Acute bronchiolitis (AB) is the most frequent reason for hospital admission in infants, constituting a significant health care and economic burden. Although in most cases AB has a benign course, a subset of cases progress to severe disease requiring admission to the paediatric intensive care unit (PICU) and respiratory support, which are associated with a considerable morbidity and mortality. It is often difficult to predict which patients will have this adverse progression, which results in the need for prolonged inpatient monitoring in certain at-risk groups. Clinical severity scales are widely used and can be helpful to make an objective and reproducible assessment of the course of the disease and the response to treatment. However, there is insufficient evidence of their usefulness in predicting the outcomes of mild-to-moderate AB.1–3 The availability of a point-of-care tool allowing us to predict which patients may have an unfavourable course would be very useful for the purpose of preventing adverse events, intensifying monitoring and optimising the use of resources (transfer to hospitals with a PICU, early initiation of respiratory support, etc.).

Lung ultrasound allows the assessment of the extent of pleuropulmonary involvement at the bedside easily and readily and without adverse effects. It has been proven to be useful and in many instances superior to chest radiography in the diagnosis of transient tachypnoea of the newborn, acute respiratory distress syndrome (ARDS), pneumonia, pleural effusion and pneumothorax.4–11 Despite the extensive evidence gathered on other acute pulmonary diseases, few studies have assessed the usefulness of lung ultrasound in AB. Studies published to date agree that the extent of lung lesions is associated with the severity of AB.12–14 It would thus be sound to hypothesise that the extent of lung involvement could be a valuable predictor of the clinical course of AB, as is the case in other diseases such as ARDS, in which it is of proven utility in predicting mortality or response to treatments such as prone ventilation or recruitment maneouvres.15–20 However, no study to date has assessed the usefulness of early lung ultrasound in predicting the clinical course of AB.

We had 2 main objectives in conducting this study: on the one hand, to analyse the association between lung ultrasound findings and severity in mild-to-moderate AB as assessed by 2 clinical severity scores used in our region, and on the other to analyse whether early lung ultrasound findings are associated with clinical outcomes in patients presenting to hospital with mild-to-moderate AB.

We included 59 patients with mild-to-moderate AB with a median age of 90 days (IQR, 30–270) and a median weight of 6.1kg (IQR, 4.8–7.9). Forty-four patients required hospital admission for a median length of stay of 5 days (IQR, 3–7.5), and 37 (84%) required oxygen therapy for a median of 3 days (IQR, 2–6). The lung ultrasound examination was performed at the time of admission to the paediatric ward in 41 patients, and in the emergency department in 18. Four patients (6.7%) were admitted to the PICU, and none required invasive mechanical ventilation. Table 1 summarises the characteristics of the patients, and Appendix B presents supplementary material.

The interrater agreement between the paediatricians that performed and interpreted the ultrasound examinations and a paediatrician experienced in lung ultrasound was substantial for the assessment of PAs (observer A kappa, 0.851; observer B kappa, 0.841) and of AIS (observer A kappa, 0.836; observer B kappa, 0.877) and moderate for the assessment of SCs (observer A kappa, 0.620; observer B kappa, 0.695). The interrater intraclass correlation coefficient (absolute agreement between individual measurements) for the total score in the LUS-Sc was 0.917 (95% CI, 0.854–0.956).

The ultrasound examinations were performed a median of 3 days (IQR, 2–4) from the onset of symptoms of AB. None of the patients were receiving respiratory support before the examination (except for conventional support with supplemental oxygen delivered via nasal prongs). The lung zones affected most frequently were the posterior paravertebral and the linea scapularis zones (Table 2). The median score in the LUS-Sc was 8 points (IQR, 5–13) in the overall sample, 6 points (IQR, 2–8) in the subset of patients discharged home from the emergency department, 9 points (IQR, 5–13.7) in the subset admitted to the paediatric ward and 17 points (IQR, 14.5–18) in the subset admitted to the PICU (P<.001) (Fig. 2). We also found statistically significant differences in the degree of AIS (discharge home, 0 [IQR, 0–2]; admission to ward, 4.5 [IQR, 3–6]; transfer to PICU, 5.5 [IQR, 5–6]; P<.001) and of SCs (discharge home, 0 [IQR, 0–2]; admission to ward, 2 [IQR, 0–3.7]; transfer to PICU, 5 [IQR, 4.2–5.7]; P=.004), but not in the severity of PAs. We did not find significant differences in severity between infants aged less than 3 months and older infants: MWDF score of 5 (IQR, 4–6) vs 4.5 (IQR, 4–7) (P=.459), and HSJD score of 6 (IQR, 4–7) vs 5.5 (IQR, 4–7) (P=.662). However, the LUS-Sc scores were higher in infants aged less than 3 months: 9 (IQR, 6–13) vs 7.5 (IQR, 2–13) (P=.069), with a higher AIS subscore (3 [IQR, 1–5] vs 1 [IQR, 0–4.5]; P=.127), and more SCs (4.5 [IQR, 3–6] vs 2 [IQR, 0–4]; P=.106), although these differences were not statistically significant.

Figure 2.

Boxplots of the distribution of the values of the lung ultrasound score by course of disease category. The bars indicate the comparison of subset pairs (discharge home vs admission to ward and admission to ward vs admission to PICU) in the multiple comparison analysis. P=.001 (Kruskal–Wallis test).

