Interventions to Increase Physical Activity in Patients with COPD: A Comprehensive Review

Abstract

It is unknown how interventions aimed at increasing physical activity (PA), other than traditional pulmonary rehabilitation, are structured and whether they are effective in increasing PA in chronic obstructive pulmonary disease (COPD). The primary aim of this review was to outline the typical components of PA interventions in patients with COPD. This review followed the PRISMA guidelines. A structured literature search of relevant electronic databases from inception to April 2014 was undertaken to outline typical components and examine outcome variables of PA interventions in patients with COPD. Over 12000 articles were screened and 20 relevant studies involving 31 PA interventions were included. Data extracted included patient demographics, components of the PA intervention, PA outcome measures and effects of the intervention. Quality was assessed using the PEDro and CASP scales. There were 13 randomised controlled trials and three randomised trials (PEDro score 5-7/10) and four cohort studies (CASP score 5/10). Interventions varied in duration, number of participant/researcher contacts and mode of delivery. The most common behaviour change techniques included information on when and where (n = 26/31) and how (n = 22/31) to perform PA behaviour and self-monitoring (n = 18/31). Significant between-group differences post-intervention in favour of the PA intervention, compared to a control group or to other PA interventions, in one or more PA assessments were found in 7/16 studies. All seven studies used walking as the main type of PA/exercise. In conclusion, although the components of PA interventions were variable, there is some evidence that PA interventions have the potential to increase PA in patients with COPD.

Keywords :

• activity monitors
• behaviour change techniques
• exercise
• outcome measures

Introduction

Exercise training is a key component of the management of chronic obstructive pulmonary disease (COPD) and is frequently delivered in the context of pulmonary rehabilitation (PR). PR typically involves structured, supervised exercise sessions over 6–12 weeks, along with patient education, psychological support and nutritional interventions (1–3). A Cochrane review has emphasised the benefits of PR in terms of statistically and clinically significant short-term improvements in exercise capacity and health-related quality of life (4). However, there is still some uncertainty on the effects of PR on long-term modification of physical activity (PA) behaviour (5). There is emerging evidence that patients with COPD only complete infrequent and short bouts of PA at moderate intensity (6). In a recent meta-analysis, involving only randomised trials and single-group intervention studies, exercise training conducted ≥4 weeks as part of PR was shown to offer statistically significant but clinically small effects on PA levels in patients with COPD (5). This suggests patients potentially need help in translating the benefits gained from PR into changes in PA behaviour (7).

To help promote the adoption and maintenance of physically active lifestyles at an individual or population level, PA interventions have been developed (8). Guidelines and position stands for clinical populations such as cancer (9), type II diabetes (10) and cardiovascular disease (11) recommend that long-term maintenance of PA is at the core of effective rehabilitation programmes. Guidelines also highlight that PA counselling should be part of a multi-layered and multidisciplinary approach (11). Although a number of PA interventions have been conducted in patients with COPD, there are no clear guidelines on the key components of PA interventions or how effective these have been in terms of long-term changes in PA levels and which PA outcome measures are most useful in assessing the efficacy of PA interventions.

The primary aim of this review was to outline the typical components of interventions other than PR focused on increasing PA levels in patients with COPD. The secondary aims were to examine the PA outcome measures and explore the efficacy of the interventions at increasing PA levels.

Methods

This comprehensive review adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (12).

Eligibility criteria

The following inclusion criteria were used to select relevant studies: any intervention which aimed to increase PA in patients with COPD using one or more of the types of interventions highlighted by Cox et al. (13) and included PA outcome measures: time spent in PA (measured in minutes per week, sessions per week, MET-minutes); energy expenditure (in calories or joules); activity counts; or step counts (13). Exclusion criteria included: abstracts; study protocols; traditional PR as defined in recent guidelines (3); maintenance programmes directly after PR which were structured similarly to PR and focused on maintaining functional capacity and outcomes of PR; and/or no PA outcome measure utilised. Studies that only recorded PA to determine adherence to the intervention were excluded.

Search and study selection

A structured literature search was undertaken on relevant electronic databases (AMED, CINAHL, Cochrane database, Embase and Medline) from inception (last search 17^th April 2014). The following search terms were used to retrieve relevant studies: (“respiratory tract disease” OR “lung disease” OR “COPD” OR “chronic obstructive lung disease” OR “chronic bronchitis” OR “emphysema”) AND (“physical activity” OR “motor activity” OR “daily life activity” OR “exercise” OR “walking” OR “ambulation” OR “lifestyle activit”. OR “lifestyle physical activit”). Results were limited by human research and English language. Two authors (JW and JB) reviewed potential full-text articles against the eligibility criteria. Where the two authors disagreed, a third author (BO’N) made the final decision. The reference listings of the selected articles were then hand-searched for additionally relevant studies.

