Correlates of physical activity behaviour in a population sample with increased cardiovascular risk and established cardiovascular disease: Cross-sectional analysis of data from the Paracelsus 10,000 prospective cohort study in Salzburg, Austria

Stefan T. Kulnik; Anna E. Carrozzo; Devender Kumar; Melanie Roth; Jens Blechert; Rik Crutzen; Patrick Langthaler; Martin Pühringer; Ludmilla Kedenko; Arne C. Bathke; Bernhard Iglseder; Bernhard Paulweber; Eugen Trinka

doi:10.1371/journal.pone.0356607

Abstract

Regular physical activity (PA) remains challenging for many people with high cardiovascular risk and established cardiovascular disease (CVD). We aimed to study correlates of PA behaviour in a population sample with increased cardiovascular risk and established CVD in Austria. We analysed cross-sectional data from the Paracelsus 10,000 study, a population-based cohort study of 40–70-year-olds in the city of Salzburg, Austria, and its surroundings. In a sub-sample of participants with increased cardiovascular risk and established CVD, we examined the associations between 29 correlates and leisure-time PA (weekly hours of walking, cycling and doing sports) as the primary PA outcome, as well as four secondary PA outcomes. We used stepwise multiple regression and machine learning models (decision trees, support vector machines, k-nearest neighbours, random forest and gradient boosting) to examine these associations. We analysed data from 1,173 participants with mean age 60.5 years (range 42–72 years). Nineteen percent of participants were female. Across the different PA outcomes and statistical analysis models, retirement, more PA during earlier life decades, and adherence to a Mediterranean diet were positively associated with PA behaviour. Greater comorbidity and depression were inversely associated with PA. In stepwise multiple regression, adjusted effect estimates on the primary PA outcome were 1.5 (SE 0.7, p = 0.0310) for retirement, 4.1 (SE 3.1, p < 0.0001) for past PA behaviour, 0.5 (SE 0.2, p = 0.0027) for Mediterranean diet, and −0.2 (SE 0.1, p = 0.0180) for depression. These findings contribute to a better understanding of the correlates of PA behaviour in populations at increased cardiovascular risk and with established CVD and may inform approaches for selective and personalised PA promotion strategies in cardiovascular care.

Citation: Kulnik ST, Carrozzo AE, Kumar D, Roth M, Blechert J, Crutzen R, et al. (2026) Correlates of physical activity behaviour in a population sample with increased cardiovascular risk and established cardiovascular disease: Cross-sectional analysis of data from the Paracelsus 10,000 prospective cohort study in Salzburg, Austria. PLoS One 21(8): e0356607. https://doi.org/10.1371/journal.pone.0356607

Editor: Seyed Aria Nejadghaderi, Kerman University of Medical Sciences, IRAN, ISLAMIC REPUBLIC OF

Received: March 25, 2026; Accepted: August 5, 2026; Published: August 19, 2026

Copyright: © 2026 Kulnik et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: The data underlying this article were provided by the Paracelsus 10,000 consortium by permission. The data set includes only those variables that were examined in this study (correlates and outcome variables) and only for the sub-group of participants with high or very high cardiovascular risk or established diagnosis of cardiovascular disease. Data will be shared on request to the corresponding author with permission of the Paracelsus 10,000 consortium. Access to the data may also be requested from the Paracelsus 10,000 consortium directly: https://salk.at/uniklinikum-salzburg/paracelsus-10000.

Funding: The author(s) received no specific funding for this work.

Competing interests: The authors have declared that no competing interests exist.

Introduction

Cardiovascular diseases (CVDs) are the leading cause of death, with an estimated 18.6 million deaths worldwide attributed to CVDs in 2019 [1]. CVDs include coronary heart disease, cerebrovascular disease, peripheral artery disease and other conditions [2]. Lack of physical activity is a critical contributing risk factor for CVDs, and regular physical activity constitutes a crucial component in the primary and secondary prevention of CVDs [3]. Despite the promotion of physical activity in both the primary and secondary prevention of CVDs, including structured and supervised cardiac rehabilitation programmes, physical activity levels remain low [4]. Therefore, it is crucial to understand critical factors that are related to and might influence physical activity behaviour, particularly in groups with increased cardiovascular risk and established diagnoses of CVD. This understanding is a crucial first step towards accounting for these factors in the development and delivery of tailored physical activity promotion strategies [5,6]. Large scale cohort studies are needed to identify and accurately estimate the associations between such factors and physical activity.

Paracelsus 10,000 (P10) is a prospective cohort study conducted in the city of Salzburg, Austria, and its surroundings [7]. Between 2013 and 2020, the study recruited a randomly selected population-based sample of 10,044 participants (5,176 women) in the age group from 40 to 70 years. The focus of P10 lies on the systematic monitoring of common non-communicable diseases and their risk factors. To this end, a large set of cross-sectional data was collected from participants, including medical history, medical examinations, biological samples, and various survey instruments [7]. This rich dataset offers an opportunity to investigate multitudes of research questions, including aspects related to exercise physiology and physical activity [8–10].

Factors that are related to and might influence physical activity behaviour in the general adult population have been well-studied and include health status, past physical activity behaviour, self-efficacy, intention to exercise, age, sex, education level, ethnic origin, body weight, social support, aspects of the physical environment and others [6,11–13]. However, these factors might differ in individuals with increased cardiovascular risk or established CVD [5]. Several systematic reviews have focused specifically on this population, yet data from the Austrian context is lacking in this literature [14]. The P10 dataset offers an opportunity to explore this research question in a cardiovascular population sample local to Austria. We therefore conducted an analysis among P10 study participants with increased cardiovascular risk (primary prevention sub-group) and with an established diagnosis of CVD (secondary prevention sub-group). We aimed to describe the associations between participant characteristics (including socio-demographic, clinical, psychological/cognitive, lifestyle, and environmental factors) as independent variables and physical activity behaviour as dependent variable, and to examine how these participant characteristics relate to physical activity behaviour. Insight into such factors can inform the tailoring and personalisation of physical activity programmes in the primary and secondary prevention of CVD [6].

Materials and methods

Details of the P10 study design have been published elsewhere [7]. In brief, 60,000 people aged 40–70 years and living in the city of Salzburg and its surroundings were randomly identified from the population register and invited by letter to participate in the study. Recruitment was stratified, aiming to include an equal number of men and women and 25%, 50% and 25% of participants in age groups 40–49 years, 50–59 years, and 60–70 years, respectively. Recruitment took place from 01 April 2013 until 31 March 2020, at which point 10,044 participants had been included in the study. Each study participant completed data collection in a single study visit on site at the University Hospital Salzburg in the presence of trained study personnel. Medical history, family history and medication were assessed by audio-recorded face-to-face interview. Study personnel specifically questioned participants whether each reported medical condition had been formally diagnosed by a medical doctor. All self-report questionnaires were completed by participants on a computer in the presence of study personnel who assisted as needed. Since autumn 2020, all study participants have been invited to return for their follow-up visit seven years after their initial study visit [7]. While the P10 study is a prospective cohort study, the data analysed for this article represents a cross-sectional dataset collected at the initial study visit.

The study received ethical approval from the research ethics committee of the County of Salzburg (reference 415-E/1521/6–2012). All study participants provided written informed consent.

For this analysis, we developed a prospective statistical analysis plan [15] and prospectively registered the analysis on the Open Science Framework (OSF) platform [16]. The purpose of this was to a priori define the research questions and corresponding statistical procedures, to increase methodological rigour and minimise the risk of selective reporting [17]. The reporting of this study follows the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement [18].

Study population

We included P10 study participants with increased cardiovascular risk (primary prevention sub-group) and with an established diagnosis of CVD (secondary prevention sub-group). The primary prevention sub-group was selected according to the SCORE2 (for age groups 40–69 years) and SCORE2-OP (for age groups 70 + years) CVD risk categorisation, which describes the 10-year risk of suffering a fatal (death due to coronary heart disease, heart failure and sudden death) or non-fatal cardiovascular event (non-fatal myocardial infarction and non-fatal stroke) [3,19,20]. The SCORE2 and SCORE2-OP risk models are based on country of residence, age, sex, systolic blood pressure, blood cholesterol (high density lipoprotein and total cholesterol) and current smoking [19,20]. For the Austrian geographic region, the high risk SCORE2/SCORE2-OP category (orange) indicates risks of 2.5% to <7.5%, 5% to <10% and 7.5% to <15% for the age groups <50 years, 50–69 years and 70 + years, respectively. The very high risk SCORE2/SCORE2-OP category (red) indicates risks of ≥7.5%, ≥ 10% and ≥15% for the age groups <50 years, 50–69 years and 70 + years, respectively [3,19,20]. All P10 study participants without established diagnosis of CVD who were in the high or very high risk SCORE2/SCORE2-OP categories were included in the primary prevention sub-group.

