Abstract
Background
The prognostic value of respiratory symptom profiles for predicting exacerbation risk and lung function decline remains unclear in mild-to-moderate chronic obstructive pulmonary disease (COPD).
Research question
Are respiratory symptom profiles associated with both exacerbation risk and lung function decline in mild-to-moderate COPD?
Study design and methods
This was a posthoc analysis of data from participants with mild-to-moderate COPD from the SubPopulations and InteRmediate Outcome Measures in COPD Study. Respiratory symptom profiles were identified through latent class analysis. Outcomes included exacerbation rates evaluated by zero-inflated negative binomial regressions, time-to-first exacerbation evaluated by Cox regression and longitudinal forced expiratory volume in 1 second (FEV1) decline evaluated by linear mixed-effects models.
Results
Among the 954 participants with mild-to-moderate COPD, five distinct respiratory symptom profiles were identified. Compared with the ‘minimal respiratory’ profile, the ‘productive cough’ profile was associated with a higher rate of any respiratory exacerbations (relative ratio [RR] 1.84; 95% confidence interval [CI] 1.29 to 2.64) and severe respiratory exacerbations (RR 2.05; 95% CI 1.12 to 3.74). Similarly, the ‘Wheeze’ profile was associated with higher rates of any (RR 1.55; 95% CI 1.12 to 2.15) and severe exacerbations (RR 1.73; 95% CI 1.00 to 2.98). The ‘nearly all respiratory symptoms’ profile was associated with a higher rate of exacerbations (any exacerbation: RR 2.12; 95% CI 1.56 to 2.89; severe exacerbations: RR 2.07; 1.23 to 3.47) and an accelerated annual FEV1 decline (−15.41 mL/year; 95% CI −30.33 to −0.51 mL/year). The ‘dry cough’ profile exhibited the lowest FEV1 trajectory despite a non-significant annual decline.
Interpretation
Respiratory symptom profiles identifies distinct prognostic outcomes in mild-to-moderate COPD. The productive cough is associated with increased exacerbation risk, while the dry cough is associated with a lower lung function trajectory.
Trial registration number
Keywords: COPD Exacerbations, COPD epidemiology
WHAT IS ALREADY KNOWN ON THIS TOPIC.
WHAT THIS STUDY ADDS
Through latent class analysis of longitudinal data from the SubPopulations and InteRmediate Outcome Measures in COPD Study cohort, we identified five distinct symptom profiles. We found that the ‘productive cough’ profile emerged as a key predictor of increased exacerbation risk, whereas the ‘dry cough’ profile was associated with an accelerated annual decline in lung function.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
This evidence highlights that respiratory symptom profiles are associated with divergent clinical impacts in mild-to-moderate COPD.
Introduction
Chronic obstructive pulmonary disease (COPD) is a highly heterogeneous lung condition and a leading cause of mortality worldwide, contributing significantly to the global economic burden.1,3 This heterogeneity is reflected in the variability of clinical features, particularly diverse respiratory symptoms, which have been shown to be associated with clinical outcomes of COPD patients.4,6 Mild-to-moderate COPD accounts for a substantial proportion of all cases, yet many of these patients lack access to precise, individualised treatment strategies. It remains unclear which specific phenotypes require targeted interventions.7 Moreover, patients with mild-to-moderate COPD often experience a more rapid decline in lung function over time.8 Previous study has demonstrated that the frequency of respiratory symptoms is associated with future exacerbation frequency in COPD. However, these studies typically characterised cough based on single symptom items or composite scores, such as COPD assessment tests, rather than using comprehensive respiratory symptom profiles.9 10 A more sophisticated and comprehensive method, such as latent class analysis (LCA), may be a more objective and holistic method for classifying symptom profiles in this population.
Different respiratory symptom profiles have been shown to be associated with adverse outcomes, including accelerated lung function decline, acute exacerbations, hospital admissions and mortality.11,16 However, much of this evidence has been based on individuals with normal lung function. For instance, using data from the Tasmanian Longitudinal Health Study and Coronary Artery Risk Development in Young Adults Study, researchers identified distinct respiratory symptom profiles through LCA.15 These profiles were associated with different rates of decline in the forced expiratory volume in 1 second (FEV1)/forced vital capacity (FVC) ratio among individuals without a clinical diagnosis of COPD. Another study found that chronic productive cough and intermittent productive cough may contribute to worse lung function trajectories.16 However, it remains unclear which specific respiratory symptom profiles should especially alert clinicians to intensify management in patients who have already been diagnosed with mild-to-moderate COPD.
To address this knowledge gap, we conducted a posthoc secondary analysis using data collected through the prospective multicentre observational study, the SubPopulations and InteRmediate Outcome Measures in COPD Study (SPIROMICS).17 Our analysis included SPIROMICS participants aged 40–80 years with spirometry-confirmed mild-to-moderate COPD who had completed a structured respiratory symptoms questionnaire at baseline, along with serial lung function measurements and regular exacerbation assessments. The study objectives were to identify distinct respiratory symptom profiles and to investigate the association between these profiles and both future acute exacerbation risk and accelerated lung function decline.