Study quality appraisal

Study quality was assessed independently by two authors (JW and JB) using the PEDro scale for randomised ­controlled trials (RCTs) and randomised trials (RTs) and the CASP scale for cohort studies. The PEDro scale is used by the Physiotherapy Evidence Database as a tool to rate the validity of over 25000 RTs, systematic reviews and clinical practice guidelines in physiotherapy (14). The PEDro scale is an 11-item scale used to rate RCTs and RTs in terms of internal validity and also interpretability of study results. The PEDro scale is scored out of 10 (item 1 is not scored), with ‘Yes’ responses gaining 1 point and ‘No’ responses gaining 0 points. RCTs and RTs were scored as either low (≤3/10), medium (4–5/10) or high (≥6/10) quality (15,16).

The CASP scale is provided by the Critical Appraisal Skills Programme to aid the appraisal of healthcare information underpinning decision making for reliability and validity in the United Kingdom (17). The CASP scale is a 12-item scale used to rate cohort studies in terms of three broad areas including: Are the results of the study valid?; What are the results?; Will the results help locally? The questions in the CASP scale are answered with either ‘No’, ‘Can't tell’ or ‘Yes’. Currently, there is no recognised numbered scoring system for this scale, therefore the authors scored the scale out of 10, with ‘Yes’ responses gaining 1 point and ‘No’ or ‘Can't tell’ responses gaining 0 points. Questions 1 and 2 are screening questions so were not scored. Where the two authors disagreed on the quality appraisal for a study, a third author (BO’N) made the final decision.

Data collection process

Data extracted included participant demographic information, recruitment methods utilised, study design, quality, intervention groups, all assessment times from baseline to final follow-up and the attrition rates from the intervention groups. Other data extracted included details on the components of the intervention(s) such as: the duration of the intervention; how the intervention was delivered to participants; the number of contacts between participants and study staff; the frequency, intensity, time and type of PA/exercise; the setting where the PA/exercise took place; and the level of PA/exercise supervision. To identify the behaviour change techniques used in the interventions, a specific taxonomy related to changing PA behaviour was utilised (18). Author JW independently chronicled the behaviour change techniques which appeared to be utilised in each intervention and identified the supporting text for each decision and this was independently verified by a second author (EC). Where the two authors disagreed, a third author (JB) made the final decision.

Data on the PA parameters used in the interventions, including between-group and within-group differences, were also extracted. Statistically significant effects of the PA interventions were measured as p ≤ 0.05. An attempt at a pooled estimate of the treatment effect of PA was unable to be conducted for all studies due to the heterogeneity of the PA interventions and PA parameters. However, to understand the magnitude of the intervention effects, effect sizes (d) were calculated for the most popular PA parameter (step counts) (19). Effect sizes were classified as trivial (≤0.2), small (>0.2), medium (>0.5) and large (>0.8) (19).

Results

Study selection

Figure 1 is a flow chart summarising how the final studies were selected. From the initial electronic searches, 12436 articles were identified. Removal of articles not relevant for the scope of this review and duplicates resulted in 69 full-text articles being selected for further reading. After reviewing the full-text articles, 49 studies were excluded after meeting one or more of the exclusion criteria. The most common exclusion criteria met were the intervention being either traditional PR or PR maintenance ­programmes (n = 9/48) and not using a PA outcome measure (n = 36/49). In total, 20 studies met the inclusion criteria and were included in the review (20–39). Within these studies, 31 interventions were identified. The studies either incorporated a PA intervention group and a control group with no PA intervention (7/20), had two or more groups comparing different PA interventions (9/20), or had one PA intervention group with no control group (4/20). No other eligible articles were ­identified from the reference lists of the selected articles.

Study quality appraisal

Tables 1a and 1b provide details on the quality of the studies. There were 13 RCTs, three RTs and four cohort studies. The RCTs and RTs were of medium-high quality; ranging from 5–7 on the PEDro scale. The most common limitations included failing to conceal allocation and insufficient blinding of the patients/therapists. The cohort studies were all scored positively in 5/10 quality indicators on the CASP scale.

