Sociodemographic Characteristics and Physical Activity in Patients with COPD: A 3-Month Cohort Study

Abstract

Decreased physical activity (PA) is associated with morbidity and mortality in COPD patients. In this secondary analysis of data from a 12-week longitudinal study, we describe factors associated with PA in COPD. Participants completed the Physical Activity Checklist (PAC) daily for a 7- to 8-day period. PA was measured monthly using the Physical Activity Scale for the Elderly (PASE). At three different time points, daily step count was measured for one week with an Omron HJ-720ITC pedometer. The 35 participants were primarily male (94%) and White (91%), with an average age of 66.5 years and FEV₁ 44.9% predicted. Common activities reported on the PAC were walking (93%), preparing a meal (89%), and traveling by vehicle (96%). PA measured by both PASE score (p = 0.01) and average daily step count (p = 0.04) decreased during follow-up. In repeated measures multivariable modeling, participants living with others had a higher daily step count (ß = 942 steps, p = 0.01) and better PASE scores (ß = 46.4, p < 0.001). Older age was associated with decreased step count (ß = −77 steps, p < 0.001) whereas White race was associated with lower PASE scores (ß = −55.4, p < 0.001) compared to non-White race. Other demographic factors, quality of life, and medications were not associated with PA. A better understanding of the role of social networks and social support may help develop interventions to improve PA in COPD.

Keywords:

• Pulmonary disease
• chronic obstructive
• physical activity
• physical exercise
• social support
• social isolation

Introduction

Chronic obstructive pulmonary disease (COPD) is highly prevalent worldwide and an important cause of morbidity and mortality.^1,2 Patients with COPD have significantly lower levels of daily physical activity (PA) when compared to healthy controls.^3,4 Among adults without chronic illness, numerous health related benefits are associated with being physically active, and there is increased recognition of the relationship between daily PA and clinical outcomes in patients with COPD.^5–10

Low PA in COPD is associated with an increased risk of acute exacerbations, hospitalizations, and mortality.^11–15 Additionally, maintaining higher levels of PA over time correlates with lower risks of COPD exacerbations and hospitalizations.^16–18 Given the importance of PA as a predictor of COPD health outcomes, there is a need to identify modifiable determinants in order to develop interventions to increase patients’ activity level.^19,20

COPD severity, dyspnea, and medical co-morbidities are associated with decreased PA.^21,22 However, prior studies that have examined the role of sociodemographic and lifestyle factors on PA such as marital status, living situation, education, smoking, alcohol use, and employment have had inconsistent results and were often not adjusted for potential confounders.^6,19,23,24 One study examining the role of social support in COPD found that better structural and functional social support was associated with increased PA in adjusted analysis.²⁴

Additional studies are therefore needed to better understand the sociodemographic and lifestyle factors associated with PA, accounting for disease severity and other confounders.^19,20 The aim of this study is therefore to examine the sociodemographic factors associated with PA in patients with COPD over a 3-month period.

Methods

Study design

This study analyzed data collected during a 12-week observational cohort study of ambulatory patients with COPD. The primary aim of the study was to examine the feasibility of using an inhaler sensor to measure the association between short-acting beta-agonist inhaler use and COPD related symptoms, air pollution, and PA patterns.²⁵

Participants

Inclusion criteria consisted of a diagnosis of COPD, short-acting beta-agonist inhaler use, age >40 years, post-bronchodilator forced expiratory volume in one second (FEV₁) to forced vital capacity (FVC) ratio <0.70 and an FEV₁% predicted ≤80%, prior tobacco use with >10 pack-year smoking history, and no exacerbations in the four weeks prior to enrollment. Exclusion criteria included a diagnosis of asthma, a modified Medical Research Council (mMRC) dyspnea score of four (too breathless to leave the house or breathless when dressing), or the need for assistance in walking.

Procedures

At baseline, participants performed spirometry and completed questionnaires regarding medical history, respiratory symptoms, PA, and health related quality of life. Baseline COPD disease severity was assessed with post-bronchodilator FEV₁ and the mMRC dyspnea scale. Sociodemographic data included age, sex, race, income, employment, education, alcohol use, enrollment season, marital status, and whether they live alone or with others.

