How do people with COPD walk? A European study on digitally measured real-world gait

Graphical abstract

Overview of the study. FEV₁: forced expiratory volume in 1 s.

Abstract

Background

The amount of walking that people with COPD do is reduced. However, data on their manner of walking (i.e. gait) are still lacking. We characterised real-world gait in COPD by assessing levels and distributions of gait parameters, and comparing them across COPD severity and with healthy peers.

Methods

549 people with COPD from seven European sites and 19 healthy older adults wore single wearable devices (either Axivity AX6 or DynaPort MoveMonitor MM+) continuously for 1 week, from which we identified walking bouts, calculated 15 digital mobility outcomes (DMOs) aggregated at the weekly level, and compared them across COPD severity levels and with healthy peers.

Results

Of the participants with COPD, 37% were female with a mean±sd age of 68±8 years and a post-bronchodilator forced expiratory volume in 1 s of 54±20% predicted. All gait DMOs were normally distributed and exhibited variability between participants (e.g. mean±sd walking speed of 0.83±0.12 m·s⁻¹, ranging from 0.48 to 1.20 m·s⁻¹). Walking speed and cadence DMOs were lower with increasing disease severity (e.g. mean±sd walking speed of 0.88±0.11, 0.85±0.12, 0.80±0.12 and 0.78±0.14 m·s⁻¹ across Global Initiative for Chronic Obstructive Lung Disease (GOLD) grades 1 to 4, p<0.001; mean±sd cadence of 93±6, 91±6 and 89±7 steps·min⁻¹ across GOLD A, B and E, p=0.013). Stride length and duration varied across COPD severity levels. Walking speed and cadence bout-to-bout variability only varied across dyspnoea severity levels. In a secondary analysis, we compared DMO data from people with COPD to a convenience sample of 19 healthy older adults (47% women, mean age 71±6 years) and found that walking speed and cadence varied between participants with COPD and healthy adults (e.g. mean±sd walking speed 0.83±0.12 versus 0.90±0.12 m·s⁻¹, p=0.041).

Conclusion

In people with COPD, gait DMOs are normally distributed and worsen as disease advances. Moreover, walking speed and cadence DMOs are significantly altered when compared to healthy peers. Further research should elucidate which DMOs can be improved with treatments to enhance mobility and reduce adverse events.

Shareable abstract

Real-world gait (i.e. how people walk) in COPD can be characterised through parameters such as walking speed, stride length or cadence, which exhibit significant differences as the disease advances. https://bit.ly/3EGIJtc

Introduction

Walking is important in COPD [1–5]. Extensive research has shown that people with COPD walk less compared to healthy counterparts [1, 6], and this has been associated with adverse outcomes including mortality and exacerbations [2, 3]. Contrary to the abundant evidence on the amount of walking performed by people with COPD, there is little information on their manner of walking (i.e. gait), particularly in daily life unsupervised settings (i.e. real-world settings) [3, 7, 8].

Gait significantly impacts well-being, survival, participation and independence in older adults and people with long-term health conditions [9, 10]. Gait can be described in terms of pace, rhythm and variability, using parameters such as walking speed, cadence or stride duration variability, respectively [11, 12]. These parameters are increasingly being used to assess gait quality in the presence of long-term health conditions [8, 11–14], but have generally been measured in controlled settings, which may not be representative of, or capture, manners of walking seen in daily life [8]. Real-world gait parameters are, nonetheless, relevant to better understand individuals’ perceptions of their own mobility [5], confidence, the metabolic demand imposed and other walking-related outcomes, such as mortality and falls.

