The questionnaire was long (found in the Supplementary Materials). Hence, a selection of the results and questions are described in the publication and the rest are included in the Supplementary Material. Results showed in this publication are based on the response of all panellists, panellists’ results analysed by specialty and country are shown in Supplementary Information 1.

A total of 144 panellists completed the first round of the questionnaire and 127 panellists completed both rounds of the questionnaire (45 GPs and 82 pulmonologists and 0 internal medicine specialists). The average age of the panellists was 46.3 years and 48.0% stated they had more than 15 years of experience treating patients with COPD. We obtained a sample of panellists representing university hospitals, non-university hospitals, primary care centres and individual practices or offices with few health professionals.

On average, each panellist reported having seen 150 COPD patients (pulmonologists saw 3 times more patients than GPs, 205 versus 49) in the last 6 months: 21.5% patients were first consultations (78.5% follow up patients) and 54.0% were current smokers.

All panellists were familiar with the GOLD (A, B, C, D) classification for treatment initiation, and 89.8% of them applied it. Both pulmonologists and GPs considered the GOLD report the most important source to guide treatment decisions.

Figure 2 shows the distribution of patients seen by the panellists in the last 6 months according to GOLD (A, B, C, D) classification and the initial maintenance treatment per category for first visit patients only (23.3%).

Distribution of COPD patients according to the GOLD (A, B, C, D) classification (a) and the initial treatment prescribed (b). GP general practitioner, LABA long-acting beta-agonists, LAMA long-acting muscarinic antagonist, ICS inhaled corticosteroids. p values refer to the differences between specialties for each statement. Statistically significant differences were considered when p value < 0.05.

Specialty did not influence the choice of initial treatment according to GOLD group. Overall, in group A (19.7%) patients, LAMA monotherapy was the main treatment initially prescribed (38.8% of the patients). While in group B (39.0%) and C (17.6%) patients, LABA/LAMA therapy (45.5% and 42.2% of the patients, respectively) was primarily chosen. In group D (23.7%) patients, the main initial treatment was triple therapy (48.8%). Statistically significant differences were observed when analysing prescriptions per country.

The most frequently used tests for stable COPD patients are (1) physical examination (every 3–6 months); (2) Clinical COPD Questionnaire (CCQ)/COPD Assessment Test (CAT) or Modified Medical Research Council (mMRC) dyspnoea Scale (every 6–12 months) and (3) blood eosinophil counts (once a year or less). GPs evaluate blood eosinophil counts for treatment initiation (51.1%) and routinely in treated patients in their clinical practice (46.7%). To a lesser extent, they use blood eosinophil counts when considering a change of therapy (33.3%) and as a biomarker for treatment decisions during exacerbations (17.8%). More pulmonologists used them for initial treatment decision (86.6%), when considering a change of therapy (52.4%), and as a biomarker for treatment decisions during exacerbations (46.3%). In the specialist care it was not commonly measured as a routine in treated patients (13.4%) (Fig. 3).

GPs general practitioners. p values refer to the differences between specialties for each statement. Statistically significant differences were considered when p value <0.05.

Both GPs (88.9%) and pulmonologists (96.3%) reported having ever withdrawn ICS treatment from a patient with COPD. Of their total ICS-treated COPD patients seen in the last year, over 22.0% were eligible for ICS withdrawal, and 17.2% were withdrawn from treatment. The main reason for withdrawal was ICS treatment no longer indicated (49.5%). Panellists reported a high percentage of successful withdrawals, without the need for ICS re-introduction and no exacerbations within 6 months of withdrawal in 75.0% of the patients. In total, 16.0% of the COPD patients needed re-introduction of ICS within 6 months of withdrawal (Table 1).

Nearly half of panellists (48.8%) relied on exacerbation frequency and blood eosinophil counts, while 35.4% relied on the frequency of exacerbations only for decision making around ICS withdrawal. Most pulmonologists (70.7%) reported that ICS withdrawal should be abrupt in patients on LABA/ICS therapy followed by LAMA/LABA introduction, while most of the GPs (68.4%) preferred gradual dose reduction and introduction of LAMA/LABA treatment at the same time. In case of patients on triple therapy, 74.0% of panellists (89.0% pulmonologist versus 46.7% GPs) agreed that ICS should be withdrawn by an abrupt dose reduction while maintaining LAMA/LABA combination (Table 2).

The results showing statistically significant difference between specialties are described as follows: (1) How confident you are with ICS withdrawal in case of pneumonia (p value 0.023); (2) ICS in patients on LABA/ICS therapy should be withdrawn by gradual dose reduction and LAMA/LABA treatment introduction at the same time (p value 0.021); (3) ICS in patients on triple therapy should be withdrawn by gradual ICS dose reduction maintaining LAMA/LABA combination (p value < 0.0001) and (4) with abrupt ICS withdrawal maintaining LAMA/LABA combination (p value < 0.0001); (5) The choice of LAMA/LABA combination when withdrawing ICS in patients who were on triple therapy is driven by the same device (p value 0.027); (6) After ICS withdrawal, patients with COPD should be monitored with planned follow-up visit and a call 1 month after ICS withdrawal (p value 0.011); (7) The reasons that could indicate the need for reintroducing ICS is exacerbations after ICS withdrawal (p value 0.012) and (8) persistent adverse events after ICS withdrawal (p value 0.041).