For the secondary prevention sub-group we included all P10 study participants who indicated one or more of the following diagnoses in the interviewer-administered medical history questionnaire (ICD-10 codes [21] in brackets): coronary heart disease (I25.1), chronic heart failure (I50), peripheral arterial disease (I73.9), abdominal aortic aneurysm (I73.1, I73.4), atrial fibrillation (I48) and stroke (I64). Out of altogether 10,044 study participants in the P10 dataset, there were 3,625 in the primary prevention sub-group and 612 in the secondary prevention sub-group.

Variables

Terminology.

In accordance with Bauman et al. [22], we acknowledge that our analysis of variables that might be related to and might influence physical activity behaviour is guided by an underlying logic of causal relationships; but that such conclusions cannot be drawn from cross-sectional observational data. We therefore use the term “correlates” to describe the variables under investigation, emphasising that these analyses present statistical associations/correlations, and not necessarily evidence of causality.

Correlates of physical activity behaviour.

From the variables available within the P10 dataset, we selected 29 to be examined as correlates of physical activity behaviour. These variables were selected because of their likely or potential relationship with individuals’ physical activity behaviour, based on published empirical research or theory [6,11–13,23]. Included were socio-demographic variables (age, sex, education level, occupation, migration background, religion, household income, marital status, number of children and retirement), clinical variables (CVD risk, comorbidity, body mass index, knee pain, hip pain and low back pain), psychological and cognitive variables (depression, daytime sleepiness and impulsivity), lifestyle variables (smoking status, alcohol use, past physical activity behaviour and Mediterranean diet) and environmental variables (number of people in the household, type of housing, size of living quarters, population density, environmental noise and air pollution).

Socio-demographic data were collected using a self-report questionnaire. CVD risk was defined as described above according to the SCORE2/SCORE2-OP risk categories and established diagnosis of CVD. In order from least to most severe CVD risk, the categories are: primary prevention sub-group with high risk, primary prevention sub-group with very high risk, and secondary prevention sub-group. Comorbidity was categorised in 0, 1–2 or ≥3 of the following diagnoses: type 2 diabetes, type 1 diabetes, coronary heart disease, chronic heart failure, peripheral arterial disease, rheumatoid arthritis, cirrhosis of the liver, chronic hepatitis, gastric or duodenal ulcer, chronic obstructive pulmonary disease, stroke, Alzheimer’s disease or other type of dementia, chronic kidney failure, pancreatic cancer, bladder cancer, lung cancer, breast cancer, brain tumour and ovarian cancer. Body mass index was calculated from clinician-measured anthropometric measurements (weight, height). Knee, hip and low back pain were measured using the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) [24] and the Oswestry Disability Index (ODI) [25]. Depression, daytime sleepiness and impulsivity were measured with the Beck Depression Inventory (BDI) [26], the Epworth Sleepiness Scale (ESS) [27] and the Barratt Impulsiveness Scale (BIS-15) [28,29]. Smoking status was recorded from self-report. Alcohol use was measured using the Alcohol Use Disorder Identification Test (AUDIT) [30]. Past physical activity behaviour was recorded from self-reported average hours per week of doing sports at ages 5–19, 20–34, 35–49, and 50 + years. Adherence to a Mediterranean diet was measured using the Mediterranean Diet Adherence Screener (MEDAS) [31]. These standard questionnaires are considered valid and reliable for the general adult population. Environmental data were collected using a self-report questionnaire, including one Likert scale question each on environmental noise (street, railway and air traffic, industrial noise, neighbours) and air pollution (street and air traffic, residential burners, industrial pollution).

Outcome variables.

The outcome of interest was physical activity behaviour, measured by self-report using a questionnaire adapted from the European Prospective Investigation into Cancer and Nutrition (EPIC) programme [32]. This questionnaire has been validated in large European cohorts [32,33] and captures not only hours per week spent walking, cycling and doing sports, but also gardening, doing handiwork/repairs and/or housework in the past year. Of note, physical activity intensity was not captured, i.e., these data represent the sum of physical activity at light, moderate and vigorous intensity levels. From this, we summed hours per week spent walking, cycling and doing sports as our primary outcome variable, based on the predictive value of leisure-time physical activity on health outcomes [34]. Additionally, we explored the secondary outcome variables Cambridge Activity Index (CAI), derived from hours per week spent cycling and doing sports plus occupational physical activity, and Recreational Activity Index (RAI), derived from hours per week spent walking, cycling and doing sports. Both indices are ordinal variables with four activity levels (active, moderately active, moderately inactive, inactive) [33]. We further explored a physical activity outcome composed of hours per week spent gardening, doing handiworks/repairs and/or housework, which represent chores and maintenance-type physical activities in and around the home, and total physical activity (hours per week of all six physical activity behaviours).

Data cleaning

Statistical analysis included only those study participants with complete datasets for all correlates and outcome variables described above. This resulted in a sample of 1,173 participants (1,018 in the primary prevention sub-group and 155 in the secondary prevention sub-group). We decided not to impute missing data because of complex patterns of missingness, and imputing a large number of correlates in this exploratory context was unlikely to provide clear statistical advantages. Moreover, imputation could also have introduced artefacts affecting the association estimates and the descriptive summaries of the analytic sample.

Statistical analysis

The statistical analysis approach included bivariate analysis, interpretable parametric regression models (i.e., multiple linear regression, ordinal logistic regression) and flexible non-parametric machine learning-based analyses for all physical activity outcome variables. Importantly, these three approaches answer distinct research questions and evaluate different aspects of the data, rather than serving as equivalent evidence for a single conclusion. Bivariate analyses serve to describe crude associations between correlates and physical activity. The parametric linear and ordinal logistic regression models enable quantification and interpretation of adjusted associations. The flexible non-parametric machine learning models evaluate global predictive performance and identify feature importance, capturing complex non-linear interactions without assuming a rigid functional form. Overall, the combination of these approaches leverages unique methodological strengths to analyse the data from descriptive, inferential and predictive perspectives. Correlates selected by the regression models are evaluated alongside the predictive feature importances from the machine learning models to provide complementary insights into the underlying patterns of physical activity behaviour.

Bivariate analysis.

Bivariate analyses were conducted for all correlates of physical activity behaviour by calculating the appropriate association statistic (Pearson’s or Spearman’s correlation coefficient or Cohen’s d) with 95% confidence interval. Pearson’s correlation coefficient was used for approximately linear associations between continuous variables with approximately symmetric distributions, whereas Spearman’s rank correlation coefficient was used for ordinal variables. Cohen’s d was used to quantify differences between binary categorical variables and the physical activity outcomes.

Parametric regression analysis.

To study the relationship between correlates (independent variables, ) and the continuous physical activity outcomes (dependent, or target variable, ) we used a multiple linear regression model. The analyses were conducted using the stats R package v4.2.3 in R statistical software v4.2.3 [35]. The linear model is defined by , with a continuous outcome variable and a normally distributed error. are the regression coefficients and represent the expected change in outcome if changes by one unit and all other variables are held constant. However, the value of changes if the set of independent variables in the model changes and is correlated with other independent variables. Thus, the variable selection is of high relevance. To iteratively examine the statistical significance of each independent variable in the regression model, we considered a stepwise regression with bidirectional variable selection based on Akaike information criterion [36].

To study the ordinal physical activity outcomes, we used an ordinal logistic regression model (proportional odds model). The model was estimated using the polr() function from the MASS package in R. This approach assumes that the relationship between each predictor and the cumulative odds of being in a higher category of physical activity is constant across thresholds. Coefficients represent the log-odds of being in a higher activity category associated with one-unit increases (or changes) in the covariates, holding all other variables constant. Variance inflation factors were examined for the full regression models (before stepwise selection) and were consistently below 2, suggesting that multicollinearity was not a concern.

Non-parametric machine learning analysis.

To prepare the data for flexible prediction-oriented machine learning models, the categorical variables were converted into numerical values using a one-hot encoding technique via Python Pandas library. One-hot encoding represents categorical variables as binary vectors, where each category is converted into a new column, and a binary value (0 or 1) indicates the presence or absence of that category in the original data. Thereafter, the dataset was split into a train and test set (80/20 ratio using the scikit-learn library) [37] for building the machine learning models. In the modelling phase, we employed a set of traditional machine learning regression and classification (for CAI and RAI) models from the scikit-learn library [37], including decision trees (DecisionTreeRegressor/Classifier), support vector machines (SVM), k-nearest neighbours (KNeighborsRegressor/Classifier), random forest (RandomForestRegressor/Classifier) and gradient boosting (XGBRegressor/Classifier). The decision to use traditional machine learning models for exploration, as opposed to advanced deep learning approaches, was made given the relatively small size of the dataset. For the regression tasks, model performance was compared in terms of mean absolute error (MAE) and root mean squared error (RMSE) on the respective physical activity outcome variable. For the classification tasks for CAI and RAI, 5-fold cross-validation micro-averaged area under the receiver operating characteristic curve (AUC) scores using the One-vs-Rest (OvR) strategy were compared. The GridSearchCV method from scikit-learn [37] was used to find the optimal hyper-parameter for the aforementioned models. All models were finally evaluated on the test data set. We also used the Shapley Additive exPlanations (SHAP) framework [38] for understanding the importance of features (independent variables) within the best-performing model. The SHAP value offers a framework for interpreting the output of machine learning models and understanding the impact of each feature on the model's performance. Importantly, SHAP values should be interpreted as contributions to model prediction, not as evidence of causal or clinical importance.

