Prognostic value of gait speed for exacerbations and mortality in COPD

Authors: Moon Jin Kim, Seohyun Kim, Hyesoo Kim, Jae Ha Lee, Chin Kook Rhee, Seoung Ju Park, Yu-il Kim, Woo Jin Kim, Kwang Ha Yoo, Tai Joon An

Abstract

Background

Gait speed, a key component of exercise capacity, has been underutilised in COPD, despite its prognostic potential. We aimed to evaluate the association between gait speed and clinical outcomes in COPD using 3-year longitudinal data from the Korean COPD Subgroup Study cohort.

Methods

Poor gait speed (<1.0 m·s^−1^) was defined by usual pace during the 6-min walk test per Asian Working Group for Sarcopenia 2019 criteria. Lung function, symptoms, acute exacerbations (AEs) and mortality were compared between gait speed groups. Analyses included propensity score-matching, quartile classification, subgroup analyses and longitudinal trajectory modelling using random coefficient models.

Results

Among 2063 participants, poor gait speed (n=831, 40.3%) was associated with older age, higher symptom burden and more previous AEs despite similar lung function. This group showed higher AE risk and frequency than the normal-speed adjusted odds ratios 1.37–1.45 for moderate and 1.64–1.65 for severe AEs; adjusted incidence rate ratios 1.24–1.36 for moderate and 1.63–1.86 for severe AEs. The 3-year mortality was significantly higher in the poor-gait-speed group (adjusted hazard ratio 2.30, 95% CI 1.42–3.73). Longitudinally, the poor-gait-speed group demonstrated persistently worse COPD Assessment Test (CAT) and St George's Respiratory Questionnaire for COPD scores at baseline, with modest CAT worsening over time (+0.44 point/year, p=0.01), while lung function decline was similar.

Conclusions

Gait speed provides a simple, integrative marker that independently predicts exacerbation risk, mortality and symptom progression in COPD.

Introduction

COPD is a progressive and irreversible respiratory disease characterised by airflow limitation and airway inflammation [1]. It imposes a considerable global burden due to its high prevalence and contributes to increased mortality [2, 3]. It significantly impairs patient's quality of life, including activities of daily living, work and leisure [4]. Among various clinical outcomes, exercise capacity is a key predictor of symptom severity, lung function deterioration, exacerbation risk and mortality [5].

COPD reduces exercise capacity through chronic airflow limitation, dynamic hyperinflation, impaired gas exchange and peripheral muscle dysfunction. These abnormalities diminish ventilatory reserve and increase the metabolic cost of ambulation. Exercise capacity and physical activity are central determinants of COPD outcomes, influencing symptoms, healthcare utilisation and survival [6]. Exacerbations further accelerate declines in walking ability underscoring the systemic nature of exercise limitation in COPD [7].

The 6-min walk test (6MWT) is routinely used in clinical practice to assess exercise capacity in COPD [8]. It is used to quantify the distance patients can walk at their usual pace [9]. However, its ability to predict long-term outcomes remains debated due to its submaximal and semi-quantitative nature and practical limitations. Some studies suggest that the 6MWT may serve as a predictor of mortality, similar to the BODE (body mass index, airflow obstruction, dyspnoea and exercise capacity) index; however, critics argue that its original design was intended to reflect the patient's “usual pace” [10–13]. While the incremental shuttle walk test (ISWT) allows for maximal effort assessment, its use in older adults is limited due to leg instability, dyspnoea and age-related factors [14].

Despite its well-established role in sarcopenia and geriatric assessments, gait speed has been relatively underexplored in the context of COPD, where endurance parameters such as 6MWT distance (6MWD) are more commonly employed [15, 16]. As gait speed may capture broader aspects of physiological reserve (integrating pulmonary, cardiovascular and musculoskeletal limitations) it has potential to complement existing prognostic tools in COPD managements [8, 17–20].

In this study, we investigated the prognostic relevance of 6MWT gait speed in a large, representative COPD cohort from South Korea.