Table 1a.  PEDro quality scores for the 16 RCT and RT studies

Reference	1. Eligibility criteria	2. Random allocation	3. Allocation concealed	4. Similar at baseline	5. Participants blinded	6. Therapists blinded	7. Assessors blinded	8. Outcome obtained from 85% of participants	9. Treatment allocated or intention-to-treat	10. Between group comparisons	11. Point measures and measures of variability	PEDro Score (range 0–10)
Nguyen (20) 2013	Y	Y(1)	N	Y(1)	N	N	Y(1)	Y(1)	Y(1)	Y(1)	Y(1)	7
Pleguezuelos (21) 2013	Y	Y(1)	N	Y(1)	N	N	N	Y(1)	Y(1)	Y(1)	Y(1)	6
Tabak (22) 2013	Y	Y(1)	Y(1)	Y(1)	N	N	N	N	Y(1)	Y(1)	Y(1)	6
Pomidori (23) 2012	Y	Y(1)	N	Y(1)	N	N	N	N	Y(1)	Y(1)	Y(1)	5
Effing (24) 2011	Y	Y(1)	N	Y(1)	N	N	N	Y(1)	Y(1)	Y(1)	Y(1)	6
Probst (25) 2011	Y	Y(1)	N	Y(1)	N	N	N	N	Y(1)	Y(1)	Y(1)	5
Berry (26) 2010	Y	Y(1)	Y(1)	Y(1)	N	N	Y(1)	N	Y(1)	Y(1)	Y(1)	7
Breyer (27) 2010	Y	Y(1)	N	Y(1)	N	N	N	Y(1)	Y(1)	Y(1)	Y(1)	6
Hospes (28) 2009	Y	Y(1)	N	Y(1)	N	N	N	Y(1)	N	Y(1)	Y(1)	5
Nguyen (29) 2009	Y	Y(1)	Y(1)	N	N	N	Y(1)	Y(1)	Y(1)	Y(1)	Y(1)	7
Nguyen (30) 2008	Y	Y(1)	Y(1)	Y(1)	N	N	Y(1)	N	Y(1)	Y(1)	Y(1)	7
Steele (31) 2008	Y	Y(1)	N	Y(1)	N	N	N	Y(1)	Y(1)	Y(1)	Y(1)	6
Varga (32) 2007	N	Y(1)	N	Y(1)	N	N	N	Y(1)	Y(1)	Y(1)	Y(1)	6
De Blok (33) 2006	Y	Y(1)	N	Y(1)	N	N	Y(1)	N	Y(1)	Y(1)	Y(1)	6
Behnke (34) 2005	N	N	N	Y(1)	N	N	N	Y(1)	Y(1)	Y(1)	Y(1)	5
Berry (35) 2003	Y	Y(1)	N	Y(1)	N	N	Y(1)	N	Y(1)	Y(1)	Y(1)	6

N = No; Y = Yes; Y(1) = Yes with 1 point.

* = not included in PEDro score.

Low = ≤3/10; Medium = 4–5/10; High = ≥6/10 (Ref. 15,16).

Table 1b.  CASP quality scores for the four cohort studies

Reference	1. Clearly focused issue	2. Appropriate method	3. Acceptable recruitment	4. Minimise bias-exposure	5. Minimise bias-outcome	6A. Confounders identified	6B. Confounders accounted for	7A. Follow-up – complete enough	7B. Follow-up – long enough	8–10. Results – precise and believable	11. Results – applicable to others	12. Results – fit with available evidence	CASP Score (range 0–10)
Moy (36) 2012	Y	Y	N	CT	Y(1)	N	Y(1)	CT	Y(1)	Y(1)	N	Y(1)	5
Moy (37) 2010	Y	Y	N	CT	N	Y(1)	Y(1)	CT	Y(1)	Y(1)	CT	Y(1)	5
Wewel (38) 2008	Y	Y	Y(1)	CT	CT	CT	Y(1)	CT	N	Y(1)	Y(1)	Y(1)	5
Nguyen (39) 2005	Y	Y	Y(1)	Y(1)	CT	CT	N	CT	Y(1)	Y(1)	CT	Y(1)	5

N = No; CT = Can't tell; Y = Yes; Y(1) = Yes with 1 point.

* = Not included in CASP score.

Study characteristics

Table 2 provides details on the characteristics of the studies. The mean age of participants recorded across all the studies was ≥55 years old with 64% (n = 794) males and 36% (n = 449) females. Participants had either moderate or severe COPD according to the National Institute for Health and Care Excellence (NICE) guidelines (2). Participants were recruited from a range of healthcare settings including hospitals (n = 11/20), PR sites (n = 5/20) and the internet (n = 3/20). Some participants were referrals from other unspecified settings (n = 5/20). The attrition rates by the end of the interventions ranged from 0–38%.

short-legendFigure 1.

Table 2.  Details of the 20 studies including recruitment, demographics, study design, quality, intervention groups, assessment times, attrition and PA instruments/parameters