Several questionnaires were administered at baseline and repeated monthly. Dyspnea was measured with the UCSD Shortness of Breath Questionnaire (SOBQ) which has a range 0 to 120 (higher scores indicated worse dyspnea).²⁶ PA was measured using the Physical Activity Scale for the Elderly (PASE), a validated questionnaire of leisure, occupational, and household activities (range 0 to 400; higher scores indicate greater PA).²⁷

During the 3-month study follow-up period, monthly in-person or telephone interviews were conducted to collect data on exacerbation symptoms and treatment, health care utilization, and to repeat the SOBQ, and PASE questionnaires. To describe the engagement and frequency of various types of PA, participants completed a Physical Activity Checklist (PAC) daily for seven to eight days approximately one month into the study.^28,29

Participants were also provided an Omron HJ-720ITC pedometer (Omron Healthcare, Inc; Bannockburn, Illinois), which has been validated in COPD.³⁰ Participants wore the pedometer for three separate seven-day periods every four weeks to measure daily step count.

Statistical analysis

The primary outcome for the study was PA, which was assessed using two different measures: the PASE questionnaire (administered four times during the 12 weeks), and steps/day (measured three times). The daily steps was an average of the number of daily steps collected by the pedometer during each of the three seven-day time periods.

Repeated measures models were used to examine the effect of sociodemographic and lifestyle variables on these two PA outcomes. First, we examined whether the two PA outcomes varied over time during the three-month follow-up period. Second, we examined the relationship between the two PA outcomes and each sociodemographic variable, adjusting for study month. For sociodemographic variables that were associated with PA at p < 0.10, we then examined potential confounders including disease severity, COPD medications, comorbid illness, enrollment season, and obesity. None of the potentially confounding variables were associated with both the sociodemographic variables and the two PA outcome variables at p < 0.10. For the final models, after adjusting for study month, we included only three sociodemographic variables: living with others (for both PA outcomes), race (for PASE only), and age (for average daily steps only). We repeated the analyses adjusting for COPD severity with age, FEV₁ and the mMRC dyspnea scale³¹ to improve precision of the estimates in the final models of the two PA outcomes.³²

All mixed model analysis was performed using PROC MIXED using SAS software 9.4 (TS1M5), copyright (c) 2016 by SAS Institute Inc., Cary, NC, USA.

Results

Patient characteristics

The study cohort was mainly male (94%) and White (91%), with a mean age of 66.5 ± 8.5 years (Table 1). Mean FEV₁% predicted was 44.9 ± 17.2 and mean mMRC dyspnea score was 2.3 ± 0.9. Home oxygen was used by 49% of participants.

Table 1. Baseline characteristics of study participants.

	N = 35
	N (%) or mean (±SD)
Demographics/Lifestyle
Age, mean (SD)	66.5	±8.5
Male sex	33	94.3
Race, White	32	91.4
Income > $20,000	18	51.4
Employed currently	3	8.6
Education, college	29	82.9
Unhealthy alcohol use	10	28.6
Enrollment period May–Sept	22	62.9
Live with others	20	57.1
Living as a couple	11	31.4
Body Mass Index, mean (SD)	29.1	±5.5
COPD severity
FEV1 % predicted, mean (SD)	44.9	±17.2
FEV1 % predicted, by severity category
50–80% (Moderate)	12	34.3
30–49% (Severe)	15	42.9
0–29% (Very severe)	8	22.9
mMRC dyspnea scale,^a mean (SD)	2.3	±0.9
0–1	7	20.0
2	10	28.6
3	18	51.4
COPD treatment
Long-acting beta-agonists, inhaled	24	68.6
Long-acting anticholinergic, inhaled	18	51.4
Inhaled corticosteroids, inhaled	21	60.0
Methylxanthine use, oral	2	5.7
Chronic corticosteroid use, oral	7	20.0
Home oxygen therapy	17	48.6
Completed pulmonary rehabilitation	10	28.6
Comorbid illness
Coronary artery disease	11	31.4
Heart Failure	3	8.6
Peripheral artery disease	4	11.4
Osteoarthritis	10	28.6
Depression	16	45.7
Anxiety	12	34.3
Post-traumatic stress disorder	11	31.4

SD = standard deviation; FEV₁ = Forced expiratory volume in 1 s; mMRC = modified medical research council.

^a Patients with mMRC = 4 were excluded from the study.

Types of daily physical activity reported by participants

Participants completed the PAC an average of 7.1 ± 0.8 days (range: 4–8 days). The most common activities that were performed on at least one day were traveling in a bus, car or train (96%), walking more than five minutes at a time (93%), and preparing a meal (89%) (Table 2). The most common activity that was performed on more than half the recorded days was walking more than 5 min at a time (75%), and meal preparation (71%). Among the 25 participants who prepared their own meal on at least one day, they prepared meals an average of 79% of the days they completed the PAC. Few participants exercised regularly with only six going to the gym at least once, of which, only three went to the gym more than half the days.

Table 2. Daily activities reported by participants on the Physical Activity Checklist (PAC).

