Changes in physical activity and all-cause mortality in COPD

Introduction

Subjects with chronic obstructive pulmonary disease (COPD) have lower levels of regular physical activity compared to subjects without COPD, even in mild-to-moderate COPD [1]. Many subjects with COPD are not able to meet the recommended minimal amount of regular daily physical activity [2–4]. This may differ between geographic areas, cultural backgrounds and/or methods used to assess daily physical activity [5].

Lower levels of physical activity have been associated with accelerated lung function decline [6], exercise intolerance [1], lower limb muscle weakness [1], lower muscle mass [7] and hospitalisations [8, 9]. In addition, subjects with COPD with very low levels of regular physical activity have the worst prognosis. This has been shown in a population-based study using self-reported levels of physical activity [8], as well as in clinical samples of subjects with COPD using objectified levels of physical activity [10].

In most COPD studies, physical activity has only been assessed at one time point [6, 8, 10–13]. However, subjects with COPD may change their physical activity behaviour over time. Only a few studies have focused on long-term changes in regular physical activity in subjects with COPD, showing a decline in physical activity in 19–26% of the subjects [14–16]. However, a common limitation to all previous studies is a lack of comparison between physical activity changes in subjects with COPD and subjects without COPD. Therefore, it is not possible to establish whether the observed changes are due to COPD progression or ageing.

A decline in regular physical activity may affect clinical characteristics in subjects with COPD. Esteban et al. [14] showed that a decline in self-reported regular physical activity at 5-year follow-up was associated with a decline in disease-specific health status in subjects with COPD. Moreover, a study in the general population found that maintaining or adopting a moderate or high level of physical activity was associated with lower risk of death [17]. Whether and to what extent changes in regular physical activity affect survival of subjects with COPD remains unknown. Nevertheless, it seems reasonable to hypothesise that a decline in regular physical activity over time may affect risk for all-cause mortality in subjects with COPD.

Therefore we sought to: 1) compare the longitudinal changes in the level of self-reported regular physical activity between subjects with and without COPD; 2) identify baseline determinants of changes in physical activity in subjects with and without COPD; and 3) identify the association between the changes in the level of self-reported regular physical activity and all-cause mortality in subjects with and without COPD.

Results

Longitudinal changes in physical activity

The proportion of the subjects with COPD with a low baseline physical activity who retained low physical activity at follow-up was significantly higher compared to the proportion of subjects without COPD (42.6 versus 31.9%, p<0.001) (fig. 2). Furthermore, the proportion of subjects with COPD with a moderate baseline physical activity level who changed their physical activity level at follow-up was significantly different compared to the proportion of subjects without COPD (decrease in physical activity: 14.9% versus 11.4%, p = 0.003; increase in physical activity: 26.4% versus 27.8%, p = 0.045) (fig. 2). Finally, the proportion of subjects with COPD with a high physical activity level at baseline who reported a decline in physical activity at follow-up was significantly higher compared to the subjects without COPD (52.6% versus 46.7%; p = 0.010) (fig. 2). All differences remained significant after correction for differences in age and sex (p<0.05). Differences in baseline characteristics of physical activity change in subjects with and without COPD are shown in tables E4 and E5.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2–

Longitudingal changes in physical activity in a) 1733 subjects with chronic obstructive pulmonary disease (COPD) (low: n = 305; medium: n = 890; high: n = 538) and b) 14 926 subjects without COPD (low: n = 1914; medium: n = 7992; high: n = 5020). *: p<0.05 compared with subjects without COPD with the same physical activity level.

Longitudinal changes in physical activity and all-cause mortality

During a median (IQR) follow-up of 17.1 (9.3–24.2) years, 5392 (64.9%) subjects died. The highest proportion of nonsurvivors was found among subjects with COPD (n = 944; 81.0%). In subjects with COPD and a low physical activity at baseline, no significant differences were found in crude and adjusted hazard ratios of mortality between subjects who remained at low physical active at follow-up and subjects who reported an increase in physical activity (table 4 and fig. 3a). Subjects with COPD and moderate to high physical activity at baseline who reported a low physical activity level at follow-up had the highest crude and adjusted hazard ratios of mortality (1.64 and 1.73, respectively, for subjects with moderate baseline physical activity, and 3.31 and 2.35, respectively, for subjects with high baseline physical activity, both p<0.001) (table 4 and figs 3b and c). In addition, subjects without COPD with a low physical activity at follow-up had the highest hazard ratio of mortality, irrespective of baseline physical activity level (p<0.05) (table 4 and figs 3d–f).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3–

Survival curves according to longitudinal changes in physical activity in a–c) subjects with chronic obstructive pulmonary disease (COPD) and d–f) without COPD and a, d) low physical activity level at baseline, b, e) moderate physical activity level at baseline and c, f) high physical activity level at baseline.

View this table:

• View inline
• View popup

Table 4– Crude and adjusted associations between longitudinal changes in physical activity and all-cause mortality in subjects with and without chronic obstructive pulmonary disease (COPD)

Discussion

This study has three major findings: 1) a decline in self-reported regular physical activity is significantly more common in subjects with COPD compared to subjects without COPD; 2) baseline determinants of longitudinal changes in self-reported physical activity are different between subjects with and without COPD; and 3) a decline in physical activity at follow-up was associated with an increased all-cause mortality risk in subjects with and without COPD. Thus, it seems that whereas an increase in physical activity was associated with reduced mortality in subjects without COPD, this was not the case in subjects with COPD who increased their activity level from low to either moderate or high.

