Nurse-performed POCUS-guided targeted chest physiotherapy in term neonates with severe pneumonia: protocol for a stepped-wedge cluster randomised trial

Abstract

Introduction

Severe pneumonia in term neonates frequently leads to acute respiratory failure requiring invasive mechanical ventilation. Point-of-care ultrasound (POCUS) can identify pulmonary consolidation at the bedside without ionising radiation, while chest physiotherapy (CPT) may help mobilise secretions and improve aeration. We aim to evaluate whether a nurse-performed POCUS-guided targeted CPT care pathway shortens the duration of invasive mechanical ventilation in term neonates with severe pneumonia.

Methods and analysis

This multicentre stepped-wedge cluster randomised trial will be conducted across three tertiary neonatal intensive care units (NICUs) over eight 3-month periods. All NICUs will begin in the control condition and then cross over sequentially to the intervention according to a randomised schedule. We plan to enrol 240 infants (approximately 80 per NICU). Eligible participants are term neonates with severe pneumonia, imaging-confirmed pulmonary consolidation and objectively defined respiratory failure requiring invasive mechanical ventilation. During intervention periods, trained bedside nurses will perform standardised lung POCUS every 8 hours and deliver targeted positioning and gentle percussion based on the ultrasound-determined location of consolidation. Control periods will involve standard care without routine nurse-performed lung POCUS; all CPT exposure will be recorded. The primary outcome is the duration of invasive mechanical ventilation among infants who survive to successful extubation. Analyses will use mixed-effects models with fixed effects for period and treatment and a random effect for cluster, supplemented by small-sample methods and prespecified covariate-adjusted and sensitivity analyses.

Ethics and dissemination

The study will be conducted in accordance with the Declaration of Helsinki. Ethics approval has been obtained from the Ethics Committee of Fujian Children’s Hospital (ethics approval number 2026ETKLRK04008). Written informed consent will be obtained from parents or legal guardians before enrolment. Findings will be disseminated through peer-reviewed publication, conference presentation and feedback to participating clinical teams.

Trial registration number

ChiCTR2600123537.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• This protocol evaluates a pragmatic nurse-led bedside intervention that could be integrated into routine neonatal intensive care unit (NICU) care if effective.

• The stepped-wedge cluster randomised design allows all participating NICUs to receive the intervention while controlling, in part, for secular trends.

• Standardised nurse training, image archiving and blinded expert quality review are built into the intervention to support fidelity.

• Because only three clusters are included, the precision of cluster-level effect estimation is limited and small-sample inference is essential.

• The intervention is a care bundle and control chest physiotherapy (CPT) remains discretionary; therefore, any treatment effect cannot be attributed to point-of-care ultrasound guidance alone, although detailed CPT and contamination data will be collected in both conditions.

Introduction

Background and rationale

Neonatal pneumonia remains an important cause of respiratory failure, intensive care admission and death worldwide.¹ It accounts for an estimated 750 000 to 1.2 million neonatal deaths annually, along with an unknown number of stillbirths.² Globally, pneumonia accounts for approximately 10% of deaths among children under 5 years of age, highlighting its substantial disease burden in early life.³ Among term neonates, severe pneumonia may rapidly progress to hypoxaemic respiratory failure requiring invasive mechanical ventilation.⁴ Although mechanical ventilation is often lifesaving, prolonged ventilation is associated with ventilator-associated complications, chronic lung injury and longer hospitalisation.^{5 6} Therefore, interventions that safely accelerate pulmonary recovery and facilitate earlier liberation from invasive ventilation may confer significant clinical and public health benefits.⁷

Point-of-care ultrasound (POCUS) has become an increasingly valuable bedside imaging modality in neonatal and paediatric respiratory care.^8–10 Compared with chest radiography, lung ultrasound can identify pulmonary consolidations, pleural abnormalities and pleural effusions with high diagnostic performance while avoiding radiation exposure and allowing serial bedside reassessment.^11–15 In neonatal critical care, clinician-performed lung ultrasound has also been shown to support timely therapeutic decision-making, including the immediate confirmation of pneumothorax and guidance for pulmonary surfactant administration.^{16 17} We hypothesise that, in the clinical nursing care of neonates with severe pneumonia, POCUS may enable real-time identification of lung consolidation and thereby support more precisely targeted respiratory care according to the infant’s current lung pathology.