Methods
Study design and participants
This analysis used data from the SPIROMICS study, a prospective multicentre observational cohort that recruited participants aged 40–80 years. The cohort included individuals with a smoking history of at least 20 pack-years, as well as non-smoking controls without pre-existing lung disease. Longitudinal follow-up consisted of annual in-person visits for up to 3 years postenrolment (visits 1–4), with spirometry performed at each follow-up appointment. Baseline assessments involved comprehensive physical examinations, standardised respiratory symptom questionnaires and postbronchodilator spirometry. Additionally, participants completed structured telephone interviews at 3-month intervals between annual visits to report any changes in smoking behaviour and the occurrence of respiratory exacerbations—defined as acute symptom worsening requiring medical intervention. The study protocol was approved by the Institutional Review Boards of all participating centres prior to data collection (online supplemental e-Table 1). All participants provided written informed consent on enrolment.
Sample and definitions
Our analysis included patients with mild-to-moderate COPD, defined by a postbronchodilator FEV1/FVC ratio <0.70 and an FEV1 ≥50% of the predicted value at baseline. Respiratory symptom profiles were classified using LCA based on structured survey questions. The questionnaire collected detailed information on the duration and frequency of respiratory symptoms, including cough, phlegm production, episodes of productive cough, wheezing, dyspnoea and sleep disturbance. To ensure the analytical robustness, we systematically harmonised all covariates prior to pooling the data. Key covariates included age, self-reported sex, smoking status (current/former), cumulative tobacco exposure (pack-years), self-identified race/ethnicity, height and body mass index.
Outcomes
The primary outcome was the annualised rate of acute exacerbations of COPD (AECOPD). AECOPD were defined as healthcare encounters—including office visits, hospital admissions or emergency department visits—resulting from deterioration due to respiratory symptoms and requiring treatment with antibiotics, systemic corticosteroids or both. Severe AECOPD were specifically defined as events requiring hospitalisation or emergency department care. Second outcomes included time to first AECOPD or severe AECOPD, annualised rate of FEV1 decline (mL/year) and annualised decline in FEV1/FVC ratio (%).
Statistical analysis
LCA, a finite mixture modelling approach, was employed to identify symptom-driven subgroups within the cohort by analysing response patterns across multiple categorical indicators. This method assumes that the observed data arise from a mixture of underlying probability distributions, enabling data-driven identification of clinically distinct phenotypes. We incorporated 18 respiratory symptom-related questions as binary indicators in the LCA models (online supplemental e-Tables 2–4), which were designed to capture a broad spectrum of respiratory symptoms experienced by participants. We sequentially fitted LCA models with 2–9 classes to determine the optimal number of latent classes. Model parameters were estimated using maximum likelihood estimation, which handled missing indicator data under the missing-at-random assumption. This approach ensured effective utilisation of all available data while minimising potential bias. To select the optimal number of classes, we primarily relied on the Bayesian information criterion and the clinical interpretability of symptom profiles. Additionally, we considered other criteria such as class size, entropy and overall model fit statistics to ensure that the chosen model was statistically robust.18 Participants were assigned to the latent class with the maximal posterior probability.
We then examined differences in demographic characteristics, anthropometric measures and spirometry results across the identified endotype clusters. For categorical variables, we used χ2 tests, while for continuous variables, we applied either one-way analysis of variance or Kruskal–Wallis tests to assess statistical significance, depending on whether data distribution assumptions were met. We employed three modelling approaches: (1) zero-inflated negative binomial regression, adjusted for age, sex, height, body mass index (BMI), race, smoking pack-years, current smoking status and history of exacerbation within the 365 days prior to enrolment;19 (2) Cox proportional hazards regression (adjusted for the same covariates) to analyse time to first exacerbation and (3) linear mixed-effects models with participant random intercepts (adjusted for the same covariates plus baseline lung function measures) to assess longitudinal lung function decline. In all models, a two-sided p value <0.05 was considered statistically significant. Analyses were performed using R software, V.4.3.1 (R Foundation for Statistical Computing), except for LAC analysis, which was conducted in Mplus software, V.8.3.
Results
Baseline participant characteristics
Among 2375 participants with complete baseline information, 954 (40.2%) met the spirometric criteria for mild-to-moderate COPD at baseline and were included in our final analysis (online supplemental e-Figure 1). The five-class model demonstrated excellent classification performance, with a mean posterior probability of 0.97, indicating high model adequacy in class assignment (online supplemental e-Table 5). Based on conditional probabilities of cough and phlegm symptoms, five distinct respiratory symptom profiles were identified and visualised in a heatmap (online supplemental e-Table 4). The distribution of symptom profiles was as follows: 287 (30.1%) exhibiting the ‘nearly all respiratory symptoms’ profile, 218 (22.8%) participants were classified as having the ‘minimal respiratory symptoms’ profile, 172 (18.0%) presented with the ‘wheeze’ profile, 158 (16.6%) showed the ‘dry cough’ profile and 119 (12.5%) demonstrated the ‘productive cough’ profile (table 1).
