Study participants

This single-center, prospective, randomized controlled trial (RCT) included the patients diagnosed with esophageal cancer and recieved surgery in Fudan University Shanghai Cancer Center between November 18, 2024 to June 30, 2025. It was approved by the Medical ethics committee of Fudan University Shanghai Cancer Center (FUSCC, No: 2310283−19) and registered at https://www.chictr.org.cn/ (Start and end time: November 18, 2024 to june 30, 2025, Approval number: ChiCTR2400092457) before patient enrollment. All participants provided written informed consent prior to enrollment and the trial adhered to the principles of the Declaration of Helsinki. Inclusion criteria: (1) Aged 18–80 years; (2) Scheduled for elective esophageal cancer surgery requiring one-lung ventilation (OLV); (3) American Society of Anesthesiologists (ASA) physical status classification I–III [16]. Exclusion criteria: (1) Preoperative pulmonary diseases such as bronchopulmonary dysplasia, respiratory distress syndrome, active upper respiratory tract infection, asthma, or tracheal/subglottic stenosis; (2) Anatomical abnormalities of the face, nose, or airway, or nasal cavity diseases (e.g., severe rhinitis, nasal septum deviation); (3) Preoperative SpO₂ < 95% on room air; (4) High risk of aspiration (e.g., severe gastroesophageal reflux disease, delayed gastric emptying).

Randomization and blinding

Eligible patients were randomized into two groups: the HFNC group and the Control group, with 1:1 ratio. Randomization was performed using the minimization method to balance the following covariates: age, sex, body mass index (BMI), ASA physical status, and type of surgery (thoracoscopic vs. open). A probabilistic element (80% probability of assignment to the group that minimizes imbalance) was incorporated to maintain the randomness of allocation. The randomization sequence was established using a computerized random number generator, and allocation concealment was ensured through the use of sequentially numbered, sealed, and opaque envelopes. To preserve allocation concealment, the randomization assignment was automatically generated by the web-based system only after the patient had completed the baseline assessment and was confirmed eligible for the study. The treating clinicians and investigators were blinded to the allocation sequence until after randomization. The randomization assignment was revealed to the research team only after the patient had been enrolled and baseline data collection was complete. Given the inherent nature of the interventions, neither the patients nor the healthcare providers responsible for administering the interventions could be blinded. Nonetheless, outcome assessors and statisticians remained blinded to group allocation for the duration of the study in order to mitigate any assessment bias.

Anesthesia management and study interventions

On the day before surgery, patients’ demographic data, medical history, and preoperative SpO₂ on room air were collected. Upon entering the operating room, patients underwent epidural puncture, central venous catheterization, and arterial catheterization for invasive monitoring. Preoxygenation with 100% oxygen at 6 L/min for 3 minutes was conducted before induction, followed by anesthesia induction with etomidate (0.3 mg/kg), propofol [target⁃controlled infusion (TCI), 2–3 μg/mL], remifentanil (TCI, 1–2 ng/mL), sufentanil (0.3 μg/kg), and rocuronium (0.6 mg/kg). After loss of eyelash reflex, a nasopharyngeal airway was placed for mask ventilation. Once neuromuscular blockade was complete, endobronchial intubation or bronchial blocker insertion was performed, with position confirmed by fiberoptic bronchoscopy. The mechanical ventilation was conducted in pressure regulated volume control (PRVC) mode, with respiratory parameters set as follows: VT = 8 ml/kg, I:E ratio = 1:2, FiO₂ = 50%, oxygen flow rate = 2 L/min, PEEP = 5 cmH₂O. EtCO₂ was maintained between 35–45 mmHg by adjusting the respiratory rate. During lateral position single-lung ventilation, respiratory parameters were set as follows: VT = 6 ml/kg, I:E = 1:2, oxygen flow rate = 2 L/min. FiO₂ was adjusted to maintain SpO₂ ≥ 95% and EtCO₂ within 35–45 mmHg. Anesthesia was maintained with sevoflurane (2%–3%) (Narcotrend® monitoring for depth at D0–D2), and surgical plethysmography index (SPI) was used to assess intraoperative analgesia with intermittent rocuronium for neuromuscular blockade. A standardized fluid therapy protocol was applied [infusion rate 2–10 mL/kg/h, 100 mL bolus if needed to maintain pulse pressure variation (PPV) at 9%–11%] [17], heart rate HR (60–90 bpm) and mean arterial pressure (MAP) (≥70 mmHg) [18] were maintained with noradrenaline or ephedrine for hypotension. Anesthesia agents were discontinued 5 minutes before surgery ended, sugammadex sodium (2 mg/kg) reversed neuromuscular blockade, and a lung recruitment maneuver (35 cmH₂O for 15 seconds) during TLV was performed before transfer to PACU.