Chest physiotherapy (CPT) has been used as an adjunctive supportive therapy in neonatal pneumonia following stabilisation and may include respiratory exercise, postural drainage, chest percussion, vibration and suctioning.^{18 19} However, evidence for its routine use in neonates remains limited, and practice varies across centres. A small non-randomised controlled study of 40 term neonates with respiratory distress reported improvements in mucociliary clearance, lung expansion, atelectasis and other chest outcomes following multimodal CPT.²⁰ Nevertheless, conventional CPT is usually guided by general clinical assessment and chest radiography rather than real-time evaluation of regional lung pathology. In the clinical nursing care of neonates with severe pneumonia, POCUS may help identify consolidated or atelectatic lung regions at the bedside and support more targeted CPT. A narrative review further suggests that thoracic ultrasound may provide more precise guidance and monitoring of respiratory therapy than auscultation or non-targeted chest radiography.²¹

In many neonatal intensive care units (NICUs), bedside nurses are well placed to combine continuous monitoring with timely clinical intervention. Although diagnostic ultrasound has traditionally been performed by physicians or sonographers, growing evidence suggests that, with structured training, non-physician clinicians—such as nurses and respiratory therapists—can reliably perform focused ultrasound examinations.^22–25 In our study, training NICU nurses to perform lung ultrasound may improve the timeliness and frequency of bedside assessment. Integrating POCUS into routine nursing practice could enable nurses to identify relevant lung findings and promptly deliver targeted supportive care, such as repositioning or gentle percussion. Evidence from other clinical settings also supports the feasibility of nurse-performed POCUS and suggests that, with appropriate training, nurses can use focused ultrasound for bedside assessment and to guide timely clinical interventions.^26–29

However, whether nurse-performed POCUS-guided targeted CPT can reduce the duration of invasive mechanical ventilation in term neonates with severe pneumonia has not been evaluated in a pragmatic multicentre trial.

Objective

The primary objective is to determine whether implementation of the nurse-performed POCUS-guided targeted CPT care pathway reduces the duration of invasive mechanical ventilation in term neonates with severe pneumonia compared with standard care without routine nurse-performed POCUS. The trial therefore evaluates the effect of the care pathway as a whole rather than the isolated effect of ultrasound guidance.

Hypothesis

We hypothesise that, compared with standard care, the nurse-performed POCUS-guided targeted CPT care pathway will reduce the duration of invasive mechanical ventilation in term neonates with severe pneumonia.

Methods and analysis

Study design

This is a multicentre stepped-wedge cluster randomised trial with three tertiary NICUs as clusters. The trial will be conducted over eight consecutive 3-month periods (24 months in total). All clusters will begin in the control condition in period 1. At the start of periods 2, 3 and 4, one NICU will cross over from control to intervention according to a computer-generated random sequence. From period 4 onwards, all clusters will be exposed to the intervention. The trial uses a repeated cross-sectional framework: different eligible infants may be enrolled in each period, and each infant will contribute data once. Figure 1 illustrates the study design.

Figure 1. Stepped-wedge cluster randomised trial schema. Each row represents one NICU cluster and each column represents one 3-month study period. White cells indicate the control phase (standard care), and grey cells indicate the intervention phase (nurse-performed POCUS-guided targeted CPT). All clusters begin in the control condition in period 1, and one cluster crosses over at the start of periods 2, 3 and 4. CPT, chest physiotherapy; NICU, neonatal intensive care unit; POCUS, point-of-care ultrasound.

Study setting

The study will be undertaken in three tertiary NICUs in China. Each participating unit provides invasive ventilation and bedside ultrasonography, with continuous specialist neonatal nursing care. Site-specific details, including staffing structures and equipment, will be documented before trial initiation.

