Chronic obstructive pulmonary disease (COPD) is the most rapidly rising cause of death in individuals >65 yrs of age, the most rapidly growing segment of the USA population.
Advancing the clinical research agenda for COPD entails careful consideration of several issues and problems: defining COPD, reliance on physiological testing, outcome measurements and comorbidities.
In addition, issues pertaining to the recent dramatic increase in chronic obstructive pulmonary disease mortality in females, smoking susceptibility and the relationship of respiratory symptoms with hyperresponsiveness, as well as chronic obstructive pulmonary disease heterogeneity, are discussed.
Chronic obstructive pulmonary disease (COPD) is a major public health problem. Recent prevalence estimates suggest that ∼15 million people have COPD; in 1996, 14.1 million people had chronic bronchitis and 1.8 million had emphysema 1. This disease is the most steeply rising cause of death in individuals >65 yrs of age, which represents the most rapidly growing segment of the USA population. Disability from COPD is expected to rise by 2020 with ageing of the population and the increase in disease prevalence. It will be the fifth-leading cause of disability-adjusted life years lost 2, and it will rise worldwide from being the sixth to the third most common cause of death 1. Excessive focus on the single environmental factor of cigarette smoking has been to the detriment of COPD research, as it has tended to minimise an understanding of genetic, developmental and other environmental risk factors that may be important in disease susceptibility and disease progression. Although cigarette smoking is most certainly the major risk factor for the development of COPD 3, only a small percentage of cigarette smokers develop the disease 4. In addition, a variable response to cigarette smoking clearly suggests genetic susceptibility.
This article briefly considers a series of important issues that need attention in COPD research. These issues include defining and diagnosing the disease, understanding the problems inherent in relying on physiological testing alone to make a diagnosis, the importance of not discounting childhood episodes as irrelevant to adult pathology, the lack of well-defined phenotypes for tracking the disease, risk factors (related to sex, hyperresponsiveness, respiratory symptoms), genetics and the important issue of comorbidities. Only careful consideration of these problems and an attempt at mitigating their ramifications will allow the advancement of the clinical research agenda for COPD and improve treatment for what is a serious, life-threatening disease.
Defining chronic obstructive pulmonary disease
One of the critical issues in COPD research has been the lack of a cohesive definition for the disease. The lack of an international standard has confounded the comparability of COPD studies. True incidence and prevalence trends may be obscured by this definitional variability, illustrated in a study by Viegi et al. 5 that compared prevalence rates of COPD in the Po Delta Valley, Italy, using European Respiratory Society (ERS) and American Thoracic Society (ATS) criteria. In subjects ≥46 yrs of age, ERS criteria suggested a 12.2% prevalence of COPD compared with a 57% prevalence by ATS criteria. Of note, clinical criteria suggested a prevalence of 28.8% 5. The Global Initiative for Chronic Obstructive Lung Disease (GOLD) represents a viable international effort to codify the definitions and criteria for diagnosis and management of COPD 6.
According to the GOLD criteria, “COPD is a disease state characterised by airflow limitation that is not fully reversible. The airflow limitation is usually both progressive and associated with an abnormal inflammatory response of the lungs to noxious particles or gases” 7. Spirometric evaluation should be used to confirm the diagnosis when suggestive symptomatology is present. Chronic cough and sputum production have a variable relationship to airflow limitation, which is the sine qua non of COPD 8. The epidemiological manifestations of the disease (chronic cough, phlegm and dyspnoea on exertion) and the pathological manifestations (enlarged and increased numbers of mucus-secreting glands, destruction of the lung parenchyma and the development of emphysema) have a statistically significant, but not a high degree of, correlation with COPD confirmed by physiological testing. Some individuals with airflow obstruction do not have severe emphysema and some individuals with emphysema do not have airflow limitations 9, 10. Epidemiological studies demonstrate that respiratory symptoms may precede or follow the development of airflow limitation by many years. The presence or absence of symptoms also depends largely on the constellation of risk factors to which an individual is exposed (table 1⇓).