Results

Fig 1 describes the flow of participants through the study. Participants included in the analysis (n = 1,173) had a mean age of 60.5 years (SD 6.3, range 42–72), and 19% were female. Participants in the secondary prevention sub-group (n = 155) had diagnoses of coronary heart disease (39%), stroke (31%), atrial fibrillation (28%), chronic heart failure (7%), peripheral arterial disease (5%) and abdominal aortic aneurysm (3%). In the primary prevention group (n = 1,018), 82% of participants had high CVD risk (orange SCORE2/SCORE2-OP risk category) and 18% had very high CVD risk (red SCORE2/SCORE2-OP risk category). The primary physical activity outcome for the entire sample amounted to a mean of 13.9 (SD 11.6) weekly hours of walking, cycling and doing sports. Participant characteristics are summarised in Table 1.

Download:

Table 1. Participant characteristics of the entire sample and according to CVD risk group.

https://doi.org/10.1371/journal.pone.0356607.t001

Download:

Fig 1. Participant flow.

https://doi.org/10.1371/journal.pone.0356607.g001

Primary physical activity outcome

The bivariate analysis for the primary physical activity outcome is presented in Table 2. In multiple regression analysis (Table 3), more past physical activity, being in retirement, adhering to the Mediterranean diet and increased air pollution showed statistically significant positive associations with physical activity behaviour. Statistically significant factors associated negatively with physical activity behaviour were a higher number of people living in the household, higher monthly household income, low mood/depression and lower occupancy housing. The factors included in the regression model explain approximately 15% of the variation in the outcome (adjusted R-squared 0.145). The residuals plot showed random distribution of residuals with no discernible pattern, providing no indication against adequate model fit.

Download:

Table 2. Bivariate association statistics for the primary physical activity outcome (weekly hours of walking, cycling and doing sports).

https://doi.org/10.1371/journal.pone.0356607.t002

Download:

Table 3. Multiple linear stepwise regression for the primary physical activity outcome (weekly hours of walking, cycling and doing sports).

https://doi.org/10.1371/journal.pone.0356607.t003

Performance indicators (MAE and RMSE) for the five machine learning models are shown in Table 4. These indicate similar performance of all five models, with XGBRegressor showing slightly superior performance (lowest MAE and RMSE comparable to the RandomForestRegressor). We therefore present the feature importance analysis for the XGBRegressor model (Fig 2). The Shapley Additive exPlanations (SHAP) plot (Fig 2a) shows how the top 15 features (correlates) contribute to the model performance. The feature value is indicated by the colour spectrum blue (lower value) to red (higher value). The direction the feature impacts on the outcome is indicated on the x-axis (positive or negative association). The features are ranked in order of importance from top to bottom (average impact of individual features on the model output magnitude, Fig 2b).

Download:

Table 4. Performance of five machine learning models in prediction of the primary physical activity outcome (weekly hours of walking, cycling and doing sports).

https://doi.org/10.1371/journal.pone.0356607.t004

Download:

Fig 2. Shapley Additive exPlanations (SHAP) plot of the XGBRegressor model.

Shown are the top 15 features contributing to the model prediction and their relative contribution. Fig 2a shows the direction of the features’ (independent variables’) impact on the model output, e.g., higher past physical activity (red data points) predominantly contributing to higher physical activity (positive portion of the x-axis). Fig 2b shows the average impact of individual features on the model output magnitude relative to each other.

https://doi.org/10.1371/journal.pone.0356607.g002

In machine learning analysis, past physical activity behaviour, Mediterranean diet, age, white collar occupation and retirement are the five positively associated features that contribute most to the model. Among the top 15 features of the XGBRegressor model, past physical activity, Body Mass Index, Mediterranean diet, depression, occupation, retirement, air pollution and income are also represented in the regression model, whereas size of living quarters, age, daytime sleepiness, impulsivity, knee pain, alcohol use and low back pain are included in the machine learning model, but not in the regression model.

Secondary physical activity outcomes

Detailed results of the regression and machine learning analyses for the secondary physical activity outcomes are given in supporting information S1 File, whereby the machine learning analyses largely align with the results from the multiple regression models. Comparing the primary physical activity outcome against CAI and RAI, the factors past physical activity behaviour, Mediterranean diet and low mood/depression are retained in the multiple regression models for both CAI and RAI. Higher income is corroborated in the model for CAI, whereas lower occupancy housing is corroborated in the model for RAI. Retirement is corroborated in the model for RAI but not retained in the model for CAI. Additional statistically significant factors in these models are current smoking (associated with less physical activity), greater comorbidity (associated with less physical activity for RAI), higher body mass index (associated with less physical activity for CAI) and self-employed occupation (associated with less physical activity for CAI).

In the comparison of the primary physical activity outcome (representing more leisure-type physical activities, or activities outside the home) against gardening, doing handiwork/repairs and housework (representing chores and maintenance-type physical activities in and around the home) and against total physical activity, several factors are retained (retirement, Mediterranean diet, low mood/depression, income, housing), but two differences are noteworthy: past physical activity is not retained in the multiple regression model for gardening, doing handiwork/repairs and housework, but remains in the model for total physical activity; and female sex is included in both models as a factor that is associated with more physical activity.

Discussion

This is the first study to analyse correlates of physical activity behaviour in a sample with increased cardiovascular risk and established CVD in Austria. Although there are some differences between the correlates included in the multiple regression analyses and in the machine learning models, both analytic approaches describe an overall coherent picture of these factors’ associations with physical activity behaviour. The main message derived from the combined analyses is that, in this age group – spanning the second half of working age to early retirement, i.e., 40–70 years – those who were retired, those who had been more physically active during earlier life decades, and those who adhered more to a Mediterranean diet were more physically active. Additional correlates were retained in some but not all the models, such as the inverse association of comorbidity and depression with physical activity, which nevertheless describe coherent trends. The inverse association between comorbidity and physical activity, for example, is well-documented in the literature for general population samples (e.g., [39]) and CVD groups (e.g., [40]) alike. Sex was not associated with the primary physical activity outcome (walking, cycling and doing sports), but female sex was associated with more physical activity for both total physical activity and gardening, doing handiwork/repairs and housework, whereby the association in total physical activity was driven by the difference between sexes in gardening, doing handiwork/repairs and housework.

When interpreting these findings, it is important to bear in mind the limitations inherent to a cross-sectional dataset. These findings represent correlations, not determinants, of physical activity behaviour and serve an exploratory or hypothesis-generating purpose. Longitudinal or interventional evidence is required to demonstrate causal relationships. Moreover, in a cross-sectional dataset the possibility of reverse causation needs to be considered, particularly for clinical variables such as body mass index, pain, comorbidity and depression, which may both influence and be influenced by physical activity.

Retirement

Our main findings largely align with other literature on correlates of physical activity behaviour. Several studies have documented that the life transition to retirement often leads to a more physically active lifestyle in the general population. A recent systematic review [41] included 36 longitudinal studies from Australia, Europe, Japan and the United States, showing a consistent increase in leisure time physical activity following retirement. The authors suggest that the absence of work activities provides retirees with an opportunity for leisure activities, including physical activity. Pooled longitudinal data [42] from 106,927 individuals across 35 countries in East Asia, Europe, North and Central America revealed that physical inactivity in retirement decreased on average by 3 percentage points (−0.03, 95% CI −0.05 to −0.01). However, studies also indicate a simultaneous increase in sedentary time, and a reduction in work- and transport-related physical activity [41]. Moreover, change in physical activity following retirement seems to be moderated by socio-demographic factors, whereby lower socio-economic positions and retirement from physically demanding occupations are associated with decrease in physical activity [41]. Contrary to this, our analysis gives some indication of an inverse relationship between socio-demographic markers (income, housing) and physical activity. Specific to CVD populations, analyses of data from 32,370 outpatients with stable coronary artery disease [40] resulted in an inverse association, with those in retirement versus full-time employment less likely to engage in vigorous physical activity (odds ratio 0.70, 95% CI 0.65 to 0.74). But this finding may not be directly comparable to our data, as our physical activity outcomes captured physical activity across all intensity levels.

The model for CAI partly contradicts the pictures presented in the other models, because it excludes retirement. This is explained by an assumption included in the formula for CAI [33]. CAI is derived from weekly hours of cycling and doing sports, plus factoring in occupational physical activity (sedentary, light, moderate or heavy manual labour). For retirees, the question about occupational physical activity does not apply and is therefore assumed to be sedentary in the formula for CAI. This assumption accounts for the elimination of the impact of retirement in the model for CAI and should be discarded.