Eligibility criteria

Inclusion criteria

1. Term neonates (gestational age ≥37+0 weeks at birth).

2. Severe pneumonia meeting the prespecified clinical, imaging and respiratory-failure criteria described below.

For this trial, severe neonatal pneumonia will be defined using a prespecified, trial-specific operational definition that will be applied consistently across all three participating NICUs. All of the following criteria must be met: (1) an acute respiratory illness with clinical features compatible with infection, such as respiratory distress, apnoea or increased airway secretions; (2) new focal or multifocal pulmonary consolidation or infiltrate demonstrated by chest radiography or clinician-performed lung ultrasound; (3) pneumonia considered by the treating neonatologist to be the principal cause of respiratory failure, with major alternative non-infectious diagnoses, including transient tachypnoea of the newborn, meconium aspiration syndrome, pulmonary oedema, isolated atelectasis and major congenital lung disease, not considered the primary explanation; and (4) objectively documented respiratory failure meeting at least one of the following prespecified criteria before invasive ventilation was initiated: (1) persistent peripheral oxygen saturation (SpO₂) <90% despite a fraction of inspired oxygen (FiO₂) ≥0.40 delivered by continuous positive airway pressure or non-invasive positive-pressure ventilation; (2) respiratory acidosis with an arterial partial pressure of carbon dioxide (PaCO₂) ≥60 mm Hg and pH ≤7.25; or (3) recurrent or prolonged apnoea requiring positive-pressure ventilation.^{30 31}

For infants who have already undergone intubation before formal screening, contemporaneous physiological and respiratory-support data obtained immediately before intubation must demonstrate that at least one of the above objective respiratory-failure criteria was met. The qualifying criterion, respiratory-support modality and settings, and the SpO₂, FiO₂ and blood-gas values nearest to the decision to intubate will be recorded on the screening case report form. Early-onset, later-onset and community-acquired neonatal pneumonia will all be eligible.

• 3 1Imaging-confirmed pulmonary consolidation in at least one lobe or region.

Imaging confirmation must be obtained by chest radiography or clinician-performed lung ultrasound at or near enrolment. Pulmonary consolidation is defined as a subpleural region with tissue-like or liver-like echotexture on lung ultrasound, or as a focal or lobar opacity on chest radiography, with or without air bronchograms.³²

• 4 1Infants must be enrolled within 48 hours of starting invasive mechanical ventilation for pneumonia.

Before enrolment, two site neonatologists will independently verify the standardised eligibility checklist and the supporting source data. Any disagreement will be resolved by a third neonatologist. The same eligibility checklist, operational definitions, case report form and source-data requirements will be used at all three participating NICUs. Only infants whose eligibility has been confirmed will be enrolled under the condition assigned to their NICU for that study period.

Exclusion criteria

1. Preterm infants (gestational age <37 weeks at birth).

2. Major congenital anomalies or comorbidities likely to materially affect respiratory management or secretion clearance (eg, congenital diaphragmatic hernia, major congenital pulmonary airway malformation or cyanotic congenital heart disease).

3. Neuromuscular disorders or severe hypotonia (eg, spinal muscular atrophy) that significantly impair the ability to clear secretions from the airway.

4. Already extubated prior to enrolment or anticipated to be extubated within a few hours.

5. Duration of intubation exceeding 48 hours prior to enrolment.

6. Absence of detectable consolidation on imaging.

7. Haemodynamic instability or contraindication to routine handling: infants receiving multiple vasoactive agents at high doses or those with conditions for which repositioning is contraindicated (eg, large sacrococcygeal teratoma).

8. Parental consent withheld or prior enrolment in this trial (no infant will be enrolled more than once).

Recruitment

Within each NICU, participant enrolment will occur continuously. Research staff, consisting of three neonatologists, will screen all newly admitted term neonates requiring mechanical ventilation for eligibility. During control periods, consent will be sought only for data collection. During intervention periods, consent will include both data collection and participation in the intervention. Given the clinical urgency, consent discussions will take place promptly after eligibility is confirmed and before initiation of the study intervention. Parents will be provided with written study information in their native language, and research staff will explain that their infant’s care may include ultrasound-guided CPT, depending on the timing of admission relative to the study schedule. Parents will be reassured that their decision to participate will not affect standard medical care and that they may withdraw at any time without consequence.

Research staff are actively involved in clinical work at the NICUs, ensuring 24-hour availability for participant recruitment, providing information to guardians and ensuring adherence to study procedures.

Cluster randomisation and allocation

The three NICUs will be randomly assigned to one of three crossover sequences, determining the order in which each NICU implements the intervention. An independent statistician, with no involvement in the trial, will generate the allocation sequence using computer-generated random numbers. To ensure allocation concealment, the randomisation schedule will be stored centrally and revealed to each NICU approximately 10 days before the scheduled crossover date, allowing time for nurse training in POCUS.

Blinding

Blinding of treating nurses is not feasible because of the nature of the intervention. Outcome assessors will not be informed of the cluster-period allocation; however, complete blinding cannot be guaranteed because intervention-related information may be present in the clinical records.