Study	Recruitment and Demographics	Design and Quality	Intervention groups	Assessment times	Attrition rate	PA Instrument/Parameters
Randomised Controlled Trials/Randomised Trials
Nguyen (20) 2013	Recruitment: Internet, referrals, letter mailings, support groups, PR sites and advertising I1; n = 43; age 68.5 (11.0) yrs; 25M/18F; FEV1 53.3 (20.4)% I2; n = 41; age 68.2 (9.9) yrs; 19M/22F; FEV1 50.6 (18.2)% I3; n = 41; age 69.3 (8.0) yrs; 24M/17F; FEV1 49.4 (19.8)%	RCT PEDro: 7	I1 – Internet-based DSMP I2 – Face-to-face DSMP I3 – General health education	Baseline (T0), Mth 3, Mth 6, Mth 12 (T1)	I1–12% (n = 38/43); I2–15% (n = 35/41); I3–10% (n = 37/41)	Exercise self-report survey. Frequency and duration of endurance and strengthening exercises for a typical week during the last 4 weeks.
Pleguezuelos (21) 2013	Recruitment: Hospital I; n = 34; age 70.2 (2.5) yrs; 34M/0F; FEV1 32.0 (1.2)% C; n = 54; age 70.5 (2.5) yrs; 54M/0F; FEV1 31.8 (1.0)%	RCT PEDro: 6	I – Urban walking circuits C – Non-urban walking circuits	Baseline (T0), Mth 9 (T1)	15% (n = 6/40)	Activity diary. Frequency and duration of daily walking
Tabak (22) 2013	Recruitment: Hospital I; n = 14; age 65.2 (9.0) yrs; 8M/6F; FEV1 48.7 (16.7)% C; n = 16; age 67.9 (5.7) yrs; 11M/5F; FEV1 56.4 (10.6)%	RCT PEDro: 6	I – Telerehabilitation programme C – Usual care	Baseline (T0), Week 1, Week 2, Week 3 (T1)	22% (n = 4/18)	Yamax DigiWalker SW-200 pedometer. Step counts
Pomidori (23) 2012	Recruitment: Hospital I1; n = 18; age 70 (9) yrs; 13M/5F; FEV1 48 (13)% I2; n = 18; age 74 (7) yrs; 14M/4F; FEV1 49 (12)%	RT PEDro: 5	I1 – Speed walking paced by a metronome I2 – Walking a known distance in a given period of time	Baseline (T0), Mth 6, Mth 12 (T1)	I1–22% (n = 5/23); I2–25% (n = 6/24)	SenseWear armband. Daily and peak EE; time spent > 3 METs
Effing (24) 2011	Recruitment: Hospital I1; n = 77; age; 62.9 (8.1) yrs; 45M/32F; FEV1 49.6 (14.2)% I2; n = 76; age; 63.9 (7.8) yrs; 44M/32F; FEV1 50.5 (17.0)%	RCT PEDro: 6	I1 – Self-management programme including COPE-active I2 – Self-management programme only	Baseline (T0), Mth 7, Mth 12 (I1–T1; I2–T2)	I1–4% (n = 3/77); I2–11% (n = 8/76)	Yamax DigiWalker SW-200 pedometer. Step counts
Probst (25) 2011	Recruitment: Referral I1; n = 20; age 65 (10) yrs; 11M/9F; FEV1 39 (14)% I2; n = 20; age 67 (7) yrs; 10M/10F; FEV1 40 (13)%	RT PEDro: 5	I1 – Calisthenics and breathing exercises I2 – Endurance and strength training	Baseline (T0), Week 12 (T1)	I1–35% (n = 11/31); I2–38% (n = 12/32)	DynaPort activity monitor and SenseWear Armband. Time spent lying, sitting, standing, walking and > 3 METs; EE; step counts
Berry (26) 2010	Recruitment: Advertising and referral I1; n = 87; age 66 (10) yrs; 47M/40F; FEV1 50.5 (20.2)% I2; n = 89; age 66 (10) yrs; 48M/41F; FEV1 53.0 (18.5)%	RCT PEDro: 7	I1 – Lifestyle activity programme I2 – Traditional exercise therapy	Baseline (T0), Mth 3 (I2–T1), Mth 6, Mth 12 (I1–T1; I2–T2)	I1–28% (n = 24/87); I2–22% (n = 20/89)	CHAMPS questionnaire. Weekly moderate PA EE
Breyer (27) 2010	Recruitment: Hospital, local clinics and advertising I; n = 30; age 61.9 (8.87) yrs; 14M/16F; FEV1 48.1 (19.1)% C; n = 30; age 59.0 (8.02) yrs; 13M/17F; FEV1 47.1 (16.3)%	RCT PEDro: 6	I – Nordic Walking C – No intervention	Baseline (T0), Mth 3 (T1), Mth 6, Mth 9 (T2)	6% (n = 2/32)	DynaPort activity monitor. Daily movement intensity (m/s2); time spent walking, standing, sitting and lying
Hospes (28) 2009	Recruitment: Hospital I; n = 18; age 63.1 (8.3) yrs; 10M/8F; FEV1 67.4 (17.5)% C; n = 17; age 61.2 (9.1) yrs; 11M/6F; FEV1 61.8 (14.4)%	RCT PEDro: 5	I – Pedometer-based exercise counselling C – Usual care	Baseline (T0), Week 12 (T1)	10% (n = 2/20)	Yamax DigiWalker SW-200 pedometer. Step counts
