Abstract

Introduction: Non-invasive ventilation (NIV) is becoming more common in the treatment of respiratory failure in children, but its effectiveness varies with the severity of the underlying disease.

Aim: The study’s objective was to assess, over a four-year period, the indications, physiological effects, and clinical outcomes of NIV in children admitted to a pediatric intensive care unit (PICU) with acute or chronic respiratory failure.

Materials and methods: This retrospective, single-center study included 25 children who received NIV between 2021 and 2025. Demographic data, underlying diagnoses, type of respiratory failure, inflammatory markers, radiological findings, NIV interface, gas exchange parameters before and after NIV, and clinical outcomes were analyzed.

Results: The median age was 3.9 years (range 0–17), with 68% being male. Neuromuscular disorders were the leading indication (55%), followed by primary pulmonary disease (37%). Spinal muscular atrophy type 1 was the most common specific diagnosis. Interface use was age-dependent: infants predominantly received nasal masks/prongs, while older children used oronasal or full-face masks. NIV produced significant improvements in gas exchange (p<0.05), indicating effective correction of respiratory acidosis and hypoxemia. Radiology and inflammatory markers distinguished two clinical phenotypes. Children with primary respiratory disease, genetic syndromes, or systemic comorbidities frequently showed pneumonic infiltrates and markedly elevated CRP/PCT, indicating infection-driven respiratory failure. In contrast, neuromuscular patients demonstrated minimal inflammatory elevation and chronic or atelectatic radiographic patterns, consistent with ventilatory pump failure. The S/F ratio at initiation strongly predicted outcome: all children with an S/F ratio <150 experienced NIV failure or death, whereas those with an S/F ratio >250 showed successful recovery. NIV duration was longest in neuromuscular disorders (mean 48 days), reflecting chronic ventilatory insufficiency.

Conclusion: NIV is effective in improving gas exchange in both acute and chronic respiratory failure in children. Clinical outcome depends on underlying diagnoses, inflammatory activity, radiologic pattern, and initial oxygenation status. The S/F ratio is a reliable predictor of NIV success.

interface, neuromuscular disease, respiratory failure

Non-invasive mechanical ventilation (NIV) has been used more often in pediatric critical care during the last 20 years. It is now considered a first-line modality for respiratory support, aiming to avoid the complications associated with invasive mechanical ventilation (IMV). NIV provides ventilatory assistance through external interfaces without the need for endotracheal intubation or tracheostomy. The complications of IMV, such as ventilator-associated pneumonia (VAP), ventilator-induced lung injury (VILI), and airway injury, are well documented and contribute to substantial morbidity and mortality among critically ill children. In comparison, NIV offers the major advantage of avoiding airway-related trauma. NIV reduces the need for sedation, shortens the duration of mechanical support, and is associated with fewer nosocomial infections and a reduced length of stay in the pediatric intensive care unit (PICU).

In the last 20 years, several studies have demonstrated that NIV is an effective and safe alternative to IMV in selected pediatric populations.^[1,2] It has shown particular benefits in patients with acute respiratory failure due to pneumonia, bronchiolitis, asthma, and acute exacerbations of chronic lung disease. NIV is also well established in children with chronic neuromuscular disorders or restrictive chest wall disease, in whom it provides long-term ventilatory assistance and prevents progressive hypercapnia. In such cases long-term NIV can improve quality of life and reduce morbidity.^[3] However, the success of NIV depends heavily on timely initiation, appropriate interface selection, optimal ventilator settings, and continuous monitoring to detect early signs of failure.

Despite the growing studies of evidence, the use of NIV in children remains challenging due to the heterogeneity of underlying diseases, variable age-related anatomy, and the need for cooperation from both patients and caregivers. Data on the effectiveness of NIV in mixed pediatric populations are still limited.

The aim of this study was to evaluate the efficacy and outcomes of non-invasive ventilation (NIV) in children admitted to the Pediatric Intensive Care Unit (PICU) with acute or chronic respiratory failure over a four-year period.

In this study, efficacy of NIV refers to the immediate physiological response to therapy, assessed by improvements in gas exchange and acid–base balance following NIV initiation, including changes in pH, PaCO₂, PaO₂, oxygen saturation, and the SpO₂/FiO₂ (S/F) ratio.