The statistician will analyse the primary outcome using masked treatment labels, with the allocation code revealed only after the primary analysis has been finalised. Any premature unblinding required for safety, ethical or regulatory reasons will be documented.

Interventions

During the intervention phase, NICUs will implement a structured protocol for nurse-performed POCUS-guided targeted CPT in addition to standard care. Representative lung ultrasound changes before and after targeted CPT are shown in figure 2.

Figure 2. Representative lung ultrasound images before and after targeted CPT. The left panel shows subpleural consolidation before targeted CPT (red arrow). The right panel, obtained after targeted CPT, shows a reduction in the extent of subpleural consolidation with re-emergence of aerated lung artefacts. CPT, chest physiotherapy.

Nurse training and competency

Before the intervention is introduced at each NICU, participating bedside nurses will undergo structured training in neonatal lung ultrasound. The programme, developed by three neonatologists and two ultrasound educators, consists of a 3-day workshop covering ultrasound physics and lung scanning techniques, followed by supervised hands-on practice involving at least 20 training infants within 7 days per nurse. Training cases will not be enrolled in the trial. Competency will be assessed on acquisition and interpretation of the pleural line, A-lines, B-lines, consolidations and static versus dynamic air bronchograms. Nurses achieving at least 90% accuracy on the standardised assessment will be certified to scan independently before their cluster crosses over. The certification date and subsequent trial scanning experience will be recorded. Refresher training and regular image-review sessions will continue throughout the intervention period. If weekly expert review identifies accuracy below 90%, the nurse will pause independent trial scanning and complete supervised remediation and recertification.

Ultrasound equipment and technique

Bedside lung ultrasound examinations will be performed using a Philips CX50 ultrasound system (Bothell, Washington, USA) with an L12-3 high-frequency linear array probe. Infants will be scanned in the supine or lateral position, as clinically tolerated, and anterior, lateral and posterior lung zones will be examined bilaterally according to a standardised 10-zone scanning protocol.³³ During each nursing shift (every 8 hours), the assigned nurse will conduct a focused lung ultrasound examination, which typically takes 5–10 min. The assessment will document the presence, extent and anatomical location of pulmonary consolidations, as well as other relevant findings such as pleural effusion or significant atelectasis. All acquired images will be stored electronically, and findings will be recorded on a standardised data collection form. If image interpretation is uncertain, real-time consultation with the on-call sonographer will be available; however, in routine practice, nurses will independently identify and localise consolidations at the bedside.

POCUS-guided CPT

CPT will be guided by the location of pulmonary consolidation identified on nurse-performed bedside lung ultrasound.

1. Targeted positioning: Based on ultrasound findings, the nurse will select a drainage position using a prespecified trial standard operating procedure that maps the anatomical location of consolidation to the recommended position. The same positioning algorithm will be used across all participating NICUs, and the identified consolidation site, selected position and positioning duration will be recorded. For example, if consolidation is identified in the right posterior lung field, the infant may be placed in the left lateral decubitus position with the affected side uppermost, as clinically tolerated.³⁴ For bilateral consolidations, positions may be alternated or the region with more extensive involvement may be prioritised. Standard safety precautions will be strictly followed during repositioning, including securing the endotracheal tube and continuously monitoring vital signs. Supportive devices such as positioning rolls will be used to maintain the infant in the ultrasound-guided drainage position for approximately 10–15 min, as tolerated. Routine repositioning every 2–3 hours will continue according to standard nursing care and will be documented separately from the protocolised POCUS-guided positioning session.

2. Gentle chest percussion: After positioning, the nurse will perform gentle percussion over the area corresponding to the identified consolidation. Percussion will be delivered using a soft neonatal percussion device according to a standardised trial-specific CPT operating procedure applied across all participating NICUs.³⁵ Each affected region will be treated for approximately 1–2 min, with an intensity sufficient to produce gentle chest-wall vibration without causing discomfort or tissue injury. Oxygen saturation and heart rate will be monitored continuously. If clinically important desaturation or bradycardia occurs, the procedure will be paused or stopped immediately and appropriate supportive measures will be provided. When multiple consolidation zones are present, each region may be treated sequentially if tolerated. Endotracheal suctioning may follow percussion when clinically indicated by audible or increased secretions.