Past physical activity behaviour

Our study corroborates positive associations between past and current physical activity behaviour that have also been observed by others. A cohort study of 712 healthy World War II veterans in the United States, for example, found a statistically significant correlation (Spearman’s rho 0.164, p < 0.01) between participation in high school sports and physical activity at age 70 + years [43]. Similarly, a general population cohort of 2,201 Australians showed some statistically significant weak correlations between childhood and adult physical activity [44]. In a cohort of 105 Swedes, physical activity during adolescence in addition to other physical performance characteristics explained 82% of the physical activity level in adulthood for women and 47% for men [45].

In CVD samples, several studies have demonstrated short-term associations between prior and current physical activity. In a study of 114 patients with myocardial infarction, physical activity approximately one week after the event correlated with physical activity 8 months after the event (Pearson’s r 0.34, 95% CI 0.17 to 0.49) [46]. In 200 patients with coronary artery disease, walking behaviour two weeks after percutaneous coronary intervention correlated with walking behaviour 6 months after the procedure (Pearson’s r 0.30, 95% CI 0.17 to 0.42) [47]. And a study of 801 patients with coronary heart disease reported a correlation in physical activity from 6 months to 12 months after hospitalisation (Pearson’s r 0.61, 95% CI 0.56 to 0.66) [48]. In contrast to these short-term correlations, previous physical activity in our study captured physical activity (specifically doing sports) across the preceding life decades, adding a long-term perspective to this literature in CVD populations.

Mediterranean diet

The emergence of Mediterranean diet as a consistently associated factor with physical activity in our analysis adds a novel finding to the literature. While the Mediterranean diet is recognised as one of the best dietary strategies for the prevention of chronic diseases and premature death [49], and there is evidence to support the synergistic effects of a Mediterranean diet combined with recommended levels of physical activity [49], we are not aware of reports of this association in a sample with increased cardiovascular risk or established CVD. In the general population, studies have found that more physical activity and healthier eating habits correlate to a certain extent, but individual heterogeneity can be considerable (e.g., [50]). In the context of our study, we suggest that both previous physical activity and Mediterranean diet could be interpreted as indicators of a person’s greater health awareness and health literacy, resulting in a more health-conscious lifestyle with higher levels of physical activity.

Gender

Contrary to our findings, female sex is most frequently reported to be negatively associated with physical activity, for example in an umbrella review [51] synthesising 11 reviews representing over 300 unique primary papers on physical activity in community-dwelling older adults. In CVD populations, female sex was an independent predictor of lower physical activity levels [40]. In a scoping review on physical activity post cardiac surgery [52] men were noted to be more physically active than women, generally walking at a faster pace and exerting more effort during exercise. Our contrary finding could be idiosyncratic to the geographic region the sample was recruited from. Austrian population statistics show that in the County of Salzburg the World Health Organisation’s physical activity recommendations are met by 26.8% and 13.9% of men in the age groups 45–59 and 60–74 years, respectively, compared to 28.5% and 25.4% of women in these respective age groups [53]. The strong association of female sex with more physical activity related to gardening, doing handiwork/repairs and housework might reflect traditional gender roles and cultural norms in this age group and geographic region of Austria.

In summary, our analysis indicates that the transition to retirement might offer an opportunity to enact a more physically active and heart-healthy lifestyle, which could be leveraged by targeted interventions for the primary and secondary prevention of CVD. Moreover, our findings might be interpreted to support the crucial importance of physical activity promotion, from early childhood and throughout the life course, as greater physical activity during earlier life decades likely facilitates the maintenance of physical activity during later life decades, also in individuals with CVD. With respect to the assessment of physical activity, our findings highlight the importance of capturing a broad range of physical activity behaviours, including gardening and household work, particularly for women, as our analysis shows that increased activities in these areas was associated with female sex. Lastly, the positive association between Mediterranean diet and physical activity presents a novel finding, and the relationship between dietary habits and physical activity could find further consideration in future research.

Strengths and limitations

The P10 study is a unique dataset in the Austrian geographic region, due to the recruitment of a large population sample and the collection of a broad range of measurements and self-report questionnaires. This can be regarded as a strength of our analysis, allowing us to examine correlates which are less commonly captured in studies of physical activity in CVD samples such as impulsiveness, daytime sleepiness, past physical activity behaviour, Mediterranean diet, air pollution, and environmental noise. Another strength of our analysis was the investigation through both regression and machine learning analyses. While regression models allow us to quantify and interpret associations, the inclusion of machine learning models provide an additional perspective by capturing potential non-linear relationships. Importantly, the overlap in key correlates identified by both approaches strengthens the robustness of our findings and increases confidence in the observed associations.

Several limitations to our study are acknowledged. Self-reported measures of physical activity are inherently subject to recall and social desirability bias. However, the EPIC physical activity questionnaire was validated in large European cohorts [32,33], and the volumes of physical activity reported in P10 are in line with device-measured physical activity reported in other studies of adults in the same age range [54]. In the inclusion criteria, we prioritised the interpretation of the relative importance of correlates and the comparison of different physical activity outcomes. We therefore included only those participants with complete datasets, excluding 72% of P10 participants with increased cardiovascular risk or established diagnosis of CVD. Nevertheless, the sample size was considered adequate for the planned multiple regression analyses relative to the number of candidate variables included in the regression models and is consistent with commonly cited heuristic sample size rules for multiple regression [55].

Participant characteristics of the excluded group are provided in supporting information S2 File. In comparison with participants included in this analysis, the excluded group has a higher proportion of women (31% versus 19%) and higher proportions of participants with the lowest education level (11% versus 6%) and the lowest income category (39% versus 25%). These differences are acknowledged as a potential source of selection bias. Of note, underrepresentation of participants with lower socio-economic status (International Standard Classification of Education levels 0–5 and blue-collar employment) was also observed for the entire P10 study sample in comparison with population statistics for the city of Salzburg [7].

We acknowledge that this analysis was reliant on the variables available within the P10 dataset. The dataset lacked a number of correlates of physical activity that are well-documented in the literature, including individual cognitive/psychological factors such as self-efficacy and intention [6,56], interpersonal factors such as social support and loneliness [51,57] and environmental factors such as the built environment [58]. We suggest that this accounts for the large proportion of the outcome variation that was not explained by our analysis models. In addition, we acknowledge that the bidirectional stepwise regression approach used for variable selection has recognised limitations, including potential model instability and sensitivity to correlations among candidate variables. Our aim, however, was not to develop an optimal prediction model but to identify a parsimonious set of correlates associated with physical activity in this exploratory analysis. Future studies could evaluate alternative penalised regression approaches, such as LASSO, ridge regression or elastic net, to assess the robustness of variable selection.

Conclusions

In this Austrian cohort of individuals with high cardiovascular risk or established diagnosis of CVD, retirement, more physical activity during earlier life decades and adherence to a Mediterranean diet were positively associated with physical activity behaviour, while greater comorbidity and depression were inversely associated with physical activity behaviour. These findings add to our understanding of correlates of physical activity behaviour in CVD populations and offer approaches for more selective prevention and intervention, e.g., leveraging the transition to retirement for physical activity promotion, or directing additional and tailored physical activity promotion offers towards those not adhering to a Mediterranean diet, those with a history of physical inactivity and those with greater comorbidity.

Supporting information

S1 File. Regression models and machine learning models for the secondary physical activity outcomes.

https://doi.org/10.1371/journal.pone.0356607.s001

(DOCX)

S2 File. Characteristics of Paracelsus 10.000 participants with increased cardiovascular risk and established cardiovascular disease: sample included in the analysis (n = 1173) versus excluded participants (n = 3064).

https://doi.org/10.1371/journal.pone.0356607.s002

(DOCX)