Upon arrival at PACU, patients were monitored with continuous electrocardiogram (ECG), non-invasive blood pressure, peripheral capillary oxygen saturation (SpO₂), and temperature. After confirming extubation criteria, the endotracheal tube was removed, and interventions were given based on randomization. No additional respiratory devices (such as PEP devices, continuous positive airway pressure, or high-frequency chest wall oscillation) were permitted during the PACU stay. The control group received conventional nasal cannula oxygen therapy at 5 L/min, with no FiO₂/flow rate adjustments unless SpO₂ < 90%. The HFNC group received oxygen therapy via the AIRVO2 device (Fisher & Paykel Healthcare, New Zealand). The initial settings were: FiO₂ = 40%, temperature = 37°C, flow rate = 10 L/min. These parameters were titrated in the PACU according to a standardized protocol: the flow rate was adjusted in 5 L/min increments (up to a maximum of 50 L/min) for SpO₂ below 92% or patient discomfort, while FiO₂ was adjusted in 10% increments (up to 100%) if hypoxemia persisted despite maximal flow. The target was to maintain SpO₂ at the patient’s preoperative baseline (typically 94–98%). Weaning was considered when patients remained hemodynamically stable for at least 30 minutes with a respiratory rate below 25 breaths/min, SpO₂ ≥ 94% on FiO₂ ≤ 40%, and no respiratory distress. The median administered settings were a flow rate of 25 L/min and FiO₂ of 40% at 30 minutes post-intervention, titrating to 30 L/min and 35% FiO₂ by PACU discharge. The total duration of PACU stay was defined as the time from PACU admission to meeting discharge criteria, which included: stable vital signs for ≥30 minutes, Aldrete score ≥9, absence of significant nausea/vomiting, and adequate pain control (numeric rating scale ≤4). The intervention duration was defined as the time from initiation of the assigned oxygen therapy to its discontinuation prior to PACU discharge. The total duration of PACU stay [120 minutes (range: 90–180 minutes) in the HFNC group vs. 115 minutes (range: 85–175 minutes) in the control group, P = 0.171] and the time receiving the assigned intervention [110 minutes (range: 80–170 minutes) in the HFNC group vs. 105 minutes (range: 75–165 minutes), P = 0.324] were comparable between the two groups.

Rescue protocol and post-PACU protocol

A standardized rescue protocol was implemented for both groups. For the control group receiving conventional oxygen therapy, escalation progressed from high-flow nasal cannula to a simple face mask and finally to non-invasive positive pressure ventilation (NIPPV) if hypoxemia (SpO₂ < 90–92%) persisted. The HFNC group underwent up-titration of flow and FiO₂ to maximum settings before escalating to NIPPV. Both groups shared identical, pre-specified criteria for NIPPV failure and endotracheal reintubation, and all staff were trained on the uniform protocol to ensure consistent management.

Upon transfer to the surgical ward, a standardized protocol guided oxygen therapy management for both groups. Oxygen was maintained if patients met any of the following criteria: SpO₂ < 94% on room air, respiratory rate > 24 breaths/min, signs of respiratory distress, or pre-existing chronic lung disease with baseline SpO₂ < 92%. Both groups transitioned to conventional nasal cannula oxygen, starting at 2–4 L/min and titrated to maintain SpO₂ ≥ 94%—increased by 1–2 L/min if SpO₂ fell below 94%, and decreased if above 98%. SpO₂ was monitored every 4 hours for the first 24 hours, then every 8 hours. Weaning was initiated when SpO₂ ≥ 94% for ≥4 hours, respiratory rate was < 22 breaths/min, and no respiratory distress was observed, involving a 50% flow rate reduction every 2 hours. Oxygen was discontinued if SpO₂ remained ≥ 94% after 2 hours on room air, but re-initiated if SpO₂ dropped below 94% within 1 hour. Escalation to non-invasive ventilation was triggered by SpO₂ < 90% despite 6–8 L/min oxygen, respiratory rate > 30 breaths/min with distress, per hospital protocol.

