Incidence of Chronic Obstructive Pulmonary Disease in a Cohort of Young Adults According to the Presence of Chronic Cough and Phlegm

Chronic obstructive pulmonary disease (COPD), a disease with a substantial and increasing socioeconomic burden (1), often has its roots many years before the diagnosis (2,3). Because effective interventions and medications for COPD are now available (4), early diagnosis may reduce prevalence, morbidity, and mortality.

The Global Initiative for Chronic Obstructive Lung Disease (GOLD) guidelines have introduced a stage 0 (normal spirometry, but presence of chronic cough or phlegm) in the COPD staging system as a tool to identify subjects at risk of developing the disease later in life (5). A recent revision made by the American Thoracic Society (ATS)/European Respiratory Society (ERS) Task Force (4) has suggested that dyspnea should also be considered among the stage 0 symptoms. However, the few prospective studies aimed at verifying the prognostic value of chronic symptoms for COPD have given contrasting results (6-8). Therefore, it remains an open question whether the presence of chronic symptoms can predict the future occurrence of COPD (9), after adjusting for the well-known risk factors, such as smoking habits.

The aims of this analysis were to assess the incidence of COPD in a cohort of young adults and to test whether chronic cough/ phlegm and dyspnea are independent predictors of the subsequent occurrence of COPD. For these purposes, an international cohort of subjects, identified in 1991-1993 in the European Community Respiratory Health Survey (ECRHS) I and assessed for COPD (10), was reassessed in 1999-2002 in the ECRHS II, using the GOLD staging system. Some of the results of this study have been previously reported in the form of an abstract (11).

METHODS

Study Design and Measurements

The ECRHS I is an international multicenter study on respiratory diseases, performed in 1991-1993 on random samples of young adults aged 20-44 yr (12). Each participant was sent a brief screening questionnaire (stage 1), and, from those who responded, a 20% random sample was invited to undergo a more detailed clinical examination (stage 2). The ECRHS II is a follow-up study of all the participants in the stage 2 of the ECRHS I, performed in 1999-2002 (the full protocol is available at www.ecrhs.org) (13). Subjects answered a standardized questionnaire administered by trained interviewers, and underwent lung function and blood tests during both ECRHS I and II. Ethical approval was obtained for each center from the appropriate ethics committee, and written consent was obtained from each participant.

The subjects eligible for the present study (Figure 1) were those who (7) participated in the ECRHS I stage 2 (1991-1993; baseline) and had valid lung function measurements (prebronchodilator values of FEV^sub 1^ and FVC) according to the ATS criterion for reproducibility (14) (additional details on lung function measurements are provided in the online supplement), (2) had an FEV^sub 1^/FVC ratio of 70% or greater at baseline, and (3) did not report having had a diagnosis of asthma at baseline.

A total of 9,839 subjects from 26 centers participating in the ECRHS II were eligible for the present study (see Table E1 in the online supplement); 6,208 (63.1%) of these subjects attended the second survey. Lung function measurements were not performed, or did not fulfill the ATS criterion, in 1,206 (19.4%) subjects. Overall, the cohort consisted of 5,002 subjects. The 4,837 eligible subjects not included in the analysis for any reason were slightly younger and had a higher prevalence of ever-smokers at baseline than those in the study cohort. Accordingly, nonparticipants had a slightly higher prevalence of respiratory symptoms than participants, but had similar lung function (additional details on the selection of subjects are provided in the online supplement).

A new case of COPD was defined as a subject who had an FEV^sub 1^/ FVC ratio of 70% or greater at baseline (ECRHS I) and an FEV^sub 1^/FVC ratio of less than 70% at the end of the follow-up (ECRHS II). The new cases who reported having had a doctor diagnosis of asthma during the follow-up were excluded from the analysis.

For each subject, the following covariates (from the ECRHS questionnaire) were considered: sex, age, chronic cough and/or phlegm, dyspnea, smoking habits, and low educational level (additional details on the definition of the covariates are provided in the online supplement).