Clinical outcome, in contrast, describes the overall patient course and final status after NIV therapy and was evaluated by clinically relevant endpoints such as NIV success or failure, need for intubation or invasive mechanical ventilation, survival, duration of NIV, requirement for long-term or home ventilation, and discharge status.

We conducted a retrospective study on all consecutive children who required non-invasive mechanical ventilation in the PICU of the Professor Ivan Mitev University Pediatric Hospital for Active Treatment in Sofia, Bulgaria, from January 2021 to February 2025. All patients were admitted with respiratory failure, defined by hypoxemia (arterial oxygen saturation <90% or PaO₂ <60 mmHg on room air) with or without hypercapnia (PaCO₂>50 mmHg), or a combination of both, accompanied by clinical signs of respiratory distress.

We used the electronic register record of patients admitted to our PICU. Assisted ventilation was used for 53 patients; among them, 25 patients (47%) required NIV for respiratory support. The extracted data included demographic characteristics of the patients, type of respiratory failure, clinical diagnosis, and laboratory results at the initiation and after NIV therapy. The decision of instituting the NIV was made by a PICU physician. NIV was considered as a treatment when the patient presented with respiratory failure, either hypercapnia (PCO₂>50 mmHg) or hy- poxemia (oxygen saturation <90%) or both SpO₂ was measured non-invasively by continuous pulse oximetry. Blood gas analysis was used to assess the PaCO₂ and PaO₂ levels from arterialized capillary or arterial samples at admission, 1 hour and 2 hours after the start of noninvasive ventilation, and thereafter every 6–12 hours or as clinically indica- ted. Arterialized capillary blood gas analysis was used as the primary tool for monitoring pH, pCO₂, and HCO₃^– trends, acknowledging its close correlation with arterial values for these parameters in hemodynamically stable patients. For the precise assessment of oxygenation status (PaO₂), arterial samples were drawn when clinically indicated. Values recorded immediately before NIV initiation and the first stable values obtained after NIV initiation were used.

Exclusion criteria were a Glasgow Coma Scale (GCS) of less than 8 or altered mental status in previously normal patients and cardiocirculatory instability.

In our pediatric intensive care unit (PICU), only conventional invasive ventilators with a dedicated non-invasive ventilation mode are available for use. The initial ventilator settings applied in our unit are detailed in Table 1. These parameters were selected in accordance with current published guidelines.^[4]

The primary ventilation modes employed were pressure-controlled ventilation (PCV) and pressure support ventilation (PSV). PCV is often favored for its superior compensation of air leaks and is generally well-tolerated. Conversely, PSV offers improved patient–ventilator synchrony and a greater ability to adapt to the patient’s own respiratory effort. The application of continuous positive airway pressure (CPAP) was reserved for specific indications, namely obstructive sleep apnea and bronchiolitis. Collected data included demographic variables (age and sex), primary clinical diagnoses, and the type and duration of NIV. The progress included the need for intubation and home ventilation. Patient’s response was assessed by oxygen saturation and blood gas analysis. Data were collected from electronic patient records before and after NIV initiation.

A total of 25 children received non-invasive ventilation (NIV) during the study period. The mean age was 6.1±6.4 years, with a median of 3.9 years (range: 0–17), reflecting a heterogeneous pediatric population. The male sex was predominant (17 boys, 68%) compared with the female (8 girls, 32%).

NIV was primarily applied in cases of neurological disorders (55%) and primary pulmonary diseases (37%). The leading cause of respiratory failure was generalized muscle hypotonia (11 children, 44%), followed by cerebral palsy (14%) and spastic quadriparesis (7%). The most frequent specific diagnosis was type 1 spinal muscular atrophy (SMA), was observed in five children (20%). Additional etiologies included genetic syndromes (n=3), epileptic encephalopathies (n=2), and myasthenia gravis (n=2). Specific genetic diagnoses among patients with neurological disabilities included Cockayne syndrome, as well as muscle hypotonia associated with both Down syndrome and Prader–Willi syndrome. In Table 2 we present the patients according to their primary respiratory diagnosis, leading to NIV and associated underlying conditions. Pneumonia was the most frequent diagnosis, accounting for 52% of cases (13/25), followed by respiratory failure in 48% (12/25). The most common comorbidities were spinal muscular atrophy and genetic syndromes. Overall, the cohort is characterized by a high prevalence of chronic neurological, genetic, and systemic conditions contributing to respiratory morbidity.