Lung ultrasound assessment

Lung ultrasound was performed by two trained anesthesiologists who had completed a standardized training program and credentialing process. Both sonographers had: (a) completed at least 50 supervised lung ultrasound examinations prior to the study; (b) achieved certification in point-of-care ultrasound from the Chinese Society of Anesthesiology; and (c) participated in a 4-hour training session on the specific modified scoring system used in this study, which included both theoretical instruction and practical hands-on scanning with standardized cases.

Ultrasound examinations were performed using a Mindray M9 ultrasound system (Mindray Bio-Medical Electronics Co., Ltd., Shenzhen, China) equipped with a high-frequency linear transducer (7–12 MHz). All examinations were conducted at T1 (before extubation), T2 (30 minutes after intervention), and T3 (before leaving PACU).

Patients were maintained in the supine position throughout the PACU stay. Each hemithorax was systematically divided into 6 regions using the anterior and posterior axillary lines as anatomical landmarks: anterior and lateral zones were further divided into upper and lower segments (2 anterior and 2 lateral regions per side), while posterior zones were divided into upper and lower segments (2 posterior regions per side), totaling 12 regions. The scanning order was standardized as follows: right anterior upper, right anterior lower, right lateral upper, right lateral lower, left anterior upper, left anterior lower, left lateral upper, left lateral lower, right posterior upper, right posterior lower, left posterior upper, left posterior lower. For posterior zone assessment in the supine position, patients were log-rolled approximately 30–45 degrees to each side with assistance from nursing staff, allowing adequate access to posterior lung fields while maintaining patient safety and stability. This maneuver enabled complete visualization of all 12 predefined regions in all patients.

A modified scoring system was used (0–3 points per region): 0 points (clear A-lines, lung sliding, 0–2 B-lines), 1 point (≥ 3 B-lines or small subpleural consolidations with smooth pleural lines), 2 points (multiple confluent B-lines or small subpleural consolidations with thickened/irregular pleural lines), 3 points (subpleural consolidations > 1 × 2 cm). Total score was sum of 12 regions (0–36 points) [19]. To ensure blinding of outcome assessors, the following procedures were implemented: (a) HFNC or conventional oxygen therapy was temporarily discontinued during the lung ultrasound scanning procedure (approximately 5–10 minutes per examination), with patients receiving standard oxygen via nasal cannula at 2–4 L/min to maintain oxygen saturation >92% during this brief period. This approach ensured that the sonographer performing the real-time scanning could not identify the intervention group based on the presence of HFNC equipment. (b) Both the on-site sonographer performing the scanning and the off-line image analyst who independently reviewed the stored images for quality control were blinded to group allocation. The sonographer was not involved in patient care or aware of the treatment assignment, and all identifying information was removed from the stored images before analysis. (c) To assess inter-rater reliability, two blinded assessors independently scored the lung ultrasound images of 20 randomly selected patients (20% of the total sample, 60 examinations). The inter-rater reliability was calculated using the intraclass correlation coefficient (ICC) with a two-way random-effects model for absolute agreement. An ICC value greater than 0.75 was considered indicative of good reliability.

