European Journal of Cardiovascular Medicine

Pneumonia remains a leading cause of morbidity and mortality among children worldwide, with a disproportionate burden in low- and middle-income countries (LMICs). In routine clinical care, chest X-ray (CXR) is commonly used to support diagnosis and guide management; however, its utility in children is constrained by limited sensitivity in early disease, inter-reader variability, challenges in differentiating viral from bacterial patterns, logistical delays, and cumulative ionising radiation exposure—limitations that are particularly relevant in resource-constrained settings. Reflecting these realities, Kazi et al. (2022), in a pragmatic scoping review focused on infants and children with severe pneumonia in LMICs, highlighted the practical role and limitations of CXR and described growing interest in lung ultrasound (LUS) as a bedside adjunct to improve diagnostic and management pathways. [1]

LUS has emerged as a radiation-free, point-of-care imaging modality capable of detecting peripheral consolidations, dynamic air bronchograms, pleural line abnormalities, B-lines, and pleural effusions—features that are frequently present in paediatric pneumonia. Evidence synthesis supports strong diagnostic performance. Pereda et al. (2015), in a meta-analysis of paediatric studies, concluded that LUS demonstrates high accuracy for diagnosing pneumonia in children and can perform comparably to CXR when standardised protocols and trained operators are used. [2] Beyond classic “high suspicion” scenarios, LUS may provide value even when clinical suspicion is low or atypical. Scheier et al. (2020) evaluated LUS in children with low clinical suspicion for pneumonia and showed that LUS can help clarify diagnosis in settings where clinical findings may be equivocal, supporting its potential role in early triage and reducing missed cases. [3]

In addition to diagnosis, LUS is increasingly studied for severity assessment and monitoring. Bobillo-Perez et al. (2020) reviewed paediatric applications of LUS and emphasised its usefulness not only for identifying pneumonia but also for estimating disease extent (e.g., number of affected zones, consolidation size, presence of effusion) and tracking response to therapy, features that can be operationalised into severity scoring approaches. [4] The scope of paediatric LUS is also expanding into critical care contexts. Uguen et al. (2024) reported on LUS as a potential diagnostic tool for ventilator-associated pneumonia in paediatric intensive care units, reinforcing the broader concept that ultrasound-based lung assessment can support timely recognition and clinical decision-making in high-acuity paediatric respiratory illness. [5]

Despite growing evidence, real-world diagnostic accuracy and severity correlation can vary across settings due to differences in operator training, patient mix, timing of imaging, and reference standards. Therefore, generating locally relevant evidence from Indian hospital settings is important to define the performance of LUS against routinely used standards and to examine whether LUS findings meaningfully reflect clinical severity at presentation.

Accordingly, this prospective study at the Indian Institute of Medical Sciences and Research, Badnapur, Jalna was undertaken to evaluate the role of lung ultrasound in the early diagnosis of paediatric pneumonia using a CXR plus final clinical diagnosis composite as the reference standard, and to assess the ability of LUS findings to stratify disease severity at presentation.