Nguyen (29) 2009	Recruitment: PR sites I1; n = 9; age 72 (9) yrs; 3M/6F; FEV1 46.7 (18.7)% I2; n = 8; age 64 (12) yrs; 3M/5F; FEV1 34.4 (15.0)%	RCT PEDro: 7	I1 – MOBILE-Coached intervention I2 – MOBILE-Self-Monitored intervention	Baseline (T0), Mth 3, Mth 6 (T1)	I1–0%; I2–0%	Stepwatch® 3 activity monitor. Step counts,% inactive time (no steps),% active time in moderate (31–80 steps/min) & vigorous activity (>80 steps/min) and peak performance (average steps/min of the best 30 min of the day).
Nguyen (30) 2008	Recruitment: Internet, referrals, letter mailings, support groups, PR sites and advertising I1; n = 19; age 68.0 (8.3) yrs; 11M/8F; FEV1 49.0 (16.8)% I2; n = 20; age 70.9 (8.6) yrs; 11M/9F; FEV1 50.3 (17.6)%	RCT PEDro: 7	I1 – Internet-based DSMP I2 – Face-to-face DSMP	Baseline (T0), Mth 3, Mth 6 (T1)	I1–27% (n = 7/26); I2–17% (n = 4/24)	Stage of motivational readiness for exercise questionnaire and exercise self-report survey. Stage of change (action or maintenance); frequency and duration of endurance, strengthening and stretching exercises for a typical week during the last 4 weeks.
Steele (31) 2008	Recruitment: Hospital and advertising I – n = 54; C – n = 57; age (I & C) 67 yrs; Gender (I & C): 98M/8F; FEV1 I – 45 (17)%; C – 41 (17)%	RCT PEDro: 6	I – Exercise adherence intervention C – Usual care	Baseline (T0), Week 12 (T1), Week 44 (T2)	22% (n = 12/54)	RT3 accelerometer and activity diary. Daily VMU; daily self-reported exercise time
Varga (32) 2007	Recruitment: NR I1; n = 22; age 61 (12) yrs; 19M/3F; FEV1 51 (16)% I2; n = 17; age 67 (10) yrs; 11M/6F; FEV1 64 (29)% I3; n = 32; age 60 (12) yrs; 25M/7F; FEV1 52 (16)%	RT PEDro: 6	I1 – Continuous training I2 – Interval training I3 – Self-paced training	Baseline (T0), Week 8 (T1)	I1–0%; I2–0%; I3–0%	Laboratory activity questionnaire. Activity level score
De Blok (33) 2006	Recruitment: PR site I; n = 10; age 65.7 (10.4) yrs; 5M/5F; FEV1 52 (22)% C; n = 11; age 62.5 (12.3) yrs; 4M/7F; FEV1 43 (13)%	RCT PEDro: 6	I – PR + lifestyle PA programme C – PR only	Baseline (T0), Week 9 (T1)	20% (n = 2/10)	Yamax DigiWalker SW-200 pedometer. Step counts
Behnke (34) 2005	Recruitment: Hospital I; n = 66; age 61.2 (8.6) yrs; 51M/15F; FEV1 41.9 (13.9)% C; n = 22; age 58.4 (6.7) yrs; 20M/2F; FEV1 46.9 (15.1)%	RCT PEDro: 5	I – Hospital-based supervised exercise training C – No intervention	Baseline (T0), Day 8 + Day 10 (T1)	0%	TriTrac-R3D accelerometer. Activity counts
Berry (35) 2003	Recruitment: Mass mailings, advertising and referral I1; n = 70; age 68.4 yrs (95% CI = 67.0-69.8); 39M/31F; FEV1 57.6% (95% CI = 53.2-62.0) I2; n = 70; age 66.9 yrs (95% CI = 65.5-68.3); 39M/31F; FEV1 59.1% (95% CI = 55.0-63.2)	RCT PEDro: 6	I1 – Long-term exercise programme I2 – Short-term exercise programme	Baseline (T0), Mth 3 (I2–T1), Mth 9, Mth 15, Mth 18 (I1–T1; I2–T2)	I1–11% (n = 8/70); I2–20% (n = 14/70)	PASE questionnaire. PASE scores outside of centre-based training only
Cohort Studies
Moy (36) 2012	Recruitment: Hospital n = 27; age 72 (8) yrs; 27M; FEV1 55 (16)%	Cohort study CASP: 5	Internet & pedometer walking programme	Baseline (T0), Mth 1, Mth 2, Mth 3 (T1)	11% (n = 3/27)	Omron HJ-720ITC pedometer. Step counts
Moy (37) 2010	Recruitment: Hospital n = 24; age 56 (7) yrs; 13M/11F; FEV1 NR	Cohort study CASP: 5	Internet and pedometer walking programme	Baseline (T0), Week 16 (T1)	33% (n = 8/24)	Omron HJ-720ITC Enhanced pedometer. Step counts
Wewel (38) 2008	Recruitment: Hospital n = 21; age 65 (9) yrs; 17M/4F; FEV1 32.3 (9.4)%	Cohort study CASP: 5	Telephone-controlled home-based training	P1 (T0), P2 (T1)	19% (n = 4/21)	ActiTrac activity monitor and Kasper & Richter GmbH & Co. pedometer. Activity counts; pedometer readings
Nguyen (39) 2005	Recruitment: I1a – Pulmonary practices, support groups, PR site, internet; I1b – Previous programme participants n = 16; age 69.1 (7.0) yrs; 8M/8F; FEV1 36 (NR)%	Cohort study CASP: 5	Internet DSMP intervention	Baseline (T0), Week 12 (T1)	30% (n = 7/23)	Exercise self-report survey. Frequency and duration of endurance, strengthening and stretching exercises for a typical week during the last 4 weeks.