While individuals with COPD often report on walking as a meaningful aspect of health [5], the role of gait in their daily lives remains largely unknown, in part because gait is not routinely assessed in clinical practice and because scientific evidence has been limited so far. However, previous research has highlighted the potential of a comprehensive walking assessment in COPD, primarily through studies conducted in laboratory settings [7, 8, 15]. Two recent reviews with meta-analyses reported that, in the laboratory, people with COPD walk less and slower than their healthy counterparts [7]; and slower walking was, independently of walking time, associated with an increased mortality risk [3]. One review also reported that other parameters, such as cadence and step length, may be reduced in COPD, but the evidence was considered limited [7]. Only one COPD study to date has investigated a single gait parameter in real-world conditions, finding that a lower cadence was associated with greater COPD severity and worse prognosis [16]. Luckily, advances in wearable technologies have prompted the development of novel methods to accurately assess the real-world gait of people living with long-term conditions, including COPD [8, 17, 18].

In this study, we aimed to comprehensively characterise daily life gait of people with COPD using digital mobility outcomes (DMOs) derived from the Mobilise-D method, a combination of a single wearable device with a processing pipeline capable of extracting a comprehensive set of gait parameters [19–21]. Specifically, we 1) assessed the levels and distribution of gait DMOs collected in the real world in people with COPD, and 2) compared them across disease severity groups. As a secondary analysis, we compared gait DMOs of people with COPD with those of healthy older adults.

Methods

A complete version of the methods is available in the supplementary material.

Study design and participants

This cross-sectional analysis used baseline COPD data from the Mobilise-D Clinical Validation Study (CVS) (ISRCTN 12051706), a prospective multicentre cohort study recruiting participants with COPD, Parkinson's disease, multiple sclerosis and proximal femoral fracture to characterise real-world walking [22]. Briefly, individuals with stable COPD from rehabilitation centres and hospitals in seven European sites (Athens, Barcelona, Grosshansdorf, Leuven, London, Newcastle and Zurich) were recruited between April 2021 and May 2022. Eligibility criteria and participant characteristics are available in the supplementary material. For the present analysis, we excluded participants with COPD who had no valid measure of walking (n=58), which resulted in a final sample of 549 individuals with COPD (90.4% of the original sample).

To explore differences between people with COPD and healthy peers in secondary analyses, we included a convenience sample of healthy older adults from the Mobilise-D Technical Validation Study (ISRCTN 12246987), a multicentre panel study recruiting participants with COPD, Parkinson's disease, multiple sclerosis, proximal femoral fracture and chronic heart failure, as well as healthy older adults, between July 2020 and July 2021 [17]. Eligibility criteria and participants’ characteristics are available in the supplementary material and elsewhere [17, 23].

Both studies were approved by the ethics committees of all participating centres, and written informed consent was obtained from all participants.

Variables and instruments

Real-world walking was objectively measured using a single wearable device at the lower back for 7 days, either the AX6 (Axivity, Newcastle upon Tyne, UK) body fixed using a custom-designed adhesive patch (n=98) or a DynaPort MoveMonitor (MM+; McRoberts, The Hague, Netherlands) body worn using a belt (n=453). Individuals were invited to use the device continuously (including at night, if willing). Both devices were metrologically equivalent and consisted of a 6 degrees of freedom inertial measurement unit with the following configuration: triaxial accelerometer with a range of ±8 g and a resolution of 1 mg, triaxial gyroscope with a range of ±2000 dps, a resolution of 70 mdps and a sampling frequency 100 Hz. Through the Mobilise-D processing pipeline, technically valid walking bout (WB)-level DMOs were derived from the raw sensor data [19–21, 24]. Briefly, sensor data were first standardised according to the protocol outlined by Palmerini et al. [19]. Next, WBs were identified using the definition of Kluge et al. [24] (i.e. a walking sequence containing at least two consecutive strides of both feet). Gait features (e.g. initial contact, cadence and stride length) were extracted for these WBs following the validated processing pipeline described in Micó-Amigo et al. [21] and Kirk et al. [20]. A valid walking measurement was defined as at least 3 days with 12 h or more wear time within waking hours (i.e. between 07:00 and 22:00) [25], and was aggregated at the weekly level following a process that combined data restrictions and statistical metrics to summarise DMOs [26]. In total, 15 DMOs of gait at the weekly level were obtained, including six DMOs of gait pace, five DMOs of gait rhythm and four DMOs of gait bout-to-bout variability. These DMOs were calculated during either shorter (i.e. lasting between 10–30 s) WBs, longer (i.e. lasting >30 s) WBs or all WBs. More details on the real-world walking assessment and definition of the 15 DMOs are provided elsewhere [7, 18–21, 24, 26]. Data on walking activity (e.g. walking duration, steps per day and number of WBs of different duration) were also obtained from this assessment.