Table 1. Baseline clinical characteristics of participants across the respiratory symptoms subclass identified by latent class analysis.
| Minimal | Wheeze | Dry cough | Productive cough | Nearly all respiratory symptoms | P value | |
|---|---|---|---|---|---|---|
| n=218 | n=172 | n=158 | n=119 | n=287 | ||
| Age—yr | 68.4 (6.9) | 66.1 (7.5) | 64.5 (8.1) | 67.1 (7.4) | 62.6 (8.3) | <0.001 |
| Male sex—no. (%) | 135 (61.9) | 90 (52.3) | 79 (50.0) | 83 (69.7) | 168 (58.5) | 0.006 |
| Black race—no. (%) | 23 (10.6) | 24 (14.0) | 18 (11.4) | 12 (10.1) | 48 (16.7) | <0.001 |
| Body mass index—kg/m2 | 26.7 (4.2) | 28.7 (5.5) | 27.8 (5.3) | 28.7 (4.8) | 27.5 (5.4) | <0.001 |
| Smoking status—no. (%) | <0.001 | |||||
| Former smoked | 170 (77.9) | 144 (83.7) | 81 (51.3) | 85 (71.4) | 98 (34.1) | |
| Current smoked | 48 (22.0) | 28 (16.3) | 77 (48.7) | 34 (28.6) | 189 (65.9) | |
| Smoking index—pack-yr | 49.0 (36.0–64.9) | 43.6 (33.0–60.0) | 48.0 (37.5–64.5) | 47.0 (36.8–62.2) | 49.0 (39.0–64.5) | 0.184 |
| Occupational history of dusts/gases/fumes—no. (%) | 74 (34.1) | 67 (39.0) | 72 (46.2) | 47 (39.5) | 138 (49.3) | 0.007 |
| Family history of respiratory diseases—no. (%) | 42 (24.0) | 51 (37.5) | 47 (33.3) | 30 (31.9) | 96 (39.8) | 0.014 |
| Use of oral steroids—no. (%) | 2 (0.9) | 3 (1.8) | 2 (1.3) | 0 (0.0) | 4 (1.4) | 0.759 |
| Use of inhaled steroids—no. (%) | 39 (17.9) | 72 (42.1) | 47 (29.9) | 42 (35.6) | 119 (41.9) | <0.001 |
| Use of inhaled bronchodilators—no. (%) | 59 (27.1) | 102 (59.6) | 81 (51.9) | 64 (54.2) | 197 (69.4) | <0.001 |
| Postbronchodilator spirometry | ||||||
| FEV1—L | 2.28 (0.73) | 2.10 (0.63) | 2.09 (0.62) | 2.20 (0.64) | 2.10 (0.62) | 0.010 |
| FEV1—% of predicted value | 79.36 (16.21) | 74.13 (14.97) | 74.03 (14.04) | 75.14 (15.60) | 71.22 (14.31) | <0.001 |
| FVC—L | 3.83 (1.02) | 3.62 (0.99) | 3.52 (0.99) | 3.79 (0.97) | 3.68 (0.99) | 0.028 |
| FVC—% of predicted value | 100.36 (15.60) | 97.32 (16.85) | 94.28 (15.54) | 97.01 (17.66) | 94.89 (15.34) | 0.001 |
| The ratio of FEV1/FVC | 0.59 (0.09) | 0.58 (0.08) | 0.60 (0.07) | 0.58 (0.09) | 0.57 (0.08) | 0.019 |
| Airflow reversibility—no. (%) | 76 (34.9) | 80 (46.5) | 67 (42.7) | 52 (43.7) | 140 (48.8) | 0.032 |
Data were presented as means (SD). P values for continuous variables were calculated by student’s t-test or the Wilcoxon rank-sum test, and p values for categorical variables were calculated by the χ2 test. Airway reversibility was defined as an FEV1 value obtained after bronchodilator use that increased by 200 ml or more and by 12% or more from the measurement obtained before bronchodilator use.
FEV1, forced expiratory volume in 1 second; FVC, forced vital capacity.
The ‘minimal respiratory symptoms’ profile served as the reference group in our primary analysis. Participants in the ‘nearly all respiratory symptoms’ profile were more likely to be younger, had the highest current smoking rate, showed the greatest occupational exposure prevalence, demonstrated the strongest family history of respiratory diseases, reported the highest usage rates of inhaled bronchodilators, presented with the lowest baseline lung function, and displayed the highest percentages of airflow reversibility. Among other subclasses, participants in the ‘wheeze’ profile were more likely to use inhaled or oral steroids, while those in the ‘dry cough’ profile typically showed lower absolute values for both postbronchodilator FEV1 and FVC (table 1).
Respiratory exacerbation outcomes
During the follow-up period, exacerbation frequency distribution varied significantly across respiratory symptom profiles (figure 1). The ‘nearly all respiratory symptoms’ profile exhibited the highest incidence of acute exacerbations, followed by the ‘productive cough’ profile, in which 52.1% of participants experienced 0–2 exacerbations per year and 4.2% experienced ≥2 exacerbations per year. In contrast, the ‘dry cough’ profile had the second-lowest exacerbation rate, with approximately 54% remaining exacerbation-free throughout the follow-up period, ranking just above the ‘minimal respiratory symptoms’ profile. Severe exacerbations were most common in the ‘nearly all respiratory symptoms’ profile (31.1%), with the ‘productive cough’ profile ranking second (23.5%).
Figure 1. Annual frequency distribution of acute and severe exacerbations during the follow-up periods across different respiratory profiles. (A) Percentage of patients with acute exacerbations, stratified by frequency: ≥2 exacerbations per year, 0–2 exacerbations per year and no exacerbations per year. (B) Percentage of patients with severe exacerbations, stratified by occurrence during follow-up periods: patients who experienced severe exacerbations and those without severe exacerbations.