	Days performed (range 4–8 days)
	≥One day		>50% of days		Proportion^a
Type of physical activity	n	%	n	%	Mean	SD
Travel in a car, bus, or train	27	(96.4)	16	(57.1)	0.59	(0.30)
Walk more than 5 min at a time	26	(92.9)	21	(75.0)	0.76	(0.26)
Prepare your own meal	25	(89.3)	20	(71.4)	0.79	(0.28)
Wash and fold laundry, sweep floors, make bed	24	(85.7)	12	(42.9)	0.56	(0.32)
Pick up mail from the mailbox	23	(82.1)	15	(53.6)	0.60	(0.24)
Visit with relatives or friends	22	(78.6)	9	(32.1)	0.51	(0.29)
Perform an errand (post office, bank, dry cleaner)	21	(75.0)	6	(21.4)	0.44	(0.25)
Walk in a grocery store	20	(71.4)	6	(21.4)	0.48	(0.30)
Unload groceries from car and carry them into home	19	(67.9)	6	(21.4)	0.45	(0.29)
Walk in a mall or store	13	(46.4)	1	(3.6)	0.27	(0.15)
Climb up stairs	12	(42.9)	9	(32.1)	0.75	(0.34)
Go to a medical appointment	9	(32.1)	2	(7.1)	0.33	(0.24)
Dig in yard or garden, chop wood, or shovel snow	8	(28.6)	1	(3.6)	0.29	(0.21)
Exercise at a gym (swim, bike, treadmill, weights)	6	(21.4)	3	(10.7)	0.50	(0.22)
Exercise at home (bike, treadmill, weights)	6	(21.4)	3	(10.7)	0.52	(0.28)

Two activities from the original PAC are not included because they did not apply to all participants: spending time with grandchildren and walking the dog.

^a Proportion of days the participant performed the activity from total number of days the PAC was completed; calculated only for participants who engaged in the activity.

Sociodemographic variables associated with physical activity

The average PASE score at baseline was 100.2 ± 48.4 and decreased over the three-month follow-up period to 82.0 ± 53.7 (p=.013) (Table 3). The average daily step count at baseline was 2,483 ± 2,141 and decreased over three months to 1,897 ± 1,500 (p < 0.10) (Table 4). Race was associated with PASE score, and age was associated with average step count. No other sociodemographic variables were associated the two PA outcomes.

Table 3. Change in measures of physical activity, and dyspnea, over the 3-month study period.

	Baseline (N = 35)		Month 1 (N = 35)		Month 2 (N = 34)		Month 3 (N = 34)
	Mean	SD	Mean	SD	Mean	SD	Mean	SD	p-value
Outcomes
PASE score	100.2	(48.4)	96.0	(62.1)	83.9	(58.1)	82.0	(53.7)	.013
Average daily steps^a	2483	(2141)	2259	(1737)	–	–	1897	(1500)	.038
SOBQ dyspnea score	52.0	(21.5)	57.9	(23.1)	54.5	(23.3)	52.7	(23.3)	.057

PASE = Physical activity scale for the elderly (higher score, better health); Average daily steps = average number of steps per day over 7-day period measured by pedometer; SOBQ = UCSD Shortness of Breath Questionnaire (higher score, worse health).

^aN = 32 at baseline, N = 30 at Month 1, N = 32 at Month 3.

Table 4. Association between individual sociodemographic variables and two physical activity outcomes, after adjusting for study month.

	PASE score^a		Average daily steps^b
Variables^c	ß	p-value	ß	p-value
Age, year	−0.5	0.66	−74	0.03
Male	13.9	0.71	−859	0.50
Race, White	−74.7	0.01	−969	0.35
Income > $20,000	14.3	0.40	224	0.70
Employed, currently	−2.9	0.92	−348	0.74
Education, college	20.8	0.36	193	0.80
Unhealthy alcohol use	14.7	0.44	443	0.49
Enrollment period May–Sept	−3.5	0.84	−794	0.18
Lives with others	40.0	0.02	1092	0.06
Lives as a couple	27.7	0.13	227	0.72
Walks a dog	10.1	0.58	498	0.42

PASE = Physical activity scale for the elderly (higher score, better health); Average daily steps = average number of steps/day over 7-day period measured by pedometer.

^aAdjusted for study month (baseline, month 1, month 2, month 3).

^bAdjusted for study month (baseline, month 1, month3); steps/day not measured at month 2.

^cEach row is a separate repeated measures model with the individual sociodemographic variable to predict the PASE score and the average daily steps after adjusting for study month.