The most commonly used NIV interface was the oronasal mask (36%), followed by the full-face mask (32%) and nasal prongs or nasal mask (28%). A single patient required alternating interfaces due to facial dysmorphism. The choice of interface correlated strongly with age—nasal masks or prongs were predominantly used in infants <1 year of age, and oronasal and full-face masks were preferred in older children (>6 years) (Fig. 1).

Figure 1.

NIV interface distribution and mean age of patients.

The analysis of blood gas (BG) parameters before and after NIV demonstrated a clear physiological improvement (Tables 3 and 4). Blood gases were analyzed from capillary samples, with arterial blood gas analysis performed when necessary.

In our cohort of 25 children on NIV, 2 (8%) met criteria for type 1 hypoxemic respiratory failure, 6 (24%) for type 2 hypercapnic respiratory failure, and 17 (68%) showed mixed or compensated gas profiles (Table 4). Before NIV initiation, the mean pH was 7.35±0.11, indicating mild acidemia with partial metabolic compensation. The mean partial pressure of carbon dioxide, pCO₂ was 60.5±29.7 mmHg, reflecting moderate hypercapnia, and the partial pressure of oxygen, pO₂, averaged 82.5±35.8 mmHg, consistent with mild hypoxemia (Table 3). The average pH of 7.35 before therapy is close to the normal range (7.35–7.45), suggesting partial compensation. The gas exchanges parameter—pCO₂ and pO₂—before NIV, showed moderate hypercapnia with mean pCO₂ ≈ 60 mmHg with substantial interindividual variability (SD ≈ 30 mmHg), consistent with a heterogeneous population ranging from mild to severe ventilatory failure. The mean pO₂ (82 mmHg) denoted borderline hypoxemia, though individual values varied widely, indicating that some patients were adequately oxygenated while others presented with significant hypoxemia. Levels of HCO₃- and base excess (BE) were within or slightly above the normal range, implying a degree of metabolic compensation for chronic CO₂ retention.

Following initiation of NIV, patients exhibited a notable improvement in acid–base parameters, reflecting effective correction of ventilatory and gas exchange abnormalities. The improvement in pCO₂ and pH was based on capillary BGA. Oxygenation parameters (pO₂ and SatO₂) are reported from a combination of capillary BGA and continuous pulse oximetry, with arterial BGA used in cases of severe hypoxemia. The mean pH increased from 7.35±0.11 before therapy to 7.43±0.08 after therapy, indicating a significant shift towards normal acid-base status (p<0.05). The reduction of pCO₂ was statistically significant (p<0.01), with the mean pCO₂ decreased from 60.5±29.7 mmHg pre-therapy to 44.8±9.5 mmHg post-therapy. This change demonstrates the efficacy of NIV in correcting respiratory acidosis through improved alveolar ventilation and CO₂ clearance. The mean pO₂ increased from 82.5±35.8 mmHg to 98.4±28.6 mmHg, reflecting improved oxygenation efficiency (p<0.05). Post-therapy HCO₃- levels showed a slight decrease (from 33.1±5.7 mmol/L to 30.4±4.9 mmol/L), while BE normalized toward zero (from +6.8±4.2 to +3.1±2.8). These changes are consistent with partial reversal of metabolic compensation following correction of hypercapnia. Peripheral oxygen saturation (SatO₂) improved modestly, from a mean of 87.64±9.54% to 93.65±5.48, though differences were not statistically significant (p>0.05) due to already near-optimal baseline values and oxygen supplementation. The cohort’s mean pre-NIV pO₂ (82.5±35.8 mmHg) reflects the mix of respiratory failure types. A significant proportion of patients (68%) had hypercapnic or mixed failure, where the primary indication for NIV was elevated pCO₂ (>50 mmHg) and respiratory acidosis, not profound hypoxemia. In these patients, pO₂ was often preserved or only mildly reduced due to supplemental oxygen. The inclusion criterion of hypoxemia (SatO₂<90% or pO₂<60 mmHg) was met in all patients, primarily through peripheral oxygen saturation (SatO₂) below 90% on room air at presentation, which rapidly improved with oxygen supplementation prior to the first BGA. This explains the pO₂ values >60 mmHg in the initial blood gas analysis for some.