Statistical Analysis

Incidence rates of COPD were estimated as the ratio of the number of new cases and the number of person-years at risk (per 1,000), which were considered equal to the length of the follow-up for each member of the cohort. The 95% confidence intervals (CIs) were computed using the Poisson distribution. The association of each predictor with the incidence of COPD was estimated by the incidence rate ratio (IRR). The homogeneity of relative risks test (15) was used for each predictor to determine if the IRRs vary across centers.

The mutually adjusted associations of all the predictors with the incidence of COPD were estimated by two-level random intercept Poisson regression models (16), with level 1 units (subjects) nested into level 2 units (ECRHS centers). The models had a dummy indicator of the occurrence of COPD as the dependent variable, a random intercept term at level 2, and all the predictor variables as fixed effects.

The statistical analysis was performed using Stata software (Stata Corporation, College Station, TX) and StatXact software (Cytel Software Corporation, Cambridge, MA).

RESULTS

The New Cases of COPD

The median length of the follow-up was 8.9 yr, ranging from a minimum of 5.8 yr to a maximum of 11.4 yr. During this period, 123 new cases of COPD occurred in our cohort. Moreover, 22 subjects with an FEV^sub 1^/FVC ratio less than 70% at the end of the follow-up were excluded from the analysis because they reported a diagnosis of asthma at the ECRHS II.

Of the 123 incident cases, 55.3% were men and 23.3% had a low educational level. At baseline, 76.9% were ever-smokers, 18.9% reported chronic cough/phlegm, and 19.7% reported dyspnea. The percentage of incident cases was strongly and inversely related to the baseline FEV^sub 1^/FVC ratio (Figure 2).

Incidence of COPD According to Individual Characteristics at Baseline

Overall, the incidence rate of COPD was 2.8 cases/1,000/yr (95% CI, 2.3-3.3); accordingly, the 10-yr cumulative incidence was 2.8%. The incidence of COPD was slightly higher in men than in women, was positively associated with age at baseline, was fivefold higher in heavy smokers than in nonsmokers, and was twofold higher in those with chronic cough/phlegm than in asymptomatic subjects (Table 1). A low educational level and the presence of dyspnea at baseline were not associated with an increased risk of developing COPD. No heterogeneity in incidence rate ratios was found among the centers involved in the study for all the factors at baseline.

The 23 incident cases of individuals with COPD (18.9%) who reported chronic cough/phlegm at baseline were very similar to the remaining incident cases with respect to the distribution of smoking habits, respiratory symptoms, and demographic characteristics at the end of the follow-up, whereas their mean FEV^sub 1^% predicted was significantly lower. The presence of chronic cough/ phlegm remained an independent and statistically significant predictor of the occurrence of COPD after adjusting for smoking habits and the other potential confounders at baseline (IRR, 1.85; 95% CI, 1.17-2.93). No interaction was found between chronic symptoms and sex.

The incidence of airflow obstruction among the 326 subjects with a diagnosis of asthma at baseline who participated in the ECRHS II (not considered in the analysis) was significantly higher (6.6 cases/1,000/yr; p < 0.001) than the incidence among the subjects in our cohort. However, the comparison of clinical and immunologic characteristics at baseline between the incident cases of individuals with COPD with or without a diagnosis of asthma showed that they represented two different phenotypes (additional details are provided in the online supplement).

Incidence of COPD According to the History of Chronic Cough/Phlegm during Follow-up

At baseline, 447 (9.2%) subjects reported chronic cough/phlegm, but only 38.0% of them confirmed the presence of these symptoms 9 yr later (Figure 3). At follow-up, 324 (7.4%) of the subjects who were symptom-free at baseline reported having had chronic symptoms during the interim period.

The incidence of COPD in subjects who confirmed the presence of chronic cough/phlegm at the end of the follow-up (9.4 cases/1,000/yr; 95% CI, 5.6-15.9) was fourfold higher than the incidence in subjects who had never reported these symptoms (2.3 cases/1,000/yr; 95% CI, 1.9-2.9).