Importantly, peak flow alone has a low specificity for the diagnosis of COPD, as decrements in peak flow can be present in other pulmonary diseases. When spirometry is performed, a diagnosis of COPD is considered to be present if the postbronchodilator forced expiratory volume in one second (FEV1) is <80% of the predicted value in combination with an FEV1/forced vital capacity (FVC) ratio of <70% 11. Recently, standards of severity have also been defined with a goal toward global uniformity in the diagnosis of COPD 11 (table 2⇓). It should be noted that the role of classifying individuals into GOLD stage 0 (with absence of airflow limitation but presence of cough and sputum) may not actually be useful for identifying the individuals “at risk” for future development of obstructive pulmonary disease. Vestbo and Lange 12 analysed 5- and 15-yr data from The Copenhagen City Heart Study using multivariate regression techniques; this analysis suggested that GOLD stage 0 did not effectively identify the future development of airways obstruction at these time points 12.
Pulmonary function testing is reproducible and FEV1 tracks consistently during growth from an early age, which can be used to track individuals over time 13 (fig. 1⇓). Pulmonary function testing is generally not administered to patients during routine outpatient visits. This greatly hinders the potential for COPD diagnosis at early stages of the disease and has direct relevance to prevalence estimates and the clinical practices of COPD 14. Spirometry has been demonstrated to be an efficient method for detection of airway obstruction in outpatient clinics in specifically targeted higher-risk groups. In a large-scale effort to demonstrate the efficacy of spirometry for screening, Zielinski and Bednarek 15 offered spirometry to 11,027 individuals >39 yrs of age with at least 11 pack-yrs of smoking. Spirometric criteria for obstructive lung disease were observed in 24.3% of the cohort. Of 2,200 participants who were lifetime never-smokers, 14.4% had demonstrable airway obstruction 15. Recent trends in the prevalence of COPD are suggested in the third National Health and Nutrition Examination Survey (NHANES III) 14. Outcome measures included a physician diagnosis of asthma, chronic bronchitis, emphysema, respiratory symptoms and low lung function. Low lung function was defined as both an FEV1/FVC ratio of <0.70 and an FEV1 <80% pred. Of the cohort of 20,050 individuals, 6.8% had low lung function as defined and 8.5% reported a diagnosis of COPD. The majority (63.3%) of those individuals with documented low lung function had no prior physician diagnosis of COPD. Data from this NHANES study also suggested that a significant amount of disease remains undiagnosed when in the mild stage, thus underestimating the actual prevalence and burden of disease 14. The classic estimates are that 10–15% of all adults >55 yrs of age have undiagnosed airflow limitation. This increases with advancing age and may reach as high as 30–40% in individuals >70 yrs of age 16, 17. In Spain, a similar observation was made regarding underdiagnosis of COPD. In a large population study of 4,045 males and females (aged 40–69 yrs), the prevalence of COPD was 9.1%; although the majority of this disease was observed in current and former smokers, there was a component of disease that was observed among nonsmokers. In 78.2% of cases there was no prior physician diagnosis of COPD. Individuals who were male, >60 yrs, more educated and had >15 pack-yrs of smoking were more likely to have received a prior diagnosis of COPD 18. In a sample of 353 individuals in a British general practice, 58 patients had obstructive lung disease noted with screening spirometry; 52% had no prior diagnosis of lung disease 19. The use of spirometry in health clinics is important to be able to estimate the true burden of COPD and its impact on health as the population ages.
Variability in patterns of clinical presentation is common in COPD, which also makes diagnosis difficult. Patients may not focus on respiratory complaints when they visit a physician. Decline of pulmonary function is present in all individuals with ageing, thus, most older persons will accommodate their physical activity to encompass their respiratory reserve. Consequently, they will not consult a physician for respiratory complaints alone or mention them to the physician during a visit. This is particularly true in Western industrialised society, in which people do not have to engage in strenuous physical activity and can thus modify their personal and occupational activities to be energy and breath conserving. It is not unusual for some intervening event such as a respiratory illness to cause a patient to visit a physician. Yet, most likely, and erroneously, the patient will attribute shortness of breath to that specific episode rather than accept the event as part of a gradual underlying, unremitting decline in pulmonary function. The net result of these patterns is that physicians are not predisposed to perform pulmonary function testing. Consequently, a substantial portion of COPD remains undiagnosed 14.
Risk factors and diagnostic obstacles
Risk considerations in COPD should be divided into risk for disease development and risk for disease progression. Intrinsic to the consideration of risk for the development of COPD are both host susceptibility and environmental factors, as enumerated in table 1⇑ and discussed below.