Participants answered interviewer- and self-administered validated questionnaires including questions on age, sex assigned at birth, gender, ethnicity, marital status, employment status, smoking, number of people living at home, use of walking aids, history of falls, breathlessness (modified Medical Research Council (mMRC) Dyspnoea Scale) [27], health-related quality of life (COPD Assessment Test (CAT) [28]), comorbidities (Functional Comorbidity Index (FCI) scale) [29] and history of exacerbations [22]. We obtained post-bronchodilator forced expiratory volume in 1 s and forced vital capacity measurements from standardised spirometry collected either in the clinical setting or during the study visit [30]. COPD severity was classified using the Global Initiative for Chronic Obstructive Lung Disease (GOLD) grades 1–4 and retrofitted to the ABE group assessment, considering symptoms and exacerbations history [31]. 6-min walk distance (6MWD) was determined following standardised methodology [32]. Height, weight and body mass index were obtained by physical examination. Physical activity experience was assessed using the Clinical visit-PROactive Physical Activity in COPD (C-PPAC) tool [33].

Statistical analysis

Power calculations and complete statistical analyses are available in the supplementary material. Sample characteristics are presented as n (%) for categorical variables and as mean±sd or median and 25th–75th percentiles (P₂₅–P₇₅) for continuous variables, depending on their distribution.

To individually characterise gait DMOs in COPD, we first assessed their levels and distribution, testing normality and symmetry using visual methods and statistical tests, and presenting them as mean±sd because of their distribution. Second, we compared DMOs across COPD severity (assessed by the GOLD 1–4 grades, the GOLD ABE groups and mMRC Dyspnoea Scale), using one-way ANOVA and building multivariable linear regression models, unadjusted and adjusting for age, sex, height, FCI score and walking duration. In a secondary analysis, we compared DMOs between COPD and healthy adults using bivariate statistics, and building multivariable linear regression models unadjusted and adjusted for age, sex, height and walking duration.

All analyses were conducted in Stata/SE 16.0 (StataCorp, College Station, TX, USA) using the latest available version of the Mobilise-D CVS database (v6.2, February 2025).

Results

Sample characteristics

At baseline, 549 individuals with COPD were included in this analysis: 37% were female, mean age was 68±8 years and mean height was 168±9 cm. 11% of participants had mild COPD, 43% had moderate COPD, 32% had severe COPD and 14% had very severe COPD based on spirometry (table 1). Participants walked a median (P₂₅–P₇₅) of 6561 steps·day⁻¹ (4005–9712 steps·day⁻¹) and reached a mean 6MWD of 416±119 m. No significant differences were identified between included and excluded participants, except for more impaired lung function and less frequent use of walking aids among included participants (supplementary table S1).

TABLE 1.