Subjects with persistent symptoms had the highest incidence of COPD, both among nonsmokers and ever-smokers at baseline, and had a nearly threefold-increased risk of developing COPD with respect to asymptomatic subjects, after adjusting for the potential confounders at baseline (IRR, 2.88; 95% CI, 1.44-5.79) (Table 2).

Incidence of COPD According to the History of Dyspnea during Follow-up

A total of 700 subjects reported dyspnea of grade 1 or higher at baseline; 39.7% of them confirmed this symptom at the end of the follow-up, whereas 396 (10.9%) of the 3,641 symptomfree subjects at baseline reported dyspnea at the end of the study (Figure 4). The incidence of COPD in subjects with persistent dyspnea (3.2 cases/1,000/yr; 95% CI, 1.6-6.5) was not significantly different from the incidence in subjects who had never reported this symptom (2.4 cases/1,000/yr; 95% CI, 1.9-3.0).

Incidence of COPD According to the History of Smoking Habits during Follow-up

A total of 67 subjects (3.2%) out of the 2,078 nonsmokers at baseline started smoking during the follow-up, whereas 281 (28.5%) of the 987 light current smokers and 153 (24.5%) of the 624 heavy current smokers gave up smoking. Changes in smoking habits were not associated with statistically significant changes in the incidence of COPD, even if quitters showed a slight decrease in incidence with respect to persistent smokers (Figure 5).

According to the lifetime pack-years smoked at the end of the follow-up, the incidence of COPD was 1.5 cases/1,000/yr (95% CI, 1.0-2.1) in nonsmokers and 2.2 cases/1,000/yr (95% CI, 1.5-3.2) and 5.7 cases/1,000/yr (95% CI, 4.4-7.2) in light and heavy smokers (both past and current), respectively.

Sensitivity Analyses

The stability of the association between chronic cough/phlegm and the occurrence of COPD was tested by repeating the analysis under the following conditions: (1) when stage 0 was defined as the presence of chronic symptoms in subjects with an FEV^sub 1^/ FVC ratio of 70% or greater and an FEV^sub 1^ of 80% predicted or greater; (2) when a more restricted definition of COPD (stage II+ of the GOLD classification) was used; (3) when all the asthma cases that occurred during the follow-up (with or without airflow obstruction) were excluded from the analysis; (4) when the association was estimated after adjusting for the presence of wheezing at baseline; (5) when a broader definition of chronic cough/phlegm was used (i.e., not requiring the presence of the symptoms for at least 3 mo/yr); and (6) when the centers with a low participation rate (< 50%) were excluded. The results of the main analysis were confirmed under each of these conditions (data not reported).

DISCUSSION

Principal Findings

We investigated the incidence of COPD in a large, international population-based cohort of young adults in relation to the presence of chronic cough/phlegm. The main findings of our analysis were as follows: (1) even in a young cohort, the yearly incidence of COPD in the subsequent 9 yr was substantially high (2.8 cases/1,000/yr) and (2) the presence of chronic cough/phlegm identified subjects with a high risk of developing airflow obstruction, despite its low stability. In particular, subjects with the persistence of the symptoms had a threefold-increased risk of developing COPD, independently of their smoking habits.

We also found that dyspnea is a very poor predictor of the occurrence of COPD in young adults, probably because it is a nonspecific symptom, usually occurring in advanced stages of the disease.

Incidence of COPD

To our knowledge, the incidence of COPD, as defined by the GOLD guidelines, has previously only been examined in two studies performed in Northern Europe on relatively small samples (~ 1,000 subjects) of adults aged 18 yr or older and followed for about 10 yr (7, 17). According to Lindberg and coworkers (7), the estimated 10-yr cumulative incidence was 13.5%, whereas the estimated 9-yr cumulative incidence was 6.1% in the study performed by Johannessen and colleagues (17). The difference in the two estimates was mainly due to the different percentage of people over age 65 yr in the studied groups. In our international, large cohort of young adults (20-44 yr), the 10-yr cumulative incidence of COPD was 2.8% (4.6% in adults aged 40-44 yr). This finding points out that COPD is a major health problem even in young adults (10, 18), who are usually not considered to be at risk. In agreement with previous findings (19), we found that the progression toward airflow obstruction is a continuous and gradual process, whereas sudden changes are extremely unlikely. Indeed, subjects who developed COPD during the follow-up already had a lower FEV^sub 1^/FVC ratio at baseline than normal subjects.