The most frequently used mode of ventilation in this study was Pressure Support Ventilation (PSV) (n=12, 48%). This mode was preferred in children with neuromuscular weakness, genetic syndromes with muscle hypotonia, and mild–moderate respiratory distress. Pressure-Controlled Synchronized Intermittent Mandatory Ventilation (PC-SIMV) was the second most used mode of ventilation (n=6, 24%) and was indicated in moderate-to-severe respiratory failure, weak spontaneous breathing, and pneumonia with muscle fatigue. Continuous positive airway pressure mode (CPAP) was used only in three of the patients with primary pulmonary disorder. Mixed modes of ventilation were used in two cases of pneumonia in children with cerebral palsy (P-SIMV/PSV) and myasthenia gravis (CPAP/PSV). Only one patient with extreme hypoxemia and a primary diagnosis of cystic fibrosis required a mode of pressure-controlled assist ventilation (P-A/C), followed by intubation and IMV.

On admission radiological findings revealed bilateral pneumonic infiltrates in 32% of children and localized parenchymal consolidation (unilateral pneumonia, atelectasis, or hypoventilated areas) in 28%. Only a minority of patients had normal chest X-rays, and very few showed chronic structural changes (Fig. 2).

Figure 2.

Distribution of chest X-ray findings on admission.

Primary respiratory disorders were associated with the highest proportion of patients with elevated inflammatory markers; over 50% in this group had elevations in both C-reactive protein (CRP) and procalcitonin (PCT). Genetic syndromes also demonstrated a substantial inflammatory component (≈ 2/3 with elevated CRP and 1/3 with elevated PCT). In contrast, neuromuscular disorders showed low inflammatory activation (only ~17% with CRP >5 mg/L and none with elevated PCT), consistent with a pathophysiology of chronic ventilatory failure rather than primary infectious pathology (Fig. 3).

Figure 3.

Distribution of inflammation markers according to primary diagnosis of patients.

The SatO₂/FiO₂ ratio (S/F) was used to assess the severity of acute respiratory distress and grade hypoxemia. In most of the cases, NIV was initiated for patients with an S/F ratio of over 250 (mild hypoxemia), while nine patients had moderate hypoxemia (S/F ratio between 150 and 250). Only seven patients (28%) presented with severe hypoxemia on admission (S/F<150). In comparing pre-therapy hypoxemia severity with clinical outcome, cases with mild-to-moderate hypoxemia showed a favorable response to NIV. Patients who required transfer to IV or long-term NIV had intermediate S/F ratios (150–220), indicating moderate hypoxemia and partial ventilatory compromise. All children with severe hypoxemic states required intubation due to failure of NIV or died (Table 5).

The duration of non-invasive ventilatory support was longest among patients with neuromuscular disorders (n=6), with a mean of 48.2±94.2 days (median 12.5). One patient with cerebral palsy required NIV for 30.0 days. In contrast, children with primary pulmonary diseases (n=1) had considerably shorter NIV courses, averaging 6.4±5.8 days (median 6), as did those with genetic syndromes (n=3), who required 6.3±3.5 days (median 6) (Fig. 4). The Kruskal–Wallis test found no statistically significant difference in NIV duration across all diagnostic groups (p=0.133), most likely due to the small sample size and high variability within the neuromuscular cohort. However, pairwise comparison revealed that patients with neuromuscular disorders required significantly longer NIV support than those with primary pulmonary or genetic conditions, reflecting the chronic nature of ventilatory insufficiency in neuromuscular disease.

Figure 4.

Distribution of mean duration of NIV therapy and diagnose of patients.