Outcome measurements

Primary outcome measurements: Total lung ultrasound score at T1 (before extubation), T2 (30 minutes after intervention), and T3 (before leaving PACU) to reflect postoperative atelectasis severity. The primary endpoint was prespecified as the change in total lung ultrasound scores from T1 to T3. The lung ultrasound scores at T2 and the group × time interaction analyzed by a mixed-effects model were considered secondary endpoints to assess the temporal profile of the intervention’s effect. Secondary outcome measurements: (1) Oxygenation parameters: Arterial blood gas samples were drawn from the indwelling arterial catheter at predefined time points (T1, T2, T3) and immediately analyzed using a standardized blood gas analyzer (Radiometer ABL90 FLEX, Denmark). The partial pressure of arterial oxygen (PaO₂) was directly measured from these samples. For the FiO₂ determination: in the HFNC group, the delivered FiO₂ was directly obtained from the AIRVO2 device display, which provides real-time FiO₂ monitoring with an accuracy of ±3%. The device was calibrated according to manufacturer specifications before each study day. In the conventional oxygen therapy group, the FiO₂ was estimated using the following validated formula for low-flow nasal cannula systems: FiO₂ = 20% + (4% × oxygen flow rate in L/min). For the standard 5 L/min flow rate used in this group, the estimated FiO₂ was 40% (20% + [4% × 5] = 40%). The incidence of hypoxic events was defined as SpO₂ ≤ 92% for >30 seconds during intervention [20]; (2) Sedation status: The Richmond Agitation-Sedation Scale (RASS) scores [21] at immediately, 15 minutes, 30 minutes after intervention were recorded. The RASS score is a structured clinical assessment tool that employs subjective evaluation to rapidly determine the level of sedation and agitation in patients. Its scale ranged from −5 (deep coma, unresponsive) to +4 (highly aggressive), with a score of 0 representing the ideal state of being awake and calm; (3) Hemodynamic parameters: HR and MAP at T1 (before extubation), T2 (30 minutes after intervention), and T3 (before leaving PACU), vasoactive drug dosage; (4) Postoperative pulmonary complications (PPCs) [22]: The diagnosis of PPCs was made by a centralized adjudication committee consisting of two independent pulmonologists and one thoracic surgeon who were blinded to group allocation. This committee reviewed all clinical, radiological, and laboratory data for each potential complication event using standardized case report forms. Committee members were not involved in patient care and had no access to information regarding treatment assignment. The specific definitions are as follows: ① Pneumonia: Required the presence of at least two of the following criteria: Fever (>38.0°C) or hypothermia (<36.0°C); Leukocytosis (>12 × 10⁹/L) or leukopenia (<4 × 10⁹/L); New or changed sputum production or purulent sputum. PLUS one of the following: New or progressive infiltrate on chest radiography or CT scan; Positive microbiological culture from bronchoalveolar lavage or protected specimen brushing; ② Pleural effusion: Required confirmation by chest radiography or ultrasound showing: New or increased pleural fluid collection; Associated with clinical symptoms(dyspnea, oxygen requirement) or requiring therapeutic intervention(thoracentesis or chest tube drainage);③ Bronchospasm: Wheezing on auscultation with prolonged expiration; Requiring bronchodilator therapy with clinical improvement; ④ Acute respiratory distress syndrome (ARDS): Acute onset within 7 days of known clinical insult; Bilateral opacities on chest imaging not fully explained by effusions, lobar/lung collapse, or nodules; Respiratory failure not fully explained by cardiac failure or fluid overload; PaO₂/FiO₂ ratio ≤300 mmHg with PEEP ≥5 cmH₂O; ⑤ Pneumothorax: Presence of air in the pleural space on chest radiography or CT scan; With or without clinical symptoms requiring intervention. All suspected PPCs cases were reviewed independently by two committee members. In case of disagreement, a third blinded adjudicator made the final determination; (5) Tolerability and adverse events related to the HFNC device: Data on adverse events related to equipment and tolerability during the patient’s stay in the PACU were collected. Adverse events included: nasal mucosa injury(defined as visible erythema, bleeding, or ulceration), abdominal distension(defined as abdominal girth increase >2 cm or patient complaint of significant bloating), nasal dryness/discomfort requiring intervention, and any other device-related complications. Tolerability was assessed using a 5-point Likert scale(1 = very uncomfortable, 2 = uncomfortable, 3 = neutral, 4 = comfortable, 5 = very comfortable) at 30 minutes post-intervention and prior to PACU discharge.

Sample size calculation

The sample size was calculated based on the primary endpoint (change in lung ultrasound score from T1 to T3). Based on our preliminary retrospective data, a mean difference of 4.0 points was assumed between groups in the change from baseline, with a standard deviation of 6.0 points for the change score. Using a two-sided α of 0.05 and 80% power, the required sample size was 37 patients per group (calculated using G*Power 3.1). To account for an anticipated 20% dropout rate, the sample size was increased to 50 patients per group (total n = 100). The calculated power for the final sample size (n = 50 per group) was 88%. Although a minimal clinically important difference (MCID) for lung ultrasound scores in this specific postoperative population has not been established, a difference of 4 points (approximately 25% reduction from baseline) was considered clinically relevant based on previous study in similar settings [23].