Relationship between Chronic Cough/Phlegm and COPD

Whereas the importance of promoting the concept of an "at-risk" population in global health policy is well recognized (5), the appropriateness of stating that chronic respiratory symptoms are markers of an ongoing pathologic process in the airways was already questioned long ago by Fletcher and Peto (19). In their study of a cohort of 792 British working men, they found that these symptoms were unrelated to airflow obstruction. Accordingly, symptoms of chronic productive cough, in the absence of airflow obstruction, were viewed as a "smoker's cough," and were regarded as an innocent symptom.

Since then, there have been several reports showing that respiratory symptoms, such as chronic cough and phlegm, are associated with a more rapid decline in FEV^sub 1^ (20-23), with an all-cause mortality excess (24-27), and with subsequent hospitalization due to COPD (23). Based on this evidence, and from a public health perspective, the GOLD guidelines suggest that the presence of symptoms of chronic cough and phlegm identifies an "at risk" stage that deserves a preventive intervention. However, there have only been a few studies directly investigating the association between respiratory symptoms and airflow obstruction assessed by spirometry. In a retrospective reanalysis of the data from the Copenhagen City Heart Study (6), Vestbo and colleagues found that changes in the prevalence of COPD reported at follow-up were unrelated to the presence of GOLD stage 0 at baseline, in agreement with the findings of Fletcher and Peto (19). In contrast, Lindberg and coworkers, who prospectively studied the incidence of COPD in people with an FEV^sub 1^/FVC ratio of 70% or greater at baseline (7), concluded that GOLD stage 0 identifies subjects at risk of developing COPD. Our prospective study confirms the latter result, even after adjusting for the main potential risk factors.

Several factors may be responsible for the difference in results among the previous studies, such as the difference in the definition of airflow obstruction (FEV^sub 1^ slope instead of FEV^sub 1^/FVC ratio in the study by Fletcher and Peto [19]), the study population (a selected group of male workers in the study by Fletcher and Peto [19]), the outcome (change in prevalence in the study by Vestbo and colleagues [6]), the exclusion criteria (subjects with asthma and those with an FEV^sub 1^/FVC ratio < 70% at baseline not excluded in the study by Fletcher and Peto [19]), and the time frame of the study (retrospective analysis of the data from the Copenhagen City Heart Study in the study by Vestbo and colleagues [6]). However, it is interesting to note that both of the two truly prospective studies (the study by Lindberg and colleagues (7) and the present analysis), which used incident cases of COPD as the outcome, and which defined the population at risk in a similar way, found a positive association between chronic symptoms and COPD.

Underlying Mechanisms

Relatively little is known about the inflammatory response in smokers' lungs and how this relates to disease susceptibility (28, 29). Moreover, the mechanism that causes cigarette smoke to induce a hyperproduction of mucus and the role of chronic mucus hypersecretion in the pathogenesis of airflow obstruction are not fully understood. The studies by Hogg on the nature of small airway obstruction (30) indicate that the progression of COPD from GOLD stage 0 to stage 4 is strongly associated with the thickening of the airway wall and each of its compartments by means of a remodeling process, which results in the accumulation of inflammatory exudates in the lumen. Impaired mucociliary clearance has been shown in young smokers, even in the absence of airflow obstruction and with a short smoking history. This occurs because of an abnormal clearance in the small airways. Patients with an advanced stage of the disease and with evident airway obstruction have mucus retention in the small peripheral airways and in the larger central airways. This cycle is further worsened during episodes of acute viral and bacterial infections, which are common in patients with chronic bronchitis (31). This topic is certainly relevant and needs to be adequately addressed in future studies.