This study aimed to describe our experience in using non-invasive ventilation (NIV) as a first-line therapy in children with respiratory failure admitted to the Pediatric Intensive Care Unit. NIV is now widely applied in pediatric practice, with reported success rates between 55% and 96%, although outcomes depend strongly on disease-specific and patient-related factors.^[5] In our cohort, the predominant causes of respiratory failure were neuromuscular diseases (NMD), characterized by respiratory muscle weakness and ineffective airway clearance. In these patients, respiratory infections frequently precipitate acute respiratory failure or exacerbate chronic respiratory insufficiency. NIV has been shown to improve both nocturnal and daytime gas exchange, enhance sleep efficiency, and prolong survival in this population.^[6] The most common diagnosis in our study was type 1 spinal muscular atrophy, the most severe form associated with profound hypotonia and chronic respiratory failure, requiring long-term home ventilation. Consistent with previous studies, NIV use in children with SMA and Duchenne muscular dystrophy has been associated with improved quality of life and reduced mortality.^[6-8] Respiratory infections represented the second most frequent indication for NIV in our series. NIV was successfully applied in cases of acute respiratory failure due to primary pulmonary disease, although severe hypoxemia remains a contraindication for its initiation.^[9-11]

An NIV interface is a device that connects the patient to the ventilator tubing. Successful NIV requires a compatible interface and securing system. One of the most common causes of NIV failure within the first few hours of treatment is patient discomfort with the interface or a lack of patient–ventilator synchrony.^[12] The choice of interface depends on the patient’s age and type of respiratory failure. Most studies recommend nasal masks as a first choice for children starting NIV.^[13,14] In Table 3, we compare our results with the current recommendations for the choice of NIV interface. Nasal masks are generally indicated for type II chronic respiratory failure or type I acute respiratory failure, but only for patients that are not severely dyspneic and can cooperate and keep their mouths closed; otherwise, air leakage would make the mask intolerable. Oronasal masks are best suited for type I and advanced type II acute respiratory failure, in which the patient cannot breathe through the nose alone, especially for dyspneic patients who tend to breathe through the mouth. In our study, oronasal and full-face masks were preferred for older children with hypercapnic respiratory failure, whereas nasal masks or prongs were used for infants.

Dual-pressure modes with continuous flow seem to be the best tolerated NIV modes. There have been very few comparative studies on the efficacy and tolerance of NIV modes, most of which have been conducted in adults.^[15] The PSV combines controlled, assisted, and spontaneous breaths, requiring the coexistence of different inspiratory and expiratory triggering mechanisms.^[16] PSV was the most common mode in patients with neuromuscular disease, chosen for the safety of its back-up mode in case of apnea and for better patient-ventilator synchrony. Studies suggest PSV is the most appropriate mode for long-term home ventilation in children with SMA.^[17] CPAP is generally preferred for primary hypoxemic, type 1 failure due to upper or lower airway obstruction. This mode delivers constant pressure to maintain airway patency.^[18] These findings align with our cohort, in which bilevel modes (PSV and PC-SIMV) were favored in neuromuscular and chronic ventilatory disorders, whereas CPAP was mainly used in hypoxemic conditions.

Although arterial blood gas (ABG) analysis remains the gold standard for evaluating gas exchange, arterial puncture is an invasive and technically demanding procedure, associated with discomfort and potential complications in approximately 11.3% of cases.^[19] In pediatric practice, arterialized capillary blood sampling is therefore routinely employed as a reliable and less invasive alternative. In this study, NIV led to a rapid and statistically significant correction of respiratory acidosis, with normalization of pH and reduction in pCO₂. These findings confirm the efficiency of NIV in improving ventilation and CO₂ clearance while also enhancing oxygenation without invasive mechanical ventilation.

The radiological and inflammatory patterns in our cohort reflected two distinct mechanisms of respiratory failure. Children with primary respiratory disease, genetic syndromes, or systemic comorbidities showed the highest inflammatory burden and the highest frequency of pneumonic infiltrates, indicating that NIV was often initiated in the setting of acute infection. In contrast, children with neuromuscular disorders had minimal inflammatory elevation and radiographs consistent with chronic or non-infectious changes—such as atelectasis or hypoventilation—supporting the notion that NIV in this group was primarily used for chronic ventilatory pump failure rather than inflammatory lung disease. Although statistical significance was not reached for individual markers, the combined radiographic and laboratory findings clearly distinguish infection-related respiratory failure from chronic neuromuscular insufficiency.