References

1. World Heart Observatory. World Heart Observatory. World Heart Observatory. https://world-heart-federation.org/world-heart-observatory/ 2024 January 2.
2. World Health Organization. Cardiovascular diseases (CVDs). https://www.who.int/news-room/fact-sheets/detail/cardiovascular-diseases-(cvds) 2022 December 20.
- View Article
- Google Scholar
3. Visseren FLJ, Mach F, Smulders YM, Carballo D, Koskinas KC, Bäck M, et al. 2021 ESC Guidelines on Cardiovascular Disease Prevention in Clinical Practice: Developed by the Task Force for Cardiovascular Disease Prevention in Clinical Practice with Representatives of the European Society of Cardiology and 12 Medical Societies with the Special Contribution of the European Association of Preventive Cardiology (EAPC). European Heart Journal. 2021;42:3227–337.
- View Article
- Google Scholar
4. Kotseva K, De Backer G, De Bacquer D, Rydén L, Hoes A, Grobbee D, et al. Lifestyle and impact on cardiovascular risk factor control in coronary patients across 27 countries: Results from the European Society of Cardiology ESC-EORP EUROASPIRE V registry. Eur J Prev Cardiol. 2019;26(8):824–35. pmid:30739508
- View Article
- PubMed/NCBI
- Google Scholar
5. Gutenberg J, Kulnik ST, Niebauer J, Crutzen R. Determinants of Physical Activity in the Cardiac Population: Protocol for a Systematic Review. JMIR Res Protoc. 2022;11(9):e39188. pmid:36069762
- View Article
- PubMed/NCBI
- Google Scholar
6. Bauman AE, Reis RS, Sallis JF, Wells JC, Loos RJF, Martin BW, et al. Correlates of physical activity: why are some people physically active and others not?. Lancet. 2012;380(9838):258–71. pmid:22818938
- View Article
- PubMed/NCBI
- Google Scholar
7. Frey V, Langthaler P, Raphaelis E, Ring-Dimitriou S, Kedenko L, Aigner E, et al. Paracelsus 10,000: An observational cohort study about the health status of the population of Salzburg, Austria. Rationale, objectives and study design. Paracelsus Proceedings of Experimental Medicine. 2023;2:1–17.
- View Article
- Google Scholar
8. Pühringer M, Ring-Dimitriou S, Stöggl T, Iglseder B, Paulweber B. Comparison of visual, automatic and semiautomatic methods to determine ventilatory indices in 50 to 60 years old adults. J Sports Sci. 2020;38(6):692–702. pmid:32089122
- View Article
- PubMed/NCBI
- Google Scholar
9. Pühringer M, Ring-Dimitriou S, Stöggl T, Iglseder B, Paulweber B. Differences in the point of optimal ventilatory efficiency and the anaerobic threshold in untrained adults aged 50 to 60 years. Respir Physiol Neurobiol. 2020;282:103516. pmid:32768502
- View Article
- PubMed/NCBI
- Google Scholar
10. Pühringer M, Ring-Dimitriou S, Iglseder B, Frey V, Trinka E, Paulweber B. Sequencing patterns of ventilatory indices in less trained adults. Front Sports Act Living. 2023;4:1066131. pmid:36755562
- View Article
- PubMed/NCBI
- Google Scholar
11. Choi J, Lee M, Lee J-K, Kang D, Choi J-Y. Correlates associated with participation in physical activity among adults: a systematic review of reviews and update. BMC Public Health. 2017;17(1):356. pmid:28438146
- View Article
- PubMed/NCBI
- Google Scholar
12. Carlin A, Perchoux C, Puggina A, Aleksovska K, Buck C, Burns C, et al. A life course examination of the physical environmental determinants of physical activity behaviour: A “Determinants of Diet and Physical Activity” (DEDIPAC) umbrella systematic literature review. PLoS One. 2017;12(8):e0182083. pmid:28787023
- View Article
- PubMed/NCBI
- Google Scholar
13. Condello G, Puggina A, Aleksovska K, Buck C, Burns C, Cardon G, et al. Behavioral determinants of physical activity across the life course: a “DEterminants of DIet and Physical ACtivity” (DEDIPAC) umbrella systematic literature review. Int J Behav Nutr Phys Act. 2017;14(1):58. pmid:28464958
- View Article
- PubMed/NCBI
- Google Scholar
14. Collado-Mateo D, Lavín-Pérez AM, Peñacoba C, Del Coso J, Leyton-Román M, Luque-Casado A, et al. Key Factors Associated with Adherence to Physical Exercise in Patients with Chronic Diseases and Older Adults: An Umbrella Review. Int J Environ Res Public Health. 2021;18(4):2023. pmid:33669679
- View Article
- PubMed/NCBI
- Google Scholar
15. Kulnik ST, Carrozzo E, Kumar D, Ring-Dimitriou S, Roth M, Blechert J. Correlates of physical activity behaviour in a cardiovascular population sample. Statistical analysis plan for a secondary analysis of data from the Paracelsus 10,000 prospective cohort study in Salzburg, Austria. Open Science Framework. 2023.
- View Article
- Google Scholar
16. Kulnik ST. Correlates of physical activity behaviour in a cardiovascular population sample. Open Science Framework (OSF) Registries. https://osf.io/j4h2z 2023. 2024 January 8.
17. Yuan I, Topjian AA, Kurth CD, Kirschen MP, Ward CG, Zhang B, et al. Guide to the statistical analysis plan. Paediatr Anaesth. 2019;29(3):237–42. pmid:30609103
- View Article
- PubMed/NCBI
- Google Scholar
18. von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP, et al. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. J Clin Epidemiol. 2008;61(4):344–9. pmid:18313558
- View Article
- PubMed/NCBI
- Google Scholar
19. SCORE2 working group, ESC Cardiovascular risk collaboration. SCORE2 risk prediction algorithms: new models to estimate 10-year risk of cardiovascular disease in Europe. European Heart Journal. 2021;42:2439–54.
- View Article
- Google Scholar
20. SCORE2-OP working group, ESC Cardiovascular risk collaboration. SCORE2-OP risk prediction algorithms: estimating incident cardiovascular event risk in older persons in four geographical risk regions. European Heart Journal. 2021;42:2455–67.
- View Article
- Google Scholar
21. ICD-10 Version: 2019. https://icd.who.int/browse10/2019/en 2023 January 25.
- View Article
- Google Scholar
22. Bauman AE, Sallis JF, Dzewaltowski DA, Owen N. Toward a better understanding of the influences on physical activity: the role of determinants, correlates, causal variables, mediators, moderators, and confounders. Am J Prev Med. 2002;23(2 Suppl):5–14. pmid:12133733
- View Article
- PubMed/NCBI
- Google Scholar
23. McGowan H, Gutenberg J, Mayr B, Sareban M, Niebauer J, Crutzen R. Determinants of physical activity behaviour in the cardiac population: a systematic review. Research Square. 2023.
- View Article
- Google Scholar
24. Bellamy N, Buchanan WW, Goldsmith CH, Campbell J, Stitt LW. Validation study of WOMAC: a health status instrument for measuring clinically important patient relevant outcomes to antirheumatic drug therapy in patients with osteoarthritis of the hip or knee. J Rheumatol. 1988;15(12):1833–40. pmid:3068365
- View Article
- PubMed/NCBI
- Google Scholar
25. Fairbank JC, Couper J, Davies JB, O’Brien JP. The Oswestry low back pain disability questionnaire. Physiotherapy. 1980;66(8):271–3. pmid:6450426
- View Article
- PubMed/NCBI
- Google Scholar
26. Beck AT, Ward CH, Mendelson M, Mock J, Erbaugh J. An inventory for measuring depression. Arch Gen Psychiatry. 1961;4:561–71. pmid:13688369
- View Article
- PubMed/NCBI
- Google Scholar
27. Johns MW. A new method for measuring daytime sleepiness: the Epworth sleepiness scale. Sleep. 1991;14(6):540–5. pmid:1798888
- View Article
- PubMed/NCBI
- Google Scholar
28. Patton JH, Stanford MS, Barratt ES. Factor structure of the Barratt impulsiveness scale. J Clin Psychol. 1995;51(6):768–74. pmid:8778124
- View Article
- PubMed/NCBI
- Google Scholar
29. Spinella M. Normative data and a short form of the Barratt Impulsiveness Scale. Int J Neurosci. 2007;117(3):359–68. pmid:17365120
- View Article
- PubMed/NCBI
- Google Scholar
30. Saunders JB, Aasland OG, Babor TF, de la Fuente JR, Grant M. Development of the Alcohol Use Disorders Identification Test (AUDIT): WHO Collaborative Project on Early Detection of Persons with Harmful Alcohol Consumption--II. Addiction. 1993;88:791–804.
- View Article
- Google Scholar
31. Schröder H, Fitó M, Estruch R, Martínez-González MA, Corella D, Salas-Salvadó J, et al. A short screener is valid for assessing Mediterranean diet adherence among older Spanish men and women. J Nutr. 2011;141(6):1140–5. pmid:21508208
- View Article
- PubMed/NCBI
- Google Scholar
32. Wareham NJ, Jakes RW, Rennie KL, Schuit J, Mitchell J, Hennings S, et al. Validity and repeatability of a simple index derived from the short physical activity questionnaire used in the European Prospective Investigation into Cancer and Nutrition (EPIC) study. Public Health Nutr. 2003;6(4):407–13. pmid:12795830
- View Article
- PubMed/NCBI
- Google Scholar
33. InterAct Consortium, Peters T, Brage S, Westgate K, Franks PW, Gradmark A, et al. Validity of a short questionnaire to assess physical activity in 10 European countries. Eur J Epidemiol. 2012;27(1):15–25. pmid:22089423
- View Article
- PubMed/NCBI
- Google Scholar
34. Bonekamp NE, Visseren FLJ, Ruigrok Y, Cramer MJM, de Borst GJ, May AM, et al. Leisure-time and occupational physical activity and health outcomes in cardiovascular disease. Heart. 2023;109(9):686–94. pmid:36270785
- View Article
- PubMed/NCBI
- Google Scholar
35. R Core Team. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. 2021.
36. Akaike H. Information Theory and an Extension of the Maximum Likelihood Principle. Selected Papers of Hirotugu Akaike. New York, NY: Springer; 1998. pp. 199–213. 10.1007/978-1-4612-1694-0_15
37. Pedregosa F, Varoquaux G, Gramfort A, Michel V, Thirion B, Grisel O. Scikit-learn: Machine Learning in Python. Journal of Machine Learning Research. 2011;12:2825–30.
- View Article
- Google Scholar
38. Lundberg SM, Lee SI. A unified approach to interpreting model predictions. Advances in Neural Information Processing Systems. Curran Associates, Inc. 2017.
39. Stojanovic M, Babulal GM, Head D. Determinants of physical activity engagement in older adults. J Behav Med. 2023;46(5):757–69. pmid:36920727
- View Article
- PubMed/NCBI
- Google Scholar
40. Biscaglia S, Campo G, Sorbets E, Ford I, Fox KM, Greenlaw N, et al. Relationship between physical activity and long-term outcomes in patients with stable coronary artery disease. Eur J Prev Cardiol. 2020;27(4):426–36. pmid:31558054
- View Article
- PubMed/NCBI
- Google Scholar
41. Xue B, Head J, McMunn A. The Impact of Retirement on Cardiovascular Disease and Its Risk Factors: A Systematic Review of Longitudinal Studies. Gerontologist. 2020;60(5):e367–77. pmid:31091304
- View Article
- PubMed/NCBI
- Google Scholar
42. Sato K, Noguchi H, Inoue K, Kawachi I, Kondo N. Retirement and cardiovascular disease: a longitudinal study in 35 countries. Int J Epidemiol. 2023;52(4):1047–59. pmid:37155837
- View Article
- PubMed/NCBI
- Google Scholar
43. Dohle S, Wansink B. Fit in 50 years: participation in high school sports best predicts one’s physical activity after age 70. BMC Public Health. 2013;13:1100. pmid:24289060
- View Article
- PubMed/NCBI
- Google Scholar
44. Cleland V, Dwyer T, Venn A. Which domains of childhood physical activity predict physical activity in adulthood? A 20-year prospective tracking study. Br J Sports Med. 2012;46(8):595–602. pmid:22144006
- View Article
- PubMed/NCBI
- Google Scholar
45. Glenmark B, Hedberg G, Jansson E. Prediction of physical activity level in adulthood by physical characteristics, physical performance and physical activity in adolescence: an 11-year follow-up study. Eur J Appl Physiol Occup Physiol. 1994;69(6):530–8. pmid:7713074
- View Article
- PubMed/NCBI
- Google Scholar
46. Luszczynska A, Sutton S. Physical activity after cardiac rehabilitation: Evidence that different types of self-efficacy are important in maintainers and relapsers. Rehabilitation Psychology. 2006;51(4):314–21.
- View Article
- Google Scholar
47. Le Grande MR, Murphy BM, Higgins RO, Worcester MUC, Parkinson A, Brown SL, et al. Physical activity and negative emotional response after percutaneous coronary intervention. Eur J Cardiovasc Prev Rehabil. 2006;13(2):254–60. pmid:16575281
- View Article
- PubMed/NCBI
- Google Scholar
48. Slovinec D’Angelo ME, Pelletier LG, Reid RD, Huta V. The roles of self-efficacy and motivation in the prediction of short- and long-term adherence to exercise among patients with coronary heart disease. Health Psychol. 2014;33(11):1344–53. pmid:25133848
- View Article
- PubMed/NCBI
- Google Scholar
49. Hershey MS, Martínez-González MÁ, Álvarez-Álvarez I, Martínez Hernández JA, Ruiz-Canela M. The Mediterranean diet and physical activity: better together than apart for the prevention of premature mortality. Br J Nutr. 2022;128(7):1413–24. pmid:34462020
- View Article
- PubMed/NCBI
- Google Scholar
50. Woolcott CG, Dishman RK, Motl RW, Matthai CH, Nigg CR. Physical activity and fruit and vegetable intake: correlations between and within adults in a longitudinal multiethnic cohort. Am J Health Promot. 2013;28(2):71–9. pmid:23458370
- View Article
- PubMed/NCBI
- Google Scholar
51. D’Amore C, Saunders S, Bhatnagar N, Griffith LE, Richardson J, Beauchamp MK. Determinants of physical activity in community-dwelling older adults: an umbrella review. Int J Behav Nutr Phys Act. 2023;20(1):135. pmid:37990225
- View Article
- PubMed/NCBI
- Google Scholar
52. Lee S, Collins EG. Factors influencing physical activity after cardiac surgery: An integrative review. Heart Lung. 2021;50(1):136–45. pmid:33143912
- View Article
- PubMed/NCBI
- Google Scholar
53. Statistik Austria. Österreichische Gesundheitsbefragung 2019 Hauptergebnisse des Austrian Health Interview Survey (ATHIS) und methodische Dokumentation. Wien: Bundesministerium für Soziales, Gesundheit, Pflege und Konsumentenschutz (BMSGPK). 2020.
54. Ramsey KA, Rojer AGM, D’Andrea L, Otten RHJ, Heymans MW, Trappenburg MC, et al. The association of objectively measured physical activity and sedentary behavior with skeletal muscle strength and muscle power in older adults: A systematic review and meta-analysis. Ageing Res Rev. 2021;67:101266. pmid:33607291
- View Article
- PubMed/NCBI
- Google Scholar
55. Green SB. How Many Subjects Does It Take To Do A Regression Analysis. Multivariate Behav Res. 1991;26(3):499–510. pmid:26776715
- View Article
- PubMed/NCBI
- Google Scholar
56. Cortis C, Puggina A, Pesce C, Aleksovska K, Buck C, Burns C, et al. Psychological determinants of physical activity across the life course: A “DEterminants of DIet and Physical ACtivity” (DEDIPAC) umbrella systematic literature review. PLoS One. 2017;12(8):e0182709. pmid:28817676
- View Article
- PubMed/NCBI
- Google Scholar
57. Garcia L, Mendonça G, Benedetti TRB, Borges LJ, Streit IA, Christofoletti M, et al. Barriers and facilitators of domain-specific physical activity: a systematic review of reviews. BMC Public Health. 2022;22(1):1964. pmid:36289461
- View Article
- PubMed/NCBI
- Google Scholar
58. Zhang Y, Koene M, Reijneveld SA, Tuinstra J, Broekhuis M, van der Spek S, et al. The impact of interventions in the built environment on physical activity levels: a systematic umbrella review. Int J Behav Nutr Phys Act. 2022;19(1):156. pmid:36550583
- View Article
- PubMed/NCBI
- Google Scholar