Methods

Data source and study population

We extracted 3-year follow-up data from Korean COPD Subgroup Study (KOCOSS) cohort, an ongoing, prospective observational cohort in South Korea that has been collecting data every 6 months from 55 referral hospitals since 2012 (ClinicalTrials.gov ID CT02800499) [21–23]. Eligible participants were aged ≥40 years, had a post-bronchodilator (BD) forced expiratory volume in 1 s (FEV1)/forced vital capacity ratio <0.7, and exhibited respiratory symptoms. Exclusion criteria included inability to complete pulmonary function tests, recent myocardial infarction or cerebrovascular event (≤3 months), pregnancy, rheumatoid diseases, active malignancy or systemic corticosteroid (≥10 mg·day^−1^) within 1 month. In addition, participants without valid 6MWD or time data were excluded. Of the 3459 COPD participants, 2063 participants who completed the 6MWT were included in the study (figure 1). The extracted variables 1) 6MWT-related gait speed (calculated from distance and time); 2) demographics and clinical age, sex, smoking status, body mass index (BMI) categories (<18.5, 18.5–22.9, 23.0–24.9 and ≥25 kg·m^−2^) and baseline heart disease (ischaemic heart diseases and heart failure); 3) COPD-related baseline blood eosinophil count (BEC) (>300 cells·μL^−1^), use of inhaled corticosteroids (ICSs), Global Initiative for Chronic Obstructive Lung Disease (GOLD) grade, history of acute exacerbations (AEs) in the previous year, COPD Assessment Test (CAT) and St George's Respiratory Questionnaire for COPD (SGRQ-C); and 4) clinical 3-year AEs and mortality data.

The 6MWT protocol and definition of poor-gait-speed group

The 6MWT, conducted according to American Thoracic Society (ATS)/European Respiratory Society (ERS) guidelines, is a well-validated submaximal test of voluntary exercise capacity with high reproducibility in COPD [24–26]. Tests were performed in a 40-m straight hallway, and standardised scripted verbal encouragement was provided at 1-min intervals. In accordance with guideline recommendations, brief voluntary rests (either standing or sitting) were permitted, and the timer continued without interruption. In addition to distance and total elapsed time, we documented whether participants completed the full 6 min or stopped early due to symptoms or clinical judgment according to guideline recommendations. Early termination was recorded as a binary completion status.

The relationship between gait speed and 6MWD was not strictly linear but demonstrated a strong monotonic association (Pearson r=0.54 and Spearman rho=0.93), as visualised by generalised additive model-based regression (supplementary figure S1). Inflection point analysis using the “gratia” package in R identified a data-driven threshold at approximately 1.11 m·s^−1^. This threshold was closely aligned with the 2019 Asian Working Group for Sarcopenia (AWGS) criteria, supporting the adoption of 1.0 m·s^−1^ as the cut-off to classify participants into normal and poor-gait-speed-groups [27].

Study outcomes

The primary outcomes were AEs and mortality. AEs were defined by an acute worsening of symptoms (cough, sputum and breathlessness) compared with the usual condition, leading to additional management with antibiotics or systemic corticosteroids [28, 29]. Moderate AEs required systemic corticosteroids or antibiotics in an outpatient setting, whereas severe AEs required emergency room visits or hospitalisation. These definitions were consistent with GOLD guideline [30]. Secondary outcomes, including FEV1, CAT and SGRQ-C, were assessed longitudinally according to baseline gait speed classification to examine differences in clinical trajectories during follow-up.

Statistical analysis

A t-test and chi-squared test was used for continuous and categorical variables, respectively. Linear mixed models assessed differences in FEV1 (%), CAT and SGRQ-C by gait speed group. Trajectories over 3 years were analysed using random coefficients. Missing data were assumed to be missing at random based on data inspection. Associations between gait speed and AEs were analysed using logistic and negative binomial regression to estimate odds ratios (ORs) and incidence rate ratios (IRRs) with 95% CIs. Adjusted ORs (aORs) and IRRs (aIRRs) were calculated after adjusting for age, sex, smoking status, BMI, underlying heart disease, baseline BEC (≥300 cells·μL^−1^), ICS use, previous AEs and GOLD grade. The same models were applied to propensity score (PS)-matched cohorts generated by 1 nearest-neighbour-matching using the same covariates. For survival analysis, Kaplan–Meier curves and the log-rank test were used to compare mortality according to gait speed categories. Sex- and age-adjusted hazard ratios (aHRs) and 95% CIs were estimated using Cox proportional hazards models. Subgroup analyses were performed according to age group (40–69 versus ≥70 years), GOLD severity (GOLD 1–2 versus 3–4) and BEC (<300 versus ≥300 cells·µL^−1^) to evaluate potential effect modification. For each subgroup, Cox proportional hazard models, including gait speed categories and interaction terms, were used to assess possible differences in associations. No additional covariates were included in subgroup analyses to minimise missing data and preserve statistical power. A two-sided p-value <0.05 was considered statistically significant. All analyses were performed using R software (v.4.1.3; https://www.r-project.org) with relevant packages including lmerTest, survival and MatchIt.