Frequency and duration

The intervention cycle, consisting of lung ultrasound assessment, targeted positioning and chest percussion, will be performed once per nursing shift as part of routine care. In NICUs with 8-hour shifts, this corresponds to three sessions per day. The intervention will continue daily until extubation. If serial lung ultrasound examinations show resolution of consolidation while the infant remains intubated, lung ultrasound monitoring will continue, but chest percussion will be withheld in the absence of a treatment target and standard supportive care will be maintained.

Additional notes

The intervention will be delivered by the bedside nurse as part of routine nursing care. Sedation will be minimised whenever clinically appropriate, although the clinical team may administer sedation when necessary under institutional protocols. All other treatments, including ventilatory support, antimicrobial therapy, fluid management and sedation, will be determined by the treating clinical team according to standard practice and will not be protocolised by the study. Clinically important incidental POCUS findings will be communicated promptly to the treating team, and any resulting diagnostic investigations or co-interventions will be recorded prospectively.

Data collection and quality assurance

All ultrasound findings and intervention procedures will be documented in the trial case report form (CRF), including operator identity and time since certification. All images will be archived electronically. One image set from each nurse will be selected randomly each week for review by an independent sonographer blinded to treatment outcomes. Image quality and interpretation agreement will be summarised by sequential intervention week to identify performance drift or a residual learning curve.

Control condition

During control periods, infants will receive standard care for severe pneumonia without routine nurse-performed POCUS or protocolised POCUS-guided CPT. Standard care at all participating NICUs includes antibiotic therapy, lung-protective ventilatory support, fluid management, nutritional support and routine nursing care in accordance with local practice.

CPT in the control condition will not be mandated by the study and will remain at the discretion of the treating team. If CPT is provided for clinical reasons, it may be delivered by nurses or respiratory therapists according to usual practice but will not be guided by scheduled bedside lung ultrasound. Every CPT episode in both conditions will be documented, including indication, positioning, percussion, suctioning, session duration and frequency.

During control periods, nurses will not perform trial-related lung POCUS or protocolised POCUS-guided CPT on enrolled participants. Nurses may undergo training during the prespecified 10-day period before their cluster crosses over, but this training will not alter the care of enrolled control participants. Diagnostic imaging, including chest radiography, will be undertaken according to standard clinical practice. Physician-performed lung ultrasound outside the trial protocol will be permitted when clinically indicated and recorded as potential contamination.

Intervention contrast and estimand

The trial is designed to estimate the effect of implementing the complete nurse-performed POCUS-guided targeted CPT care pathway vs standard care. Because the intervention combines scheduled ultrasound assessment, ultrasound-targeted positioning and protocolised percussion, whereas CPT in the control condition remains discretionary, the design cannot isolate the incremental effect of ultrasound guidance from that of greater CPT intensity or standardisation. Detailed CPT exposure data will be collected in both conditions to characterise the achieved treatment contrast. Any analyses accounting for CPT dose or individual components will be exploratory and will not be interpreted as estimating a causal effect attributable to POCUS alone.

Microbiological and antimicrobial data

Microbiological data from clinically indicated tests performed before or within 24 hours after enrolment, preferably before initiation or modification of antimicrobial therapy, will be recorded prospectively. These tests may include blood culture, endotracheal aspirate gram stain and bacterial culture with susceptibility testing and respiratory nucleic-acid testing. No additional invasive sampling will be performed solely for research purposes. Aetiology will be classified as bacterial, viral, mixed or undetermined using prespecified criteria by two neonatologists blinded to study condition, with disagreement resolved by a third reviewer. The undetermined category will include infants with negative or insufficient microbiological testing, and the completeness of microbiological testing will be reported.

For systemic antibacterial and antiviral therapy, the CRF will record the agent, route, timing, treatment modifications during the first 72 hours after enrolment and total duration. Among infants with a bacterial isolate, the empirical regimen at enrolment will be classified as concordant or non-concordant with subsequent in vitro susceptibility results. Antimicrobial selection and modification will remain at the discretion of the treating team and will not be influenced by study allocation.

Outcomes

Primary outcome

Duration of invasive mechanical ventilation among infants who survive to successful extubation: This outcome is defined as the time from intubation for pneumonia to the first successful extubation. Successful extubation is defined as remaining free from invasive mechanical ventilation for at least 72 consecutive hours. Reintubation within 72 hours will be considered part of the same ventilation episode. Ventilation duration will be calculated in hours and converted to days by dividing by 24. Infants who die before successful extubation will not contribute to the primary outcome analysis because the outcome is not defined for these infants; deaths will instead be captured in the prespecified mortality and ventilator-free-day outcomes. For infants transferred before successful extubation, outcome information will be obtained from the receiving hospital whenever feasible; otherwise, the primary outcome will be treated as missing rather than censored.

