxygen therapy; CI, confidence interval; PPCs, postoperative pulmonary complications; PaO2/FiO2, ratio of arterial oxygen partial pressure to fractional inspired oxygen; POD, postoperative day; SD, standard deviation; 6MWT, 6-minute walk test; IQR, interquartile range; OR, odds ratio

^a Effect estimates represent OR for PPCs and mean differences (HFNC - COT) for continuous outcomes

^b Ordinal logistic regression

^c χ2 test

^d t test

^e Mann-Whitney U test

Fig. 2.

Postoperative pulmonary complication severity and risk difference by group. (A) Distribution of modified pulmonary complication scores in the COT and HFNC groups. PPCs were graded on a 0–5 scale adapted from Kroenke et al. Grade 0 indicated no symptoms; grade 1, mild findings such as dry cough, abnormal auscultation, or low-grade fever with a normal chest radiograph; grade 2, moderate symptoms including productive cough, hypoxemia (SpO₂ < 90%), or radiologically confirmed atelectasis with fever or abnormal lung signs; grade 3, major complications requiring intervention (e.g., pneumonia, pneumothorax, thoracentesis or reintubation < 48 h); and grade 4, invasive ventilation ≥ 48 h. The most severe PPC per patient was recorded. (B) Absolute risk difference for moderate-to-severe PPCs (grade ≥ 2). The absolute risk difference is expressed as percentage points (HFNC - COT). Point represents risk difference and horizontal bar indicates 95% CI. COT, conventional oxygen therapy; HFNC, high-flow nasal cannula; PPCs, postoperative pulmonary complications; CI, confidence interval

Moderate-to-severe PPCs (grade ≥ 2) occurred in 23.1% (25/104) of HFNC patients and 39.6% (42/106) of COT patients, corresponding to an absolute risk difference of -15.6% (95% CI -28.0 to -3.2; P = 0.010) (Fig. 2B).

Sensitivity analyses adjusted for COPD, smoking history, and TNM stage yielded consistent findings, with HFNC remaining associated with lower PPC severity (adjusted ordinal OR 0.48, 95% CI 0.26–0.88; P = 0.018) (e-Table 2).

In-hospital outcomes

In-hospital outcomes were similar between groups (Table 2). The PaO2/FiO2 ratio on POD 2 was 297.38 ± 82.55 mmHg in the HFNC group and 289.28 ± 58.36 mmHg in the COT group (mean difference 8.10, 95% CI -11.32 to 27.52; P = 0.413). 6MWT distance at discharge (320.82 ± 58.82 vs. 334.15 ± 61.61 m) and the change of 6MWT from baseline (-115.98 ± 51.61 vs. -112.07 ± 61.84 m) also did not differ significantly. Median hospital stay was 5 days (IQR 4–6) in both groups (P = 0.103). No intervention-related serious adverse events were observed. Two patients in the COT group required escalation of care during hospitalization. One patient was admitted to the ICU because of postoperative respiratory failure. Another patient developed a postoperative bronchopleural fistula and required reintubation with invasive mechanical ventilation for more than 48 h; according to the prespecified PPC grading system, this event was classified as grade 4 PPC. These findings were consistent with a lower burden of clinically relevant PPCs and fewer escalation events during hospitalization.

Secondary longitudinal respiratory symptom outcomes

Secondary longitudinal respiratory symptom outcomes showed significant group-by-time interactions during the 6-month follow-up (Table 3). The adjusted probability of clinically meaningful dyspnea (mMRC-DS ≥ 2) differed over time (interaction P < 0.001). At discharge, the probability was lower in the HFNC group (14.0% vs. 26.6%; risk difference − 12.6%, 95% CI -21.2% to -4.1%; adjusted P = 0.024), with no significant differences at later time points (e-Table 3; Fig. 3A).

Table 3.

Time-by-group interaction tests across all study outcomes

Outcome	Model	Interaction P value
mMRC-DS	Mixed-effects ordinal logistic model	< 0.001
LCQ-MC	Linear mixed-effects model
Physical		< 0.001
Psychological		0.417
Social		< 0.001
Total score		0.004
sCSS	Mixed-effects ordinal logistic model
Daytime score		< 0.001
Nighttime score		0.100
Total score		< 0.001
MDASI-LC	Linear mixed-effects model
Activity		< 0.001
Mood		0.053
Work (including housework)		0.787
Relations with other people		< 0.001
Walking		< 0.001
Enjoyment of life		0.025

mMRC-DS, the modified Medical Research Council dyspnea scale; LCQ-MC, the Leicester Cough Questionnaire Mandarin Chinese version; sCSS, the simplified cough symptom score; MDASI-LC, the MD Anderson Symptom Inventory for Lung Cancer

Fig. 3.

