Introduction

Inhaled corticosteroids (ICS) are powerful anti-inflammatory and immunosuppressive drugs that inhibit the expression of various pro-inflammatory genes. They have shown clear beneficial effects in patients with eosinophilic airway diseases, including asthma and eosinophilic chronic obstructive pulmonary disease (COPD).1 2

Bronchiectasis, a chronic lung disease characterised by abnormal widening of the airways and excessive sputum production, is driven in part by excessive airway inflammation, which is predominantly neutrophilic.3 4 ICS are widely used in patients with bronchiectasis, but small randomised trials have not shown clear benefits, with some observational studies suggesting potentially worse outcomes in ICS users with bronchiectasis.5–9 Patients with bronchiectasis are at high risk of bacterial infection and neutrophilic inflammation is known to be associated with resistance to the anti-inflammatory effects of corticosteroids.3 10 Given these considerations, current bronchiectasis guidelines advise against the use of ICS for the treatment of bronchiectasis, except in those with co-existing asthma, allergic bronchopulmonary aspergillosis (ABPA) and a subset of patients with eosinophil-dominant COPD and/or frequent exacerbations.11 However, it is important to note that these current guidelines are based on limited evidence. Despite these guideline recommendations, ICS are still used frequently, including in patients without a history of asthma, COPD or ABPA. As such, determining the clinical impact of ICS on bronchiectasis patient outcomes is of great clinical relevance. Determining the frequency of ICS use and the clinical features associated with it may aid in ensuring greater adherence to guidelines and reduce variation in care.

Recent evidence suggests the existence of an eosinophil-driven endotype of bronchiectasis accounting for ~20% of all bronchiectasis cases after asthma has been excluded.12 13 As blood or sputum eosinophilia is linked to ICS response in other eosinophilic respiratory diseases, this raises the question of whether ICS may prove clinically beneficial in treating this eosinophilic subset of patients with bronchiectasis. A post hoc analysis of a randomised controlled trial (RCT) of ICS in bronchiectasis found improved quality of life in patients with elevated blood eosinophil counts, however further evidence is required and most studies to date have included few patients.14

The European Bronchiectasis Registry (EMBARC) collects data from patients with bronchiectasis from >30 countries internationally, including the recording of treatment patterns.9 In this analysis, we investigated the prevalence of ICS use within a large, heterogeneous bronchiectasis cohort to understand the clinical features and outcomes associated with ICS use.

Methods

EMBARC is a prospective observational study including CT-confirmed patients with bronchiectasis from >30 countries worldwide. A detailed study protocol has been previously published alongside baseline characteristics of the study population.9 15

Data collection

Enrolment into EMBARC began in January 2015. Patients enrolled up to May 2022 were included for the present analysis. Patient data were collected annually using a standardised case report form incorporating clinical data, demographics, comorbidities, medications, aetiological testing, microbiology, radiology and lung function. The attending physician/medical team recorded bronchiectasis aetiology and the presence of co-existing asthma, COPD or ABPA. The results of sputum samples from stable state and exacerbation in the previous 12 months were used to identify bacterial infections. Radiological severity was assessed using the patient’s most recent CT scan, following the modified Reiff scoring system as previously described.16 Disease severity was evaluated using the Bronchiectasis Severity Index (BSI).17 Exacerbations were defined as use of antibiotics for acute respiratory symptoms and were recorded using patient history, hospital and prescription records.

Patient subgroups

ICS use is not recommended in the European Respiratory Society (ERS) guidelines for patients without a diagnosis of asthma, COPD or other diseases, such as ABPA.11 We therefore classified patients for analysis as either ‘bronchiectasis without indication for ICS’ or ‘patients with current indications for ICS’, based on the presence/absence of a physician-recorded diagnosis of asthma, COPD and/or ABPA.

‘Bronchiectasis without recommendation for ICS’ was defined by guidelines as a primary diagnosis of bronchiectasis in the absence of another airway disease where ICS use is recommended according to current treatment guidelines, that is, ABPA, asthma and/or COPD. ABPA was classified using the newly modified diagnostic guidelines introduced by the International Society for Human and Animal Mycology (ISHAM).18 Sensitivity analyses were performed by excluding all patients with comorbid asthma, COPD and/or ABPA, alongside those with missing test results for more than one of the ISHAM criteria given the inability to accurately assign ABPA status.

