Acute bronchiolitis (AB) is the most common lower airway infection in infants, with an annual incidence of 10% in infants under 12 months of age and a hospital admission rate in Spain of between 2 and 5% of all affected patients.1,2 During the epidemic periods, AB constitutes an important healthcare burden in both the primary care and hospital settings.3 Overall, AB is estimated to be the most frequent cause of admission in infants less than one year of age.4 During the 1980s and 1990s, hospital admissions due to AB increased 2.4-fold in the United States,5 with no evidence of increased mortality. At that time, it was suggested that some of these admissions might be unnecessary and could be attributable to other factors.6 Later, in the period between 2000 and 2009, admissions due to AB in infants under 12 months of age began to decrease slightly, although the hospitalisation costs increased by 34%, at the expense of the more serious cases requiring admission to intensive care and mechanical ventilation.4

Adequate assessment of the clinical condition of a patient with AB is of great importance for the paediatrician, since it constitutes the basis of the decision-making process. Severity scales, applied objectively and with rigour, should prove useful in evaluating the clinical course of the patients and the efficacy of the prescribed treatments. While such scales are routinely used, they are supported by observational studies with limitations; as a result, no concrete scale is recommended in the clinical practice guides.7–9

One of the most widely-used scales in our setting, the Wood–Downes–Ferrés score (WDF), was designed to assess respiratory failure in patients with severe asthma.10,11 However, a Spanish group has recently proposed a new scale, known as the Acute Bronchiolitis Severity Scale (Escala de Severidad de la Bronquiolitis Aguda, ESBA), based on different clinical parameters specific of AB and which has been evaluated in infants under one year of age, with good interobserver agreement.12

The main objective of this study is to determine whether the ESBA shows a better correlation to the severity of AB than the WDF. As secondary objectives, an analysis is made of the mean stay due to bronchiolitis, admission to the Intensive Care Unit (ICU), the need for ventilation, and the incidence of respiratory acidosis.

The study included a total of 201 patients, of whom 134 (66.7%) were males. Eight infants were excluded from the study: one male with Down's syndrome, three patients whose informed consent was not obtained, three premature infants with a gestational age of <35 weeks, and one infant with clinical manifestations consistent with serious systemic bacterial infection.

The epidemiological and clinical characteristics of the study sample are described in Table 1. The median patient age was 2.3 months (IQR=1.3–4.4), and the median maternal age was 31 years (IQR=26.7–35). The mean duration of the disease at the time of admission was 2.3 SD2.6 days.

Thirteen patients suffered bronchiolitis considered to be severe according to the variable severe condition. The mean WDF score was 5.45 SD1.69 points, versus 5.07±2.18 points in the case of the ESBA. Three patients in our sample (1.5%) scored above 9 points on the ESBA, and 22 (10.9%) scored above 7 points on the WDF. The mean patient stay was 4.30 SD 2.14 days. The results of the bivariate analysis are shown in Table 2.

According to the defining criteria used (see Material and methods), a total of 13 cases were regarded as corresponding to severe AB. The median age of these patients was 1.1 months (IQR=0.4–1.6), which was significantly younger than the median age of the rest of the patients in the sample. In these infants with severe bronchiolitis the WDF and ESBA scores were significantly higher than in the rest of the patients. The mean of WDF score for non-severe bronchiolitis was 5.32 SD1.60 and for severe 7.31±1.88, p<0.01. For the ESBA score: non-severe bronchiolitis was 4.91 SD2.11 and for severe 7.31SD 1.88, p<0.01.

We measured the agreement among scales creating a matrix in which the columns represent the different levels of severity (mild, moderate, severe) as Table 2 shows. We obtained a weighted kappa coefficient of −0.17.

Tables 3 and 4 show the data corresponding to the sensitivity, specificity, PPV, NPV and likelihood ratios of the scales, according to the cut-off points in the original studies.

Comparisons were made of the ROC curves of each scale in order to determine which instrument is more accurate in assessing severe bronchiolitis. The results obtained are presented in Fig. 1.

Figure 1.

Comparison of the ROC curves of the WDF and ESBA scales.

(0.05MB).

The values of the AUC corresponding to the two scales were 0.79 (IC 95% 0.73–0.85) for the WDF score; and 0.82 (IC 95% 0.75–0.87) for ESBA. The difference between the two areas was 0.02 (95%CI: 0.01–0.15), which proved non-significant (p=0.72).