Adjusted longitudinal trajectories of respiratory symptoms and functional interference by group. (A) mMRC-DS; (B) LCQ-MC total score; (C) MDASI-LC activity item; (D) MDASI-LC walking item. Adjusted predicted trajectories of study endpoints in the HFNC and COT groups across study time points, including the preoperative assessment, at discharge, and 7 days, 1 month, 2 months, 3 months, and 6 months after discharge. Points represent adjusted predicted values derived from mixed-effects models, and error bars indicate 95% CIs. For the mMRC-DS, points represent the adjusted predicted probability of clinically meaningful dyspnea (mMRC-DS ≥ 2) estimated from a mixed-effects ordinal logistic regression model. LCQ-MC total scores and MDASI-LC item scores represent adjusted predicted values estimated from linear mixed-effects models. P values correspond to the overall group-by-time interaction from the respective mixed-effects models. mMRC-DS, the modified Medical Research Council dyspnea scale; LCQ-MC, the Leicester Cough Questionnaire Mandarin Chinese version; MDASI-LC, the MD Anderson Symptom Inventory for Lung Cancer. HFNC, high-flow nasal cannula; COT, conventional oxygen therapy; CI, confidence interval. * P < 0.05 after Bonferroni adjustment for six postoperative time points (discharge, 7 days, 1, 2, 3, and 6 months)

For LCQ-MC, the physical, social, and total scores showed significant interactions (all P < 0.05). At discharge, the physical domain score was higher in the HFNC group (adjusted mean difference 0.41, 95% CI 0.26 to 0.56; P < 0.001). At 7 days after discharge, HFNC patients had higher physical (0.34; 95% CI 0.18 to 0.50), social (0.47; 95% CI 0.28 to 0.66), and total scores (0.75; 95% CI 0.05 to 1.45), whereas no differences were observed thereafter (Fig. 3B, e-Figure 1 A-C, e-Table 4).

For cough severity (sCSS), significant interactions were observed for daytime and total scores (both P < 0.001), but not for nighttime scores (P = 0.100) (Table 5). HFNC was associated with lower odds of higher daytime cough scores at discharge (OR 0.02, 95% CI 0.01 to 0.08; P < 0.001) and at 7 days after discharge (OR 0.14, 95% CI 0.05 to 0.43; P = 0.004). The total score also favored HFNC at 7 days after discharge (OR 0.32, 95% CI 0.17 to 0.61; P = 0.003), with no significant differences beyond 1 month (e-Table 5).

Secondary symptom-related functional interference outcomes

Secondary analyses of MDASI-LC functional interference showed differential recovery trajectories between groups. Significant group-by-time interactions were observed for activity and walking interference (both P < 0.001) (Table 3).

At discharge, adjusted mean scores for activity (2.85 vs. 3.75; adjusted mean difference − 0.90; 95% CI -1.19 to -0.61; P < 0.001) and walking (3.11 vs. 4.07; adjusted mean difference − 0.96; 95% CI -1.25 to -0.68; P < 0.001) were lower in the HFNC group.

The difference in activity interference persisted at 7 days after discharge (adjusted mean difference − 0.87; 95% CI -1.16 to -0.58; P < 0.001) (Fig. 3C–D).

At 1 month after discharge, interference with relations with others was also lower in the HFNC group (adjusted mean difference − 0.57; 95% CI -0.78 to -0.36; P < 0.001). No significant between-group differences were observed for other interference domains (mood, work, or enjoyment of life) after Bonferroni correction at any time point (e-Table 6, e-Figure 2 A–D).

Discussion

In this randomized, open-label trial of older adults (aged 60–80 years) undergoing VATS lobectomy or segmentectomy, early prophylactic HFNC reduced the in-hospital burden of PPCs compared with COT. HFNC was associated with lower odds of greater PPC severity (ordinal OR 0.504) and an absolute risk reduction of 15.6% in moderate-to-severe PPCs (grade ≥ 2). Beyond complication reduction, HFNC was associated with a lower burden of PPCs and more favorable early postoperative respiratory recovery, particularly around discharge and during the first postoperative week. In contrast, conventional physiological outcomes, including PaO2/FiO2 on POD 2, 6MWT performance at discharge, and length of hospital stay, did not differ between groups. Collectively, these findings suggest that potential clinical benefits of prophylactic HFNC in older adults may be better captured by complication burden and patient-centered recovery metrics rather than isolated physiological indices.