Secondary outcomes

1. Ventilator-free days at day 10 (VFD-10) and day 28 (VFD-28): These outcomes are defined as the number of days during the first 10 or 28 days after enrolment on which the infant is alive and free from invasive mechanical ventilation. Infants who die before day 10 or day 28 will be assigned a value of 0 for the corresponding outcome.

2. Extubation failure: This outcome is defined as unplanned reintubation within 72 hours after a planned extubation. This outcome will be analysed among infants who undergo at least one extubation attempt.

3. Length of hospital stay: This outcome is defined as the number of days from hospital admission to discharge or death.

4. All-cause mortality before discharge: This outcome will be recorded for all enrolled infants and analysed as a safety outcome.

5. Adverse events related to the intervention: The following adverse events will be actively monitored: unintended extubation during nursing procedures; clinically significant oxygen desaturation or bradycardia during ultrasound assessment or CPT requiring intervention; skin injury or rib fracture potentially related to percussion; and delays in other aspects of care attributable to the intervention.

6. Process measures: These will include adherence to scheduled ultrasound assessments, agreement between nurse-performed ultrasound findings and expert image review, use of CPT in the control group and time spent on intervention-related activities per shift.

All outcomes will be prespecified before unblinded analysis in a detailed statistical analysis plan. Outcome data will be obtained from the medical record, nursing documentation and trial-specific CRFs. Two independent research assistants will collect outcome data from medical charts while blinded to intervention assignment to minimise bias.

Study timeline

Before each cluster crosses over, nurses at that site will complete training and pilot testing during the 10-day pre-crossover implementation window. Training cases will not be enrolled in the trial, and the protocol intervention will not be delivered to enrolled participants until the scheduled crossover date. During period 1, all NICUs will remain in the control condition and baseline data will be collected. At the beginning of periods 2, 3 and 4, one NICU will sequentially cross over to the intervention condition according to the randomised schedule. From period 4 onwards, all NICUs will be in the intervention condition. Patient recruitment and outcome assessment will continue throughout all study periods. The schedule of enrolment, interventions and assessments is shown in table 1. No other planned changes to clinical protocols will occur during the study period.

Table 1. Schedule of enrolment, interventions and assessments.

Study activity	Screening/ enrolment	Baseline (at enrolment)	Each nursing shift until extubation	72 hours after extubation	Day 10/day 28 after enrolment	Discharge/ death
Eligibility assessment	X
Written informed consent	X
Allocation according to cluster-period	X
Demographic and perinatal data		X
Clinical assessment and ventilator status		X	X	X	X	X
Imaging confirmation of lung consolidation		X
Nurse-performed lung ultrasound (intervention periods only)			X
Targeted positioning and gentle chest percussion (intervention periods only)			X
Routine clinical care		X	X	X	X	X
Baseline illness severity (modified enrolment SNAP-II and respiratory variables)		X
Microbiological sampling and results		X
Antimicrobial treatment		X	X			X
CPT exposure and components (both conditions)			X
Adverse event monitoring			X	X	X	X
Primary outcome: duration of invasive mechanical ventilation			X	X
VFD-10 and VFD-28					X	X
Extubation failure				X
Length of hospital stay and mortality before discharge						X

CPT, chest physiotherapy; SNAP-II, Score for Neonatal Acute Physiology II; VFD, ventilator-free days.

Sample size calculation

The sample size was estimated for the primary outcome, duration of invasive mechanical ventilation, under a stepped-wedge cluster randomised design. We assumed that the intervention would reduce the mean duration of invasive mechanical ventilation by 2 days, which was considered clinically meaningful. Based on prior NICU data, the mean duration of invasive mechanical ventilation in the control condition was assumed to be approximately 7 days, with an SD of 3 days. Because empirical intracluster correlation coefficient (ICC) estimates for this outcome in comparable neonatal stepped-wedge trials are limited, an ICC of 0.01 was used for the primary sample size calculation, with a sensitivity scenario assuming an ICC of 0.02.

