Common Radiological Features on Chest X-Rays of Infants With Bronchiolitis: Do They Support Management?

Abstract

Bronchiolitis is a viral respiratory tract infection among infants commonly caused by respiratory syncytial virus (RSV) and is a leading cause of hospital admission among infants.

The United Kingdom (UK) guidance recommends against ‘routinely performing chest x-rays (CXRs) in children with bronchiolitis as changes on chest radiographs can mimic pneumonia’ but should be considered if intensive care is being proposed. The quality of evidence supporting this statement is low [1]. Despite this, CXRs are regularly used in the assessment of bronchiolitis [2]. Rius-Peris et al. found fewer than 10% of CXRs showed radiological evidence of complicated bronchiolitis while Schuh et al. found fewer than 1% of CXRs yield an alternative diagnosis [3, 4].

Our retrospective study investigated the rate of radiographically defined complicated bronchiolitis. It also investigated the rate of alternative radiographic diagnoses among infants with bronchiolitis admitted to a single UK paediatric hospital over 9 years.

We retrospectively reviewed health records of infants with bronchiolitis assessed at the Children's Hospital for Wales between 1 October 2015 and 31 March 2023. The details of our data collection strategy can be found in our previously published work [5].

All CXRs were reported by a consultant paediatric radiologist at the time of imaging. Researchers classified reports as ‘benign’ or ‘pathological’, this was performed by a consultant paediatric radiologist (AE) if the report was ambiguous.

Reports mentioning at least one of the following terms: normal, hyperinflation, perihilar infiltrates, diffuse interstitial markings, bronchial wall thickening or linear atelectasis; meant the CXR was classified as ‘benign’ and deemed ‘simple bronchiolitis’. CXRs were classified as ‘pathological’ and therefore ‘complicated bronchiolitis’ if the report mentioned: pulmonary infiltrates, collapse, consolidation, airspace opacification, nonlinear atelectasis, pleural effusion, pneumothorax, cardiomegaly or foreign body [3].

Data were analysed using IBM SPSS Statistics(v29.0.2). Chi-squared testing was performed to compare differences between groups. A p value of < 0.05 was chosen to show statistical significance.

Over 9 years, 4072 patient episodes for bronchiolitis were included, of which 59% were male. The mean age was 4.96 months, 1997 (49%) of episodes resulted in a ward admission, 234 (6%) were admitted to PICU. RSV (53%) and rhinovirus (25%) were the most common viruses. Nasogastric feeding was used in 895 episodes (22%) and intravenous hydration 191 (5%). High-flow nasal cannula (HFNC) was used in 483 (12%) episodes, continuous positive airway pressure (CPAP) in 146 (4%) and 118 (3%) required invasive ventilation. Antibiotics were prescribed in 408 (10%) of episodes.

A total of 431 patient episodes had a CXR (11%), of which 248 (58%) were pathological and 183 were benign (42%). Those infants with a pathological CXR were significantly more likely to be younger than 3 months, admitted for over 48 h, have RSV or receive antibiotic therapy, see Table 1.

Of the 248 pathological CXRs, 239 (93%) had consolidation or airspace shadowing and 88 (36%) showed collapse or nonlinear atelectasis. Pleural effusion was identified in 11 (4%) episodes, pneumothorax in 7 (3%), cardiomegaly in 1 (0.4%), no CXR showed a foreign body but 85 had multiple abnormalities.

Antibiotics were prescribed in 408 (10%) of patient episodes, of these 228 (56%) had undergone a CXR. Patients with pathological CXRs were given antibiotics in 63% of episodes. All pathological findings, except for pleural effusion and cardiomegaly were associated with increased antibiotic use.

Our retrospective analysis of infants with bronchiolitis revealed 10% of episodes involved a CXR. Although our CXR rate is similar to other studies, we found a much higher percentage of pathological CXRs (almost 60%) compared with previous research which found < 10% to be pathological [3-5]. One reason for this may be researchers in Rius-Peris et al. [3] were unblinded to the bronchiolitis diagnosis and therefore more likely to designate CXRs as benign, compared with our radiologists who received limited clinical information and so were less prone to confirmatory bias. Further, Schuh et al. [4] only recruited patients with non-severe bronchiolitis which could have lowered the rate of pathological CXRs.

We found an antibiotic prescription rate of 11% which is similar to previous studies, however, only 64% of infants with a pathological CXR were prescribed antibiotics [2]. This suggests the CXR is interpreted alongside the clinical assessment when deciding whether to prescribe antibiotics. Therefore, it is likely a significant proportion of infants with pathological CXRs were deemed low-risk of bacterial infection.

There are several limitations to this study; first, it is only a single-centre study. Second, the retrospective data collection relied on electronic discharge summaries accurately detailing the medical interventions. Also, we were unable to collect data on the clinical characteristics of patients such as vital signs, work of breathing or blood tests, which may have helped to distinguish between bronchiolitis and bacterial pneumonia. Without these data, it is difficult to determine the number of patients with bronchiolitis as opposed to those with a true bacterial pneumonia.

Our study demonstrates an appropriately low use of both antibiotics and CXRs. However, we found a high proportion of infants who received a CXR had pathological changes, but this did not always result in antibiotic administration.

We conclude CXRs should not routinely be used in the assessment of bronchiolitis but instead may assist clinical decision-making for infants admitted with bronchiolitis where there is a clinical deterioration or a prolonged admission.

C. Towriss: conceptualization, methodology, data curation, formal analysis, writing – original draft, investigation, project administration, validation. A. Evans: conceptualization, investigation, writing – review and editing, validation, methodology, formal analysis, project administration. T. A. Betts: methodology, data curation, project administration, writing – review and editing, investigation. A. E. Darby: methodology, project administration, data curation, writing – review and editing, investigation. M. Edwards: conceptualization, methodology, data curation, investigation, validation, supervision, visualization, project administration, writing – original draft, writing – review and editing.

The authors declare no conflicts of interest.