Living with others was associated with an increased PASE score (β = 37.2, p < 0.0001) in the final model, adjusting for race and study month (Table 5A). After also adjusting for COPD severity (age, FEV₁ and mMRC), the PASE score was 46.4 (p < 0.0001) points higher for those who lived with others (Table 5A). White race compared to non-White race was associated with a worse PASE score (β= −55.4, p < 0.0001). In the model to predict average daily step count (Table 5B), living with others (β = 798, p = 0.03) and age were associated with increased daily step count in a model adjusting for study month. After further adjustment for disease severity with FEV₁ and mMRC, participants who lived with others walked 942 more steps (p = 0.008) compared to those living alone, and increasing age was associated with 77 fewer (p = 0.0003).

Table 5. Multivariable association between sociodemographic variables and two physical activity outcomes.

(A) PA outcome: PASE score
	Model 1^a		Model 2^a
	ß	p-value	ß	p-value
Lives with others	37.2	<0.0001	46.4	<0.0001
Race, White	−70.7	<0.0001	−55.4	<0.0001
Age, per year			−0.3	0.44
Study month
Month 0	Ref	Ref	Ref	Ref
Month 1	−4.2	0.72	−4.2	0.67
Month 2	−17.1	0.15	−16.7	0.09
Month 3	−18.9	0.11	−18.5	0.06
FEV1 % predicted
<30%			Ref	Ref
30–49%			−10.1	0.29
50–79%			−8.5	0.40
mMRC dyspnea scale^b
0–1				Ref
2			−37.7	0.0007
3			−71.4	<0.0001
(B) PA outcome: Average daily steps
	Model 1^a		Model 2^a
	ß	p-value	ß	p-value
Lives with others	798	0.03	942	0.008
Age, per year	−68	0.001	−77	0.0003
Study month
Month 0	Ref	Ref	Ref	Ref
Month 1	−209	0.62	−156	0.69
Month 3	−573	0.17	−537	0.17
FEV1 % predicted
<30%			Ref	Ref
30–49%			−203	0.64
50–79%			663	0.14
mMRC dyspnea scale^b
0–1			Ref	Ref
2			646	0.18
3			−592	0.18

PASE = Physical activity scale for the elderly (higher score, better health); Average daily steps = average number of steps/day over 7-day period measured by pedometer; FEV₁ = forced expiratory volume in one second (higher score, better health); mMRC = modified Medical Research Council dyspnea scale (higher score, worse health).

^aModel 1: Multivariable model adjusted for study month plus the variables associated with physical activity (p < 0.10) in Table 4. Model 2: Multivariable model adjusted for the variables in model 1 plus COPD severity variables: Age, FEV₁ % predicted and mMRC dyspnea scale.

^bPatients with a mMRC score of 4 (too dyspneic to leave the house or breathless when dressing) were excluded from this study.

Unlike living with others, living as a couple was not associated with PA in unadjusted analysis to predicted either PASE score (β = 27.7, p = 0.13) or average daily step count (β = 227, p = 0.72) and was therefore not included in the final models (Table 4).

Discussion

In this three-month longitudinal study of participants with moderate-to-severe COPD, we found that among sociodemographic characteristics, living with others was associated with both objectively measured higher average daily step counts and higher levels of self-reported PA measured with the PASE instrument. Increasing age was associated with lower average daily step count, whereas White race was associated with lower PASE scores. Other characteristics such as education, employment, income, sex, unhealthy alcohol use, enrollment period, or dog walking were not associated with PA in this analysis.

The finding that living with others is associated with increased PA is consistent with prior research in non-COPD populations which showed that social support in older adults is associated with increased PA, particularly social support from family members.³³ In addition to the fact that these studies were not performed among persons with COPD, there was variability in the measurement of social support and PA.³³

When examining the sociodemographic and lifestyle factors associated with PA in COPD, a review by Gimeno-Santos et al. described one study that showed being married was associated with lower energy expenditure and found that most studies did not adjust for potential confounders and that the findings of these studies were not consistent.¹⁹ Similar to our study, increasing age was found to be associated with low physical activity although the quality of evidence in this prior study was noted to be poor. Our study supports the finding that increased age is associated with decreased average daily step count among those with COPD. The review by Gimeno-Santos also noted that there may be differences in PA between cultural groups, although these results are also not consistent. We found White participants had lower self-reported physical activity than non-White participants, although this finding is limited by the small sample size with only three non-White participants.