Study design, setting, and duration A prospective diagnostic accuracy study was conducted in the Department of Paediatrics at the Indian Institute of Medical Sciences and Research (IIMSR), Badnapur, Jalna, over 6 months. The study evaluated the diagnostic performance of lung ultrasound (LUS) for paediatric pneumonia and its utility in severity assessment, using a composite reference standard (CXR + final clinical diagnosis). Study population and eligibility Children presenting to the paediatric outpatient/emergency services with suspected pneumonia were screened. Inclusion criteria • Age 1 month to 12 years • Clinical suspicion of pneumonia (e.g., fever and/or cough with fast breathing, respiratory distress, abnormal chest auscultation, or clinician suspicion warranting imaging) • Parent/guardian provided written informed consent (and assent where applicable) Exclusion criteria • Known chronic lung disease likely to confound imaging interpretation (e.g., cystic fibrosis, bronchiectasis) • Hemodynamic instability precluding imaging as per protocol • Children already on mechanical ventilation at presentation (unless the protocol explicitly included them) • Refusal of consent Sample size and sampling technique All eligible children presenting during the study period were enrolled using consecutive sampling until the end of 6 months. The sample size was determined by feasibility within the study duration, with the intent to achieve adequate numbers for estimating sensitivity and specificity with acceptable precision in a real-world paediatric setting. Index test: Lung ultrasound (LUS) protocol LUS was performed at presentation or as early as feasible after clinical evaluation (ideally within the first few hours of presentation/admission). A portable ultrasound machine with a high-frequency linear probe (7–12 MHz) was used for pleural and superficial lung assessment; a curvilinear probe could be used in older children or for deeper lesions when required. A standardized scanning approach was followed (e.g., 6-zone or 12-zone protocol across anterior, lateral, and posterior areas bilaterally). Each hemithorax was assessed systematically. LUS diagnostic features suggestive of pneumonia included: • Subpleural consolidation (with or without dynamic air bronchograms) • Focal B-lines or confluent B-lines in a localised distribution • Pleural line irregularity overlying an abnormal area • Pleural effusion (simple/complex) where present A lung ultrasound examination was considered positive for pneumonia if it demonstrated either (i) a subpleural consolidation (with or without dynamic air bronchograms) or (ii) a focal interstitial pattern (focal/confluent B-lines) with pleural-line abnormality in a localized distribution, with or without pleural effusion Severity assessment using LUS For severity assessment, LUS findings were recorded as: • Number of involved zones (out of total scanned zones) • Presence/absence of pleural effusion • Consolidation size category (e.g., <1 cm, 1–2 cm, >2 cm or as recorded in cm) • Bilateral involvement (yes/no) A pragmatic LUS severity measure was operationalized as either a zone-based score (higher score = greater involvement) or categorical severity based on extent (e.g., limited unilateral vs multilobar/bilateral and/or effusion). This LUS-based severity was compared with clinical severity outcomes. Reference standard (composite) The reference standard was a composite of Chest X-ray (CXR) plus final clinical diagnosis. • CXR (AP/PA view as per age and feasibility) was interpreted by the radiologist/clinician as pneumonia present/absent and pattern (lobar/bronchopneumonia/interstitial) with pleural effusion if present. • Final clinical diagnosis was made by the treating pediatrician using clinical course and available investigations (including response to therapy when relevant). A child was classified as reference-standard positive for pneumonia when the composite assessment supported pneumonia (CXR consistent with pneumonia and/or final clinical diagnosis consistent with pneumonia in the clinical context). Discordant cases were resolved by predefined clinical adjudication criteria documented in the protocol. Blinding To reduce interpretation bias: • LUS operators recorded findings on a structured form before reference standard interpretation. • CXR was interpreted without access to LUS results where feasible. • Final clinical diagnosis was made as per routine care; the extent of blinding to LUS could be stated depending on workflow practicality. Data collection A structured proforma captured: • Demographics: age, sex • Clinical findings: symptom duration, respiratory rate, SpO₂, fever, chest indrawing, auscultation findings • Treatment and outcomes: oxygen requirement, admission/PICU requirement, length of stay (if admitted) • Imaging variables: detailed LUS findings and CXR impression Statistical analysis Data were analyzed using standard statistical software. Diagnostic accuracy (primary analysis): LUS performance for pneumonia diagnosis was evaluated against the composite reference standard using a 2×2 table to calculate: • Sensitivity, specificity • Positive predictive value (PPV), negative predictive value (NPV) • Overall accuracy All estimates were reported with 95% confidence intervals. Agreement analysis: Agreement between LUS and CXR interpretation was assessed using Cohen’s kappa. Severity assessment (secondary analysis): Association between LUS severity markers (e.g., zone involvement, consolidation size, effusion) and clinical severity indicators (oxygen requirement, PICU admission, length of stay, or severity category) was tested using appropriate statistical tests (chi-square/Fisher’s exact for categorical variables; t-test/Mann–Whitney for continuous variables). Where feasible, logistic regression was planned to identify independent predictors of severe disease. A p-value <0.05 was considered statistically significant. Ethical considerations Institutional Ethics Committee approval was obtained prior to study initiation. Written informed consent was obtained from parents/guardians (and assent from older children where applicable). Confidentiality was maintained through anonymized coding. LUS is non-ionizing and was performed as an adjunct; CXR was ordered as per standard clinical indication and institutional practice.

Participant inclusion and clinical profile

Over the 6-month study period, 150 children with suspected pneumonia were enrolled consecutively; all underwent LUS and CXR and were adjudicated using the composite reference standard (CXR + final clinical diagnosis). Using this reference, 60.0% (90/150) were classified as pneumonia. Baseline demographic and clinical characteristics stratified by reference diagnosis are summarized in Table 1.

Compared with children without pneumonia, those adjudicated as pneumonia had a more severe physiological profile at presentation—higher temperature and respiratory rate with lower SpO₂—and were more likely to have chest indrawing and require supplemental oxygen (Table 1). Median length of stay was also longer among pneumonia cases. Overall, clinical severity distribution in the cohort showed a substantial burden of moderate-to-severe illness, supporting evaluation of LUS not only for diagnosis but also for severity assessment.

In modality-to-modality comparison, agreement between LUS and CXR was moderate (κ = 0.54), with discordance more frequently due to LUS positive/CXR negative findings than the reverse (counts: 26 vs 8).