Compared with the ‘minimal respiratory symptoms’ profile, participants in the ‘productive cough’ profile had a significantly higher rate of any respiratory exacerbations (relative ratio (RR) 1.84; 95% CI 1.29 to 2.64; p<0.001) and severe respiratory exacerbations (RR 2.05; 95% CI 1.12 to 3.74; p=0.020) during the follow-up period. Similarly, the ‘nearly all respiratory symptoms’ profile showed a significantly higher rate of any exacerbation (RR 2.12; 95% CI 1.56 to 2.89; p<0.001) and severe exacerbations (RR 2.07; 95% CI 1.23 to 3.47; p=0.006) when compared with the ‘minimal respiratory symptoms’ profile (table 2).
Table 2. Respiratory exacerbation outcomes across symptom profiles.
| Unit | Minimal | Wheeze | Dry cough | Productive cough | Nearly all respiratory symptoms | |
|---|---|---|---|---|---|---|
| Unadjusted | ||||||
| Any respiratory exacerbation (through the entire follow-up visit) | RR | Ref. | 2.38 (1.68 to 3.36) p<0.001 |
2.19 (1.54 to 3.13) p<0.001 |
2.38 (1.63 to 3.47) p<0.001 |
3.50 (2.57 to 4.75) p<0.001 |
| Severe respiratory exacerbation (through the entire follow-up visit) | RR | Ref. | 2.62 (1.51 to 4.55) p=0.001 |
2.24 (1.26 to 3.97) p=0.006 |
2.53 (1.39 to 4.63) p=0.003 |
3.78 (2.31 to 6.19) p<0.001 |
| Time to first exacerbation | HR† | Ref. | 2.03 (1.48 to 2.76) p<0.001 |
1.77 (1.27 to 2.46) p=0.001 |
2.11 (1.51 to 2.94) p<0.001 |
2.43 (1.83 to 3.22) p<0.001 |
| Time to first severe exacerbation | HR† | Ref. | 2.92 (1.52 to 5.62) p=0.001 |
1.93 (0.94 to 3.98) p=0.074 |
2.57 (1.26 to 5.24) p=0.010 |
3.08 (1.66 to 5.71) p<0.001 |
| Adjusted‡ | ||||||
| Any respiratory exacerbation (through the entire follow-up visit) | RR | Ref. | 1.55 (1.12 to 2.15) p=0.009 |
1.40 (0.99 to 1.98) p=0.058 |
1.84 (1.29 to 2.64) p=0.001 |
2.12 (1.56 to 2.89) p<0.001 |
| Severe respiratory exacerbation (through the entire follow-up visit) | RR | Ref. | 1.73 (1.00 to 2.98) p=0.049 |
1.23 (0.68 to 2.22) p=0.497 |
2.05 (1.12 to 3.74) p=0.020 |
2.07 (1.23 to 3.47) p=0.006 |
| Time to first exacerbation | HR† | Ref. | 1.45 (1.05 to 2.00) p=0.023 |
1.32 (0.94 to 1.87) p=0.111 |
1.87 (1.33 to 2.63) <0.001 |
1.67 (1.22 to 2.28) p=0.001 |
| Time to first severe exacerbation | HR† | Ref. | 2.16 (1.11 to 4.22) p=0.024 |
1.42 (0.67 to 2.99) p=0.358 |
2.17 (1.04 to 4.53) p=0.039 |
2.13 (1.08 to 4.17) p=0.028 |
Zero-inflated negative binomial regression models.
Proportional hazards regression models.
Adjusted for age, sex, height, BMI, race, smoking pack-years and current smoking status.
RR, relative ratio.
Furthermore, participants in the ‘productive cough’ profile had a shorter time to first any respiratory exacerbations (HR 1.87; 95% CI, 1.33 to 2.63; p<0.001) and severe respiratory exacerbations (HR 2.17; 95% CI 1.04 to 4.53; p=0.039) compared with the ‘minimal respiratory symptoms’ profile. Similarly, the ‘nearly all respiratory symptoms’ profile exhibited a significantly higher risk of any exacerbations (HR 1.67; 95% CI 1.22 to 2.28; p=0.001) and severe exacerbations (HR 2.13; 95% CI 1.08 to 4.17; p=0.028) (table 2).
Rate of lung function decline
We observed accelerated annual declines in FEV1 and the FEV1/FVC ratio across all respiratory symptom profiles, including ‘minimal respiratory symptoms’, ‘wheeze’, ‘dry cough’, ‘productive cough’ and ‘nearly all respiratory symptoms’. The adjusted average rates of FEV1 decline were −46.96, –39.32, −54.75, –52.31, −62.37 mL/year, respectively, while the adjusted average rates of FEV1/FVC ratio decline were −0.60%, −0.55%, −1.06%, −0.84% and −1.15% /year, respectively (online supplemental e-Table 6). Compared with participants in the ‘minimal respiratory symptom’ profile, those in the ‘nearly all respiratory symptoms’ profile showed a steeper average rate of FEV1 decline (−15.41 mL/year; 95% CI −30.33 to −0.51 mL/year) (table 3). Although the average rate of FEV1 decline in the ‘dry cough’ profile was not significantly different from that in the ‘minimal respiratory symptoms’ profile (−7.79 mL/year; 95% CI −24.96 to 8.75 mL/year), the FEV1 trajectory during the follow-up periods remains the lowest (figure 2). Similarly, the average rate of FEV1/FVC ratio decline was steeper in the ‘nearly all respiratory symptoms’ profile (−0.55% /year; 95% CI −0.90 to −0.20% /year) and ‘dry cough’ profile (−0.45% /year; 95% CI −0.84 to −0.07% /year) compared with the ‘minimal respiratory symptom’ profile (table 3). The FEV1/FVC ratio trajectory during the follow-up period was lowest in the ‘nearly all respiratory symptoms’ profile, followed by the ‘wheeze’ profile (figure 2).