Abbreviations – 95% CI: 95% Confidence interval; CHAMPS: Community Health Activities Model Program for Seniors physical activity questionnaire; C: Control; DSMP: Dyspnoea Self-Management Programme; EE: Energy Expenditure; F: Female; M: Male; FEV1: Forced expiratory volume in one second; I: Intervention; METs: Metabolic Equivalents; Min: Minutes; Mth: Month: MOBILE; Mobilising Support for Long-term Exercise programme; m/s²: Metres per second squared; n: number; NR: Not reported; P1: 2-week baseline period; P2: 2-week intervention period; PA: Physical activity; PASE: Physical Activity Scale for the Elderly; PR: Pulmonary rehabilitation; RCT: Randomised controlled trial; RT: Randomised trial (no control); T0: Baseline assessment; T1: Post-intervention follow-up period (short-term); T2: Last follow-up period after T1 (long-term); VMU: Vector Magnitude Units; yrs: Years of age

*Baseline characteristics of all randomised participants.

**Baseline characteristics of completed participants only.

Intervention components

Detailed overview of the different intervention components are provided in the online Supplementary Table S1. Intervention duration ranged from 10 days to 18 months. Interventions were short-term ≤ 3 months (n = 17/31); medium-term 4–6 months (n = 5/31); or long-term > 6 months (n = 9/31). Modes of delivery included one or more of the following: face-to-face exercise sessions (n = 13/31), telephone (n = 13/31), face-to-face PA/exercise counselling sessions (n =10/31), PA/exercise diaries (11/31) and internet (n = 6/31). Individual interventions included one mode (10/31), two modes (9/31), or three or more modes of delivery (12/31). The frequency of contacts with the health professional varied according to the intervention duration; ranging from one contact over an 8-week programme (32) to 234 contacts over an 18-month programme (35).

The frequency, intensity, time and type of PA/exercise were not comprehensively recorded and varied greatly between interventions. PA/exercise was generally prescribed for ≥ 3 times per week (n = 25/31). In the interventions which clearly reported on intensity (n =21/31), this was prescribed using Borg scale ratings from 3 to 6 (n = 12/21); intensity ratings of 60%–90% of maximal effort based on exercise test results (n = 9/21) or a combination of both (n = 4/21). When the length of the PA/exercise sessions was reported (n = 24/31), most sessions lasted 10–45 minutes (n = 17/24) with the longest sessions lasting 60 minutes (n = 7/24).

In the interventions which clearly reported on weekly prescribed PA/exercise time (n = 22/31), this was 60–119 minutes (n = 4/22), 120–180 minutes (n = 16/22) or >180 minutes (n = 2/22). The most popular types of PA/exercise prescribed included walking (n = 26/31), cycling (n = 14/31) and resistance/strength training (n = 14/31). A range of PA/exercise settings were used including home and community (n = 18/31); research centre/hospital (n = 9/31); or a combination of home and research centre/hospital (n = 4/31). The PA/exercise sessions were either fully supervised (n = 9/31), fully unsupervised (n = 18/31) or had a mixed level of supervision (n = 4/31).

The mean number of behavioural change techniques reported in the interventions was six (range 1–13). Table 3 highlights the behaviour change techniques reported in the 31 interventions. The most popular included the following: information on where and when (n = 26/31) and instruction on how (n = 22/31) to perform the behaviour; prompt self-monitoring of behaviour (n = 18/31); goal setting (behaviour) (n = 15/31); setting graded tasks (n = 12/31); and facilitating social comparison (n = 11/31). Facilitating social comparison involves explicitly drawing attention to others’ performance to elicit comparisons (18). Pedometers were used to enable self-monitoring in 8/31 interventions with 5/8 of these interventions also explicitly using the pedometer as a motivational/behavioural reinforcement tool. One or more psychological theories for PA behaviour change were explicitly identified in seven studies including: Social Cognitive Theory (n = 5/7); Self-Regulation Theory (n = 3/7) and Self-­Efficacy Theory (n =2/7).

Table 3.  Number of behaviour change techniques used across the 31 interventions

Behaviour Change Technique	Number	Behaviour Change Technique	Number
Information on where and when to perform the behaviour (20–27,29–32,34–38)	26	Prompt rewards contingent on effort/progress towards behaviour (22)	1
Instruction on how to perform the behaviour (20–27,29,31,32,34–38)	22	Demonstrate the behaviour (36)	1
Prompt self-monitoring of behaviour (20,22,23,24,26,28,29,30,31,32,33,36,37,39)	18	Environmental restructuring (31)	1
Goal setting (behaviour) (20,22,26,27,28,29,30,31,33,36,37,39)	15	Information about others’ approval	0
Set graded tasks (24–26,28,31–33,36,37)	12	Normative information about others behaviour	0
Facilitate social comparison (24–27,32,35,37)	11	Goal setting (outcome)	0
Prompt practice (20–22,24,26,30,31,38)	11	Action planning	0
Use of follow-up prompts (20,23,24,26,30,31,39)	10	Prompt review of outcome goals	0
Plan social support/social change (20,26,29–31,36,37,39)	10	Provide rewards contingent on successful behaviour	0
Provide feedback on performance (20,22,29,30,35–37,39)	10	Shaping	0
Prompt self-monitoring of behavioural outcome (20,29,30,39)	7	Prompting focus on past success	0
Prompt review of behavioural goals (22,24,26,28,33,36,39)	7	Agree behavioural contract	0
Motivational interviewing (20,28,30,33)	6	Prompt identification as a role model	0
Barrier identification/problem solving (29,31,36,37,39)	5	Prompt anticipated regret	0
Prompting generalisation of a target behaviour (21,26,31,33,38)	5	Fear arousal	0
Stress management/emotional control training (20,30,39)	5	Prompt self-talk	0
Information on individual consequences of behaviour (24,31,36)	4	Prompt imagery	0
Teach to use prompts/cues (22,26)	2	Time management	0
Relapse prevention/coping planning (28,31)	2	General communication skills training	0
Information on general consequences of behaviour (20)	1	Stimulate anticipation of future rewards	0

*Michie et al. 2011 (18).