A subgroup analysis was performed to evaluate ICS use and outcomes in patients with documented peripheral blood eosinophilia. During the study period, EMBARC did not record absolute blood eosinophil counts but includes whether patients have levels above the local laboratory normal range (typically >400 cells/µL but >500 cells/µL in some cases). Individuals with missing blood eosinophil data were excluded from further analyses of response stratified by blood eosinophil counts.

Long-term clinical outcomes

Patients in EMBARC are followed up every 12 months for up to 5 years for assessment of long-term clinical outcomes.9 15 Statistical analysis of relevant end points takes into account the duration of follow-up, including loss to follow-up. Relevant clinical outcomes were survival, exacerbation frequency and frequency of hospitalisation due to severe exacerbation.

Statistical analysis

Summary data are presented as median with IQR unless otherwise stated. All hypothesis tests are two-sided with assumptions assessed graphically. Comparisons of continuous data between two groups were conducted using Mann-Whitney U test. Comparisons of more than two groups were performed by Kruskal-Wallis test. To compare differences in categorical data between groups, χ2 test was performed. Due to the very large sample size of the EMBARC, comparison of data (categorical or continuous) between groups may highlight small differences as highly statistically significant, which may not necessarily reflect clinical significance. Exacerbation frequency and frequency of severe exacerbations requiring hospitalisation over time were analysed using a negative binomial model, with time in study included as an offset. For exacerbation frequency and frequency of hospitalisation analyses, we calculated incident rate ratios (IRR) after adjusting for potential confounding effects of age, sex, geographical region, body mass index (BMI), diabetes, Pseudomonas aeruginosa infection and forced expiratory volume in 1 s (FEV1) by including these as covariates. Similarly, survival analysis measuring all-cause mortality was performed using Cox proportional hazards regression with adjusted HRs calculated after inclusion of age, sex, BMI, P. aeruginosa infection, FEV1, smoking status, geographical region and comorbidities as covariates; the proportional hazards assumption was verified using a time-dependent covariate in the Cox model. No adjustments were made for multiple comparisons, but interpretation of results takes into account the large sample size of the EMBARC, which may lead to significant p values for effect sizes that are not clinically relevant. All analyses used SPSS V.28.0.0.0 or GraphPad Prism V.9.4.0.

Results

19 324 patients were enrolled from 31 countries. Table 1 shows the patient characteristics of those who were prescribed ICS versus those who were not prescribed ICS at the time of enrolment. Of the 10 109 patients (52.3%) receiving ICS in the EMBARC, 89.5% were receiving ICS in combination with at least one bronchodilator.

Table 1

Detailed patient characteristics of individuals with bronchiectasis prescribed ICS versus those not prescribed ICS

Frequency of ICS use between countries

Figure 1 shows the percentage of ICS users per country across the 31 countries contributing to the EMBARC. Countries with the highest rates of ICS use included Pakistan (85%), North Macedonia (83%), Finland (73%) and India (63%), whereas Kyrgyzstan had the lowest rates of ICS use (17%). The absolute frequency of ICS use in each country is shown in online supplemental table 1.

Figure 1

Map showing the frequency of inhaled corticosteroids (ICS) use across different European and non-European countries. Figure created with BioRender (2024).

Table 1 indicates that ICS users appear to have a more severe bronchiectasis phenotype compared with those not using ICS, evidenced by increased radiological severity, a higher daily sputum volume with an increased need for regular airway clearance and an increased use of other therapies including long-term macrolides and inhaled/oral antibiotics. Providing further evidence for greater disease severity, figure 2 shows significant differences in BSI scores (p<0.0001) and FEV1% predicted (p<0.0001) between ICS users and non-users, with ICS users also having an increased frequency of frequent exacerbation (≥3 exacerbations) and severe exacerbations requiring hospitalisation in the 12 months prior to baseline.

Figure 2

The relationship between ICS use and bronchiectasis disease severity parameters including FEV1 (A) BSI (B) and proportion of patients experiencing exacerbation/hospitalisation in the previous year (C). Red=+ICS, blue=−ICS. Data shown as median. FEV1 are shown as percentage predicted. BSI, Bronchiectasis Severity Index; FEV1%p, forced expiratory volume in 1 s percentage predicted; ICS, inhaled corticosteroid. ****P<0.0001.

Table 1 also highlights that those with a history of COPD as well as asthma, ABPA and other classically eosinophil-driven respiratory comorbidities, such as nasal polyps and rhinosinusitis, are more likely to be prescribed ICS.