Childhood asthma and increased airways responsiveness may also represent the most important events that mark susceptibility for lung function decline and subsequent development of COPD 20. During childhood, pulmonary function increases and given that symptoms are loosely correlated with pulmonary function, asthma-type symptoms tend to decline with lung growth. Childhood events are important in that they reduce maximal lung growth and may precipitate early decline 21 (fig. 2⇓). A retrospective study of asthma incidence rates in Rochester, MN, USA, showed that 40% of children who wheezed during their first year of life went on to develop asthma and that 90% of all asthma was diagnosed by age 6 yrs 22.
Most adults who are susceptible to lung damage from smoking have also had some childhood episode or illness that marked them as susceptible for developing COPD 23, which is of concern with regard to trends in COPD. Data from the National Health Interview Survey revealed that self-reported asthma rates have increased by 75% during the years 1980–1994. The most significant increase was in children from birth to 4 yrs of age, whose rates were noted to increase by 106% 24. These trends may influence the future burden of disease secondary to COPD. Unfortunately, adult patients with respiratory problems are often unable to recall those episodes from the distant past. Consequently, both patients and physicians regard current symptoms to be new-onset adult disease, when, in fact, the symptoms may represent a recrudescence of disease that began early in life.
Although different aspects of indoor air pollution such as chronic biomass or wood smoke may be relevant to the development of obstructive lung disease 25–28, globally, cigarette smoking is the most important environmental risk that contributes to COPD 3. Factors that appear to be important host characteristics with regard to susceptibility to smoking-related COPD include genetic predisposition, female sex, the presence of underlying asthma or airways responsiveness and, possibly, the presence of underlying allergy 29, 30. When a patient with such host susceptibilities begins to smoke cigarettes, disease is produced at a much lower dose (as measured by pack-yrs of cigarette smoking) 4. Since individuals vary in their genetic susceptibility to cigarette smoking, there is substantial variation in the pulmonary function for any given dose of cigarette smoking 4 (fig. 3⇓). Regardless of smoking status, however, clinicians are much more likely to diagnose males with COPD and females with asthma or asthmatic bronchitis. In a longitudinal study, females >40 yrs of age are often deemed “asthmatic” when compared with males of the same age and with similar symptoms 31. The combination of recall bias on the part of the patient and diagnostic bias on the part of the physician consequently creates diagnostic uncertainty and potential misdiagnosis of COPD 32. Feenstra et al. 33 recently showed that there will be an increase in the burden of COPD between 1994–2015. This will result in a 43% increase in prevalence of COPD for males and a 142% increase for females.
Allergic phenotypes represent immediate hypersensitivity to antigen and may manifest as an elevation in total or specific immunoglobulin E, elevated total eosinophil counts and positive skin-test reactions to specific allergens. Skin-test reactivity increases until age 15 yrs 34 and the decline in reactivity after age 35 yrs confounds the measurement of this intermediate phenotype in individuals suspected to have COPD. Although the findings of association between skin-test reactivity and FEV1 have not been consistent 35–37, Gottlieb et al. 37 investigated this prospectively and found that positive skin-test reactivity predicted increased rates of decline in FEV1 and FEV1/FVC ratio. It has also been suggested that immediate type-I hypersensitivity is a risk factor for the development and progression of COPD 30. In this model, atopy may influence childhood asthma and thus limit maximally attained lung growth, accelerate FEV1 decline and enhance interaction with cigarette smoke, thus promoting lung function decline and the development of COPD 30.
Airway hyperresponsiveness to methacholine may help identify individuals susceptible to the development of COPD and those susceptible to progressive lung function decline 38. This intermediate phenotype is not only a cardinal feature in asthma, but also an important finding in a subset of individuals with COPD 30, 37, 38. Baseline levels of lung function, age and smoking history all play a role in this phenotype. In some individuals, airway hyperresponsiveness predicts an acceleration in lung function decline and the development of COPD 38; in others, this phenotype may predict COPD-related mortality 39. In at-risk individuals, airway responsiveness may be demonstrated in the majority of instances when it is actually measured 40. However, methacholine challenge is often avoided in individuals already suspected of having low lung function.