Sociodemographic, clinical and functional baseline characteristics of 549 people with COPD in the European Mobilise-D Clinical Validation Study

Characteristic
Subjects (n)	549
Recruitment site
Athens	48 (8.7)
Barcelona	150 (27.2)
Grosshansdorf	132 (24.0)
Leuven	108 (19.6)
London	25 (4.5)
Newcastle	47 (8.5)
Zurich	41 (7.4)
Sex assigned at birth: female	202 (36.8)
Gender: women	199 (36.7)
Age (years)	67.6±8.0
Ethnicity (white)	545 (99.3)
Height (cm)	168.3±9.3
Weight (kg)	78.1±17.5
BMI (kg·m−2)	27.5±5.4
BMI categories
Underweight	16 (2.9)
Normal weight	167 (30.7)
Overweight	195 (35.9)
Obese	167 (30.7)
Marital status (married)	332 (61.0)
Employment status (retired)	374 (68.8)
Living alone (yes)	154 (28.3)
Pack-years	47 (36–64.5)
Dyspnoea (mMRC grade 0–4)	2 (1–2)
HRQoL (CAT score 0–40)	14 (9–19)
6MWD (m)	416±119
FEV1 (L)	1.49±0.66
FEV1 (% pred)	53.7±20.5
FVC (L)	3.06±1.0
FVC (% pred)	84.4±19.7
Airflow limitation (GOLD grade 1–4)#
1	62 (11.3)
2	235 (42.8)
3	178 (32.4)
4	74 (13.5)
GOLD classification (ABE groups, 2023)¶
A	107 (20.0)
B	322 (60.1)
E	107 (20.0)
FCI (0–18 score)	4 (3–5)
Self-reported diagnosis of comorbidities
Arthritis	169 (31.2)
Osteoporosis	78 (14.4)
Chronic heart failure	72 (13.3)
Heart attack	58 (10.7)
Neurological diseases	9 (1.7)
Diabetes	83 (15.3)
Depression	131 (24.2)
Anxiety	95 (17.5)
Visual impairment	197 (36.4)
Hearing impairment	87 (16.1)
Obesity	165 (30.4)
Hypertension (yes)	228 (41.5)
ICS alone	328 (61.0)
LABA or LAMA alone	474 (87.8)
LABA or LAMA and ICS, in combination	313 (58.0)
Any falls in the 12 months prior to the study inclusion (yes)	115 (21.1)
Wearable device used
Dynaport MM+	452 (82.3)
Axivity AX6	97 (17.7)
Walking aids (yes)	44 (8.0)
Long-term oxygen supplementation (yes)	42 (7.7)
Physical activity experience (C-PPAC scores, 0–100)
Amount	62.6±19.3
Difficulty	71.6±15.8
Total score	67.1±15.1
Any moderate exacerbations in the 12 months prior to study inclusion (yes)	147 (26.8)
Any severe exacerbations in the 12 months prior to study inclusion (yes)	64 (11.7)
Walking activity
Number of steps (steps·day−1)	6561 (4005–9712)
Walking duration (h·day−1)	1.17 (0.74–1.73)
Number of WBs (WB·day−1)	293.3±137.0
Number of WBs >10 s (WB·day−1)	121 (83–170)
Number of WBs >30 s (WB·day−1)	19 (11–33)
WB duration (s)	8.9 (8.2–9.7)

Data are presented as median (P₂₅–P₇₅), mean±sd or n (%), unless otherwise stated. Some variables had missing values: 89 in employment status, 25 in mMRC dyspnoea, 15 in CAT, 6 in 6MWD, 13 in GOLD ABE groups, 7 in FCI and self-reported comorbidities, 11 in ICS, 11 in LABA, 11 in LAMA, 7 in physical activity experience difficulty, 9 in physical activity experience amount and total score. 6MWD: 6-min walk distance; BMI: body mass index; CAT: COPD Assessment Test; C-PPAC: Clinical visit-PROactive Physical Activity in COPD; FCI: Functional Comorbidity Index; FEV₁: forced expiratory volume in 1 s; FVC: forced vital capacity; GOLD: Global Initiative for Chronic Obstructive Lung Disease; HRQoL: health-related quality of life; ICS: inhaled corticosteroid; LABA: long-acting β-agonist; LAMA: long-acting muscarinic antagonist; mMRC: Modified Medical Research Council Dyspnoea Scale; P₂₅: 25th percentile; P₇₅: 75th percentile; WB: walking bout. ^#: 1 (mild): FEV₁ ≥80% pred; 2 (moderate): 50%≤FEV₁<80% pred; 3 (severe): 30%≤FEV₁<50% pred; 4 (very severe): FEV₁ <30% pred; ^¶: Group A: low symptom severity, low exacerbation risk; Group B: high symptom severity, low exacerbation risk; Group E: high exacerbation risk.