Statistical analysis

The SPSS 26.0 software (IBM Corp, Chicago, IL, USA) was used for statistical analysis. All analyses were conducted according to the intention-to-treat (ITT) principle. This means that all randomized participants were analyzed in the groups to which they were originally assigned, regardless of protocol deviations, treatment crossover, or withdrawal. The primary and secondary efficacy outcomes were analyzed using the full ITT population (n = 100). For safety and tolerability analyses (e.g., adverse events, device tolerability), the data from all participants who received at least part of the assigned intervention were included. The primary analysis used a linear mixed-effects model to estimate the between-group difference in the change in LUS scores from T1 to T3, adjusted for OLV duration. A hierarchical testing strategy controlled Type I error. Missing data were handled using maximum likelihood.

Normality of continuous data was tested with Shapiro-Wilk test; normally distributed data (mean ± SD) were compared with independent samples t-test (between groups) or repeated-measures ANOVA (within groups over time); non-normally distributed data (median, Q1–Q3) were compared with Mann-Whitney U test (between groups) or Kruskal-Wallis H test (within groups over time). Categorical data (n, %) were compared with χ² test or Fisher’s exact test. For longitudinal analyses, both repeated-measures ANOVA and linear mixed-effects models (LMM) were employed. Sphericity was assessed using Mauchly’s test, and Greenhouse-Geisser corrections were applied when the assumption was violated. Additionally, a linear mixed-effects model with random intercepts for subjects and fixed effects for group, time, and their interaction was used as the primary analytical approach. This model accounted for within-patient correlation, handled missing data, and allowed for adjustment for OLV duration as a covariate. Least-squares means (LSM) with 95% confidence intervals were derived from the LMM. To control for Type I error due to multiple endpoints and time points, a hierarchical testing strategy was employed. The primary endpoint (change in lung ultrasound score from T1 to T3) was tested first at α = 0.05. Only if this primary comparison was statistically significant were subsequent secondary comparisons interpreted. For secondary outcomes involving multiple comparisons, false discovery rate (FDR) correction was applied. Two-tailed P < 0.05 was statistically significant. In addition to p-values, effect sizes were reported with 95% confidence intervals (95% CIs) to enhance the interpretability of findings. For continuous outcomes (e.g., lung ultrasound scores, PaO₂/FiO₂ ratios), the mean difference (MD) between groups was calculated along with 95% CIs. For dichotomous outcomes (e.g., hypoxemia, pulmonary complications), risk ratios (RRs) or risk differences (RDs) with 95% CIs were computed. In addition to the unadjusted analysis, a multivariable logistic regression model was performed to assess the independent effect of the intervention on the primary composite outcome of 7-day PPCs. Goodness-of-fit of the model was assessed using the Hosmer-Lemeshow test, and multicollinearity was checked using variance inflation factors (VIFs).

Comparison of the clinical characteristics between two groups

This study screened 120 patients scheduled for esophageal cancer surgery between November 2024 and June 2025, with 100 finally included in the study. Specifically, 50 patients were assigned to the high-flow nasal cannula oxygen therapy group (HFNC group) and 50 to the conventional nasal cannula oxygen therapy group (control group). The flow chart of participants is presented in Fig 1. The baseline characteristics were generally well balanced between the two groups, as shown in Table 1.

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Comparison of the clinical characteristics between two groups.

https://doi.org/10.1371/journal.pone.0348511.t001

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. The flow chart of participants.

https://doi.org/10.1371/journal.pone.0348511.g001

Comparison of the total lung ultrasound scores Between two groups

The inter-rater reliability for lung ultrasound scoring was excellent, with an ICC of 0.92 (95% CI: 0.85–0.96), indicating high consistency between the two assessors. For the primary endpoint, the reduction in lung ultrasound score from T1 to T3 was significantly greater in the HFNC group (mean change: −5.6 points) than in the control group (mean change: −1.4 points), with a between-group mean difference of −4.2 points (95% CI: −4.7 to −3.7; P < 0.001). The linear mixed-effects model with adjustment for OLV duration confirmed a significant group × time interaction (F[2, 196] = 85.3, P < 0.001). Least-squares mean lung ultrasound scores (95% CI) from the model were as follows: at T1, HFNC group 13.2 (11–18) vs. control group 13.1 (11–17); at T2, HFNC group 8.7 (7–12) vs. control group 12.7 (10–16); at T3, HFNC group 7.6 (6–10) vs. control group 11.7 (10–14). The estimated mean difference (95% CI) in the change from T1 to T3 between groups was −4.2 points (−4.7 to −3.7).