ICS use and microbiology

We observed significantly increased frequencies of various respiratory pathogens at baseline in ICS users compared with non-users, including P. aeruginosa (n=2004, 19.8% vs n=1292, 14.0%, p<0.001), Haemophilus influenzae (n=1633, 16.2% vs n=1247, 13.5%, p<0.001) and Streptococcus pneumoniae (n=726, 7.2% vs n=505, 5.5%, p<0.001) among others. Surprisingly, we observed a reduced frequency of non-tuberculosis mycobacterium (NTM) culture in ICS users compared with non-users (n=228, 2.3% vs n=348, 3.7%, p<0.001). Full microbiology results are shown in online supplemental table 2.

ICS use in those with bronchiectasis not recommended ICS within the EMBARC

We next identified a cohort of patients with bronchiectasis without a current indication for ICS (ie, those with a primary diagnosis of bronchiectasis in the absence of asthma, COPD and/or a known case of ABPA) and stratified this cohort by ICS use to determine the effect of ICS on individuals with bronchiectasis not currently recommended ICS treatment.11 Of the 19 324 enrolled patients, we observed 5801 (30.0%) patients with a history of asthma, 4899 (25.3%) with COPD and 608 (3.1%) patients with a known case of ABPA, as such 9609 individuals (49.7%) were excluded for further analysis. Table 2 shows the characteristics of the resulting population following exclusion.

Table 2

Detailed patient characteristics of individuals with bronchiectasis in the absence of asthma, COPD and/or ABPA receiving ICS versus those not receiving ICS

We found that 32.7% of the resulting cohort of patients with bronchiectasis without a current indication for ICS were receiving ICS therapy (table 2). Table 2 highlights similar findings to our previous analysis of the whole EMBARC cohort, highlighting ICS users as more severe compared with non-users. Sensitivity analysis excluding all patients with asthma, COPD, known ABPA and those with insufficient serological testing results for diagnosing ABPA, also showed similar results (online supplemental table 3).

Overall, frequencies of known bronchiectasis aetiologies and comorbidities remained relatively consistent between ICS users and non-users. However, an increased frequency of primary ciliary dyskinesia (PCD) (n=178, 5.6% vs n=196, 3.0%) and tuberculosis (TB) (n=363, 11.4% vs n=623, 9.5%) was observed among ICS users. Frequency of comorbid osteoporosis (n=366, 11.5% vs n=648, 9.9%, p=0.016), rhinosinusitis (n=664, 21.2% vs n=1062, 16.6%, p<0.001) and nasal polyps (n=193, 6.2% vs n=258, 4.0%, p<0.001) was also significantly higher among ICS users compared with non-users. In contrast, a significantly reduced frequency of comorbid liver disease (n=11, 0.3% vs n=45, 0.7%, p=0.030) and cancer (n=274, 8.6% vs n=678, 10.4%, p=0.005) was observed among ICS users. Full details on aetiology and comorbidity can be found in online supplemental table 4.

The impact of ICS on long-term bronchiectasis outcomes

First examining the entire bronchiectasis population, including patients with a diagnosis of asthma, COPD and/or ABPA, we investigated differences in exacerbation frequency, the frequency of hospitalisation due to severe exacerbation and mortality between ICS users and non-users during long-term follow-up. Long-term clinical outcomes were evaluated over a cumulative total of 26 213 years of patient follow-up (range 0–5 years per patient).

We observed a small but statistically significant increase in exacerbation frequency in ICS users in comparison with non-ICS users, irrespective of adjustment for potential confounders including age, sex, geographical region, BMI, diabetes, P. aeruginosa infection and FEV1 (IRR 1.09, 95% CI 1.07 to 1.11, p<0.001 for unadjusted vs IRR 1.10, 95% CI 1.08 to 1.13, p<0.001 for adjustment). Interestingly, following adjustment for the above confounders, a small but statistically significant reduction in hospitalisation frequency was observed in ICS users (IRR 0.93, 95% CI 0.88 to 0.98, p=0.008).

There were 1077 deaths (5.6% of all participants) during follow-up. Risk of all-cause mortality, prior to adjustment, was significantly increased among ICS users compared with non-users (HR 1.24, 95% CI 1.09 to 1.40, p<0.001), however this association was not significant following adjustment for potential confounders (age, sex, BMI, P. aeruginosa infection, FEV1, smoking status, geographical region and number of comorbidities) (HR 1.03, 95% CI 0.90 to 1.18, p=0.645).