On considering our study sample, we could establish new cut-off points for the scales that were able to more precisely assess severe bronchiolitis. In this regard, we selected cut-off points with an optimum relationship between sensitivity and specificity, afforded by the Youden index (J). The data are shown in Table 5.

Accordingly, in the sample of patients admitted to hospital due to AB in our study, the cut-off points offering the best relationship between Se and Sp were found to be lower than those proposed in the original studies. On lowering the threshold for defining bronchiolitis as severe (5 points in the case of WDF and 6 points in the case of ESBA), the scales were able to classify a larger number of patients as presenting severe bronchiolitis.

The WD scale in principle does not seem to be a good tool for the clinical evaluation of AB, since it was designed and published in 1972 for the assessment of respiratory failure in only 18 paediatric patients with asthma crisis.10 Valuable parameters such as respiratory frequency or heart rate stratified according to age, the presence of crackles or oxygen saturation were not contemplated in the mentioned publication. In the present study, the WDF scale yielded a high false negative rate with the original cut-off points, and therefore does not seem reliable for detecting severe bronchiolitis. On applying the cut-off points obtained from the sample, the WDF scale was seen to be able to detect more seriously ill patients, increasing the absolute number of true positive results.

On the other hand, the ESBA12 offered the opportunity of working with an instrument exclusively designed for application to bronchiolitis, with good interobserver agreement (kappa index 0.682), employed in patients of our geographical setting, admitted to hospital and under one year of age. This scale contemplated the evaluation of signs such as crackles and wheezing (rated according to extent), and age stratification of respiratory frequency and heart rate – these being very objective measures. However, on analysing our large sample of infants with the cut-off points offered by the authors of the instrument, we found that the ESBA scale did not classify as severe any of those cases of bronchiolitis requiring non-invasive ventilation, transfer to the ICU or respiratory acidosis. With regard to the inclusion criteria, their patients did not differ very much from our own sample of infants, although the Se and Sp values published by the investigators did not coincide with our own values. The criteria used for admitting patients to the paediatric ICU or for deciding their destination may have been different from those of our own study. Nevertheless, the study of Ramos et al. lacks true validation of the scale with a sufficient sample of patients, classification based on the ESBA scale and the measurement of Se, Sp and an ROC curve with the different cut-off points. In this respect, on performing that analysis with our study sample, the ESBA scale yielded Se=84.6% and Sp=78.7, which differ considerably from the values reported by the authors of the original study. A very interesting finding was the recording of LR+=3.98, which indicates that an ESBA score of >6 obtained in a hospitalised patient with bronchiolitis implies an almost four-fold increase in the pre-test probability that the case of bronchiolitis is truly severe. Considering a pre-test probability of 12%, the post-test probability would be about 47.7% with an ESBA score of >6. On analysing the behaviour of the WDF scale, LR+ was even lower, with only a two-fold increase in the probability that the case of bronchiolitis is severe in the presence of a score of >5. Therefore, with the same prevalence of severe illness among the admitted patients, the post-test probability is about 24% if the WDF score is >5.

Both scales yielded low positive predictive values and also low LR+ and LR− values; as a result, neither instrument appears to be valid for discriminating those infants in our sample of patients that suffer truly severe bronchiolitis.

Although the positive and negative predictive values of the ESBA scale were higher than in the case of the WDF scale, the areas under the ROC curve were not significantly different; we therefore cannot conclude that one instrument is better than the other in differentiating severe cases of the disease.

In the context of the hospitalisation of these patients, it seems advisable to ensure the diagnosis of all seriously ill infants, since anticipating respiratory failure before it manifests is crucial. It is therefore worth ensuring higher Se values. For this reason, in our sample the cut-off points obtained with the ROC curves allow the negative results of the WDF and ESBA scales to more reliably discard severe cases.

In our study we only recorded 13 cases of severe bronchiolitis (6.4% of the total) – this being consistent with the data published in the literature, where the estimated paediatric ICU admission rate is 5–16%.8 However, our recorded transfer rate was 2.5%, which is lower than the percentages found in the literature. A number of factors may explain this observation, such as the exclusion of patients with background disorders capable of causing eventual bronchiolitis to be more severe, or the fact that three of the hospitals were equipped with non-invasive ventilation techniques such as CPAP or HFO therapy – which are useful for the management of severe bronchiolitis.15,16 Likewise, some cases of bronchiolitis could not be recruited and were lost, since the patients were transferred before they could be enrolled or before informed consent could be obtained from the parents.