Similar to other studies that have shown that hypoxemia or the presence of acute respiratory distress syndrome predicts NIV failure, our study identified that FiO₂ at initiation as a marker for failure.^[20-23] The severity of hypoxemia is a key predictor of failure.^[24] The oxygenation index, estimated by the ratio of arterial oxygen partial pressure to the fraction of inspired oxygen (PaO₂/FiO₂ or P/F ratio), is a standard measure. The ratio of pulse oximetric saturation to the fraction of inspired oxygen (SpO₂/FiO₂ or S/F) is a less invasive method of assessing hypoxemia severity, with effectiveness comparable to the P/F ratio. The SpO₂/FiO₂ ratio before NIV initiation was a powerful predictor of outcome. In our study, patients with baseline S/F ratio >200–250 showed high likelihood of recovery, while those below 150 frequently experienced NIV failure or mortality. Thus, the S/F ratio is an accessible and reliable index for early stratification of patients’ prognosis during non-invasive ventilatory therapy.

The mean duration of NIV in patients with primary pulmonary diseases, respiratory failure in genetic syndromes, or other causes—cancer, metabolic diseases, etc.—was 6.3 days, which is similar to the results of a study by Ferdous et al.^[25] Neurological disorders are a major indication for long-term mechanical ventilation (LTMV), as chronic respiratory failure frequently develops from progressive muscle weakness. Previous studies report that 20–56% of children with neuromuscular disease require LTMV.^[26-28] In our cohort, children with neuromuscular disorders and cerebral palsy demonstrated the longest duration of NIV therapy, reflecting the severity of their ventilatory compromise. All patients with spinal muscular atrophy required continuation of NIV at home, consistent with evidence that NIV improves tidal ventilation, unloads respiratory muscles, and enhances gas exchange in chronic neuromuscular conditions.^[29,30]

Non-invasive ventilation can be an effective respiratory support in both acute and chronic respiratory failure when applied with appropriate clinical judgment. Careful attention to patient selection, interface choice, and timing of initiation is essential to optimize outcomes. Continuous close monitoring of clinical signs, gas exchange, and oxygenation indices such as the S/F ratio plays a critical role in guiding therapy and predicting success.

This study has several limitations. First, the sample size was relatively small, which limited the statistical power of the findings. Second, the study was conducted in a single center, and the results reflected local clinical practice patterns. Larger, multicenter studies are needed to validate these observations and strengthen the conclusions.

Ethical statement

• The authors declared that no clinical trials were used in the present study.
• The authors declared that no experiments on humans or human tissues were performed for the present study.
• The authors declared that written informed consent was obtained from the parents or legal guardians of all minor patients for the publication of this study and any accompanying images. The signed informed consent forms are safely deposited at the Professor Ivan Mitev University Pediatric Hospital of Active Treatment, Sofia, Bulgaria and available for review upon request.
• The authors declared that no experiments on animals were performed for the present study.
• The authors declared that no commercially available immortalized human and animal cell lines were used in the present study.

Conflict of interest

The authors have declared that no competing interests exist.

Artificial Intelligence (AI) use

The authors accept full responsibility for the content of the manuscript, including the disclosure of any use of AI. During the preparation of this work, the authors used an AI agent for grammar checking and language style editing.

Funding

No funding was reported.

Author contributions

LB conceptualized the manuscript, drafted it, served as the study’s attending physician, and conducted preliminary statistics. LI, TT, and BB contributed ideas for the manuscript, assisted with manuscript revisions, and served as the study patients’ attending physicians. GP conceptualized the manuscript, oversaw its completion, and performed the final statistics and revision.

Author ORCIDs

Lilia Bozadzhieva https://orcid.org/0009-0004-9573-0838

Lyubomila Ilarionova https://orcid.org/0009-0005-6145-0157

Blagomir Zdravkov https://orcid.org/0009-0006-0644-8748

Guergana Petrova https://orcid.org/0000-0001-8168-742X

Data availability

All of the data that support the findings of this study are available in the main text.