Results were similar in patients without a current recommendation for ICS. Unadjusted analysis showed a statistically significant increase in exacerbation frequency in ICS users compared with non-users (dotted line) (IRR 1.31, 95% CI 1.21 to 1.43, p<0.001), an effect which persisted following adjustment for the relevant exacerbation-related confounders as previously described (IRR 1.19, 95% CI 1.10 to 1.30, p<0.001). ICS use also associated with increased hospitalisation frequency (IRR 1.24, 95% CI 1.09 to 1.42, p=0.001) but this effect was not significant following confounder adjustment (IRR 1.04, 95% CI 0.90 to 1.12, p=0.626).

There were 389 deaths during follow-up in this subset of patients (4.0% of patient subgroup). Both unadjusted and confounder adjusted analysis showed a significantly increased all-cause mortality risk among ICS users compared with non-users (HR 1.39, 95% CI 1.14 to 1.70, p=0.001 for unadjusted analysis vs HR 1.28, 95% CI 1.03 to 1.60, p=0.026 for adjusted analysis). All outcomes are summarised in figure 3, with the outcomes of the entire bronchiectasis population with stratified by the presence of asthma, COPD, ABPA or a combination of such respiratory comorbidities summarised in online supplemental figure 1.

Figure 3

Long-term outcomes, including mortality, hospitalisation for severe exacerbations and overall exacerbation frequency in patients with bronchiectasis overall, and patients with bronchiectasis without a current indication for ICS (ie, those with a primary diagnosis of bronchiectasis in the absence of asthma, COPD and ABPA) who are receiving ICS. Data are shown as IRR or HR (mortality) with 95% CIs. Data shown for frequency of exacerbation and frequency of hospitalisation have undergone adjustment for age, sex, BMI, geographical region, diabetes, Pseudomonas aeruginosa and FEV1. Data shown for mortality have undergone adjustment for age, sex, BMI, P. aeruginosa, smoking status, geographical region, number of comorbidities and FEV1. Dotted line represents ICS non-users. *P<0.05; **p<0.01; ***p<0.001. ABPA, allergic bronchopulmonary aspergillosis; BMI, body mass index; COPD, chronic obstructive pulmonary disease; FEV1, forced expiratory volume in 1 s; ICS, inhaled corticosteroids; IRR, incident rate ratio.

The impact of ICS use on those patients with bronchiectasis not currently recommended ICS and with evidence of peripheral blood eosinophilia

Blood eosinophil data were available for 4385 patients without a history of asthma, COPD or ABPA. 8.9% of individuals receiving ICS had evidence of eosinophilia. A similar distribution was seen among those not receiving ICS, where 5.9% of individuals had eosinophilia. Further details regarding patient characteristics related to blood eosinophil counts can be found in online supplemental table 5.

For the purposes of outcome analysis, patients were divided into three groups. The reference group consisted of patients with bronchiectasis without evidence of eosinophilic inflammation and not treated with ICS. We then investigated whether the presence of eosinophilic inflammation, the use of ICS or the combination of both were associated with different outcomes.

With regard to exacerbation frequency, ICS users with elevated blood eosinophils had a significantly reduced frequency of exacerbation (RR 0.70, 95% CI 0.59 to 0.84, p<0.001, figure 4A). We also observed that ICS non-users with elevated eosinophils had a modest, but non-significant, increase in exacerbation frequency compared with our reference group (RR 1.17, 95% CI 1.00 to 1.38, p=0.053, figure 4A).

Figure 4

The clinical impact of ICS on exacerbation frequency (A), hospitalisation frequency (B) and mortality risk (C) of those with bronchiectasis without a current indication for ICS use (ie, minus asthma, COPD and/or ABPA) with evidence of eosinophilic inflammation. Data are shown as IRR or HR (mortality) with 95% CIs. Reference line (x=1.0) represents the reference group, that is, individuals with bronchiectasis without a current indication for ICS with normal blood Eos levels who are not receiving ICS. Data shown for frequency of exacerbation and frequency of hospitalisation have undergone adjustment for age, sex, BMI, geographical region, diabetes, P. aeruginosa and bronchiectasis disease severity in the form of FEV1. Data shown for mortality have undergone adjustment for age, sex, BMI, P. aeruginosa, smoking status, geographical region, number of comorbidities and disease severity in the form of FEV1. ABPA, allergic bronchopulmonary aspergillosis; BMI, body mass index; COPD, chronic obstructive pulmonary disease; Eos, eosinophil; FEV1, forced expiratory volume in 1 s; ICS, inhaled corticosteroids; IRR, incident rate ratio.