(0.07MB).

We found a statistically significant correlation between the LUS-Sc scores, especially the degree of AIS and SCs, and all clinical parameters except the respiratory rate (Table 3). The strongest correlations were between the total LUS-Sc score and the scores in the MWDF (Spearman rho, 0.504, P<.001) and the HSJD scales (Spearman rho, 0.518; P<.001). Fig. 3 presents the scatterplots for these correlations.

Figure 3.

Scatterplots showing the correlation between the LUS-Sc and the clinical severity scores.

(0.09MB).

For every 5-point increase in the LUS-Sc, there were increases in the probability of admission to the PICU (OR, 2.5 [95% CI, 1.1–5.9]; P=.035), in length of stay by 1.2 days (95% CI, 0.55–1.86; P<.001) and in duration of oxygen therapy by 0.87 days (95%CI, 0.26–1.48; P=.006). These associations remained significant after adjusting for sex, weight, age, duration of AB symptoms and the presence of risk factors for severe AB (Table 4).

In this preliminary study, we found that lung ultrasound findings were correlated to symptom severity in mild-to-moderate AB and had a certain value in the prediction of the subsequent course of the disease. However, our findings could not prove that they added to the information provided by clinical severity scores for the purpose of anticipating an unfavourable progression.

Very few studies have used lung ultrasound in the context of AB, despite it being a method widely used in the assessment of patients with acute pulmonary diseases such as pneumonia, pleural effusion or lung oedema, in which its diagnostic yield exceeds that of plain X-rays while avoiding exposure to ionising radiation.5–7,12

In 2011, in a study comparing ultrasound and X-ray findings in patients with AB, Caiulo et al. found that the extent of SCs and AIS was associated with disease severity, and that the number and grading of SCs paralleled the scores in the Downes scale.12 In 2015, Basile et al. assessed the usefulness of ultrasound abnormalities in patients with AB, in terms of the presence and extent of SCs and the degree of AIS, for identification of patients requiring oxygen therapy.13 Our results were consistent with those of these studies, and we have also found evidence that abnormal ultrasound features exhibit a similar linear correlation with 2 validated clinical severity scores.26

We ought to mention that our study did not find a linear correlation between respiratory rate (which is widely considered a sensitive marker of respiratory involvement) and lung ultrasound findings. We do not have an obvious explanation for this result. It is possible that the greater variability and the lack of specificity of respiratory rate (due to its association with fever, nasal obstruction, agitation in the patient, etc.) combined with the small sample size contributed to this unexpected result.

Ultrasound findings such as PAs, SCs or the presence of AIS are not specific for AB and overlap significantly with the findings in other diseases with diffuse interstitial involvement, such as bronchopulmonary dysplasia or ARDS.6,27 However, our study and others in the previous literature are a proof of concept and suggest that the degree of sonographic abnormality reflects the severity of the pathophysiological process underlying AB to some extent. We know that the most severe cases of AB are those with presentations similar to that of ARDS, with severe interstitial involvement and a decrease in lung compliance as opposed to a pattern of obstruction and air trapping. It is possible that the early detection of reduced aeration by lung ultrasound could be useful in predicting the course of disease. In our study, we observed that patients that ended up requiring admission to PICU had more extensive involvement in the early ultrasound examination compared to patients discharged home or admitted exclusively to the paediatric ward, and also found that ultrasound was of some value in identifying patients that eventually required longer lengths of stay and more days of oxygen therapy. However, our statistical models did not show that these associations remained significant after adjusting for the clinical severity scores. This is not sufficient reason to definitely rule out the use of ultrasound in AB, but should rather encourage further research on its potential role in this disease. For instance, we observed that in patients with a similar severity determined on the basis of clinical features, ultrasound revealed a greater compromise of lung aeration in infants aged less than 3 months compared to older infants, which could contribute to the more severe course of disease observed in the younger set, in whom alveolar and interstitial involvement may be more prevalent, contrary to the predominance of obstruction and air trapping in older infants.

All current clinical practice guidelines agree that routine use of diagnostic tests is not recommended in AB.1,2 However, ultrasound examination could be useful in cases where despite performing the routine clinical assessment the need for inpatient observation remains unclear. Sonographic evidence of extensive involvement should alert us to the possibility of progression to severe disease. Thus, in our study, a LUS-Sc score of more than 13.5 points had a sensitivity of 100% and a specificity of 82% in predicting admission to the PICU.

We found a moderate correlation between the clinical severity scores and the degree of lung involvement evinced by early ultrasound examination in infants with mild-to-moderate AB. The most useful sonographic findings were those concerning the degree of AIS and SC. Lung ultrasound findings may be associated with clinical course and be useful to identify patients at risk of progression to severe disease.

XIX Clinical and Epidemiological Research Grant Competition—2016, Fundación Ernesto Sánchez Villares, Sociedad de Pediatría de Asturias, Cantabria y Castilla y León. Project 02/2016: Utilidad de la ecografía pulmonar en la valoración de la gravedad y pronóstico de la bronquiolitis aguda. Principal investigator: Elia Zoido Garrote. Grant of 3000 euro.

The authors have no conflicts of interest to declare.