Table 3. Examining the association between respiratory symptom class on baseline FEV1 and FEV1 longitudinal change in study cohort.
| Annualised rate of FEV1 change (mL/yr) (95% CI) | P value | Annualised rate of FEV1 to FVC ratio change (%) (95% CI) | P value | |
|---|---|---|---|---|
| Unadjusted | ||||
| Minimal | Ref. | Ref. | ||
| Wheeze | 7.55 (−9.47 to 24.57) | 0.385 | 0.04 (−0.36 to 0.44) | 0.844 |
| Dry cough | −7.42 (−25.34 to 10.50) | 0.417 | −0.37 (−0.79 to 0.06) | 0.090 |
| Productive cough | −6.14 (−26.21 to 13.93) | 0.549 | −0.27 (−0.74 to 0.21) | 0.267 |
| Nearly all respiratory symptoms | −12.10 (−28.25 to 4.06) | 0.142 | −0.56 (−0.89 to -0.19) | 0.004 |
| Adjusted | ||||
| Minimal | Ref. | Ref. | ||
| Wheeze | 7.64 (−8.20 to 23.46) | 0.345 | 0.56 (−0.32 to 0.43) | 0.769 |
| Dry cough | −7.79 (−24.36 to 8.75) | 0.358 | −0.45 (−0.84 to to 0.07) | 0.022 |
| Productive cough | −5.35 (−23.82 to 13.13) | 0.571 | −0.24 (−0.67 to 0.19) | 0.272 |
| Nearly all respiratory symptoms | −15.41 (−30.33 to −0.51) | 0.043 | −0.55 (−0.90 to −0.20) | 0.002 |
Adjusted for age, sex, height, BMI, race, smoking pack-years, current smoking status and baseline lung function as applicable to the specific variable of interest.
FEV1, forced expiratory volume in 1 second; FVC, forced vital capacity.
Figure 2. Annual lung function decline among respiratory symptom profiles during the follow-up durations. (A) Unadjusted annual FEV1 decline. (B) Adjusted annual FEV1 decline. Adjusted for baseline age, gender, race, body mass index, height, smoking pack-years and smoking status. (C) Unadjusted annual FEV1/FVC decline. (D) Adjusted annual FEV1/FVC decline. Adjusted for baseline age, gender, race, body mass index, height, smoking pack-years and smoking status. FEV1, forced expiratory volume in 1 second; FVC, forced vital capacity.
Discussion
Secondary analysis of data from the SPIROMICS study revealed an inverse association between distinct respiratory symptom profiles and both future acute exacerbations risk and lung function decline. Our principal findings were as follows: (1) the ‘nearly all respiratory symptoms’ profile and the ‘productive cough’ profile were significantly associated with increased frequency and risk of respiratory exacerbations (both total and severe); (2) the ‘nearly all respiratory symptoms’ profile was associated with accelerated FEV1 decline and progressive FEV1/FVC ratio reduction and (3) the ‘dry cough’ profile was associated with accelerated FEV1/FVC ratio decline and persistently lower FEV1 trajectory. Collectively, these findings provide novel evidence that identifying respiratory symptom profiles represents a critical and useful approach for reducing acute exacerbation risk and mitigating accelerated lung function decline in mild-to-moderate COPD.
While prior studies have established that chronic respiratory symptoms predict accelerated lung function decline, they have not examined symptom profiles in relation to the risk of AECOPD.11 20 These studies were also limited by either focusing on populations with normal lung function or relying on manual symptom classification approaches. The clinical relevance of such an investigation is amplified by the recognition that exacerbation-like events are significant drivers of morbidity and health impairment, irrespective of COPD status.21 22 Therefore, it is critical to bridge this gap by systematically characterising symptom clusters and evaluating their utility in predicting AECOPD risk. A prospective study of 2749 adults that manually categorised symptoms found that cough-related symptoms were associated with future obstructive physiology and radiographic emphysema, but this analysis examined individual symptoms (chronic cough) in isolation rather than symptom clusters.11 Such individual symptom classifications mask substantial underlying heterogeneity, as cough may present alone or in combination with wheezing and/or phlegm production—distinct clinical presentations that likely carry different prognostic implications. Through LCA, we found that specific respiratory symptom profiles confer a higher risk of both frequent exacerbations and lung function decline. These findings advance prior work by demonstrating that multidimensional symptom profiling can better identify high-risk subgroups for targeted monitoring and account for underlying heterogeneity. Another study, which included 975 employed, working-age individuals with current cough in Finland, used a data-driven approach (ie, K-means cluster analysis) to identify cough clusters. However, this study lacked data to associate these clusters with longitudinal clinical characteristics.20
In our analysis, chronic productive cough emerged as the most clinically significant symptom profile, demonstrating strong associations with both the highest frequency of acute exacerbations and the highest frequency of severe exacerbations. Previous study reported that frequent productive cough—defined using two questions from the St George’s Respiratory Questionnaire—was more likely to be associated with moderate or severe exacerbations over the subsequent 12 months.23 Similar findings have been corroborated by other studies.24,26 However, these prior investigations have predominantly focused on single respiratory symptoms (eg, chronic productive cough). In contrast, our study systematically compared prognostic differences among multiple respiratory symptom profiles, thereby providing a more comprehensive understanding of their clinical implications.
