Introduction

Cardiovascular diseases (CVDs) are the leading cause of the global disease burden, which is expected to rise due to population growth and ageing [1]. CVDs are the most prevalent non-communicable diseases and a major cause of mortality and morbidity [2], impacting individuals’ quality of life and economic growth [3]. CVDs encompass a range of conditions, with some of the most common being angina pectoris and myocardial infarction, as major clinical manifestations of coronary artery disease [2], as well as heart failure, stroke, and atrial fibrillation and flutter [1]. Physical activity (PA) is essential in secondary prevention of CVDs, as it helps prevent recurrent events [4] and is linked to lower rehospitalization [5] and risk of mortality [6]. However, factors associated with PA, ranging from demographic to health-related characteristics, remain underexplored in individuals with CVDs [7]. In this context, device-measured PA should be preferred when investigating associations with PA, as it provides a more precise quantification of PA compared to self-reported PA by limiting recall and response bias [8], although evidence from such studies remains limited.

Correlates of PA have been examined in the general adult population, with demographic factors such as age and sex among the most commonly studied [9,10]. Some studies report an inverse relationship between age and PA [11–13] and a positive association with being male [11,12], while other findings regarding sex are inconclusive [13–15]. In populations with CVDs, an inverse relationship of age and PA was reported in systematic reviews, where PA was assessed both through self-report and device-based measures in patients with heart failure [16] and post-stroke [17]. Socioeconomic factors are also investigated as correlates of PA, including education and income, which are both determining factors for individuals regarding the capacity and opportunity to make informed health decisions [18]. In adults from the general population, the majority of studies report a positive association between socioeconomic factors and predominantly self-reported PA [9,10]. However, Bauman and colleagues [9] point out that this positive association primarily applies to low- and middle-income countries, whereas it is inconsistent or even inverse in high-income countries. An American study with nearly 17,000 self-identified post-stroke patients analyzing the relationship between self-reported PA and socioeconomic factors, found that higher education was associated with a higher PA level [19].

Furthermore, PA is related to lifestyle factors such as smoking, alcohol consumption and dietary habits. A systematic review including mixed populations found that a “healthy lifestyle” cluster commonly includes being physically active, alongside non-smoking and adherence to dietary and alcohol guidelines [20]. These lifestyle factors are key modifiable risk factors in the secondary prevention of CVDs, as their improvement can considerably reduce adverse health outcomes in people with CVDs [21]. Another important factor associated with PA is multimorbidity, which is the co-occurrence of several, often chronic diseases [22]. Multimorbidity is likely to occur in people with CVDs [23] and affects both mental and especially physical health [24]. When chronic diseases occur alongside a CVD, they are also referred to as comorbidities [23]. Recent cross-sectional studies suggest that having multiple chronic diseases is associated with less time spent in PA [25,26].

PA is associated with multiple factors that have predominantly been explored in general or mixed populations. Research focusing on individuals with CVDs is limited, often constrained by small sample sizes or reliance on self-reported PA measures. A better understanding of factors associated with PA within CVD populations is essential for identifying distinguishing characteristics between individuals with low and high PA levels. Such knowledge can inform the development of more targeted interventions to promote PA and hence support secondary prevention in people with CVDs. Therefore, this study aims to examine the associations between accelerometer-measured PA and demographic, socioeconomic, lifestyle factors, as well as comorbidity in individuals diagnosed with CVDs.

Materials and methods

Results

The final sample included 1,484 participants (32% women) with a median age of 60.2 years, and their characteristics are presented in Table 1, as well as the characteristics for each PA tertile. Across the whole sample, the median time spent in MVPA per day was 45 minutes. The majority completed secondary school (48%) and had no financial strain (85%). Around half of the participants were either regular/occasional smokers (12%) or former smokers (42%), 8% of participants had harmful alcohol consumption, and around half of the participants had an unhealthy diet (51%). The majority had no comorbidity (71%). The median time between the diagnosis and enrolling in SCAPIS was 5.8 years.

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Characteristics of the whole study sample and the PA tertiles showing the frequencies and percentages (in parentheses), except for moderate-to-vigorous physical activity; Missing values were present for education, financial strain, smoking status and alcohol consumption.

https://doi.org/10.1371/journal.pone.0352673.t001

The associations between PA levels and demographic, socioeconomic, lifestyle factors and comorbidity are shown in Table 2. Being in the oldest age group was significantly associated with lower odds of having a higher PA level compared to the youngest age group (OR = 0.59; 95% CI [0.44, 0.77]). Being female was significantly associated with lower odds of having a higher PA level compared to being male (OR = 0.73; 95% CI [0.59, 0.91]). Being a regular/occasional smoker was significantly associated with lower odds of having a higher PA level compared to non-smokers (OR = 0.37; 95% CI [0.26, 0.52]). Having a healthy diet was significantly associated with higher odds of having a higher PA level compared to having an unhealthy diet (OR = 1.84; 95% CI [1.46, 2.31]). Having one comorbidity (OR = 0.61; 95% CI [0.48, 0.77]) or two or more comorbidities (OR = 0.45; 95% CI [0.30, 0.68]) was significantly associated with lower odds of being in a medium or high PA category compared to having no comorbidities. Financial strain, level of education and alcohol consumption had no significant association with the PA level.