An important consideration is that airway hyperresponsiveness may be a cause and not a result of COPD. A 25-yr study in the Netherlands demonstrated that increased airway hyperresponsiveness is an independent risk factor for decline in FEV1 41. In a study of individuals with early-onset COPD, those with increased airway hyperresponsiveness had an accelerated decline in FEV1 42. It is still debatable whether hyperresponsiveness improved by anti-inflammatory therapy may improve the course of the disease. One study has shown that inhaled corticosteroids improved hyperresponsiveness after 4 yrs of treatment. Despite this improvement, pulmonary function decline did not cease 43.
Respiratory symptoms have been associated with lower levels of pulmonary function 44–47, although patients may underestimate the morbidity associated with their disease. In a large international telephone screening of 201,921 households (in the USA, Canada, France, Italy, Germany, the Netherlands, Spain and the UK), 3,265 individuals with probable obstructive lung disease were identified. For example, among individuals with the most severe breathlessness and limitation (unable to leave home secondary to dyspnoea), 35.8% described their disease as mild; of those unable to ambulate for a few minutes on level ground, 60.3% described their disease as mild 48. Although the impairment associated with chronic cough was observed predominantly in males 44, it has been reported that females and males with chronic cough and chronic mucus hypersecretion have an accelerated decline in FEV1 45, 46 and greater mortality from COPD 47. Whether these symptoms are markers of pathological alteration of the airways is not known.
The relationship between symptoms and hyperresponsiveness has been studied in Vlagtwedde/Vlaardingen in the Netherlands. The more severe hyperresponsiveness a subject had without respiratory symptoms, the more likely they were to develop respiratory symptoms in later life 49. Furthermore, in a subsequent analysis, it has been shown that individuals with hyperresponsiveness and eosinophilia are more prone to develop respiratory symptoms 50. Thus, both hyperresponsiveness and eosinophilia may constitute risk factors for the development of symptoms and are worthwhile measuring in studies assessing COPD risk. These parameters may also be useful for assessing the impact of prevention and intervention strategies.
Sex and sex differences
A difference has always existed in the prevalence of COPD in males and females, with males having higher rates. The gap between males and females has been narrowing, however, and the increased rates of females smoking cigarettes in the last 20–30 yrs have been associated with steadily increasing rates of COPD in females. Figure 4⇓ illustrates that females have been catching up with males rapidly in COPD mortality, particularly during the last decade 51. Mannino et al. 52 analysed data concerning death from 1979–1993 and found that although mortality rates for males with COPD have started to stabilise, these rates were continuing to increase among females.
There are alternative explanations for increasing COPD mortality among females. One proposal is that females are more susceptible than males to the effects of cigarette smoke. For any given level of smoking, females have a lower FEV1 53. Furthermore, there have been significantly greater rates of FEV1 decline among female smokers compared with male smokers, as shown in an epidemiological setting in the Netherlands 54. Finally, a sex difference in severe, early-onset COPD has been demonstrated 55. In the offspring of 84 individuals with severe, early-onset COPD, there was a markedly elevated prevalence (71.4%) in females. Current or exsmoking female first-degree relatives appeared to have a significantly greater risk of a low FEV1, with odds ratios of 1.91 for FEV1 <80% pred and 3.65 for FEV1 <30% pred. These data suggest that females may be more susceptible to the development of COPD. Interestingly, there was no significant risk for chronic bronchitis in females, although there was a risk in males. Chen et al. 56 showed that females have more severe small airway disease for a similar level of smoking (fig. 5⇓). Emphysema occurs predominantly in the small airways and this finding may suggest a greater likelihood for females to develop emphysema when smoking. Leynaert et al. 57 demonstrated that females were more likely to have airway hyperresponsiveness than males; this could not be explained by sex differences in the presence of asthma, respiratory symptoms, atopy or lung function. A bronchoconstrictive response to methacholine was associated with heavy tobacco consumption only in females. They concluded that the excess of airway hyperresponsiveness in females is not due to their having a smaller lung function size or airway calibre than males, but to a greater susceptibility to the effects of cigarette smoking 57.
Airway hyperresponsiveness is a risk factor for both the development and progression of COPD. The more severe the hyperresponsiveness, the more rapid the decline in pulmonary function 35, 42, 58, 59. Thus, the higher prevalence of airways hyperresponsiveness in females may indeed explain the difference between males and females with respect to smoking-induced COPD. Females in the Lung Health Study had greater airway hyperresponsiveness to methacholine than males with mild COPD 59. Notably, sex differences in this group of individuals with established COPD were not significant with respect to the influence of hyperresponsiveness on lung function decline. Conversely, females have a greater degree of hyperresponsiveness to methacholine irrespective of pulmonary function 60. Thus, it is still possible that the enhanced severity of hyperresponsiveness in females may contribute to their greater susceptibility to cigarette smoke and other environmental risk factors that influence pulmonary function.