[ref1] 1. World Heart Observatory. World Heart Observatory. World Heart Observatory. https://world-heart-federation.org/world-heart-observatory/ 2024 January 2.

[ref2] 2. World Health Organization. Cardiovascular diseases (CVDs). https://www.who.int/news-room/fact-sheets/detail/cardiovascular-diseases-(cvds) 2022 December 20.
View Article
Google Scholar

[3] View Article

[4] Google Scholar

[ref3] 3. Visseren FLJ, Mach F, Smulders YM, Carballo D, Koskinas KC, Bäck M, et al. 2021 ESC Guidelines on Cardiovascular Disease Prevention in Clinical Practice: Developed by the Task Force for Cardiovascular Disease Prevention in Clinical Practice with Representatives of the European Society of Cardiology and 12 Medical Societies with the Special Contribution of the European Association of Preventive Cardiology (EAPC). European Heart Journal. 2021;42:3227–337.
View Article
Google Scholar

[6] View Article

[7] Google Scholar

[ref4] 4. Kotseva K, De Backer G, De Bacquer D, Rydén L, Hoes A, Grobbee D, et al. Lifestyle and impact on cardiovascular risk factor control in coronary patients across 27 countries: Results from the European Society of Cardiology ESC-EORP EUROASPIRE V registry. Eur J Prev Cardiol. 2019;26(8):824–35. pmid:30739508
View Article
PubMed/NCBI
Google Scholar

[9] View Article

[10] PubMed/NCBI

[11] Google Scholar

[ref5] 5. Gutenberg J, Kulnik ST, Niebauer J, Crutzen R. Determinants of Physical Activity in the Cardiac Population: Protocol for a Systematic Review. JMIR Res Protoc. 2022;11(9):e39188. pmid:36069762
View Article
PubMed/NCBI
Google Scholar

[13] View Article

[14] PubMed/NCBI

[15] Google Scholar

[ref6] 6. Bauman AE, Reis RS, Sallis JF, Wells JC, Loos RJF, Martin BW, et al. Correlates of physical activity: why are some people physically active and others not?. Lancet. 2012;380(9838):258–71. pmid:22818938
View Article
PubMed/NCBI
Google Scholar