The relationship between different criteria and values combined creating patient profiles and the initial COPD treatment choice was assessed using a discrete choice experiment (see details in Fig. 1). Smoking status was not statistically significant in the model, and therefore not included. Details on the results found are presented in Table 3.

Results of the multivariate analysis indicate that severe symptoms, 1 or more exacerbations leading to hospitalisation and blood eosinophil counts ≥300 cells/µl were associated with not initiating treatment with LAMA monotherapy. A patient profile with severe symptoms leads to the initiation with LAMA/LABA therapy. In contrast, blood eosinophil counts ≥300 cells/µl results in the decision to not initiating treatment with LAMA/LABA. The presence of 1 or more exacerbations leading to hospitalisation and blood eosinophil counts ≥300 cells/µl are drivers for panellists to prescribe LABA/ICS therapy. Meanwhile, the presence of moderate to severe symptoms and 1 or more exacerbations leading to hospitalisation influence panellists to prescribe triple therapy. The relative weight of lung function values did not reach significance in the multinomial model, although it was significant in the univariate analysis (results shown in Supplementary Table 7).

Panellists were asked about the importance they attributed to each criterion for the selection of the initial COPD treatment (Fig. 4). The patient’s ability to inhale (93.7% of agreement), previous exacerbations (92.1%), breathlessness (89.0%), and the type of inhaler (80.3%) were found the most important aspects considered by panellists for the selection of an initial COPD treatment. Comorbidities, smoking status, price or level of reimbursement was considered important by much fewer panellists for COPD treatment decision. Nearly two-third (65.4%) of panellists (57.8% of GPs and 69.5% of pulmonologists) considered blood eosinophil counts as important, although this did not reach the pre-defined consensus threshold of 70% (Fig. 4).

Importance of each criterion for the selection of initial COPD treatment.

Most panellists (84.3%) agreed that LAMA/LABA improve breathlessness compared to LAMA and LABA/ICS. LAMA/LABA was also considered to improve quality of life (79.5%) and physical activity (73.3%) compared to LAMA, but no consensus (pre-defined as 70% agreement) was reached on this when compared to LABA/ICS. Similarly, most panellists (85.0%) agreed that LAMA/LABA can be the initial treatment for some patients, and there was no consensus that LAMA should be the initial COPD treatment before prescribing LAMA/LABA. There was a clear disagreement (78.0%) with the statement in the questionnaire if LABA/ICS should be prescribed as initial treatment before LAMA/LABA. No significant differences were found between specialities regarding the above.

Panellists believed that LABA/ICS treatment may decrease exacerbations to a greater extent than LAMA/LABA in patients with a high exacerbation risk and high eosinophil blood counts, but not in case of low exacerbation risk and low blood eosinophil counts. When considering prescribing ICS, panellists generally considered blood eosinophil counts, exacerbation risk, and comorbidities (70.1% agreement), and generally took into account any uncertainty of concomitant asthma diagnosis (77.1%). There was consensus regarding the importance of considering comorbidities when selecting COPD treatment (78.7%) and that long-term ICS prescriptions should be re-assessed to verify their efficacy (73.2%). Panellists did not agree on the potential risks of ICS to increase bruising or individual comorbid diseases such as tuberculosis, diabetes and osteoporosis. Statements such as ICS being less beneficial in smokers or that treatment with ICS increases the risk of pneumonia (69.3% agreed) did not reach consensus among panellists. No significant differences were found between specialities regarding the risks and benefits of ICS treatment in COPD patients.

The majority of panellists (79.5%) agreed that ICS is over-prescribed in the primary care setting and felt confident to withdraw ICS when no longer indicated (87.4%) or due to a lack of response (77.2%). Both specialties agreed that GPs generally felt insecure about ICS withdrawal (70.9%).

Panellists believed that increasing the use of LAMA/LABA in patients for whom LABA/ICS is not indicated could increase clinical response to treatment (82.7% of agreement), better COPD control (82.7%), and a reduction in exacerbations (70.1%). In patients with further exacerbations with high eosinophil count, stepping up the inhaled treatment from LAMA/LABA to triple therapy may result in better COPD control (84.3%), prevent exacerbations (83.5%), and increase clinical response to treatment (81.9%). Panellists also believed that reducing the use of non-indicated ICS in COPD patients could result in a reduction of bacterial exacerbations (78.0%), a better outcome of co-existing diseases (76.4%), and a reduction of adverse events such as infections (76.4%). No significant differences were found between specialities.

Consensus was reached regarding the need for resources to optimise management and treatment of COPD patients, such as: tobacco cessation programmes (85.0% of agreement), recommendations for treatment management of multi-morbid patients (81.9%), access to a validated tool that helps to identify patients that would benefit from ICS (80.3%), among others. In terms of the economic impact of COPD treatment, the panellists agreed that exacerbations (92.9%) and pneumonia (85.0%) represent a major financial burden for healthcare expenditures. In general, no significant differences were found between specialities regarding the clinical impact of treatment optimisation.