The underlying mechanism linking chronic productive cough to future exacerbations remains unclear. One plausible explanation is that AECOPD diagnosis requires sustained symptom worsening; thus, patients with baseline chronic cough and sputum production may more readily meet diagnostic thresholds during acute worsening. Alternatively, these symptoms may represent a distinct COPD endotype characterised by persistent mucus hypersecretion. Recent evidence supports this hypothesis, demonstrating elevated airway mucin concentrations (eg, MUC5AC, MUC5B) in patients with chronic bronchitis, which suggests a potential role for mucin biomarkers for COPD subtyping.27 Elevated sputum mucin concentrations have been independently associated with exacerbation risk in COPD patients.28 This association may be attributed to mucus plugging-induced airway obstruction or secondary bacterial colonisation, providing a biologically plausible explanation for the association between chronic productive cough and exacerbation susceptibility.
Although not reaching conventional significance thresholds, we observed that patients with mild-to-moderate COPD presenting with dry cough tended to have lower FEV1 levels and a more rapid decline in FEV1/FVC ratio throughout the follow-up period compared with those with other respiratory symptom profiles—a finding consistent with prior study in the overall COPD population.29 The decline in FEV1/FVC ratio carries significant clinical implications in mild-to-moderate COPD.30 31 Previous studies have reported significant differences in FEV1/FVC ratio among patients with different symptom scores in this population.31 As the STAR grade increases—a classification system based on FEV1/FVC ratio, the worsening of pulmonary structure, the rate of FEV1 decline and the frequency of exacerbations all become more pronounced. These findings highlight the clinical importance of early identification of FEV1/FVC ratio decline in managing mild-to-moderate COPD. The presence of a dry cough may serve as a potential clinical marker for patients at risk of rapid functional deterioration, warranting closer monitoring. Early intervention with dual bronchodilators may be considered for those with dry cough-predominant mild-to-moderate COPD.
The mechanistic basis for dry cough-associated progression in COPD may involve multiple pathways. First, chronic dry cough in COPD may result from ciliary motility dysfunction, leading to impaired mucus clearance. This defect creates a vicious cycle: retained mucus exacerbates airway inflammation and epithelial damage, which in turn accelerates lung function decline.32 33 Second, comorbidities such as gastro-oesophageal reflux, which usually cause chronic dry cough, may promote airway irritation and inflammatory progression, both of which can contribute to a decline in lung function.34 Notably, neutrophilic airway inflammation characteristic of COPD appears to play a dual role in this process. Elevated levels of neutrophils and associated chemotactic cytokines have been observed in participants with non-asthmatic chronic dry cough, and those inflammatory factors may accelerate lung function deterioration.35 36
Based on the identified symptom profiles, we propose the following clinical management recommendations: for the ‘nearly all respiratory symptoms’ profile, intensive intervention including dual bronchodilators with potential inhaled corticosteroids is warranted, accompanied by close monitoring every 3–6 months. The ‘productive cough’ profile would benefit from mucus-clearing strategies with mucolytic agents and infection prevention measures, including appropriate vaccinations, to reduce the risk of acute exacerbations. For the ‘dry cough’ profile, while exacerbation risk is moderate, close monitoring of lung function decline is essential, with consideration of dual bronchodilators to preserve lung function. Finally, the ‘minimal respiratory symptoms’ profile should maintain standard COPD care with routine monitoring and risk education.
This study has several notable strengths. First, our analysis was based on a large, prospective cohort, which enabled the evaluation of diverse clinical outcomes across different respiratory symptom profiles. Second, the application of data-driven LCA represents a methodological advance over conventional symptom assessment methods. This analytical approach facilitates the identification of multidimensional symptom patterns rather than isolated symptoms, thereby capturing the complex interplay of clinical manifestations and their collective prognostic implications. Third, by focusing on mild-to-moderate COPD populations, we address an important research gap, as this understudied subgroup may derive particular benefit from early symptom pattern recognition and targeted interventions. Nevertheless, certain limitations warrant consideration. Although the cohort size ensures statistical robustness, the generalisability of our findings requires external validation using standardised protocols across diverse populations. While LCA methodology provides valuable insights, the clinical translatability and temporal stability of identified symptom profiles need to be confirmed through longitudinal replication studies. Furthermore, the relatively short follow-up duration may limit the comprehensive characterisation of the prognostic value of symptom patterns for long-term outcomes.
Conclusion
Our study provides important insights into the clinical characteristics of mild-to-moderate COPD, with particular emphasis on the impact of different respiratory symptom profiles. We found that the ‘productive cough’ profile was associated with a higher risk of future respiratory exacerbations, while the ‘dry cough’ profile was associated with accelerated lung function decline. These results highlight the importance of comprehensive symptom profiling in clinical decision-making, enabling personalised interventions to mitigate exacerbation risk and slow disease progression in patients with mild-to-moderate COPD.
Supplementary material
Footnotes
Funding: National Natural Science Foundation of China (82270044, 82170042, 82200044, 82300059), Science and Technology Program of Guangzhou (202201020451), National Key Research and Development Program of China (2017YFC1310600, 2022YFF0710802), Shenzhen Science and Technology Program (JCYJ20210324114400002), State Key Laboratory of Respiratory Disease, Guangzhou Medical University (SKLRD-Z-202317, SKLRD-OP-202401), China Postdoctoral Science Foundation (2022M720915) and the China International Medical Foundation (Z-2017-24-2301) supported our study. The funder had no role in the study design, data collection, analysis, interpretation or the writing of the manuscript.