In accordance with the hierarchical testing strategy, since the primary endpoint was significant (P < 0.001), interpretation of secondary endpoints was permitted. Post-hoc analyses at each time point revealed that lung ultrasound scores were similar at T1 (HFNC group: 13.2 ± 1.6 vs. control group: 13.1 ± 1.6; mean difference: 0.1, 95% CI: −0.5 to 0.7; P = 0.802). However, at T2 (30 minutes post-intervention), the HFNC group showed significantly lower lung ultrasound scores compared to the control group (8.7 ± 1.2 vs. 12.7 ± 1.5; mean difference: −4.0, 95% CI: −4.4 to −3.6; P < 0.001 after FDR correction). This difference was maintained at T3, with the HFNC group having lower scores (7.6 ± 1.0 vs. 11.7 ± 1.1; mean difference: −4.1, 95% CI: −4.5 to −3.7; P < 0.001 after FDR correction). The changes in total ultrasound scores for both groups are presented in Fig 2.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Changes in total lung ultrasound scores over time between the two groups.

Data are presented as mean ± SD. Statistical analysis was performed using a linear mixed-effects model (LMM) with random intercepts (adjusted for one-lung ventilation duration) and False Discovery Rate (FDR) correction for multiple comparisons. N: No significant difference between groups at T1 (P = 0.802, FDR-adjusted). ***: P < 0.001 between the HFNC group and control group at T2 and T3 (FDR-adjusted). ###: P < 0.001 for within-group comparisons (T2 vs. T1, T3 vs. T1) in the HFNC group; no significant within-group changes were observed in the control group (P = 0.180, FDR-adjusted).

https://doi.org/10.1371/journal.pone.0348511.g002

Comparison of the oxygenation parameters between two groups

PaO₂/FiO₂ ratios were comparable between groups at baseline (T1) (HFNC group: 285 ± 45 mmHg vs. control group: 288 ± 42 mmHg; mean difference: −3, 95% CI: −12–6; P = 0.389). At T2, the HFNC group had significantly higher PaO₂/FiO₂ ratios compared to the control group (325 ± 38 mmHg vs. 295 ± 35 mmHg; mean difference: 30, 95% CI: 24–36; P < 0.001 after FDR correction). This improvement was sustained at T3, with the HFNC group maintaining higher PaO₂/FiO₂ ratios (340 ± 32 mmHg vs. 305 ± 30 mmHg; mean difference: 35, 95% CI: 29–41; P < 0.001 after FDR correction). The linear mixed-effects model for PaO₂/FiO₂ ratios also showed a significant group × time interaction (F[2, 196] = 42.8, P < 0.001). Least-squares mean PaO₂/FiO₂ ratios (95% CI) were: at T1, HFNC group 283 (251–330) vs. control group 286 (262–326); at T2, HFNC group 322 (299–365) vs. control group 293 (274–320); at T3, HFNC group 338 (297–372) vs. control group 303 (284–335). The estimated mean difference (95% CI) in the change from T1 to T3 between groups was 37 mmHg (32–42). The changes in PaO₂/FiO₂ ratio for both groups are presented in Fig 3.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Changes in PaO₂/FiO₂ ratio over time between the two groups.

Data are presented as mean ± SD. Statistical analysis was performed using a linear mixed-effects model (LMM) with random intercepts (adjusted for one-lung ventilation duration) and False Discovery Rate (FDR) correction for multiple comparisons. N: No significant difference between groups at T1 (P = 0.389, FDR-adjusted). ***: P < 0.001 between the HFNC group and control group at T2 and T3 (FDR-adjusted). ###: P < 0.001 for within-group comparisons (T2 vs. T1, T3 vs. T1) in the HFNC group; no significant within-group change was observed in the control group (P = 0.162, FDR-adjusted).

https://doi.org/10.1371/journal.pone.0348511.g003

During the intervention period, 6 cases (12.0%) in the HFNC group experienced hypoxic events, whereas 14 cases (28.0%) occurred in the control group. The median time to first rescue intervention in the control group was 25 minutes (IQR: 15−40 minutes) after PACU admission. The median duration of NIPPV in the control group was 120 minutes (IQR: 90−180 minutes). The incidence of hypoxic events was significantly lower in the HFNC group compared to the control group (RR: 0.43, 95% CI: 0.18 to 0.99, P = 0.046), with a risk difference of −16.0% (95% CI: −30.2% to −1.8%). Notably, all hypoxic events in the HFNC group were alleviated by increasing oxygen flow rate (to 50L/min), whereas 6 cases in the control group required NIPPV therapy to correct hypoxaemia. No patients in either group required reintubation during the PACU stay or within the first 24 hours postoperatively.