[17] View Article

[18] PubMed/NCBI

[19] Google Scholar

[ref7] 7. Frey V, Langthaler P, Raphaelis E, Ring-Dimitriou S, Kedenko L, Aigner E, et al. Paracelsus 10,000: An observational cohort study about the health status of the population of Salzburg, Austria. Rationale, objectives and study design. Paracelsus Proceedings of Experimental Medicine. 2023;2:1–17.
View Article
Google Scholar

[21] View Article

[22] Google Scholar

[ref8] 8. Pühringer M, Ring-Dimitriou S, Stöggl T, Iglseder B, Paulweber B. Comparison of visual, automatic and semiautomatic methods to determine ventilatory indices in 50 to 60 years old adults. J Sports Sci. 2020;38(6):692–702. pmid:32089122
View Article
PubMed/NCBI
Google Scholar

[24] View Article

[25] PubMed/NCBI

[26] Google Scholar

[ref9] 9. Pühringer M, Ring-Dimitriou S, Stöggl T, Iglseder B, Paulweber B. Differences in the point of optimal ventilatory efficiency and the anaerobic threshold in untrained adults aged 50 to 60 years. Respir Physiol Neurobiol. 2020;282:103516. pmid:32768502
View Article
PubMed/NCBI
Google Scholar

[28] View Article

[29] PubMed/NCBI

[30] Google Scholar

[ref10] 10. Pühringer M, Ring-Dimitriou S, Iglseder B, Frey V, Trinka E, Paulweber B. Sequencing patterns of ventilatory indices in less trained adults. Front Sports Act Living. 2023;4:1066131. pmid:36755562
View Article
PubMed/NCBI
Google Scholar

[32] View Article

[33] PubMed/NCBI

[34] Google Scholar

[ref11] 11. Choi J, Lee M, Lee J-K, Kang D, Choi J-Y. Correlates associated with participation in physical activity among adults: a systematic review of reviews and update. BMC Public Health. 2017;17(1):356. pmid:28438146
View Article
PubMed/NCBI
Google Scholar

[36] View Article

[37] PubMed/NCBI

[38] Google Scholar

[ref12] 12. Carlin A, Perchoux C, Puggina A, Aleksovska K, Buck C, Burns C, et al. A life course examination of the physical environmental determinants of physical activity behaviour: A “Determinants of Diet and Physical Activity” (DEDIPAC) umbrella systematic literature review. PLoS One. 2017;12(8):e0182083. pmid:28787023
View Article
PubMed/NCBI
Google Scholar

[40] View Article

[41] PubMed/NCBI

[42] Google Scholar

[ref13] 13. Condello G, Puggina A, Aleksovska K, Buck C, Burns C, Cardon G, et al. Behavioral determinants of physical activity across the life course: a “DEterminants of DIet and Physical ACtivity” (DEDIPAC) umbrella systematic literature review. Int J Behav Nutr Phys Act. 2017;14(1):58. pmid:28464958
View Article
PubMed/NCBI
Google Scholar

[44] View Article

[45] PubMed/NCBI

[46] Google Scholar

[ref14] 14. Collado-Mateo D, Lavín-Pérez AM, Peñacoba C, Del Coso J, Leyton-Román M, Luque-Casado A, et al. Key Factors Associated with Adherence to Physical Exercise in Patients with Chronic Diseases and Older Adults: An Umbrella Review. Int J Environ Res Public Health. 2021;18(4):2023. pmid:33669679
View Article
PubMed/NCBI
Google Scholar

[48] View Article

[49] PubMed/NCBI

[50] Google Scholar

[ref15] 15. Kulnik ST, Carrozzo E, Kumar D, Ring-Dimitriou S, Roth M, Blechert J. Correlates of physical activity behaviour in a cardiovascular population sample. Statistical analysis plan for a secondary analysis of data from the Paracelsus 10,000 prospective cohort study in Salzburg, Austria. Open Science Framework. 2023.
View Article
Google Scholar

[52] View Article

[53] Google Scholar

[ref16] 16. Kulnik ST. Correlates of physical activity behaviour in a cardiovascular population sample. Open Science Framework (OSF) Registries. https://osf.io/j4h2z 2023. 2024 January 8.

[ref17] 17. Yuan I, Topjian AA, Kurth CD, Kirschen MP, Ward CG, Zhang B, et al. Guide to the statistical analysis plan. Paediatr Anaesth. 2019;29(3):237–42. pmid:30609103
View Article
PubMed/NCBI
Google Scholar

[56] View Article

[57] PubMed/NCBI

[58] Google Scholar

[ref18] 18. von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP, et al. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. J Clin Epidemiol. 2008;61(4):344–9. pmid:18313558
View Article
PubMed/NCBI
Google Scholar

[60] View Article

[61] PubMed/NCBI

[62] Google Scholar

[ref19] 19. SCORE2 working group, ESC Cardiovascular risk collaboration. SCORE2 risk prediction algorithms: new models to estimate 10-year risk of cardiovascular disease in Europe. European Heart Journal. 2021;42:2439–54.
View Article
Google Scholar

[64] View Article

[65] Google Scholar

[ref20] 20. SCORE2-OP working group, ESC Cardiovascular risk collaboration. SCORE2-OP risk prediction algorithms: estimating incident cardiovascular event risk in older persons in four geographical risk regions. European Heart Journal. 2021;42:2455–67.
View Article
Google Scholar

[67] View Article

[68] Google Scholar

[ref21] 21. ICD-10 Version: 2019. https://icd.who.int/browse10/2019/en 2023 January 25.
View Article
Google Scholar

[70] View Article

[71] Google Scholar

[ref22] 22. Bauman AE, Sallis JF, Dzewaltowski DA, Owen N. Toward a better understanding of the influences on physical activity: the role of determinants, correlates, causal variables, mediators, moderators, and confounders. Am J Prev Med. 2002;23(2 Suppl):5–14. pmid:12133733
View Article
PubMed/NCBI
Google Scholar

[73] View Article

[74] PubMed/NCBI

[75] Google Scholar

[ref23] 23. McGowan H, Gutenberg J, Mayr B, Sareban M, Niebauer J, Crutzen R. Determinants of physical activity behaviour in the cardiac population: a systematic review. Research Square. 2023.
View Article
Google Scholar

[77] View Article

[78] Google Scholar

[ref24] 24. Bellamy N, Buchanan WW, Goldsmith CH, Campbell J, Stitt LW. Validation study of WOMAC: a health status instrument for measuring clinically important patient relevant outcomes to antirheumatic drug therapy in patients with osteoarthritis of the hip or knee. J Rheumatol. 1988;15(12):1833–40. pmid:3068365
View Article
PubMed/NCBI
Google Scholar

[80] View Article

[81] PubMed/NCBI

[82] Google Scholar

[ref25] 25. Fairbank JC, Couper J, Davies JB, O’Brien JP. The Oswestry low back pain disability questionnaire. Physiotherapy. 1980;66(8):271–3. pmid:6450426
View Article
PubMed/NCBI
Google Scholar

[84] View Article

[85] PubMed/NCBI

[86] Google Scholar

[ref26] 26. Beck AT, Ward CH, Mendelson M, Mock J, Erbaugh J. An inventory for measuring depression. Arch Gen Psychiatry. 1961;4:561–71. pmid:13688369
View Article
PubMed/NCBI
Google Scholar

[88] View Article

[89] PubMed/NCBI

[90] Google Scholar

[ref27] 27. Johns MW. A new method for measuring daytime sleepiness: the Epworth sleepiness scale. Sleep. 1991;14(6):540–5. pmid:1798888
View Article
PubMed/NCBI
Google Scholar

[92] View Article

[93] PubMed/NCBI

[94] Google Scholar

[ref28] 28. Patton JH, Stanford MS, Barratt ES. Factor structure of the Barratt impulsiveness scale. J Clin Psychol. 1995;51(6):768–74. pmid:8778124
View Article
PubMed/NCBI
Google Scholar

[96] View Article

[97] PubMed/NCBI

[98] Google Scholar

[ref29] 29. Spinella M. Normative data and a short form of the Barratt Impulsiveness Scale. Int J Neurosci. 2007;117(3):359–68. pmid:17365120
View Article
PubMed/NCBI
Google Scholar

[100] View Article

[101] PubMed/NCBI

[102] Google Scholar

[ref30] 30. Saunders JB, Aasland OG, Babor TF, de la Fuente JR, Grant M. Development of the Alcohol Use Disorders Identification Test (AUDIT): WHO Collaborative Project on Early Detection of Persons with Harmful Alcohol Consumption--II. Addiction. 1993;88:791–804.
View Article
Google Scholar