Ethical approval

All participants provided written informed consent, with the confidentiality of their personal information being ensured. The ethics committees at each participating medical centre approved the study protocol of KOCOSS.

Results

Baseline characteristics

Table 1 summarises the baseline characteristics of the study population. Nearly 93% of participants were male (n=1916). The poor-gait-speed group was older than the normal-speed group (70.1±8.1 versus 67.7±7.4 years; p<0.01). Current smoking was less prevalent in the poor-speed group (20.2% versus 29.1%; p<0.01). A history of exacerbations during the previous year was more frequent in the poor-speed group (29.2% versus 17.1%; p<0.01). Mean baseline FEV1 (%) was comparable between groups (p=0.89); however, GOLD stage distribution revealed a higher proportion of more severe airflow limitation in the poor-speed group (GOLD 29.5% versus 25.6%; GOLD 7.9% versus 4.7%). No significant group differences were observed for BMI, BEC or ICS use. After PS matching, baseline characteristics were well balanced across groups, with persistent differences remaining in symptom burden and previous exacerbations. In quartile-based analyses, lower gait speed quartiles were associated with older age, higher symptom burden and worse lung function (all p<0.01), as well as progressively higher AE frequency and 3-year mortality (supplementary table S1).

Lung functions according to gait speed

Over 3 years, the group with normal 6MWT gait speed maintained relatively stable lung function, whereas the poor-speed group exhibited a decline. Notably, a significant difference emerged at year 2 (normal: 58.8±17.0%; 54.9±18.6%, p<0.01), whereas differences at years 0, 1 and 3 were not significant (figure 2a). Quartile analysis revealed a significant trend in FEV1 (%) over time, with Q4 consistently showing the highest values from baseline to year 3 (all p<0.05), whereas differences among Q1 to Q3 were not substantial (figure 2b).

CAT and SGRQ-C according to gait speed

CAT and SGRQ-C scores exhibited a significant and consistent trend according to 6MWT gait speed (figure 2c–f). Poor-speed group consistently showing higher CAT and SGRQ-C scores across all time points (p<0.01). Quartile analyses demonstrated a progressive association, with lower gait speed quartiles corresponding with worse baseline (Q1: 15.8±8.6 to Q4: 12.3±7.3, p<0.01) and SGRQ-C (Q1: 36.1±23.9 to Q4: 24.3±16.9, p<0.01). These differences progressively widened over time and were most pronounced by year 3 (p<0.01).

Longitudinal lung function and symptoms trajectories according to gait speed group

In longitudinal analyses using random coefficient models, no significant baseline difference in lung function was observed between groups, and both groups showed similar lung function trajectories over 3 years without significant group-by-time interaction. In contrast, symptom burden was consistently higher in the poor-gait-speed group at baseline across all measures. For CAT scores, the poor-gait-speed group showed higher baseline values (estimate: +1.8, p<0.01) with a modest worsening over time (+0.1±0.2 per year, p=0.01). SGRQ-C total and subdomain scores were also significantly higher at baseline in the poor-speed group (total: +6.8; +3.7; +8.1; +7.1; all p<0.001), but no significant differences in longitudinal slopes were observed (table 2).