Comparison of the sedation status (RASS scores) between two groups

At each time point following intervention (immediately, 15 minutes, and 30 minutes post-intervention), the RASS scores mainly remained at ‘0 points (alert and calm)’ across both patient groups, with no statistically significant differences between groups at any time point (Immediately: P = 0.900; 15 minutes: P = 0.842; 30 minutes: P = 0.603). Furthermore, neither group exhibited RASS scores ≤ −2 (moderate sedation or above) or ≥ + 2 (agitation) (Table 2).

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Comparison of the sedation status (RASS scores) between two groups.

https://doi.org/10.1371/journal.pone.0348511.t002

Comparison of the hemodynamic parameters between two groups

There were no statistically significant differences in HR or MAP between the two groups at T1 (HR: 70.26 ± 10.12 vs. 70.12 ± 11.66, P = 0.949; MAP: 91.84 ± 11.68 vs. 87.26 ± 13.29, P = 0.070), T2 (HR: 74.86 ± 11.13 vs. 70.86 ± 11.54, P = 0.081; MAP: 95.14 ± 10.48 vs. 93.48 ± 12.67, P = 0.477), and T3 (HR: 77.58 ± 13.38 vs. 76.54 ± 12.09, P = 0.684; MAP: 95.10 ± 10.80 vs. 94.96 ± 11.51, P = 0.950). Moreover, the two groups of patients used similar vasoactive drugs dosage (noradrenaline: P = 0.803; ephedrine: P = 0.502) during the intervention period (Table 3).

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Comparison of the hemodynamic parameters and between two groups.

https://doi.org/10.1371/journal.pone.0348511.t003

Comparison of the PPCs within 7 days between two groups

Within 7 days postoperatively, the overall incidence of PPCs was significantly lower in the HFNC group (5/50, 10.0%) compared to the control group (18/50, 36.0%), with a RR of 0.28 (95% CI: 0.11 to 0.69, P = 0.002). Specifically, the incidence of pneumonia was 4.0% (2/50) in the HFNC group versus 16.0% (8/50) in the control group (RR: 0.25, 95% CI: 0.06 to 1.06), and the incidence of pleural effusion was 6.0% (3/50) versus 20.0% (10/50) (RR: 0.30, 95% CI: 0.09 to 0.99). The results of the multivariable logistic regression analysis, adjusted for age, sex, BMI, ASA classification, duration of one-lung ventilation, and surgical approach, confirmed the protective effect of HFNC therapy. After adjustment, the HFNC group remained associated with a significantly lower odds of developing 7-day PPCs compared to the control group (adjusted odds ratio [aOR]: 0.24, 95% CI: 0.08 to 0.73; P = 0.011). The multivariable model demonstrated adequate goodness-of-fit (Hosmer-Lemeshow test: χ² = 7.12, P = 0.523), and no significant multicollinearity was detected (all VIFs < 2.0) (Table 4).

Download:

• PNG
  larger image
• TIFF
  original image

Table 4. Comparison of the postoperative pulmonary complications within 7 days between two groups.

https://doi.org/10.1371/journal.pone.0348511.t004

Comparison of adverse events and tolerability between two groups

No serious adverse events related to HFNC therapy were observed in either group. In the HFNC group, 2 patients (4.0%) experienced mild nasal mucosa erythema that resolved spontaneously without intervention, while 1 patient (2.0%) in the control group developed similar mild nasal irritation (P = 0.559). No cases of abdominal distension, significant nasal bleeding, or ulceration occurred in either group. Regarding tolerability assessment, the HFNC group reported significantly higher comfort scores compared to the control group at both 30 minutes post-intervention (4.2 ± 0.6 vs. 3.7 ± 0.8, P < 0.001) and prior to PACU discharge (4.3 ± 0.5 vs. 3.8 ± 0.7, P < 0.001). The proportion of patients reporting “comfortable” or “very comfortable” experiences was higher in the HFNC group (86.0% vs. 68.0%, P = 0.013).