The trial includes three NICUs observed over eight 3-month periods in a cross-sectional stepped-wedge design, with different infants enrolled in each period. On the basis of historical admission data, we estimated that each NICU would contribute approximately 8–10 eligible infants per period. Sample size was estimated using simulation, with 1000 simulations per scenario, based on a mixed-effects model including random cluster effects and fixed period effects.

The simulation indicated that a total sample size of 210 infants would provide approximately 85% power to detect a 2-day reduction in mean ventilation duration at a two-sided α level of 0.05, assuming an ICC of 0.01. Under a more conservative ICC of 0.02, the estimated power remained approximately 80%. To allow for uncertainty in recruitment and incomplete outcome data, we plan to enrol 240 infants in total, corresponding to approximately 80 infants per NICU over the study period.

No formal interim analysis or sample size re-estimation is planned. Recruitment will be monitored throughout the trial.

Statistical methods

All analyses will be prespecified in a statistical analysis plan finalised before unblinding. Infants will be analysed according to the condition assigned to their NICU at enrolment, regardless of intervention adherence or protocol deviations. The primary analysis will be restricted to infants who survive to successful extubation, whereas mortality and VFD analyses will include all enrolled infants. Statistical tests will be two sided with a significance level of 0.05, and effect estimates will be reported with 95% CIs. Analyses will be conducted using R or SAS.

For the primary outcome, duration of invasive mechanical ventilation among infants who survive to successful extubation, the main analysis will use a linear mixed-effects model appropriate for a repeated cross-sectional stepped-wedge cluster randomised design. The model will include fixed effects for intervention condition and study period to account for secular trends, and a random intercept for cluster to account for within-cluster correlation. Given the small number of clusters, small-sample inference will be applied using the Kenward-Roger correction. The treatment effect will be reported as the adjusted mean difference in duration of invasive mechanical ventilation between the intervention and control conditions, with its 95% CI and p value. If model assumptions are notably violated, sensitivity analyses will be undertaken using an appropriate transformation of the outcome or a rank-based approach.

Secondary outcomes will be analysed using methods appropriate to their distribution. VFDs at day 10 and day 28 will be analysed using linear mixed-effects models with the same fixed and random effects structure as the primary analysis; if these outcomes are markedly skewed, sensitivity analyses using alternative distributional assumptions or rank-based methods will be undertaken. Extubation failure will be analysed among infants with at least one extubation attempt using a mixed-effects logistic regression model including fixed effects for intervention condition and study period and a random intercept for cluster. Length of hospital stay will be analysed using a linear mixed-effects model; if substantial skewness is present, a log transformation or other appropriate alternative model will be considered. All-cause mortality before discharge and intervention-related adverse events will be summarised by study condition and analysed descriptively; where event frequency permits, mixed-effects logistic or Poisson regression models will be used as appropriate. Process measures, including adherence to scheduled ultrasound assessments, agreement between nurse-performed ultrasound findings and expert review, use of CPT in the control condition and time spent on intervention-related activities, will be summarised descriptively and will not be subjected to formal hypothesis testing for efficacy.

A prespecified covariate-adjusted analysis of the primary outcome will include three a priori baseline prognostic variables: modified enrolment Score for Neonatal Acute Physiology II (SNAP-II), Oxygen Saturation Index (OSI) and duration of invasive ventilation before enrolment. OSI will be calculated as (FiO₂×mean airway pressure×100)/SpO₂ using values obtained during a clinically stable period. A sensitivity analysis will additionally include the four-level aetiological category (bacterial, viral, mixed or undetermined). Among infants with a bacterial isolate, empirical antimicrobial concordance with susceptibility results will be explored separately. Exploratory subgroup analyses will assess potential effect modification by microbiologically confirmed bacterial versus viral aetiology using a treatment-by-aetiology interaction term; these analyses are not powered for definitive subgroup conclusions.

To assess a possible residual learning curve after certification, expert-rated image quality and interpretation agreement will be summarised according to time since nurse certification and cumulative scanning experience. As a sensitivity analysis, the primary analysis will be repeated after excluding infants enrolled during the first 4 weeks following each cluster crossover. These findings will be interpreted cautiously because only three clusters are included.

Prespecified sensitivity analyses will assess robustness to alternative outcome distributions. Given the limited number of clusters, cluster-period summaries will also be presented to assess the consistency of the findings with the primary individual-level analysis.