Gait pace

People with COPD walked at a mean speed of 0.67±0.07 m·s⁻¹ during shorter WBs and 0.83±0.12 m·s⁻¹ during longer WBs. Their maximum walking speed was 0.90±0.13 m·s⁻¹ during WBs >10 s and 0.99±0.18 m·s⁻¹ for longer WBs. Stride length was 91±9 cm during shorter WBs and 106±12 cm during longer WBs (table 2). All walking speed and stride length parameters were normally distributed.

TABLE 2.

Distribution of real-world gait digital mobility outcomes (DMOs) in 549 people with COPD

	Definition	Mean±sd	Minimum–maximum
Gait pace
Walking speed in shorter (10–30 s) WBs (m·s−1)	The distance covered by the whole body per second, assessed in WBs between 10 and 30 s	0.67±0.07	0.46–0.88
Walking speed in longer (>30 s) WBs (m·s−1)	The distance covered by the whole body per second, assessed in WBs longer than 30 s	0.83±0.12	0.48–1.20
Maximum walking speed in WBs >10 s (m·s−1)	The maximum (P90) distance covered by the whole body per second, assessed in WBs of more than 10 s	0.90±0.13	0.55–1.27
Maximum walking speed in longer (>30 s) WBs  (m·s−1)	The maximum (P90) distance covered by the whole body per second, assessed in WBs of more than 30 s	0.99±0.18	0.54–1.47
Stride length in shorter (10–30 s) WBs (cm)	Length of two consecutive steps, assessed during WBs between 10 and 30 s	91±9	60–117
Stride length in longer (>30 s) WBs (cm)	Length of two consecutive steps, assessed during WBs longer than 30 s	106±12	68–146
Gait rhythm
Cadence in all WBs (steps·min−1)	Step frequency during a period of time (mins), calculated in all WBs	85±4	74–97
Cadence in longer (>30 s) WBs (steps·min−1)	Step frequency during a period of time (mins), calculated in WBs longer than 30 s	91±7	70–111
Maximum cadence in longer (>30 s) WBs  (steps·min−1)	Maximum (P90) step frequency during a period of time (mins), calculated using WBs longer than 30 s	100±9	73–124
Stride duration in all WBs (s)	Time elapsed between the initial contacts of two consecutive footfalls of the same foot, assessed in all walking WBs	1.30±0.06	1.06–1.49
Stride duration in longer (>30 s) WBs (s)	Time elapsed between the initial contacts of two consecutive footfalls of the same foot, assessed in WBs longer than 30 s	1.26±0.09	0.95–1.57
Gait bout-to-bout variability
Walking speed bout-to-bout variability in longer   (>30 s) WBs (%)	Bout-to-bout variability of walking speed, assessed in WBs of more than 30 s	17±5	2–37
Stride length bout-to-bout variability between  longer (>30 s) WBs (%)	Bout-to-bout variability of stride length, assessed in WBs of more than 30 s	12±4	2–27
Cadence bout-to-bout variability (%)	Bout-to-bout variability of cadence, assessed in all WBs	12±1	8–17
Stride duration bout-to-bout variability (%)	Bout-to-bout variability of stride duration, assessed in WBs	14±2	9–26

Based on Kluge et al. [24], a walking bout (WB) is a walking sequence containing at least two consecutive strides of both feet (e.g. R-L-R-L-R-L or L-R-L-R-L-R). Some variables had missing values: 4 in walking speed bout-to-bout variability in longer (>30 s) WBs, stride length bout-to-bout variability between longer (>30 s) WBs. P₉₀: 90th percentile.