PPCs remain common and clinically important after lung resection, associated with increased morbidity and healthcare use [4, 21]. Although reported incidence varies with definitions and surveillance strategies, contemporary studies and systematic reviews consistently identify PPCs as a key determinant of delayed recovery and a high-priority target for perioperative intervention [22–24]. Evidence suggests HFNC benefits are most pronounced in higher-risk patients and when outcomes reflect clinically meaningful deterioration [25–27]. A large randomized noninferiority trial in cardiothoracic surgical patients demonstrated that HFNC was noninferior to intermittent noninvasive ventilation for treatment failure, supporting its use as a primary post-extubation strategy [28]. Meta-analytic evidence indicates prophylactic HFNC reduces reintubation risk in higher-risk subgroups [29]. These physiological effects suggest that HFNC may function as a form of low-level noninvasive respiratory support, although the present study was not designed to comprehensively evaluate detailed ventilatory or acid-base physiological parameters.

In contrast, evidence specific to thoracic surgery remains heterogeneous. Randomized and pilot studies have reported inconsistent effects of prophylactic HFNC, likely due to differences in baseline risk, protocols, and endpoints. Several randomized and pilot studies have not demonstrated consistent reductions in PPCs or physiological outcomes. For example, a randomized trial incorporating prophylactic HFNC into an enhanced recovery pathway did not improve 6MWT-measured recovery despite shorter hospital stay and higher patient satisfaction [30]. A small pilot study similarly reported no reduction in PPCs and frequent discontinuation due to discomfort, highlighting the roles of tolerability and limited statistical power [11]. A more recent trial also failed to reduce diaphragmatic dysfunction or respiratory complications overall, although potential benefit was suggested in mechanistically defined subgroups [13]. Meta-analyses in lung resection populations have not shown consistent improvements in post-extubation hypoxemia, suggesting that oxygenation-based endpoints may not fully capture clinically meaningful benefit [31].

Few prior trials have focused on older adults, who have reduced physiological reserve and increased susceptibility to PPCs [8, 32, 33]. The present study therefore addresses an important evidence gap by targeting patients aged 60–80 years, a clinically high-risk group in whom preventive respiratory strategies may yield greater absolute benefit.

Our findings suggest that HFNC effects in older adults undergoing VATS are better captured by a complication burden framework together with secondary longitudinal analyses of respiratory symptoms and symptom-related functional interference. By integrating ordinal PPC severity with these secondary symptom outcomes, this approach provides a broader assessment of postoperative recovery. Because follow-up could be discontinued after complete symptom resolution according to the prespecified protocol, these secondary longitudinal findings should be interpreted cautiously. Notably, benefits were concentrated in dyspnea and cough, key determinants of functional recovery and QOL after lung resection yet were not paralleled by differences in conventional physiological indices. These findings underscore the importance of aligning outcome selection with clinically meaningful recovery processes.

Potential mechanisms underlying the observed clinical benefits of HFNC may include improved humidification, secretion clearance, and reduced work of breathing. Prior physiological studies have suggested that HFNC may reduce nasopharyngeal dead space and generate low-level positive airway pressure [10]. However, the present study was not designed to systematically evaluate detailed ventilatory or acid-base physiological parameters. These effects may be particularly beneficial in older adults, who are vulnerable due to declines in respiratory muscle strength, cough efficacy, and airway clearance [34, 35]. In the immediate post-extubation period following lung resection, when pain, shallow breathing, secretion retention, and atelectasis risk are most pronounced, HFNC may mitigate the progression from mild impairment to clinically significant PPCs by reducing work of breathing and facilitating airway clearance [8].

The absence of between-group differences in PaO2/FiO2 on POD 2 and in 6MWT outcomes does not diminish the observed reductions in PPC burden and early symptom improvement. Oxygen therapy in both groups was titrated to similar SpO2 targets, which may attenuate point-in-time oxygenation differences even if HFNC confers underlying benefits in work of breathing or secretion clearance [36]. Furthermore, the temporal pattern of benefit, predominantly observed around discharge and during the first postoperative week, corresponds to the period of greatest respiratory vulnerability, when the physiological effects of HFNC are most clinically relevant. In contrast, longer-term recovery is more strongly driven by tissue healing, rehabilitation, and resolution of postoperative inflammation, irrespective of the early oxygen delivery strategy. This trajectory is consistent with prior longitudinal observations showing that cough-related QOL changes most prominently early after surgery and subsequently converges over time [37].

