Chronic obstructive pulmonary disease (COPD) is a major global public health problem [1]. Despite significant advances in our understanding of the pathobiology of the disease and the availability of novel treatments, COPD currently ranks third as a global cause of death [1]. Here, we issue a call to action to improve this situation by addressing two major bottlenecks: diagnosis and management of COPD (Fig. 1).

Fig. 1.

Two key bottlenecks in the diagnosis and management of COPD.

Image generated by ChatGPT at the request of the authors on April 14th, 2026.

First, about 70% of patients with COPD worldwide remain undiagnosed; consequently, they are not treated [2]. A recent study in primary care showed that, among patients “labelled” as having COPD, half had not undergone spirometry, meaning that their diagnosis was exclusively “clinical”; among those who had undergone spirometry, half showed no evidence of airflow obstruction and were therefore misdiagnosed and potentially mistreated [3]. Furthermore, 47% of newly diagnosed patients had missed diagnostic opportunities during the prediagnostic period, defined as more than 30 days between the first health care contact for respiratory symptoms and a formal spirometry-confirmed diagnosis [4]. This delay is largely driven by clinical inertia, whereby symptomatic triggers – such as unscheduled visits or the empirical prescription of antibiotics and inhalers – are managed without confirmatory spirometry [5]. There are also potential implementation barriers in routine primary care settings, including access to spirometry, resource limitations, and variability across health care systems.

A recent cluster analysis identified three distinct phenotypes during this period: “Frequent Attenders/High-Risk,” “Pauci-symptomatic–Preserved,” and “Silent Decliners.” The first group was characterised by severe airflow obstruction, the highest symptom burden, and a markedly elevated frequency of prediagnostic exacerbations. The “Pauci-symptomatic–Preserved” group presented with relatively preserved lung function and minimal symptom burden but exhibited a high rate of diagnostic delay. Finally, “Silent Decliners” had severe airflow obstruction despite only moderate symptom expression, making reactive, symptom-based screening alone insufficient [5]. All these diagnostic gaps have severe clinical consequences, including a 15.5-fold higher risk of exacerbations and a 4.3-fold increase in respiratory mortality compared with individuals without the disease [6].

Reducing underdiagnosis and misdiagnosis of COPD requires better education of patients, health care workers, administrators, and society in general about the burden of COPD. Artificial intelligence may help identify patients with respiratory symptoms and/or risk factors, such as smoking, in electronic health records [1] or in the general population [7], as well as support the proper interpretation of spirometry [8]. In any case, improving outcomes requires a shift from reactive management to proactive, stratified case finding [1]. Adopting data-driven, personalised diagnostic tools is essential to intervene before the irreversible consequences of late diagnosis take hold [1].

Second, after an appropriate diagnosis, follow-up is necessary to assess disease progression and adopt the necessary therapeutic measures [1]. The Global Initiative for Chronic Obstructive Lung Disease (GOLD) 2026 report discusses two different concepts that may partially overlap in addressing “disease activity” in COPD: disease stability, defined as a low disease activity state with no exacerbations, no worsening of symptoms, and no accelerated loss of lung function [9]; and clinical control (CC), defined as a state of low disease activity, with no exacerbations and no worsening of symptoms, together with low impact on the patient, defined by symptoms below a prespecified threshold and FEV1 above a threshold value [10].

The RADAR score is a newly proposed tool to assess the degree of CC in routine clinical practice [11]. It includes four core items: use of rescue medication, acute exacerbations, dyspnoea, and activity. Each item is assigned a numerical value, yielding a total score ranging from 0 to 8 points. Collectively, assessment of these four core items can provide important information on the future risk of a composite outcome that includes hospitalisations, emergency department visits, and mortality after 12 months of follow-up [11]. Patients who achieved CC, defined as a RADAR score of 0–1 points, showed a significantly lower incidence of adverse outcomes and reported better quality of life after 12 months of follow-up compared with those with partial CC, defined as a RADAR score of 2–3 points, or poor CC, defined as a RADAR score of 4 or more points [11].

Similarly, a recent study in primary care showed that poor CC is common and underrecognised in patients with COPD receiving maintenance inhaled therapy, largely because of suboptimal clinician perception and failure to address key modifiable determinants [12]. Similarly, the EPOCAS study, a multicentre observational study conducted across primary care centres in Asturias, Spain, involving 249 patients with COPD, showed that nearly 40% of primary care patients exhibited suboptimal CC: 26.4% had partial control and 13.4% had poor control [13]. Importantly, in 68.4% of uncontrolled patients, no therapeutic changes were introduced [13].

RADAR can serve as an objective “impact” meter that may help break this inertia by supporting specific therapeutic measures based on treatable traits, such as treatment adherence, correct inhaler use, vaccination, comorbidities, and level of physical activity [14]. However, there is currently no evidence that implementing these measures improves health outcomes. Furthermore, in patients with advanced COPD, CC can be very difficult to achieve because of irreversible structural lung damage and a high symptom burden.

In summary, a more proactive diagnostic approach is necessary to address the currently unacceptable rates of underdiagnosis, misdiagnosis, and delayed diagnosis [1]. Similarly, assessment of clinical control, and intervention when needed, is necessary during follow-up to improve the well-being and prognosis of these patients [1].