PA outcomes and intervention effects

All studies measured PA at baseline and post-intervention (n = 20/20). Assessment of PA included objective (pedometers n = 7/20 and accelerometers n = 7/20) and subjective tools (questionnaires n = 6/20 and activity diaries n = 2/20). The PA parameters included objective (step counts n = 8/20, activity counts n = 3/20, energy expenditure n = 2/20 and time spent in different PA n = 4/20) and subjective parameters (estimated energy expenditure n = 1/20 and time spent in different PA n = 6/20).

Between-group differences in PA were measured in 16/20 studies. Statistically significant between-group differences post-intervention in favour of the PA intervention group (I1), compared to a control group or to other PA interventions, in one or more PA assessments were found in 7/16 studies (21,23,24,27,28,31,34) including the following PA parameters: step counts (2/7); activity counts (1/7); objective energy expenditure (1/7); time spent in different objective PA (2/7); time spent in different subjective PA (2/7). There were similarities in the intervention components; particularly regarding the type of PA/exercise with walking included in all seven studies. Other components of the interventions varied. The duration of the intervention lasted 12 weeks (3/7 studies) or 12 months (2/7 studies).

Modes of delivery included face-to-face exercise sessions (4/7 studies), paper diaries (4/7 studies), face-to-face instruction /education sessions (4/7 studies), telephone calls or texts (3/7 studies) and individual face-to-face exercise counselling (2/7 studies). Intensity of PA/exercise was 70–80% of maximum heart rate in 3/7 studies and in addition to walking (7/7 studies), other types of PA/exercise included cycling (2/7 studies) and resistance training (2/7 studies). Common BCTs included information on where and when to perform the behaviour (6/7 studies) and instruction on how to perform the behaviour (6/7 studies). Other BCTs included were prompt self-monitoring of behaviour (4/7 studies), use of follow-up prompts (3/7 studies), set-graded tasks (3/7 studies) and goal setting (behaviour) (3/7 studies). Pedometers were used as self-monitoring tools in 2/7 studies.

The frequency of contacts, choice of PA/exercise setting, frequency and time per week spent in PA/exercise and PA/exercise supervision were variable between the seven studies. In the most commonly used PA parameter (step counts), six studies allowed for the magnitude of the short-term intervention effects to be calculated. One study (33) found a large effect (>0.8), three studies (22,24,28) found small effects (>0.2) and two studies (25,29) found trivial effects (≤ 0.2) of the PA intervention compared to the control/other PA intervention groups. Statistically significant differences in favour of the control group versus the PA intervention group (I1) were found in one study (29). Statistically significant between-group differences with short-to-medium follow-up (≤ 6 months) post-intervention in favour of the PA intervention group were found in 1/2 studies (27).

Within-group differences in PA were reported in 14/20 studies (10 RCTs and RTs and four cohort studies). Statistically significant improvements pre- versus post-intervention in one or more PA assessments were measured in 11/14 studies (20,21,23,26-28,30,32,36-38). A statistically significant decline in one or more PA assessments pre- versus post-intervention was measured in one study (25). Statistically significant improvements with short-to-medium follow-up (≤ 6 months) post-intervention was measured in 1/2 studies (27).

Discussion

This comprehensive review, which has followed a rigorous methodology, has highlighted a variety of components used in PA interventions in COPD and has identified a range of behaviour change techniques. A range of objective (pedometers and accelerometers) and subjective (questionnaires and activity diaries) PA outcomes were used and there is evidence that PA interventions have some effect on PA levels (step counts).

Patients with COPD are traditionally offered PR which has a strong evidence base for effectiveness (1–4). Surveys of PR have highlighted that there are not enough programmes available; there are currently less than 1.5% of patients with COPD annually receiving PR in the UK (40,41). Usually patients with moderate to severe COPD are targeted (1,3). The majority of programmes are outpatient-based and are supervised by clinicians (40,41).

This structured and supervised format of PR does not meet the needs of all patients, with high numbers of drop-outs and non-adherence (42–44), yet alternative options for increasing PA for patients with COPD currently do not seem to be offered. Studies have found that PR may not always lead to improved PA levels in COPD (45) and the benefits can diminish towards baseline by 6 and 12 months (1,42). Almost half of the studies in this review demonstrated a significant improvement in PA levels in the PA intervention group versus the control group. These positive results from PA interventions are supported by other reviews in different populations such as those being treated for cancer (46), type 2 diabetes (47) and cardiovascular disease (48).