We saw a similar trend with hospitalisation frequency, where ICS users with elevated blood eosinophils had a reduced frequency of hospitalisation compared with the reference group (RR 0.56, 95% CI 0.35 to 0.90, p=0.016, figure 4B).

There were 151 deaths during follow-up among this subset of patients. ICS non-users with elevated blood eosinophils had no significantly increased mortality risk compared with the reference group (HR 1.50, 95% CI 0.74 to 3.01, p=0.262, figure 4C). ICS users with elevated blood eosinophils also showed no significant difference in mortality risk (HR 0.62, 95% CI 0.23 to 2.41, p=0.619, figure 4C).

Discussion

For the majority of those with bronchiectasis, with the exception of those with asthma, eosinophilic COPD and/or ABPA, treatment with ICS is not currently recommended due to concerns over increased infections and other corticosteroid-related adverse events.11 In addition, previously conducted RCTs exploring ICS use in bronchiectasis reported little clinical benefit.19–21

Our data confirm that ICS are still widely used despite recommendations from the ERS, British Thoracic Society and other evidence-based guidelines for bronchiectasis.11 22 The headline data show that approximately half of all patients with bronchiectasis, and one-third of patients without a diagnosis of asthma, COPD or ABPA, use ICS. Our EMBARC data show that ICS is among the most widely used of all treatments for people with bronchiectasis. Understanding patterns of ICS use across the world and the potential clinical impact is therefore important. The recent description by Shoemark et al of an eosinophilic endotype of bronchiectasis that may theoretically benefit from ICS therapy,13 further emphasises the importance of understanding the role of ICS in bronchiectasis.

The results presented here, in over 19 000 patients with bronchiectasis, highlight those receiving ICS treatment appear more clinically severe compared with those not receiving ICS treatment. Likewise, we made similar observations of increased severity among those receiving ICS without a history of asthma, COPD and/or known ABPA. We also observed an increased use of other respiratory medications, including antibiotics, bronchodilators and airway clearance techniques, in those receiving ICS. That clinicians may try ICS in patients with more severe bronchiectasis is not unexpected and may reflect the wide availability of ICS and a lack of alternative treatment options.23 However, as it is not possible to infer causality from observational studies such as this,24 further investigations are needed to better understand the direction of causality, that is, whether ICS use promotes bronchiectasis severity or whether bronchiectasis severity promotes ICS use.

In addition, sputum culture revealed an increased frequency of respiratory pathogens in those receiving ICS, including P. aeruginosa, H. influenzae and S. pneumoniae; pathogens frequently linked to bronchiectasis severity. As it is not possible to infer causality from the data presented in this study24 whether the higher frequency of infection among those receiving ICS reflects bias by indication, whereby ICS are used in patients with more severe disease including more frequent infections, or whether ICS increases the risk of certain infections, is unclear. The latter is clear for certain infections such as NTM and TB,25 26 and our lack of association between ICS use and NTM isolation may reflect the cessation of ICS in those with a clinical suspicion of NTM pulmonary disease or following NTM isolation prior to diagnosis. ICS use alters the lung microbiome in COPD seemingly by dampening antimicrobial immune responses and promoting overgrowth of potentially pathogenic organisms while reducing overall bacterial diversity.27 28 There are currently no prospective studies of the effect of ICS on the lung microbiome in bronchiectasis.

Bias by indication is a major problem when examining long-term outcomes in observational datasets. Patients with bronchiectasis receiving ICS were more severe, and despite extensive adjustment for markers of severity, unmeasured confounding may persist, artificially showing ‘harm’ associated with ICS. Our results showing higher exacerbation rates and other poor outcomes with ICS use in the general bronchiectasis population should therefore be treated with caution. In the overall bronchiectasis population, following adjustment for confounders, ICS users had a significantly increased frequency of exacerbation but a significantly lower frequency of hospitalisation. When looking at the long-term outcomes of those with bronchiectasis who did not have a current recommendation for ICS, prior to adjustment, ICS use similarly associated with a significantly increased risk of exacerbation, hospitalisation and mortality, but the effect on hospitalisation was lost following adjustment and was comparable with our reference group. Therefore, we can only report that ICS users have a higher likelihood of poor outcomes but cannot confirm if this is causal or the result of unmeasured confounding. Nevertheless, the difference in results between the overall population and the population excluding COPD, asthma and ABPA is intriguing, suggesting that the ‘harm’ associated with ICS use in the overall population is masked by an apparent benefit of ICS in those with asthma, COPD or ABPA.