Airway size is another factor possibly contributing to differences between sex susceptibility to the effects of cigarette smoke. Between males and females at a young age, differences already exist that may account for the enhanced effect of smoking in females. Young females have larger airways in relation to lung size than young males 61 and specific airway resistance is lower at any given height in females (fig. 6⇓). Thus, relevant to the development of COPD, smoke particles may impact males and females differently.
Most COPD is not secondary to α1-antitrypsin deficiency, the only currently proven genetic risk for this disease 62, 63. Instead, most COPD represents the expression of a complex genetic disease, resulting from an interaction between host and environmental factors 64. Most studies of COPD genetics have focused upon intermediate phenotypes and pulmonary function parameters. Twin studies have shown that the concordance of lung function was larger in monozygotic than in dizygotic twins 65. In a cohort of individuals with COPD, Silverman et al. 66 studied 104 first-degree relatives of 44 probands with severe, early-onset COPD. First-degree relatives of probands had significantly lower FEV1 and FEV1/FVC ratios when compared with controls having similar smoking histories. Reduced spirometry in first-degree relatives was found in current or exsmokers only 66. These reports suggest both familial aggregation of COPD (independent of protease inhibitor type) and the importance of gene-by-environment interactions between cigarette smoke and still unidentified genetic influences. Recent publications have found multiple genetic loci involved in spirometric phenotypes associated with COPD 65, 67–70. The variability in gene-to-gene and gene-by-environment interactions, both in the susceptibility and the progression of COPD, constitutes the challenge of dissecting the genetics of COPD.
Environmental risk factors
Cigarette smoke is the most important environmental challenge that influences lung function decline and the development of COPD. The dose/effect relationship between cigarette smoke and lung function decline depends upon when an individual is exposed. Among those smokers already with a low FEV1, lung injury and decrease in lung function secondary to cigarette smoking is more impressive. In the Lung Health Study 42, middle-aged smokers who continued smoking for 5 yrs had further losses of FEV1 of several hundred millilitres. Importantly, however, COPD has been identified in nonsmokers as well, with 4% of males and 5% of females reporting physician-diagnosed COPD. Therefore, other related environmental risks are important contributors. Occupational and air pollution exposures may result in the development of an obstructive phenotype 71, 72, and environmental awareness and control should be an ancillary feature addressed in minimising patient risk for the development and progression of COPD.
Risk for disease exacerbation
It is still debatable whether exacerbations of COPD constitute a risk factor for accelerated decline in lung function 73. However, exacerbations constitute a large individual burden and have a dramatic impact on the quality of life and need for healthcare 74. Risk factors for exacerbations have not been studied extensively, although a wide variety have been proposed. Bacterial 75, 76 and viral 77 infections are important factors that influence exacerbations of COPD. Exacerbation may be manifest as increased sputum production with the development of purulent features and increased dyspnoea. In one study, daily wheeze and bronchitic symptoms contributed to a higher frequency of exacerbation 74. Kessler et al. 78 reported that risk factors for hospitalisation included a low body mass index and poor performance on a 6-min walk test, but gas exchange impairment and pulmonary and haemodynamic dysfunction were the most important risks. Importantly, one study suggests that many patients ultimately hospitalised with COPD exacerbation have modifiable characteristics that impact on susceptibility (such as lack of influenza vaccination, improper use of medications, ongoing cigarette smoke exposure) 79. These deficiencies in outpatient management may be addressed to help limit disease exacerbation. In this same study, previous hospital admissions, lower FEV1 and underprescription of long-term oxygen therapy were independently associated with a higher risk of admission for a COPD exacerbation 79. Longitudinal studies are urgently needed to assess relevant determinants of COPD-related exacerbations and hospitalisation.
Disease outcome measurement
Despite the understanding of the host susceptibility factors, COPD diagnosis remains a challenge. Frequently, COPD is diagnosed only when the disease has already progressed. It is not possible to measure lung tissue damage or small airway disease at an early stage. This early stage is precisely when a person should be warned about the negative effects of environmental stimuli and when a physician could intervene to limit the progression of these changes in lung tissue and airway walls.