Lung ultrasound patterns among LUS-positive examinations

Among children with a positive LUS (n=95), the dominant pattern was parenchymal consolidation with pleural-line abnormalities, frequently accompanied by focal interstitial changes (Table 4). Consolidation was identified in 78/95 (82.1%), and dynamic air bronchograms were present in 52/78 (66.7%) of those with consolidation, supporting an alveolar phenotype in most LUS-positive scans. Pleural-line abnormalities were recorded in 77/95 (81.1%), and focal B-lines in 64/95 (67.4%) (Table 4).

Markers of greater sonographic extent were also observed: pleural effusion was present in 15/95 (15.8%), and bilateral involvement in 30/95 (31.6%). The overall burden of sonographic involvement among LUS-positive children was moderate-to-high (median zones involved 5 [IQR 3–7]; median LUS score 11 [IQR 6–14]). Consolidation size among those with consolidation (n=78) was most commonly >2 cm (46.2%), followed by 1–2 cm (30.8%) and <1 cm (23.1%).

When stratified by reference-standard status, LUS false positives (LUS+/reference–, n=10) tended to show a lower extent of involvement (median LUS score 3 [IQR ~1–3]) than true positives (LUS+/reference+, n=85; median LUS score 12 [IQR 8–14]), with bilateral involvement uncommon in false positives (Table 4).

Table 1. Baseline characteristics by composite reference standard (N = 150)

Diagnostic performance of lung ultrasound for pneumonia (primary outcome)

Using the composite reference standard (CXR + final clinical diagnosis), lung ultrasound (LUS) correctly identified 85 of 90 pneumonia cases and correctly ruled out pneumonia in 50 of 60 non-pneumonia cases (Table 2). Overall diagnostic performance metrics are summarised in Table 3.

Table 2. 2×2 diagnostic table: LUS vs composite reference standard (N = 150)

Table 3. Diagnostic accuracy of LUS for paediatric pneumonia (vs composite reference)

*Consolidation size category calculated among children with consolidation (overall n=78; true positives n=73; false positives n=5).

Among LUS-positive examinations (n=95), consolidation was present in 78 (82.1%); the remaining LUS-positive cases met the predefined positivity criteria based on focal interstitial changes with pleural-line abnormality (with/without effusion) in the absence of a discrete consolidation.

Severity assessment using lung ultrasound (secondary outcome)

LUS demonstrated a clear gradient with clinical severity and downstream care needs. Median LUS score increased stepwise across clinical severity categories, indicating progressively greater sonographic involvement with more severe illness (Figure 2). Children who required supplemental oxygen had substantially higher LUS scores than those who did not, and LUS scores were also higher among children requiring PICU admission, supporting the role of LUS as an early bedside tool for severity phenotyping. Consistent with this, higher LUS scores were associated with longer hospital stay.

For prediction of oxygen requirement, the LUS score showed good discrimination with an area under the ROC curve (AUC) of 0.766 (Figure 1). In an adjusted model, LUS score remained an independent predictor of oxygen requirement (aOR 1.17 per 1-point increase, 95% CI 1.09–1.26), indicating a clinically meaningful rise in risk with increasing sonographic burden.

Figure 1. Receiver operating characteristic (ROC) curve of lung ultrasound (LUS) score for predicting oxygen requirement in children with suspected pneumonia

Table 4. Distribution of LUS features among examinations classified as LUS-positive for pneumonia (n=95)

ROC curve showing the ability of the LUS severity score (0–24) measured at presentation to discriminate children who required supplemental oxygen during care. The area under the curve (AUC) was 0.77, indicating good discrimination.

Figure2.LUS Score Distribution by Clinical Severity

Lung ultrasound (LUS) score by clinical severity category. Box-and-whisker plot showing the distribution of LUS severity scores (0–24) across mild, moderate, and severe presentations; higher median scores and wider spread are observed with increasing clinical severity.

Limitations

This was a single-centre study conducted over 6 months, which may limit generalizability. The composite reference standard (CXR + clinical diagnosis) can introduce incorporation/verification bias, and CXR itself is an imperfect comparator. LUS is operator-dependent; interobserver variability and the absence of formal blinded dual-reading may have influenced estimates. Finally, the sample size limited precision for subgroup analyses and for evaluating less frequent outcomes (e.g., PICU admission).

In this prospective diagnostic accuracy study, lung ultrasound demonstrated high sensitivity (94.4%) and good overall accuracy (90.0%) for diagnosing paediatric pneumonia compared with a CXR plus clinical composite reference standard. LUS findings and scores also correlated with clinical severity and oxygen requirement,

supporting its utility not only for early diagnosis but also for bedside risk stratification. Incorporating LUS into routine paediatric evaluation—alongside standardized training and structured reporting—may reduce reliance on radiography and facilitate timely management in resource-constrained settings.