Figures 1–3 show that all six walking speed and stride length DMOs decreased with increasing disease severity as per GOLD grades, GOLD ABE groups and mMRC grades and these differences remained statistically significant after adjustment in all cases, except for stride length during longer WBs across GOLD ABE groups (supplementary tables S2–S4).

FIGURE 1.

Distribution of gait digital mobility outcomes (DMOs) across Global Initiative for Chronic Obstructive Lung Disease (GOLD) grades 1–4. Shorter walking bouts (WBs) were 10–30 s. Longer WBs were >30 s. p-trends shown across COPD severity stages obtained from linear regression models with COPD severity stages as a continuous variable NA: not applicable. ^#: unadjusted; ^¶: adjusted for age, sex, height, Functional Comorbidity Index and walking duration.

FIGURE 3.

Distribution of gait digital mobility outcomes (DMOs) across Modified Medical Research Council (mMRC) Dyspnoea Scale. Shorter walking bouts (WBs) were 10–30 s. Longer WBs were >30 s. p-trends across mMRC grades correspond to linear regression models, with mMRC Dyspnoea Scale grades as a continuous variable. NA: not applicable. ^#: unadjusted; ^¶: adjusted for age, sex, height, Functional Comorbidity Index and walking duration.

FIGURE 2.

Distribution of gait digital mobility outcomes (DMOs) across Global Initiative for Chronic Obstructive Lung Disease (GOLD) ABE groups. Shorter walking bouts (WBs) were 10–30 s. Longer WBs were >30 s. p-trends across COPD severity stages obtained from linear regression models with COPD severity stages as a continuous variable. NA: not applicable. ^#: unadjusted; ^¶: adjusted for age, sex, height, Functional Comorbidity Index and walking duration.

Gait rhythm

People with COPD walked at a mean cadence of 85±4 steps·min⁻¹ during all WBs, and 91±7 steps·min⁻¹ during longer WBs. Their maximum cadence in longer WBs was 100±9 steps·min⁻¹. The mean stride duration was 1.30±0.06 s during all WBs and 1.26±0.09 s during longer WBs (table 2). All cadence and stride duration parameters were normally distributed.

All cadence DMOs were lower, and stride duration during longer bouts was longer, with increasing COPD severity, in both unadjusted and adjusted analyses. No differences across GOLD grades or GOLD ABE groups were observed for stride duration in all bouts (figures 1–3, supplementary tables S2–S4).

Gait bout-to-bout variability

Walking speed bout-to-bout variability was normally distributed in COPD, and it was significantly different across GOLD grades, GOLD ABE groups and mMRC grades. Only differences by mMRC grades remained statistically significant after adjustment (figures 1–3, supplementary table S4). Stride length bout-to-bout variability was normally distributed in COPD, and it was not significantly different from that in healthy adults. Observed differences by severity groups disappeared in adjusted models (figures 1–3, supplementary tables S2–S4).

Cadence bout-to-bout variability was normally distributed in COPD, and it was significantly different across GOLD grades, GOLD ABE groups and mMRC grades. Only differences by mMRC grades remained statistically significant after adjustment. Stride duration bout-to-bout variability was normally distributed in COPD, and it was not different from that in healthy adults nor across disease severity groups (figures 1–3, supplementary tables S2–S4).

Secondary analysis: differences between COPD and healthy older adults

A secondary analysis included healthy older adults: 47% were female, mean age was 71±6 years and mean height was 166±10 cm. The healthy adults were older than the participants with COPD, without any other statistically significant difference (supplementary table S5).