Implications

Although showing that the presence of chronic cough/phlegm identifies subjects with a high risk of developing airflow obstruction, particularly when the symptoms are persistent, our results do not imply that they are a necessary condition, or that the maximum benefit from prevention would be attained if the intervention was limited to people with these symptoms. Indeed, despite the fact that, in our study, the cumulative incidence of COPD among symptomatic subjects was higher than that among ever-smokers (5.1 vs. 3.4%), the absolute number of incident cases of subjects with chronic cough/phlegm at baseline was low, representing only 18.9% of the new cases of the disease.

Accordingly, our results confirm that, from a public health perspective, the prevention of tobacco use and the promotion of smoking cessation is the main strategy to prevent the occurrence and reduce the burden of COPD (in fact, 76.9% of the incident cases represent subjects who were ever-smokers). In addition, our results point out that the presence of chronic cough/ phlegm is not an innocent symptom, but is an early marker of airflow obstruction.

In summary, our results suggest that (1) in epidemiologic research, given the poor stability and reproducibility of the questions on chronic cough/phlegm, the persistence of symptoms should be accurately investigated by using specific questions about their onset, duration, and variability over time, and (2) in clinical practice, the repeated assessment of these symptoms may be particularly useful in identifying subjects at a higher risk of developing COPD.

Strengths and Weaknesses of the Study

The main strength of our study is that it relies on the follow-up of a large population-based cohort made up of young adults. The young age of our cohort allows the investigation of the early phases of COPD, which are crucial to identifying the group of subjects that could benefit from a preventive action, while the international nature of the cohort guarantees a wide generalization of our results.

All the subjects reporting a diagnosis of asthma at baseline were excluded, as were the incident cases of individuals with airflow obstruction who had a diagnosis of asthma during the follow-up. Accordingly, even if we cannot rule out that some cases of undiagnosed asthma, or of other rare conditions (e.g., bronchiectasis and cystic fibrosis), could be present in our cohort, it is extremely unlikely that our incidence estimates are biased because of the presence of subjects with a diagnosis other than COPD. Moreover, the association between chronic cough/ phlegm and the occurrence of COPD reported in our study was stable on the basis of a wide variety of sensitivity analyses.

One limitation of our study is the use of prebronchodilator spirometric values for defining COPD, instead of postbronchodilator values, as suggested by the GOLD guidelines. However, Johannessen and coworkers (32) found that the assessment of associations with prognostic or risk factors is not influenced by the type of lung function assessment (pre- or post-bronchodilation), even if the prevalence of COPD defined without bronchodilation may be overestimated. Moreover, in contrast with Johannessen and coworkers, we excluded from the analysis all the cases of asthma that could show the highest difference between pre- and post-bronchodilation measurements. Therefore, our estimates of the COPD incidence are only expected to be overestimated to a minor extent. However, we had only two measurements of lung function over a 9-yr period, and this could have an unpredictable impact on our results.

Another point that could raise some criticism is the use of the GOLD definition for airflow obstruction. Recently, the new ATS/ERS guidelines on Interpretative Strategies for Lung Function Tests proposed a new definition of airflow obstruction, as a reduction of FEV^sub 1^/FVC ratio below the fifth percentile of the predicted value (33), rather than below a fixed value (0.7). The lower limit of the normal range of the FEV^sub 1^/FVC ratio usually decreases with age (34), and seems to reduce the misclassification of subjects with respect to a fixed cut-off, particularly in older populations than the one studied in the present analysis (35). However, we analyzed our data according to the GOLD and ATS/ERS guidelines because the new definition of COPD has not yet been formally accepted.

As in other longitudinal studies, the participation rate was quite low in some centers. However, this should only have influenced our findings to a minor extent, as they were confirmed when the centers with the lowest participation rate were excluded.

Conclusions

In our large, international cohort of young adults, the incidence of COPD is substantial. The presence of chronic cough/phlegm is an early marker of COPD in a nonignorable proportion of patients, independently of smoking habits. In clinical practice, the persistence of chronic symptoms should be accurately investigated.