The careful characterization of children with SA is considered an important step towards improving the knowledge of this small but very challenging group of patients [2, 5]. This is one of the few studies describing the distinguishing features of SA in children and adolescents in Europe. Hospital admissions for asthma, emergency-department visits during the past year, symptoms triggered by physical activity, lower spirometric values, and worse HR-QoL, but not well-being index, were differentiating features of SA versus NSPA. Current signs and symptoms possibly associated with atopy (i.e. rhinoconjunctivitis and eczema), eosinophil count, total IgE levels, and FENO values were similar in children with SA and NSPA, while lifetime sensitization to food allergens was an independent factor associated with SA. Home environment was similar in both groups, even though children with SA had a borderline higher prevalence of smoking parents.
SA in children is a challenging disorder with significant public health implications [1, 2, 19]. Unsurprisingly, in this study the occurrence of hospital admissions and emergency-department visits for asthma was a discriminating feature of SA versus NSPA. Nevertheless, 20 % of patients with SA had never been hospitalized, 71 % and 54 % were not hospitalized or admitted to the emergency department during the year preceding the study, and none had been admitted to intensive care units during the previous year. These findings are consistent with an earlier report from Sweden , where criteria for patient inclusion were similar to ours, while children enrolled in 5 specialized USA centers participating in the Severe Asthma Research Program  and those recruited in 12 specialized French centers at the beginning of Omalizumab treatment  had more severe exacerbations or at least required a larger utilization of health care services. Furthermore, most of our patients with SA regularly play a sport and are as physically active as their peers with NSPA.
Previous studies have highlighted an association between increasing asthma severity in children and both reduced HR-QoL and parents’ work attendance [10, 21, 22]. Asthma symptoms not only affect children physically, but also impair them and their families socially and emotionally [10, 21, 22]. In line with previous observations, this study demonstrates that poor HR-QoL strongly discriminates SA from NSPA. Interestingly, in our study girls with SA reported a significantly better HR-QoL than boys. Such gender distinction could reflect different psychological responses to limitations imposed by asthma (e.g. on physical activity) rather than actual differences in the disease itself. Moreover, our patients with SA had similar Wellbeing Indexes to peers with NSPA, with no gender differences. Of note, both groups had a median index very close to the poor wellbeing threshold (set at 13), the cut-off value below which it is recommended to test the patient for depression . This finding has never been reported in children with SA and is consistent with the observation that the prevalence of anxiety and depressive disorders is significant among asthmatic patients, particularly adolescents [19, 23]. A more in-depth analysis of psychological aspects in children and adolescents with SA would be worthy also for identifying specific interventions that could help reduce asthma morbidity.
According to some [8, 10] but not all  previous studies, most of our SA patients were sensitized to aeroallergens, with no prevalence difference from the NSPA group. Nevertheless, we highlight the novel finding that lifetime sensitization to food allergens was more frequent in SA than in NSPA. Although children with asthma show a strikingly high prevalence of food sensitization [24, 25], the majority have no clinical food allergies , which is consistent with the presence of current food allergies in only a third of our children sensitized to food. It is well known that children with food sensitization have increased asthma morbidity, with a higher hospitalization frequency and greater need for steroid medications , and even if they have developed tolerance to food allergens by school-age, previous sensitization still represents a risk factor for later asthma development [27, 28]. Our results confirm and extend these findings by demonstrating that lifetime sensitization to food allergens is an independent risk factor for SA.
Reduction in FEV1 is often used to define childhood asthma severity in treatment guidelines  and clinical studies [7, 9, 29]. Interestingly, while our SA patients showed worse spirometric measures than their NSPA peers, more than half had a normal FEV1, indicating that a reduction in FEV1 is an insensitive measure of SA. Indeed, contrary to adults, spirometry may be a poor predictor of asthma severity in children, and previous studies examining the relationship between FEV1 and the childhood asthma severity level have demonstrated very weak correlations between lung function and asthmatic symptoms [19, 29, 30]. FENO levels were similar in both groups, in line with some previous studies showing no significant increase in FENO values in SA children [9, 21, 29]. As all our subjects were on regular steroids maintenance therapy, the lack of any difference in FENO between SA and NSPA may be explained by the anti-inflammatory effects of the treatment.
Unlike previous studies [8–10], a number of comorbidities (i.e. rhinoconjunctivitis, obesity, and symptoms of sinusitis and of gastroesophageal reflux), family characteristics (namely, parental asthma and education), and environmental exposures (i.e. dampness at home and exposure to heavy traffic) were similar in SA versus NSPA patients. Exposure to smoking at home was indeed more prevalent in patients with SA, although the difference did not reach statistical significance possibly because of the low sample size.
We used a national online web-based system to collect a large number of personal, family and environmental data from children with SA and NSPA enrolled in various Italian centers. The main strengths of this system are that it allows for collecting huge amounts of longitudinal data, and ideally enables inclusion of foreign patients for larger international studies. Another strength of our study is that, unlike previous reports [6–8, 10], we included only children with refractory asthma or in whom treatment of comorbidities has been addressed as per the ERS/ATS Guidelines definition .
This study has limitations. Firstly, the study was not designed to assess the prevalence of SA in Italy, which would have been difficult to achieve considering the extension of the country. Secondly, the number of patients recruited was small and this prevented us from drawing definite conclusions on the lack of difference in a few variables between patients with SA and NSPA. However, our sample size is similar to that of previous European studies [8–10]. Finally, exposure to smoke and adherence to prescribed medication were assessed via self-reporting. Nonetheless, the lack of any adherence differences between SA and NSPA suggests that this shortcoming did not have much impact on our findings.