After adjusting for age, sex, height and total daily walking duration, compared to healthy older adults, people with COPD walked slower during longer WBs (0.83 m·s⁻¹ versus 0.90 m·s⁻¹, p=0.041) and had a slower maximum walking speed during WBs >10 s (0.90 m·s⁻¹ versus 0.98 m·s⁻¹, p=0.018) and longer WBs (0.99 m·s⁻¹ versus 1.12 m·s⁻¹, p=0.015). No significant differences were observed in walking speed during shorter WBs or in stride length DMOs (figure 4, supplementary table S6).

FIGURE 4.

Differences in gait digital mobility outcomes (DMOs) between 549 people with COPD and 19 healthy older adults. Shorter walking bouts (WBs) were 10–30 s. Longer WBs were >30 s. p-values correspond to linear regression models. ^#: unadjusted; ^¶: adjusted for age, sex, height and walking duration.

In unadjusted analyses, compared to healthy older adults, people with COPD had a lower cadence and lower maximum cadence during longer WBs, but only the latter remained statistically significant after adjustment (100 steps·min⁻¹ versus 107 steps·min⁻¹, p=0.039). No differences were found for cadence during all WBs or stride duration DMOs (figure 4, supplementary table S6).

Cadence bout-to-bout variability in COPD was significantly different from healthy adults in unadjusted analysis, but not when adjusted for age, sex, height and total daily walking duration. Walking speed, stride length and stride duration bout-to-bout variability were not different between COPD and healthy adults (figure 4, supplementary table S6).

Discussion

Using state-of-the-art devices and algorithms, this study provides novel evidence on real-world gait manner (i.e. how people walk in daily life) in individuals with COPD. We found that 1) real-world gait DMOs are quantifiable in COPD and exhibit between-person variability and 2) virtually all real-world gait DMOs worsen with increasing disease severity. Secondary analysis considering a small convenience sample of healthy peers showed that real-world walking speed and cadence DMOs are altered in people with COPD compared to healthy older adults, particularly during longer WBs.

Key findings and comparison with previous studies

Our findings when characterising gait DMOs in COPD require discussion. First, all 15 gait DMOs are measurable in the presence of COPD, even in individuals with very severe disease, in whom very slow walking is expected. Second, all gait DMOs followed symmetrical, normal distributions, similarly to those previously reported during laboratory assessments [16, 23, 34], supporting their construct validity. In agreement with previous literature on older adults and other mobility-impairing conditions [35–37], our analyses suggest that people with COPD have a higher walking speed during longer WBs compared to shorter WBs, likely reflecting contextual factors and the intention of the walking episodes rather than on a manifestation of the condition. Third, all gait DMOs exhibited variability between subjects. For instance, walking speed during longer WBs ranged from 0.48 to 1.20 m·s⁻¹, values comparable to the pace of an older adult with mobility impairment and to a healthy adult who walks briskly, respectively [38]. This variability suggests that gait DMOs can provide valuable insights for characterising COPD. Of note, the observed values for some DMOs, such as walking speed, stride length and cadence, were lower than those reported in previous COPD studies conducted in laboratory, controlled settings [34, 39, 40], and similar to those conducted either under guided free-living conditions (i.e. during short, unsupervised assessments of 2.5 h where individuals were encouraged to complete specific tasks) [23] or in the single previous study assessing real-world cadence in COPD [16]. This supports the notion that the way people walk under direct observation in laboratory settings (walking capacity) is different from what they habitually do in the real world (walking performance), and emphasises the importance of real-world gait assessment for the comprehensive characterisation of COPD.

Our findings indicate that, with increasing airflow limitation and breathlessness, and with the combined assessment of levels of symptoms and frequency of previous exacerbations, walking gets more affected. People with more severe COPD walk slower than those with milder cases, take shorter and longer-lasting strides, maintain a lower cadence, and exhibit reduced bout-to-bout variability in walking speed, stride length and cadence. Observed differences were statistically significant for DMOs calculated during shorter and longer WBs, and after adjusting for walking duration. Notably, we did not find significant differences in stride duration bout-to-bout variability with increasing disease severity. These novel real-world findings align with existing COPD studies conducted under controlled conditions and with the real-world study of cadence, which reported slower walking and lower cadence with greater disease severity [5, 41]. While the specific mechanisms connecting airflow limitation, symptoms burden, history of exacerbations and perhaps other clinical non-respiratory characteristics to gait changes are still unknown, greater alterations in more severe cases could be attributed to reduced functional capacity caused by increased symptoms burden and a higher number of disease complications and/or inflammatory sequelae associated with advanced disease stages [31].