We found that living with others is associated with increased physical activity measured both with the self-reported PASE score and with objectively measured average daily step count. This is consistent with another study that examined the association between social support and PA measured with a triaxial activity monitor and found a similar magnitude of increased number of steps per day (942 steps in our study) for those who lived with someone compared to those who live alone.²⁴ This would be considered a clinically important difference in step count in COPD, which ranges from 350 to 1100 steps/day and is associated with a difference in medical events such as worsening breathing, emergency department visits, and hospitalizations.³⁴ This study by Chen et al. examined the association between living with others and average daily step count in a larger sample of COPD patients and adjusted for additional characteristics including psychological symptoms, sex, education, income, employment status, alcohol use, comorbidity, home oxygen use, and BODE score. Our findings help to support the findings from this initial study by Chen et al., and extend the results to self-reported PASE scores, which may be more practical to collect than objective daily step counts.

Social support can be both functional support such as emotional, instrumental appraisal, and information, or can be structural support as described by social networks.³⁵ Living with others is a type of structural support, and interestingly in our study, living with someone was more strongly associated with PA than living as a couple. This suggests that the benefit of cohabitation might arise from the positive effects of general social support mechanisms rather than specific types of social bonds. One possible mechanism was suggested by Mesquita et al. in a study that found participants with COPD living with someone more physically active are more likely to have higher levels of moderate to vigorous PA compared to those living with someone who is more sedentary.³⁶ Loneliness is associated with decreased social network index and is also associated with decreased PA.³⁷ A better understanding of how living with others impacts exercise and engagement in PA will therefore help to develop future interventions to increase PA.

Increasing or maintaining PA in COPD has been identified as a clinical target because it is an important prognostic factor for long term hospitalization risk and mortality.^12,38 PA in COPD patients has been characterized by differences in both the quantity and quality of the activities as compared to the general population.³⁹ To examine the types of PA that patients with COPD engage in, we administered the PAC for seven days to better quantify engagement in common activities more associated with light, moderate, or vigorous activity. We found that few participants reported exercising at a gym or at home, and that the most common activities were walking and common household activities such as traveling, preparing meals, or getting the mail. Notably it has been shown that conventional recommendations for consecutive periods of moderate-to-vigorous PA may be impractical for COPD patients who may have decreased stamina, and they may favor bursts of moderate-to-vigorous or more prolonged periods of light-to-moderate PA.^22,39

One unexpected observation in this study was a significant decline in PA over the three-month study period measured with both the PASE score and pedometer step count. An observational study by Bergman et al. of 50 healthy adults (mean age 31) examined how many days are needed to establish a baseline PA level using an accelerometer.⁴⁰ The authors found that 9 days are needed for light intensity, 33 days for moderate intensity and 302 days for vigorous intensity of PA. The study by Bergman did not focus on a chronically ill population such as our population of COPD patients, but this finding suggests that one explanation for the decline in PA seen in our study is the Hawthorne effect, in which subjects behave differently when initially observed but over time return to their usual PA. Another possibility is that exacerbations may have caused transient decreases in PA. Importantly, the relationship between sociodemographic variables and PA did not change over time in our study.

This study did have several limitations. This was an analysis of data from a pilot study with a small sample size, limiting the power to detect associations between participant factors and PA. Step count was measured using a pedometer worn for seven days during three different time periods, an approach consistent with other studies but that may not represent patients’ typical PA. The decline in PA measured both by PASE score and by average daily step count raises the possibility that the engagement and intensity was not at usual levels during the 3-month study period.⁴⁰ Although developed for elderly patients, there is limited testing of the PASE specifically in COPD and there is not an established minimally clinically important difference. However, a questionnaire-based measure such as the PASE may provide important additional information when objectively measured PA using pedometers or accelerometers are not practical. Further validation of this measure in COPD is needed. Strengths of this study include a well-characterized sample of patients with COPD with close follow-up and adjustment for disease severity and sociodemographic factors.

Conclusions

Participants with COPD who live with others have subjectively and objectively measured increased levels of PA compared to those who live alone. Given the documented benefits of PA on morbidity, mortality, and natural progression of this very common chronic medical condition, interventions should consider incorporating social support to improve engagement in PA to improve patient outcomes.

Disclaimer

The views expressed in this article are those of the authors and do not necessarily reflect the position or policy of the Department of Veterans Affairs or the United States Government.

Acknowledgements

Prior to this research human subjects approval was obtained by the VA Puget Sound Institutional Review Board (ID: 00408) and all participants provided written informed consent. This work was supported by Pilot Project Award #10-299 from the United States Department of Veterans Affairs, Health Services Research and Development Program.

Disclosure of interest

In the last three years, Dr. Fan has received grant funding from the Department of Veterans Affairs, the National Institutes of Health, the Firland Foundation, Patient-Centered Outcomes Research Institute. The other authors report no conflict of interest.