Provenance and peer review: Not commissioned; externally peer reviewed.
Patient consent for publication: Consent obtained directly from patient(s).
Ethics approval: This study involves human participants. This multicentre study was conducted with the approval of the institutional review boards at all participating sites, including Columbia University (IRB-AAAE9315), Johns Hopkins University (NA_00035701/CR00018131), National Jewish Health (19970), Temple University (21416), University of Alabama at Birmingham (120906004), University of California, Los Angeles (10001740), University of California, San Francisco (10-03196), University of Illinois (2013-0939), University of Iowa (2013088719), University of Michigan (HUM00036346), University of Utah (00027298) and Wake Forest University (00012805). All participants provided written informed consent upon enrolment. Participants gave informed consent to participate in the study before taking part.
Patient and public involvement: Patients and/or the public were not involved in the design, conduct, reporting or dissemination plans of this research.
Data availability statement
Data are available upon reasonable request.
References
- 1.Mannino DM, Roberts MH, Mapel DW, et al. National and Local Direct Medical Cost Burden of COPD in the United States From 2016 to 2019 and Projections Through 2029. Chest. 2024;165:1093–106. doi: 10.1016/j.chest.2023.11.040. [DOI] [PubMed] [Google Scholar]
- 2.Global Strategy for the Diagnosis, Management, and Prevention of Chronic Obstructive Pulmonary Diseas (2024 REPORT) 2024. https://goldcopd.org/2024-gold -report/ Available.
- 3.Xu J, Murphy SL, Kochanek KD, et al. Mortality in the United States, 2021. NCHS Data Brief. 2022;2022:1–8. [PubMed] [Google Scholar]
- 4.de Torres JP, Marin JM, Martinez‐Gonzalez C, et al. The importance of symptoms in the longitudinal variability of clusters in COPD patients: A validation study. Respirology . 2018;23:485–91. doi: 10.1111/resp.13194. [DOI] [PubMed] [Google Scholar]
- 5.Miravitlles M, Izquierdo JL, Esquinas C, et al. The variability of respiratory symptoms and associated factors in COPD. Respir Med. 2017;129:165–72. doi: 10.1016/j.rmed.2017.06.017. [DOI] [PubMed] [Google Scholar]
- 6.Contoli M, Baraldo S, Conti V, et al. Airway inflammatory profile is correlated with symptoms in stable COPD: A longitudinal proof‐of‐concept cohort study. Respirology . 2020;25:80–8. doi: 10.1111/resp.13607. [DOI] [PubMed] [Google Scholar]
- 7.Çolak Y, Nordestgaard BG, Lange P, et al. Prognosis of Patients with Chronic Obstructive Pulmonary Disease Not Eligible for Major Clinical Trials. Am J Respir Crit Care Med. 2022;206:271–80. doi: 10.1164/rccm.202110-2441OC. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Tantucci C, Modina D. Lung function decline in COPD. Int J Chron Obstruct Pulmon Dis. 2012;7:95–9. doi: 10.2147/COPD.S27480. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Li X, Song Q, Cheng W, et al. The clinical characteristics and outcomes of different inhaled therapies in chronic obstructive pulmonary disease patients with frequent cough. Ann Med. 2023;55:2304107. doi: 10.1080/07853890.2024.2304107. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Papi A, Hughes R, Del Olmo R, et al. Relationships between symptoms and lung function in asthma and/or chronic obstructive pulmonary disease in a real-life setting: the NOVEL observational longiTudinal studY. Ther Adv Respir Dis. 2024;18:17534666241254212. doi: 10.1177/17534666241254212. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Kalhan R, Dransfield MT, Colangelo LA, et al. Respiratory Symptoms in Young Adults and Future Lung Disease. The CARDIA Lung Study. Am J Respir Crit Care Med. 2018;197:1616–24. doi: 10.1164/rccm.201710-2108OC. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Hamad G, Rigby A, Morice AH. Prognostic significance of chronic respiratory symptoms in individuals with normal spirometry. Eur Respir J. 2020;55:1902093. doi: 10.1183/13993003.02093-2019. [DOI] [PubMed] [Google Scholar]
- 13.Woodruff PG, Barr RG, Bleecker E, et al. Clinical Significance of Symptoms in Smokers with Preserved Pulmonary Function. N Engl J Med. 2016;374:1811–21. doi: 10.1056/NEJMoa1505971. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Balte PP, Chaves PHM, Couper DJ, et al. Association of Nonobstructive Chronic Bronchitis With Respiratory Health Outcomes in Adults. JAMA Intern Med. 2020;180:676–86. doi: 10.1001/jamainternmed.2020.0104. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Perret JL, Bui DS, Pistenmaa C, et al. Associations between life-course FEV1/FVC trajectories and respiratory symptoms up to middle age: analysis of data from two prospective cohort studies. Lancet Respir Med. 2025;13:130–40. doi: 10.1016/S2213-2600(24)00265-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Zhang J, Lodge CJ, Walters EH, et al. Association of novel adult cough subclasses with clinical characteristics and lung function across six decades of life in a prospective, community-based cohort in Australia: an analysis of the Tasmanian Longitudinal Health Study (TAHS) Lancet Respir Med. 2024;12:129–40. doi: 10.1016/S2213-2600(23)00340-5. [DOI] [PubMed] [Google Scholar]