Whether these significant improvements in PA result in minimum clinically important differences (MCID) for patients with COPD is unknown, as no data on MCID and PA parameters in COPD currently exists. Data from other clinical populations may help provide estimates of MCID for some of the PA parameters. For example, Motl and colleagues (49) have shown that a change of 779 steps per day appears to represent a MCID in patients with multiple sclerosis. Although it is unknown whether this MCID would be similar in patients with COPD, 5/8 studies that measured pre- versus post-intervention daily step counts in this review had an increase which may represent a clinically important difference (24,28,33,36,37).

Future research needs to establish MCIDs for PA parameters in COPD. To quantify the magnitude of the treatment effects on step counts, we calculated effect sizes for six studies. One study measured a large effect of the PA intervention whereas the other five studies measured small or trivial effects of the PA intervention on step counts compared to the control. Considering most patients with COPD typically exhibit low levels of daily PA (50), it is important to recognise that relatively small increases in PA are still likely to be effective in conferring important health benefits compared with larger PA increases in already more active populations (51).

The components of all the PA interventions were variable in terms of duration, format of delivery, contacts and the behaviour change techniques utilised. This makes it difficult to distinguish which components are likely to optimise the benefit of the PA interventions. However, focusing on the intervention components in the seven studies measuring significant between-group differences in favour of the PA intervention appeared to show some patterns. It is currently unclear what intensity, length of intervention and mode of delivery (e.g., internet or face-to-face) is required to support change in PA behaviour in COPD. A whole diversity of behaviour change techniques were used with the most common including providing information and instruction on the behaviour, self-monitoring and goal setting in terms of the PA behaviour. It is not feasible to include every behaviour change technique so careful attention should be given to the theoretical framework of the PA intervention to ensure compatibility between the intervention format and planned behaviour change techniques (18). Investment is important in future studies that explore optimal PA intervention/models.

The prescribed/recommended PA/exercise appeared to be based on established guidelines in COPD (1,3) and/or using general PA recommendations (52). In general, this included moderate aerobic PA/exercise for 150–180 minutes per week. Almost half of the interventions also included resistance training to provide a more holistic approach to PA/exercise training. Most of the interventions promoted walking which has proven physical and psychological health benefits (53). Walking can be completed in a range of settings, has a very low injury risk and can be completed at variable speeds. The approach taken by Berry and colleagues in gradually reducing patients’ dependency on staff and structured centre-based exercise training towards independently self-regulating daily life PA could be practical and potentially help in the long-term adoption of PA behaviour change (26).

The majority of studies in this review used an objective tool to assess PA (pedometer or accelerometer). They report variable parameters including step counts, activity counts, energy expenditure and time spent in different PA and this heterogeneity in reporting has limited the ability to conduct a pooled estimate of the treatment effect on PA levels. In order to provide a more comprehensive assessment of treatment effects, studies should include an objective outcome measure and report step counts and/or time spent in different intensities (including sedentary behaviour) at a minimum, and also energy expenditure when available (54).

Pedometers can be used to help patients monitor themselves to promote self-regulation and set goals by giving direct feedback (55), whereas most accelerometers do not provide direct feedback. Conversely pedometers can only enable estimation of daily step counts, whereas accelerometers provide a more accurate and comprehensive estimate of PA and sedentary behaviour in terms of volume and intensity. The subjective measures used in studies in this review included mostly in-house questionnaires and diaries. The limitations of questionnaires and diaries as outcome measures is well documented and subjective estimates of PA usually do not relate well to objective assessment (56, 57). Within COPD, it is likely that questionnaires and diaries are more useful as an aid for goal setting and providing supplementary information on PA rather than providing an accurate PA assessment (58).

This review updates published reviews in this area (5,59,60) and provides a current synthesis of the literature related to PA interventions specifically in COPD. The strength of this review was that it followed the PRISMA systematic approach in terms of having a detailed search strategy, independent assessment of study eligibility, data extraction and quality assessment by two research team members. Study quality was moderate-high with no low quality studies which also helped to strengthen the review findings.

We were unable to pool the results of all studies together due to the large heterogeneity, in terms of the PA intervention and the reporting of the PA outcome assessment. However, we were able to calculate effect sizes for some studies to calculate the magnitude of the PA intervention treatment effects. This review focused on PA interventions and PA outcomes but did not explore the impact of PA on other important outcomes such as quality of life and exercise capacity. The lack of details provided in most of the papers made it difficult in some circumstances to determine whether certain behaviour change techniques were being utilised and resulted in reporting bias due to potentially relevant techniques not being classified (61).

Globally there is a clear message that PA is important for improving health in chronic disease (62). However, patients with COPD face specific challenges when trying to break the cycle of breathlessness and physical inactivity. Support to adopt a more physically active lifestyle could lead to improvements in symptoms such as reduced exercise capacity, breathlessness and depression (63,64). The potential for low cost, accessible PA interventions is an attractive option for healthcare ­clinicians.

Conclusions

This review has shown that although the components of PA interventions were variable, there is some evidence that PA interventions have the potential to increase PA in patients with COPD. This review justifies investment in research that would explore the role of PA interventions in patients with COPD.

Declaration of Interest Statement

Author JW is funded by the Department for Employment and Learning Northern Ireland (DELNI) as part of his PhD. The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.

Supplementary materials are available in the online version of this article.