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Odds ratios from the multiple ordinal logistic regression with physical activity level (low, medium, high) as the dependent variable. The odds ratios indicate the likelihood of being in a higher physical activity tertile compared with a lower tertile for each independent variable relative to its reference category (Ref). (Results are pooled across five imputations).

https://doi.org/10.1371/journal.pone.0352673.t002

Based on the statistically significant results of the multiple ordinal logistic model, the predictive probabilities showed that participants in the oldest age group had a 44.9% probability of having a low PA level and a 24.4% probability of having a high PA level. Women were more likely to have a low (42.3%) rather than a high (26.5%) PA level. Regular/occasional smokers were more likely to have a low (52.4%) than a high (18.6%) PA level. Participants with a healthy diet were more likely to have a high PA level (36.1%) compared with a low PA level (32%). Participants with one comorbidity were more likely to have a low (40.2%) rather than a high (27.8%) PA level, and those with two or more comorbidities were more likely to have a low (47.0%) than a high (22.6%) PA level.

Discussion

This cross-sectional study investigated the associations between accelerometer-measured PA and demographic, socioeconomic, lifestyle factors and comorbidity in individuals diagnosed with CVDs from a large Swedish cohort. Lower PA levels were significantly associated with older age, female sex, regular or occasional smoking, and having one or more comorbidities. Conversely, a healthy diet was significantly associated with higher PA levels.

The most notable findings of our study are that older age (≥ 60 years), female sex, regular or occasional smoking, and having two or more comorbidities were among the factors associated with the highest probabilities of a low PA level in people with CVDs. This indicates that demographic, lifestyle factors and comorbidities are strongly associated with PA levels in this population. The strong negative association with smoking and the positive association with a healthy diet highlight the clustering of lifestyle behaviors, where individuals engaging in one healthy behavior are more likely to engage in others [20]. Interestingly, harmful alcohol consumption showed no association with PA. This lack of association may partly reflect the small proportion of participants reporting harmful alcohol use (8%). Additionally, the use of a modified AUDIT version, compared to the original instrument, may have increased the likelihood of higher scores, as only two response options were available for the last two items, potentially leading to misclassification and an attenuated association with PA. The relationship between comorbidities and PA in our study is in line with results from previous studies, including the whole SCAPIS cohort [25,26], as well as studies including participants with coronary artery disease [41,42]. Stewart and colleagues report that study participants with poorer general health, cardiac and non-cardiac comorbidities were likely to be less physically active [42]. However, in Huang and colleagues’ study [41], the negative association between comorbidities and PA only applied to men. The negative association that we observed between PA and age confirms findings from studies including adults from the general population [9,10] as well as people with CVDs [16,17]. Age has a changing association with PA across the lifespan, whereby PA levels may fluctuate during middle adulthood before becoming lower in older age [43].

In this study, female sex was significantly associated with lower PA levels. This aligns with findings from general population studies, where women report lower PA compared to men [11,12]. Findings in CVD populations are inconsistent, with no significant associations between sex and PA [16] or weak correlations [17]. Viewing the whole SCAPIS sample, Ekblom-Bak and colleagues found that men engage in more vigorous PA than women, whereas women engage in more light PA [26], suggesting that the intensity of PA plays a role when comparing between sexes.

The level of education and financial strain showed no association with PA. However, the association between completion of secondary school and PA was close to the threshold for statistical significance, which should be interpreted with caution. The limited socioeconomic variability in the sample, where the majority had completed at least secondary school (83%) and reported no financial strain (85%), as well as the elevated VIF for education, may have affected the precision of the estimates. Arora and colleagues found that higher education was associated with a higher PA level in participants with stroke [19], which was based on bivariate analyses and did not account for confounding factors. Our multivariable approach suggests that education may contribute to the PA level, but its role appears limited when other factors are adjusted for.

A strength of this study is the use of data from the large population cohort SCAPIS, which includes accelerometer-measured PA from 27,890 participants [26]. Additionally, SCAPIS data are linked to national registry data, which enables the retrieval of the CVD diagnoses. Compared with the general Swedish population, SCAPIS participants show a significant but small difference in education (higher education in the SCAPIS cohort) and proportion of never-smokers (lower in the SCAPIS cohort) [26], and the alcohol consumption appears to be higher in the SCAPIS cohort [44]. Using multiple ordinal logistic regression enabled us to investigate associations between several factors and PA, providing a more realistic picture compared to bivariate analyses. MI served as an adequate method to retain observations with valid accelerometer data and missing values in the independent variables. Measuring PA using accelerometers reduces recall and response bias, compared to self-reported measures [8]. Hip-worn accelerometers can measure total body movement [45], but are limited in capturing activities, such as cycling, water-based and upper-body activities [46]. We used MVPA as an overall measure of PA and divided participants into tertiles to distinguish between PA levels. A limitation of this approach is that individuals near the tertile cut points may fall into different groups despite having similar MVPA values, reducing variability and statistical power. Nevertheless, we consider this approach practical as it facilitates interpretation and comparison across PA levels. We are aware that sedentary behavior is an important movement behavior alongside PA. However, the focus of this study was PA as a single movement behavior and its relationship with other relevant variables. Future research could apply compositional analysis, considering a combined distribution of sedentary behavior and PA, to provide a more comprehensive understanding of movement behaviors in CVD populations. Importantly, the associations in this study are correlational and not to be interpreted as causal.

Conclusions

This study addressed the gap in PA correlates in individuals with CVDs by using accelerometer-measured PA from a large Swedish cohort. Lower PA was significantly associated with older age, female sex, regular or occasional smokers, and comorbidities, while a healthy diet was significantly associated with higher PA levels. Financial strain, level of education and alcohol consumption showed no association. These findings highlight characteristics linked to PA levels, which may inform the design of future research aimed at CVD populations with low activity profiles in need of PA promotion. Further research should investigate potential mediating and moderating pathways to clarify how these factors interact and influence PA levels.

Supporting information

Acknowledgments

We would like to acknowledge the help of Biobank Sweden and the local biobank facilities for their services in handling of biological samples and biobanking.