Missing outcome data are expected to be limited because outcome data will be obtained from routine medical records and trial CRFs. Death before successful extubation will not be treated as missing; these infants will be excluded from the prespecified survivor primary analysis and included in mortality and VFD analyses. If missing outcome or covariate data are non-negligible, their extent and patterns will be described and multiple imputation will be considered in a sensitivity analysis.

The statistical analysis plan will be finalised before unblinding and will be available from the corresponding authors on reasonable request.

Data collection, management and monitoring

Baseline data recorded before the first intervention session will include gestational age, birth weight, sex, delivery mode, 5-minute Apgar score, postnatal age at enrolment, early-onset (≤72 hours after birth) or later-onset disease, timing of symptom onset, antimicrobial exposure before enrolment, duration of invasive ventilation before enrolment and vasoactive support. Overall physiological severity at enrolment will be assessed using a modified enrolment SNAP-II, calculated using the original SNAP-II scoring criteria and the most abnormal physiological values recorded during the 12 hours ending at enrolment.³⁶

Baseline respiratory severity variables will include ventilator mode, FiO₂, peak inspiratory pressure, positive end-expiratory pressure, mean airway pressure, set respiratory rate, tidal volume per kilogram, stable SpO₂, the most recent arterial partial pressure of oxygen (PaO₂), PaCO₂ and pH, oxygenation index when arterial PaO₂ is available, OSI in all infants and the number and anatomical distribution of consolidation zones on baseline imaging.

During both study conditions, the frequency, duration and components of CPT, physician-performed lung ultrasound, ventilator management and antimicrobial exposure will be recorded. During intervention periods, nurse-performed ultrasound findings, intervention delivery, adherence to scheduled assessments and intervention-related adverse events will also be documented. Outcome data will be obtained from routine medical records, nursing documentation and trial-specific CRFs.

All trial data will be entered into a secure password-protected study database accessible only to authorised study personnel. Each participant will be assigned a unique study identification number, and directly identifiable information will be stored separately from the analytical dataset. Range checks, consistency checks and periodic source-data verification will be performed to promote data quality. All lung ultrasound images will be archived electronically for quality assurance and linked to the study database using the participant study number.

Data-collection personnel will be trained before study initiation using standardised operating procedures. Selected CRFs and source documents will be reviewed regularly by the coordinating study team. One ultrasound image set from each nurse will be selected randomly each week for blinded expert review of image quality and interpretation accuracy. The nurse’s certification date, scan operator and cumulative trial experience will be retained to support the planned learning-curve analyses.

Harms and auditing

All adverse events related to study procedures will be recorded throughout the trial. Particular attention will be paid to intervention-related events, including unintended extubation, clinically significant oxygen desaturation or bradycardia during ultrasound assessment or CPT, skin injury, rib fracture and delays in other aspects of care attributable to the intervention. Serious adverse events considered related to study procedures will be reported promptly to the relevant institutional review boards in accordance with local requirements.

No formal independent audit is planned. Any important protocol modifications will be documented and submitted to the relevant ethics committees and communicated to participating sites before implementation.

Ancillary and post-trial care: All participants will continue to receive standard NICU care throughout and after trial participation. Because the intervention is embedded in routine clinical care and is considered low risk, no special post-trial intervention is planned. Any medical management required for trial-related harm will be provided according to institutional policies and local regulations.

Patient and public involvement

Patients and the public were not involved in the design, conduct, reporting or dissemination plans of this study. Given the acute nature of severe neonatal pneumonia and the intensive care setting, direct patient or parent involvement in trial design was not feasible. However, parents or legal guardians will receive detailed study information during the consent process, and the findings of the study will be disseminated through peer-reviewed publication and conference presentation.

Ethics and dissemination

The protocol has been approved by the Ethics Committee of Fujian Children’s Hospital (ethics approval number 2026ETKLRK04008) and registered with the Chinese Clinical Trial Registry (ChiCTR2600123537). The protocol, guardian information and consent materials, consent procedures and data-confidentiality measures received ethics approval. Any modification that may affect trial conduct or participant safety will be submitted to the ethics committee and communicated to participating sites before implementation. Written informed consent will be obtained from a legal guardian before enrolment. Results will be submitted to peer-reviewed journals and presented at relevant national and international conferences. After publication of the primary findings, de-identified participant-level data, the data dictionary and statistical code will be made available through ResMan, subject to ethics and institutional approvals and applicable data-protection requirements. Authorship will reflect substantive contributions to study design, conduct, analysis and reporting.