The proportion of subjects with COPD with a low physical activity level was significantly higher compared to subjects without COPD, indicating a more inactive lifestyle in subjects with COPD. This finding is consistent with earlier studies [3, 11, 22–24]. The proportions of subjects with COPD with a high physical activity level (29%) may be somewhat surprising compared to subjects without COPD (31.7%). Then again, the CCHS cohort is a sample of the general population. This may explain, at least in part, that the proportion of COPD patients with high baseline physical activity level is comparable with the proportion of subjects without COPD who report a high baseline physical activity level. Indeed, COPD patients with Global Initiative for Chronic Obstructive Disease stage I and II were highly prevalent.

In addition, the proportion of subjects with COPD with a decline in physical activity was higher compared to the proportion of subjects without COPD, irrespective of physical activity level at baseline (Tables 3 and 4). This finding is novel as previous studies on the long-term changes in physical activity did not include a non-COPD control group [14–16].

From cross-sectional studies it is well known that physical activity is associated with clinical, sociodemographic and behavioural factors [1, 5, 11, 14, 22, 23, 25, 26]. However, data on determinants of longitudinal changes in regular physical activity are limited. The current results indicate that determinants of longitudinal changes in self-reported physical activity differ between subjects with and without COPD. Changes in physical activity in subjects without COPD are determined by a large variety of factors, including clinical data, sociodemographic data and the presence of comorbidities, whereas in subjects with COPD the variables with the highest likelihood for changes in physical activity are sex, the degree of airflow obstruction and smoking status. This may indicate that subjects with COPD are mainly limited by the consequences of COPD and that other variables have a limited additional impact on changes in physical activity. Then again, possible confounders, such as diet and objectified comorbidities, were not assessed, which may also partly explain the greater decline in physical activity in subjects with COPD. Determination of clinical and demographic correlates of physical activity may be important for identifying subjects at risk of developing an inactive lifestyle, but also for designing a specific and adequate physical activity programme for these subjects.

Previously, using the present population, Garcia-Aymerich et al. [8] showed that subjects with COPD and a very low to low baseline physical activity level had a higher mortality risk than subjects with COPD and a moderate to high physical activity level. However, changes in physical activity behaviour were not taken into account. Our analyses are the first to show an association between long-term changes in regular physical activity and all-cause mortality in subjects with COPD. A decline to a low physical activity level was associated with the highest all-cause mortality risk in subjects with and without COPD. Moreover, it seems that being moderately physically active may already have important beneficial consequences in subjects with and without COPD. However, in subjects with COPD with a low physical activity level at baseline, no beneficial effect of increased physical activity at follow-up was seen. This suggests that once subjects with COPD become physically inactive, the high mortality risk cannot be reversed by increasing activity levels. Thus, in subjects with COPD it is of clinical importance to remain at a moderate to high physical active level.

Given the association between reduced physical activity and important health-related outcomes [8, 10, 25], as well as the findings of this study, an adequate physical activity level should already be encouraged in subjects with mild COPD. Structured exercise programmes seem to be beneficial in increasing physical activity; however, results are not uniform and long-term data are lacking [27]. Since it is important that subjects with COPD integrate physical activity in their daily life, education and counselling should be an important component of programmes targeting physical activity. Indeed, it has previously been shown that activity monitor-based counselling has a beneficial effect on physical activity and health-related outcomes in patients with chronic diseases [28]. Stimulating physical activity and prevention of a sedentary lifestyle in the earliest stages of COPD may not only result in less adverse health-related consequences and a reduction in healthcare utilisation and costs [8], but also in better survival.

The results of our study may be affected by several limitations. The inclusion of subjects with at least two consecutive examinations may have caused a certain degree of selection bias, as we will have selected the healthiest subjects for analyses. Furthermore, the assessment of physical activity by a questionnaire may represent a subjective level of physical activity. However, it has previously been shown that self-reported measures are fairly accurate of changes in physical activity in large epidemiological studies [29], and it is probably the only way to collect repeated data in such large sample sizes over time. Moreover, the large sample size in this study compensates any degree of nondifferential misclassification of subjects [30]. Finally, the questionnaire used in the CCHS to assess physical activity fulfils the criteria of a patient-reported outcome questionnaire [19]. Questions and answer options could easily be interpreted and completed by the participants, subjects’ physical activity behaviour was measured without interpretation from anyone else, and this questionnaire was able to detect clinically relevant changes in physical activity (fig. 3). As such, it provides important results to subjects with COPD and their healthcare providers in order to make proper, patient-centred treatment decisions, including lifestyle modifications [31].

The used sample is regarded as representative for the Danish population. It might be argued whether our results can be generalised to other populations, for example those who cycle less. With regard to the possibility of residual confounding factors, our results have been adjusted for all available risk factors. It is plausible that other factors not included in the CCHS data (e.g. diet and objectified comorbidities) may also contribute to changes in physical activity. In addition, we did not adjust for changes in FEV₁ and BMI, since we believe that these are mediators rather than confounders. Nevertheless, additional analyses showed that the small nonsignificant changes in FEV₁ and BMI over time do not contribute significantly to the final multivariate model as presented in Tables 4, E8 and E9.

Although a longitudinal study design can limit the effect of reverse causality, it might still be present. However, a previous study showed that earlier reported associations from the CCHS were not due to time-dependent confounding [32].

To conclude, our results show that subjects with COPD not only reported a lower physical activity level compared to subjects without COPD, but are also more likely to decrease their level of physical activity over time. Moreover, subjects with a decline to a low physical activity level at follow-up had a higher all-cause mortality risk than subjects who remained or increased regular physical activity compared to baseline. However, the beneficial effect of increasing physical activity did not apply to subjects with COPD who were physically inactive at baseline. Therefore, these observational data suggest that it is important to assess and encourage regular physical activity at the earliest stages of COPD in order to maintain a physical activity level that is as high as possible, as this is associated with a better prognosis.