Strengths of this study include randomized allocation, the focus on an older thoracic oncology population underrepresented in prior HFNC trials, systematic daily assessment of PPC severity, and prespecified longitudinal evaluation of symptom burden and functional interference using validated instruments. The use of an ordinal PPC severity endpoint enhances sensitivity to clinically meaningful shifts in complication burden that may be obscured when PPCs are analyzed as a binary outcome. Although the sample size calculation was based on PPC incidence, the use of an ordinal severity outcome may provide greater statistical efficiency and clinical interpretability by capturing shifts across the full spectrum of complication severity.

Limitations

Several limitations should be acknowledged. First, this was a single-center, open-label trial, which may limit generalizability and introduce potential performance and detection bias. In addition, although PPCs were assessed using predefined criteria, the ordinal PPC grading system incorporated both objective and semi-subjective clinical components, particularly within lower severity grades, and some degree of assessment bias therefore cannot be completely excluded. Second, although PPC severity was prespecified as the primary outcome before enrollment, the sample size calculation was based on binary PPC incidence estimates derived from preliminary retrospective data rather than assumptions specific to ordinal-outcome modeling. Therefore, the study was not formally powered for the ordinal primary analysis, although analyses of clinically relevant PPCs yielded findings consistent with the primary results. Third, follow-up assessments for symptom-related outcomes could be discontinued after complete symptom resolution according to a prespecified protocol-defined stopping rule. Because follow-up discontinuation was related to outcome status, some bias in the longitudinal symptom analyses cannot be excluded despite the use of mixed-effects models. Finally, detailed HFNC adherence data, physiological variables related to ventilation and acid-base status, and patient comfort or tolerability were not systematically collected. Consequently, dose–response relationships and mechanistic pathways underlying the observed clinical effects could not be fully evaluated.

Conclusions

In conclusion, among older adults undergoing VATS lobectomy or segmentectomy for NSCLC, early prophylactic HFNC reduced the burden and severity of in-hospital PPCs. Secondary longitudinal analyses also suggested improvements in early postoperative respiratory symptoms and symptom-related functional recovery, although these findings should be interpreted cautiously because follow-up could be discontinued after complete symptom resolution according to the prespecified protocol. These findings suggest a potential benefit of prophylactic HFNC in higher-risk older surgical patients and highlight the importance of incorporating patient-centered outcomes when evaluating perioperative respiratory support. Importantly, future research should aim to better delineate the balance between HFNC and conventional oxygen therapy, identify the optimal timing and patient subgroups most likely to benefit, and advance more individualized approaches to postoperative respiratory support. Until then, HFNC may represent a useful preventive respiratory support option for selected older adults or higher-risk postoperative patients rather than for routine use in all patients undergoing lung resection.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

NSCLC

Non–small cell lung cancer

VATS

Video-assisted thoracoscopic surgery

PPCs

Postoperative pulmonary complications

QOL

Quality of life

HFNC

High-flow nasal cannula

COT

Conventional oxygen therapy

ICU

Intensive care unit

SpO2

Peripheral oxygen saturation

COPD

Chronic obstructive pulmonary disease

FiO2

Fraction of inspired oxygen

POD

Postoperative day

6MWT

6-minute walk test

mMRC-DS

The modified Medical Research Council dyspnea scale

LCQ-MC

The Leicester Cough Questionnaire Mandarin Chinese version

sCSS

The simplified cough symptom score

MDASI-LC

The MD Anderson Symptom Inventory for Lung Cancer

SD

Standard deviation

IQR

Interquartile range

CI

Confidence interval

OR

Odds ratio

Author contributions

RJL, XJP and WSC were involved in research conception and study design. RJL, HYY and GMY were responsible for data collection, entry and analysis. RJL and GMY prepared the manuscript. GMY revised the manuscript. ML, YQC and CMX contributed to data collection and interpretation. XHT, XJP and WSC critically revised the manuscript. All authors read and approved the final manuscript.

Funding

The present study was supported by Finance Department of Fujian Province Project (Grant number: 20250929) and Joint Funds for the Innovation of Science and Technology, Fujian Province (Grant number:2023Y9352).

Data availability

The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request. Data sharing is subject to institutional ethics approval and data protection regulations.

Declarations

Ethics approval and consent to participate

The study was approved by local Ethics Committee and written informed consent to data analysis was obtained by all studied patients.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.