COPD is defined physiologically on the basis of pulmonary function level, and spirometry has become the primary end-point for clinical trials. Although knowledge is increasing 80, 81, relatively little is known about the inflammatory processes in COPD. Unlike other diseases, where there are viable biomarkers to monitor the inflammatory process, COPD lacks established intermediate phenotypes that can be utilised to track the disease progression and serve as surrogate markers and outcome variables in clinical trials. This is a tremendous problem both for determining the economic and social burdens of COPD, as well as determining efficacy and effectiveness in clinical studies. One such monitoring tool may be induced sputum or exhaled air condensate. Sputum eosinophilia appears to be present in some individuals with COPD and has been associated with a better oral corticosteroid response 82. To date, no study has investigated prospectively whether this can be utilised in clinical practice to assess inhaled steroid responders. It has been well established that exhaled hydrogen peroxide (H2O2) is increased in patients with COPD 83. One study investigated the effect of 600 mg N-acetylcysteine (NAC) on exhaled air condensate with H2O2 and lipid peroxidation products 83. After 9 and 12 months of treatment, the NAC group exhaled 23-fold and 26-fold less H2O2 than those who received placebo. No significant effect of NAC administration on lipid peroxidation products was noted over the whole treatment period. These results indicate that long-term oral administration of NAC attenuates H2O2 formation in the airways of COPD subjects and provides support for the antioxidant action of the drug. However, further studies are necessary to estimate the clinical significance of this finding 84.
One other potential biomarker of COPD is high-resolution computed tomography (HRCT), which harnesses the anatomical and pathological features of COPD together with cutting-edge technology. In fact, refinement of this technique may prove useful in addressing the presence of risk for lung obstruction before spirometric abnormalities develop. Sanders et al. 85 and Santis et al. 86 showed that 68–80% of smokers had demonstrable evidence of emphysema on HRCT in the presence of normal spirometry. Although HRCT as a clinical technique is still being developed, it may provide a much-needed link between the anatomy and physiology and allow for earlier recognition and management of disease susceptibility factors.
Finally, COPD is a heterogeneous disease. Some patients with COPD have evidence of bronchitis in the airway wall. Others have small and almost undetectable airway disease with conventional pulmonary function testing and others may have a loss of elasticity in lung tissue. It is clear that these disease characteristics suggest different underlying disease processes. Continuing to consider these disparate groups as one group may obscure risk factors for disease development or progression. In grouping these different phenotypes, a clear understanding of the variable expression of COPD is lost. Consequently, the opportunity to select adequate measures to address the underlying differences is also lost. Therefore, better tools are needed to assess the various aetiological and pathogenic aspects of COPD. The goals are to slow and, ultimately, to prevent the development of emphysema.
In general, COPD has its onset in individuals during their 60s and 70s. Smoking, which is the major risk factor for COPD, is also a risk factor for atherosclerosis, cancer and coronary artery disease, and thus the potential for severe and significant comorbidities influencing the course and natural history of COPD is substantial. Most studies have been unable to characterise the COPD population accurately with regard to their comorbidities. Subsequently, the influence of these comorbidities on the results of clinical trials and the natural history of disease progression is poorly understood.
COPD is a major public health problem. Although reliable physiological testing can define disease severity and occurrence, this testing is not uniformly applied to high-risk populations. There are major uncertainties with regard to diagnoses based on symptoms because symptoms that mark the susceptible adult are usually not remembered or reported at the time of disease onset. An additional major concern is diagnostic misclassification. Risk factors other than smoking, such as airway hyperresponsiveness and atopy/allergy, need to be studied further, as do sex and genetic differences that may impact susceptibility.
Finally, investigators lack repeatable, reliable markers of airway inflammation and burden of disease, independently of spirometric testing that would allow monitoring of disease progression and effects of therapy in clinical trials. High-resolution computed tomography may provide a much-needed biomarker for use in chronic obstructive pulmonary disease investigations, but its role has still to be established. Most chronic obstructive pulmonary disease presents in older individuals with other diseases and thus assessment of the impact of comorbid conditions on the natural history of chronic obstructive pulmonary disease are urgently needed to place this disease in better context. The future clinical agenda for chronic obstructive pulmonary disease research will depend on the ability to solve these problems.
- Received August 21, 2002.
- Accepted February 20, 2003.
- © ERS Journals Ltd