In secondary analyses, we found that some parameters of real-world gait are impaired in individuals with COPD when compared to a small convenience sample of healthy older adults. Specifically, during longer WBs, people with COPD exhibited slower walking, lower cadence and reduced bout-to-bout variability in speed and cadence. In contrast, we did not find significant differences in either walking speed or cadence during shorter WBs, or in stride length, stride duration or their bout-to-bout variability. This pattern of differences based on WB duration aligns with previous findings observed under controlled conditions [23]. It implies that accurately characterising gait in COPD requires the consideration of WB duration, because impairments may not be apparent during shorter walking episodes. While the mechanisms connecting walking difficulties with longer but not shorter WBs require further investigation, we may reason that COPD symptom burden may be more manageable during brief WBs, and prompt individuals to adjust their gait manner during longer daily activities.

Study implications

Our results are relevant to research, clinical practice and public health. Future studies assessing real-world walking in COPD, either considering it as an exposure, intervention or outcome variable, should go beyond traditional parameters of walking amount, e.g. step count, because we have shown that gait is also impaired in the presence of this condition, even after adjusting for walking duration (i.e. a measure of walking amount). Building upon our findings, longitudinal data on gait parameters may be used to unravel unknown clinical consequences of COPD. However, future research may elucidate which specific DMOs hold higher clinical relevance in the presence of COPD and which have higher potential to be used and interpreted in regular clinical settings. For healthcare professionals, attention to gait may help assess the disease's impact on daily life and, eventually, the effectiveness of pharmacological and non-pharmacological therapies. Importantly, by addressing changes in gait and acting on them, clinicians might be able to tackle key determinants of falls, disability and mortality in this population. Finally, our findings point to considering gait impairment, and not only walking amount reduction, when developing and implementing interventional studies and public health programmes for active and healthy ageing.

Strengths and limitations

Our study has several strengths. First, we used a novel approach to characterise real-world walking in COPD, using objective measures provided by a wearable device capturing gait performance in ecologically valid environments [21]. Second, we recruited participants from different settings across Europe, thus providing a wide spectrum of disease severity and increasing the external validity of our findings. Our analysis also has shortcomings. First, environmental factors (e.g. neighbourhood walkability, rurality, weather conditions) may affect the real-world gait of people with COPD but were not controlled for in this study and should be the focus of future research. Second, the selection of a convenience sample of healthy older adults and the unequal sample sizes between COPD and healthy peers may hinder the identification of meaningful differences. However, we calculated the power of our sample to identify differences between groups and evaluated variances in all analyses. Third, healthy adults were recruited during the COVID-19 pandemic, which may have influenced their daily activities and walking behaviour, likely altering differences in DMOs between groups. Finally, although a change in walking behaviour is theoretically possible when objectively assessing real-world walking, this has been extensively tested in COPD and there is no evidence of a Hawthorne effect in this population [42, 43].

Conclusions

Real-world gait DMOs are normally distributed in people with COPD and worsen as disease advances. Walking speed and cadence DMOs are significantly altered in COPD compared to healthy peers. Further research should elucidate which DMOs can be improved with treatments to enhance mobility and reduce adverse events.

Shareable PDF

This PDF extract can be shared freely online.

Shareable PDF

ERJ-02303-2024.Shareable

Supplementary material

Please note: supplementary material is not edited by the Editorial Office, and is uploaded as it has been supplied by the author.

Supplementary material

ERJ-02303-2024.Supplement