- 17.Couper D, LaVange LM, Han M, et al. Design of the Subpopulations and Intermediate Outcomes in COPD Study (SPIROMICS) Thorax. 2014;69:491–4. doi: 10.1136/thoraxjnl-2013-203897. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Weller BE, Bowen NK, Faubert SJ. Latent Class Analysis: A Guide to Best Practice. Journal of Black Psychology. 2020;46:287–311. doi: 10.1177/0095798420930932. [DOI] [Google Scholar]
- 19.Han MK, Quibrera PM, Carretta EE, et al. Frequency of exacerbations in patients with chronic obstructive pulmonary disease: an analysis of the SPIROMICS cohort. Lancet Respir Med. 2017;5:619–26. doi: 10.1016/S2213-2600(17)30207-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Koskela HO, Selander TA, Lätti AM. Cluster analysis in 975 patients with current cough identifies a phenotype with several cough triggers, many background disorders, and low quality of life. Respir Res. 2020;21:219. doi: 10.1186/s12931-020-01485-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Fortis S, Strand M, Bhatt SP, et al. Respiratory Exacerbations and Lung Function Decline in People with Smoking History and Normal Spirometry. Am J Respir Crit Care Med. 2025;211:957–65. doi: 10.1164/rccm.202401-0023OC. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Hou D, Liu Z, Li X, et al. Impacts of COPD exacerbation history on mortality and severe cardiovascular events among patients with COPD in China: a retrospective cohort study. Respir Res. 2025;26:252. doi: 10.1186/s12931-025-03316-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Hughes R, Rapsomaniki E, Janson C, et al. Frequent productive cough: Symptom burden and future exacerbation risk among patients with asthma and/or COPD in the NOVELTY study. Respir Med. 2022;200:106921. doi: 10.1016/j.rmed.2022.106921. [DOI] [PubMed] [Google Scholar]
- 24.Kim V, Han MK, Vance GB, et al. The chronic bronchitic phenotype of COPD: an analysis of the COPDGene Study. Chest. 2011;140:626–33. doi: 10.1378/chest.10-2948. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Burgel P-R, Nesme-Meyer P, Chanez P, et al. Cough and sputum production are associated with frequent exacerbations and hospitalizations in COPD subjects. Chest. 2009;135:975–82. doi: 10.1378/chest.08-2062. [DOI] [PubMed] [Google Scholar]
- 26.Kim V, Zhao H, Regan E, et al. The St. George’s Respiratory Questionnaire Definition of Chronic Bronchitis May Be a Better Predictor of COPD Exacerbations Compared With the Classic Definition. Chest. 2019;156:685–95. doi: 10.1016/j.chest.2019.03.041. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Kesimer M, Ford AA, Ceppe A, et al. Airway Mucin Concentration as a Marker of Chronic Bronchitis. N Engl J Med. 2017;377:911–22. doi: 10.1056/NEJMoa1701632. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Radicioni G, Ceppe A, Ford AA, et al. Airway mucin MUC5AC and MUC5B concentrations and the initiation and progression of chronic obstructive pulmonary disease: an analysis of the SPIROMICS cohort. Lancet Respir Med. 2021;9:1241–54. doi: 10.1016/S2213-2600(21)00079-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Koo H-K, Park S-W, Park J-W, et al. Chronic cough as a novel phenotype of chronic obstructive pulmonary disease. Int J Chron Obstruct Pulmon Dis. 2018;13:1793–801. doi: 10.2147/COPD.S153821. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Bhatt SP, Nakhmani A, Fortis S, et al. FEV(1)/FVC Severity Stages for Chronic Obstructive Pulmonary Disease. Am J Respir Crit Care Med. 2023;208:676–84. doi: 10.1164/rccm.202303-0450OC. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Calverley PMA. A STAR Is Born: A New Approach to Assessing Chronic Obstructive Pulmonary Disease Severity. Am J Respir Crit Care Med. 2023;208:647–8. doi: 10.1164/rccm.202306-1106ED. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Whitsett JA. Airway Epithelial Differentiation and Mucociliary Clearance. Ann Am Thorac Soc. 2018;15:S143–8. doi: 10.1513/AnnalsATS.201802-128AW. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Shah A, Shoemark A, MacNeill SJ, et al. A longitudinal study characterising a large adult primary ciliary dyskinesia population. Eur Respir J. 2016;48:441–50. doi: 10.1183/13993003.00209-2016. [DOI] [PubMed] [Google Scholar]
- 34.Baldomero AK, Wendt CH, Petersen A, et al. Impact of gastroesophageal reflux on longitudinal lung function and quantitative computed tomography in the COPDGene cohort. Respir Res. 2020;21:203. doi: 10.1186/s12931-020-01469-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Jatakanon A, Lalloo UG, Lim S, et al. Increased neutrophils and cytokines, TNF-alpha and IL-8, in induced sputum of non-asthmatic patients with chronic dry cough. Thorax. 1999;54:234–7. doi: 10.1136/thx.54.3.234. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Córdoba-Lanús E, Baz-Dávila R, Espinoza-Jiménez A, et al. IL-8 gene variants are associated with lung function decline and multidimensional BODE index in COPD patients but not with disease susceptibility: a validation study. COPD. 2015;12:55–61. doi: 10.3109/15412555.2014.908831. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Data are available upon reasonable request.