The low number of patients with a severe condition generates a high imprecision estimation of the scales’ sensitivity and questions the results after modifying the cut-off points, even though they improve the estimated sensitivities.

The weaknesses of our study include the existence of multiple observers involved in patient assessment, and who did not receive specific instruction other than a training session about the study. In this regard, in relation to the degree of interobserver agreement, differences among studies are noted, although on measuring a series of clinical signs in patients with respiratory problems, the observed degree of agreement is low.17–19 It therefore seems advisable to use the largest possible number of objective variables that are less likely to differ between measurements.

One of the parameters of the ESBA scale – the inspiration/expiration ratio – appears to be particularly difficult to evaluate. The original study on this scale already identified this ratio as the weakest parameter in terms of interobserver agreement. In patients with high respiratory frequencies it can be difficult to establish whether inspiration and expiration are similar or whether one is prolonged with respect to the other. Furthermore, this ratio is not a vital parameter commonly considered in paediatric physical examination.8,9

Among the limitations of the study, results could change according to the gold standard used as a synonym for severity.

The low weighted kappa index found in the study shows the low concordance between measures found between scales.

Other investigations have attempted to validate and compare severity scales in application to bronchiolitis. A study carried out in the United States between 2007 and 2009 evaluated the suitability of two scales for assessing the severity of bronchiolitis using hospital admission and the duration of hospital stay as discrimination measures.20 These scales were the Respiratory Distress Assessment Instrument (RDAI) and The Children's Hospital of Wisconsin Respiratory Score (CHWRS), which yielded low areas under the ROC curves of 0.51 and 0.68, respectively. The study concluded that the RDAI is more limited than the CHWRS in defining the clinical condition of the patient at a concrete timepoint during hospital admission. In this regard it can be deduced that prior training is needed in order to establish reliable scoring, and that the scales can be useful for determining the efficacy of treatments or for assessing the time course referred to one same individual.

The RDAI scale has also been tested for predicting the risk of hospital admission and hospital stay in a multi-centre study involving patients with bronchiolitis, where oxygen saturation <94%, an RDAI score of >11 and a respiratory frequency of >60rpm were found to be the best predictors of hospital admission and prolonged stay in hospital.21

The scales used to score patient respiratory condition have been widely used to assess the efficacy of the different treatments used for bronchiolitis. However, few of these instruments have been specifically validated for this purpose. One of the best attempts to validate the modified Wood's Clinical Asthma Score (M-WCAS) for bronchiolitis was carried out by a group of investigators in a hospital between 2010 and 2011, with the assessment of usability, reproducibility and validity of the WDF scale in patients hospitalised because of bronchiolitis and under 24 months of age.22 This study effectively confirmed a correlation between high scores on the scale and admission to intensive care.

This same scale has been used to evaluate the efficacy of Heliox in patients admitted to the ICU, with satisfactory outcomes, since such treatment resulted in a reduction in the score of the instrument, a shortened ICU stay and a decrease in respiratory frequency and heart rate.23 However, none of these studies established comparisons with a scale exclusively designed for application to this disease.

On the other hand, bronchiolitis is a changing and dynamic disorder; point measurements therefore may be imprecise in assessing the final patient outcome, since they only reflect the moment of the disease in which the measurements were obtained. Measurements obtained in different moments could be useful to determine evolution changes in one patient, but not necessarily indicate the final course of the illness.

Most of the clinical data defined in the different clinical guidelines as parameters to be considered on deciding the admission of a patient with bronchiolitis are referred to young patient age, the background disease condition, the presence of apnoea and the degree of tachypnoea.9,24–27 In our study the bivariate analysis found these parameters to be consistent with the descriptions found in the literature, although our results also underscored the relevance of fever and patient rejection of food.

Recently, NICE guidelines for bronchiolitis have been published and recommend further research into the effectiveness of Paediatric Early Warning Score (PEWS) in predicting deterioration for infants with bronchiolitis. The Guideline Development Group encourages assessing scores that could predict bronchiolitis deterioration.28

In conclusion, clinical scales do not seem to be precise in assessing the severity of patients with bronchiolitis. Nevertheless, in our sample the ESBA scale with new cut-off points appears to be superior to the WDF scale in classifying patients as being in a severe condition, although the differences failed to reach statistical significance.

New validation studies are needed to confirm whether our new cut-off points better classify severe hospitalised infants with AB.

The authors have no conflict of interest to declare.