Exacerbations according to gait speed group

Moderate exacerbations were consistently more frequent in the poor-gait-speed group across all follow-up periods (e.g. 46.5% versus 33.3% at 1 year). Severe AEs showed a similar pattern (e.g. 13.7% versus 7.9% at 1 year) (table 1). The poor-speed group was associated with the occurrence of moderate AEs during the 1- and 2-year follow-up periods (1-year aORs 1.45, 95% CI 1.08–1.94; 2-year aORs 1.37, 95% CI 1.00–1.86; both p<0.05) (table 3). The risk for severe AEs was similarly increased over 2 and 3 years (2-year aORs 1.64, 95% CI 1.10–2.44; 3-year aORs 1.65, 95% C: 1.11–2.46; both p=0.01). The poor-speed group consistently exhibited a higher frequency of both moderate and severe AEs (aIRRs for moderate AEs: 1.24–1.36; for severe AEs: 1.63–1.86; all p<0.05), except for 3 years (p=0.05). These patterns remained consistent in the PS-matched cohort (moderate AE aORs for 1 1.50; severe AE aORs: 1.55–1.61; moderate AE aIRRs: 1.29–1.31; severe AE aIRRs: 1.51–1.68). Similar trends were observed in supplementary quartile-based analyses, showing progressively higher exacerbation risk and frequency with slower gait speed (supplementary table S2).

3-Year survival analysis

Kaplan–Meier survival curves (figure 3a) demonstrated higher 3-year mortality in the poor-gait-speed group, with 48 deaths (6.5%) compared with 26 deaths (2.4%) in the normal-speed group (p<0.01). This pattern was further confirmed by cumulative incidence plots (figure 3b). After adjusting for age and sex, poor gait speed was associated with a significantly increased risk of mortality (aHR 2.30, 95% CI 1.42–3.73; log-rank p=0.001). In quartile-based analyses, the lowest quartile (Q1) showed a significantly higher mortality risk compared with the highest quartile (Q4) (aHR 2.68, 95% CI 1.26–5.69; p=0.01), with a dose–response relationship across quartiles (log-rank p<0.001; figure 3c).

In subgroup analyses, the adverse prognostic impact of poor gait speed was consistently observed across age groups, BEC and GOLD stages (table 4). Notably, among participants aged ≥70 years, poor gait speed was associated with a markedly elevated mortality risk (aHR 5.32, 95% CI 2.72–10.39; p<0.01), with consistent patterns across other subgroups.

Discussion

This study demonstrates that 6MWT-derived gait speed is a robust and clinically meaningful predictor of adverse outcomes in COPD. Utilising a large-scale, prospective Korean cohort, we found that slower gait speed, particularly below the clinically meaningful threshold of 1 m·s^−1^, was associated with increased symptom burden, higher exacerbation risk, more rapid lung function decline and greater mortality over 3 years of follow-up. These associations remained significant after adjustment for confounders and were confirmed in matched analyses.

The prognostic value of the 6MWD in COPD is well established [31, 32]. However, the 6MWD reflects not only physiological capacity but also motivation, pacing strategy, endurance and other effort-dependent factors that introduce variability unrelated to underlying functional reserve [26]. In our cohort, 6MWD and gait speed were highly correlated (ρ=0.93), yet early termination contributed to modest patients who did not sustain the full 6 min had a shorter denominator for speed calculation. Additionally, because brief voluntary rests are permitted without pausing the timer, two individuals may walk similar distances but different walking times, resulting in different average speeds. Thus, gait speed captures intrinsic walking capacity more directly than the absolute distance alone. To further clarify this relationship, we compared the 360-m 6MWD threshold with the <1.0 m·s^−1^ gait-speed criterion (supplementary table S3). Risk estimates for exacerbations and mortality were remarkably consistent in direction and magnitude across the two approaches. This close concordance suggests that the prognostic information embedded in low 6MWD is largely attributable to impaired gait speed itself, reinforcing gait speed as a more fundamental physiological marker.

Gait speed offers a simplified, integrative assessment of physical reserve, reflecting the combined performance of the respiratory, cardiovascular and musculoskeletal systems [5, 8, 19, 33, 34]. Its reproducibility and ease of implementation have led to its adoption as a primary measure in sarcopenia, and similar advantages apply to COPD [35]. Because gait speed is feasible even in patients who cannot complete more demanding tests such as the ISWT or a full 6MWT, it is particularly suited for older adults and frail populations [36]. Together, these strengths underscore its utility as a complementary (and in many clinical settings, more feasible) functional assessment tool.