When determining the effect of ICS on those with differing blood eosinophil levels, we observed a strong, significant reduction in exacerbation frequency among ICS users with eosinophilia compared with both ICS users and non-users with normal blood eosinophil levels. We also observed that ICS non-users with eosinophilia had a modest (although not significant) increase in exacerbation frequency compared with ICS non-users with normal eosinophil levels. These data may be considered more convincing of a true benefit of ICS, as bias by indication would tend to bias the data towards a worse outcome in ICS users. An important limitation of our study is the lack of specific eosinophil cut-off data. To date, most studies use blood eosinophil counts >300 cells/µL as a threshold for eosinophilic bronchiectasis.12 13 Our study uses a higher threshold as determined by local laboratories (>400–500 cells/µL). The benefit of ICS in other diseases increases with increasing eosinophil count and so our approach would tend to increase the apparent benefit of ICS in this subgroup. In turn, given concerns regarding the instability of eosinophil counts over time,29 it is important to note that blood eosinophil data recorded at one time point (baseline) was used to classify eosinophilic or non-eosinophilic patients.

While our study has the strengths of a large-scale, multicentre, observational cohort, we do acknowledge limitations of our study. Due to the very large sample size of the EMBARC, between-group comparisons may highlight small differences as highly statistically significant, which may not necessarily reflect clinical significance. In turn, while we eliminated all known cases of ABPA according to the ISHAM criteria, there were many patients who had not undergone diagnostic testing for ABPA or had inadequate testing data to confirm an ABPA diagnosis. As such, the decision to include these individuals and fully benefit from the available data in the registry may have consequently skewed our findings if undiagnosed ABPA was present. To account for this, we performed sensitivity analyses where all individuals not tested for ABPA or those with inadequate serological testing data were removed, and our findings showed similar results to our reported findings here. Similarly, while those with physician-recorded diagnosis of asthma and/or COPD were excluded, those with undiagnosed asthma/COPD may have been included in our analyses given that a physician’s diagnosis is subjective and creates challenges for reliably excluding all cases of co-existing asthma/COPD. This is an important limitation of our study, particularly in reference to our results suggesting a positive effect of ICS in those with elevated eosinophils, where elevated blood eosinophils are currently recognised as a predictor of ICS response in such conditions. Improvements in aetiology/comorbidity testing and diagnosis are therefore necessary to ensure the inclusion/exclusion of appropriate patients in future clinical research.

Importantly, different forms of ICS have different pharmacological properties and therefore factors, such as retention time within the lung, differ between forms of ICS.30 The patients receiving ICS in this study were prescribed varying forms of ICS of differing doses at their physician’s discretion. Therefore, we cannot comment on whether there are differential effects between different forms and/or doses of ICS or confirm the reasons behind a patient’s prescription of ICS therapy, even in those not currently recommended therapy according to treatment guidelines. We also do not have details on the length of time patients were receiving ICS prior to baseline, therefore we cannot comment on if treatment duration influences clinical outcomes. In turn, while we acknowledge literature reporting differential effects of dual ICS and long-acting beta-agonist (LABA) therapy compared with ICS alone on patient outcomes in those with asthma,31 and the variation in bronchiectasis patient outcomes between different dual ICS-LABA therapies,32 we did not explore this effect here.

Lastly, while we report a possible benefit of ICS in those with eosinophilic bronchiectasis, it would be inappropriate to entirely dismiss the apparent harm associated with long-term ICS use.26 33 Therefore, without the support of large-scale RCTs, clinicians should remain cautious of prescribing ICS to those with bronchiectasis in the absence of asthma, COPD and/or ABPA.

Nonetheless, here we show, using a large-scale, multicentre bronchiectasis registry, that ICS use is common in bronchiectasis, even in the absence of corticosteroid-responsive respiratory comorbidities, and that ICS users appear to have a more severe bronchiectasis phenotype. Notably, our data showing that ICS users with eosinophilia have a significantly reduced frequency of exacerbations and hospitalisations compared with those with normal blood eosinophil counts, with or without ICS, aligns with reports from post hoc analyses of previously unsuccessful RCTs of ICS in bronchiectasis, which identify eosinophilia as a strong predictor of positive clinical effects of ICS in this population.34 Our data suggest the urgent need for new RCTs of ICS in patients with eosinophilic bronchiectasis.