Our study aligns with previous research demonstrating gait speed's prognostic value in both geriatric and COPD populations, extending these findings to an Asian cohort [8, 33, 35, 37]. Slower gait speed in COPD likely reflects disease-specific physiological impairments beyond general frailty [38]. Limited ventilatory reserve, a tendency toward dynamic hyperinflation, and peripheral muscle dysfunction reduce ambulatory efficiency and heighten vulnerability to physiologic stressors, mechanisms that plausibly link reduced gait speed to exacerbations and mortality [39]. Consistent with these mechanisms, even among patients with similar baseline lung function, slower gait speed independently identified those at higher risk, emphasising its role as a valuable risk stratifier. Furthermore, subgroup analyses demonstrated that the prognostic value of gait speed remained consistent across various clinical contexts, including age, blood eosinophil count and GOLD stage. In patients aged ≥70 years, poor gait speed (<1 m·s^−1^) was associated with a more than five-fold increase in mortality risk compared with those with preserved gait speed. Similar patterns were observed across eosinophil subgroups and GOLD severity levels, suggesting that gait speed provides robust prognostic information independent of conventional risk stratifiers. These findings reinforce the clinical utility of gait speed as an integrative marker of physical reserve and frailty across diverse COPD phenotypes. By adopting the AWGS 2019 criteria, we ensured consistency with established ageing and frailty frameworks, which supported our use of the 1.0 m·s^−1^ cut-off [27].

Our study demonstrated several key strengths. First, gait speed is easy to calculate, requires no additional equipment beyond the standard 6MWT and reflects the integrated function of respiratory, musculoskeletal and neurological systems. This makes it feasible and broadly applicable in various clinical settings, even for patients unable to complete maximal effort tests. Second, our study benefits from long-term follow-up using real-world data from a large, representative national cohort. The consistent associations across multiple clinical outcomes, including lung function, symptom burden (CAT and SGRQ-C), exacerbation rates and survival, were robust across both continuous and categorical gait speed assessments (<1 m·s^−1^ threshold and quartile stratification). These findings remained significant after adjustment for confounders and were further validated in PS-matched analyses, enhancing the reliability of our results. Third, unlike previous retrospective studies that identified largely non-modifiable prognostic factors such as age, sex, FEV1 and comorbidities (e.g. heart failure), our study highlights gait speed as a clinically accessible surrogate marker. As a measurable indicator of physical performance, gait speed offers a practical target for intervention. Fourth, gait speed can be derived directly from 6MWT distance and time, allowing inclusion of patients unable to complete the full 6MWT, and offering a feasible, widely applicable marker even in routine clinical settings.

This study has several limitations. First, although we adopted the established threshold of 1.0 m·s^−1^ based on sarcopenia and frailty guidelines, data-driven inflection point analysis in our cohort identified a closely aligned threshold (∼1.11 m·s^−1^), supporting the appropriateness of this cut-off. Second, the study cohort was predominantly male (about 93%), which may limit the generalisability of our findings to female COPD populations. However, low smoking rates among women in many Asian countries, including South Korea, lead to male-dominant COPD cohorts [40]. There was no significant difference in sex distribution between gait speed groups, and results remained consistent after adjustment or matching for sex, suggesting minimal risk of sex-related bias. Although gait performance may differ between men and women, many functional metrics, including the widely used 1 m·s^−1^ threshold in sarcopenia, apply unified cut-offs to enhance clinical usability and standardisation. Our findings support the potential value of such a common threshold; however, future validation studies with larger female representation will be essential to establish generalisability and explore whether sex-specific refinements are warranted. Third, as an observational study, causal inference is limited. Nonetheless, the real-world data provide valuable insights into the prognostic utility of gait speed in clinical practice, highlighting the need for future interventional studies to evaluate its role as a modifiable target. Fourth, although gait speed was objectively measured, physical performance may still be affected by non-respiratory comorbidities. However, the KOCOSS database excluded individuals with major mobility-limiting conditions (e.g. hemiplegia or recent cerebrovascular accidents), thereby minimising potential confounding. Fourth, both 6MWD and gait speed may be influenced by day-to-day variability in symptoms, such as fluctuating dyspnoea or fatigue. Although measurements were obtained under standardised conditions, the potential impact of within-subject variability cannot be completely excluded.

Conclusion

This study demonstrates that gait speed is a simple and clinically meaningful predictor of adverse outcomes in COPD. Its integration into routine assessment may improve risk stratification, facilitate early identification of high-risk individuals, and support personalised care. As gait speed reflects overall functional capacity, it may serve as both a prognostic marker and a therapeutic target. Future research should validate these findings and explore interventions to improve gait speed and prognosis.