[104] View Article

[105] Google Scholar

[ref31] 31. Schröder H, Fitó M, Estruch R, Martínez-González MA, Corella D, Salas-Salvadó J, et al. A short screener is valid for assessing Mediterranean diet adherence among older Spanish men and women. J Nutr. 2011;141(6):1140–5. pmid:21508208
View Article
PubMed/NCBI
Google Scholar

[107] View Article

[108] PubMed/NCBI

[109] Google Scholar

[ref32] 32. Wareham NJ, Jakes RW, Rennie KL, Schuit J, Mitchell J, Hennings S, et al. Validity and repeatability of a simple index derived from the short physical activity questionnaire used in the European Prospective Investigation into Cancer and Nutrition (EPIC) study. Public Health Nutr. 2003;6(4):407–13. pmid:12795830
View Article
PubMed/NCBI
Google Scholar

[111] View Article

[112] PubMed/NCBI

[113] Google Scholar

[ref33] 33. InterAct Consortium, Peters T, Brage S, Westgate K, Franks PW, Gradmark A, et al. Validity of a short questionnaire to assess physical activity in 10 European countries. Eur J Epidemiol. 2012;27(1):15–25. pmid:22089423
View Article
PubMed/NCBI
Google Scholar

[115] View Article

[116] PubMed/NCBI

[117] Google Scholar

[ref34] 34. Bonekamp NE, Visseren FLJ, Ruigrok Y, Cramer MJM, de Borst GJ, May AM, et al. Leisure-time and occupational physical activity and health outcomes in cardiovascular disease. Heart. 2023;109(9):686–94. pmid:36270785
View Article
PubMed/NCBI
Google Scholar

[119] View Article

[120] PubMed/NCBI

[121] Google Scholar

[ref35] 35. R Core Team. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. 2021.

[ref36] 36. Akaike H. Information Theory and an Extension of the Maximum Likelihood Principle. Selected Papers of Hirotugu Akaike. New York, NY: Springer; 1998. pp. 199–213. 10.1007/978-1-4612-1694-0_15

[ref37] 37. Pedregosa F, Varoquaux G, Gramfort A, Michel V, Thirion B, Grisel O. Scikit-learn: Machine Learning in Python. Journal of Machine Learning Research. 2011;12:2825–30.
View Article
Google Scholar

[125] View Article

[126] Google Scholar

[ref38] 38. Lundberg SM, Lee SI. A unified approach to interpreting model predictions. Advances in Neural Information Processing Systems. Curran Associates, Inc. 2017.

[ref39] 39. Stojanovic M, Babulal GM, Head D. Determinants of physical activity engagement in older adults. J Behav Med. 2023;46(5):757–69. pmid:36920727
View Article
PubMed/NCBI
Google Scholar

[129] View Article

[130] PubMed/NCBI

[131] Google Scholar

[ref40] 40. Biscaglia S, Campo G, Sorbets E, Ford I, Fox KM, Greenlaw N, et al. Relationship between physical activity and long-term outcomes in patients with stable coronary artery disease. Eur J Prev Cardiol. 2020;27(4):426–36. pmid:31558054
View Article
PubMed/NCBI
Google Scholar

[133] View Article

[134] PubMed/NCBI

[135] Google Scholar

[ref41] 41. Xue B, Head J, McMunn A. The Impact of Retirement on Cardiovascular Disease and Its Risk Factors: A Systematic Review of Longitudinal Studies. Gerontologist. 2020;60(5):e367–77. pmid:31091304
View Article
PubMed/NCBI
Google Scholar

[137] View Article

[138] PubMed/NCBI

[139] Google Scholar

[ref42] 42. Sato K, Noguchi H, Inoue K, Kawachi I, Kondo N. Retirement and cardiovascular disease: a longitudinal study in 35 countries. Int J Epidemiol. 2023;52(4):1047–59. pmid:37155837
View Article
PubMed/NCBI
Google Scholar

[141] View Article

[142] PubMed/NCBI

[143] Google Scholar

[ref43] 43. Dohle S, Wansink B. Fit in 50 years: participation in high school sports best predicts one’s physical activity after age 70. BMC Public Health. 2013;13:1100. pmid:24289060
View Article
PubMed/NCBI
Google Scholar

[145] View Article

[146] PubMed/NCBI

[147] Google Scholar

[ref44] 44. Cleland V, Dwyer T, Venn A. Which domains of childhood physical activity predict physical activity in adulthood? A 20-year prospective tracking study. Br J Sports Med. 2012;46(8):595–602. pmid:22144006
View Article
PubMed/NCBI
Google Scholar

[149] View Article

[150] PubMed/NCBI

[151] Google Scholar

[ref45] 45. Glenmark B, Hedberg G, Jansson E. Prediction of physical activity level in adulthood by physical characteristics, physical performance and physical activity in adolescence: an 11-year follow-up study. Eur J Appl Physiol Occup Physiol. 1994;69(6):530–8. pmid:7713074
View Article
PubMed/NCBI
Google Scholar

[153] View Article

[154] PubMed/NCBI

[155] Google Scholar

[ref46] 46. Luszczynska A, Sutton S. Physical activity after cardiac rehabilitation: Evidence that different types of self-efficacy are important in maintainers and relapsers. Rehabilitation Psychology. 2006;51(4):314–21.
View Article
Google Scholar

[157] View Article

[158] Google Scholar

[ref47] 47. Le Grande MR, Murphy BM, Higgins RO, Worcester MUC, Parkinson A, Brown SL, et al. Physical activity and negative emotional response after percutaneous coronary intervention. Eur J Cardiovasc Prev Rehabil. 2006;13(2):254–60. pmid:16575281
View Article
PubMed/NCBI
Google Scholar

[160] View Article

[161] PubMed/NCBI

[162] Google Scholar

[ref48] 48. Slovinec D’Angelo ME, Pelletier LG, Reid RD, Huta V. The roles of self-efficacy and motivation in the prediction of short- and long-term adherence to exercise among patients with coronary heart disease. Health Psychol. 2014;33(11):1344–53. pmid:25133848
View Article
PubMed/NCBI
Google Scholar

[164] View Article

[165] PubMed/NCBI

[166] Google Scholar

[ref49] 49. Hershey MS, Martínez-González MÁ, Álvarez-Álvarez I, Martínez Hernández JA, Ruiz-Canela M. The Mediterranean diet and physical activity: better together than apart for the prevention of premature mortality. Br J Nutr. 2022;128(7):1413–24. pmid:34462020
View Article
PubMed/NCBI
Google Scholar

[168] View Article

[169] PubMed/NCBI

[170] Google Scholar

[ref50] 50. Woolcott CG, Dishman RK, Motl RW, Matthai CH, Nigg CR. Physical activity and fruit and vegetable intake: correlations between and within adults in a longitudinal multiethnic cohort. Am J Health Promot. 2013;28(2):71–9. pmid:23458370
View Article
PubMed/NCBI
Google Scholar

[172] View Article

[173] PubMed/NCBI

[174] Google Scholar

[ref51] 51. D’Amore C, Saunders S, Bhatnagar N, Griffith LE, Richardson J, Beauchamp MK. Determinants of physical activity in community-dwelling older adults: an umbrella review. Int J Behav Nutr Phys Act. 2023;20(1):135. pmid:37990225
View Article
PubMed/NCBI
Google Scholar

[176] View Article

[177] PubMed/NCBI

[178] Google Scholar

[ref52] 52. Lee S, Collins EG. Factors influencing physical activity after cardiac surgery: An integrative review. Heart Lung. 2021;50(1):136–45. pmid:33143912
View Article
PubMed/NCBI
Google Scholar

[180] View Article

[181] PubMed/NCBI

[182] Google Scholar

[ref53] 53. Statistik Austria. Österreichische Gesundheitsbefragung 2019 Hauptergebnisse des Austrian Health Interview Survey (ATHIS) und methodische Dokumentation. Wien: Bundesministerium für Soziales, Gesundheit, Pflege und Konsumentenschutz (BMSGPK). 2020.

[ref54] 54. Ramsey KA, Rojer AGM, D’Andrea L, Otten RHJ, Heymans MW, Trappenburg MC, et al. The association of objectively measured physical activity and sedentary behavior with skeletal muscle strength and muscle power in older adults: A systematic review and meta-analysis. Ageing Res Rev. 2021;67:101266. pmid:33607291
View Article
PubMed/NCBI
Google Scholar

[185] View Article

[186] PubMed/NCBI

[187] Google Scholar

[ref55] 55. Green SB. How Many Subjects Does It Take To Do A Regression Analysis. Multivariate Behav Res. 1991;26(3):499–510. pmid:26776715
View Article
PubMed/NCBI
Google Scholar

[189] View Article

[190] PubMed/NCBI

[191] Google Scholar

[ref56] 56. Cortis C, Puggina A, Pesce C, Aleksovska K, Buck C, Burns C, et al. Psychological determinants of physical activity across the life course: A “DEterminants of DIet and Physical ACtivity” (DEDIPAC) umbrella systematic literature review. PLoS One. 2017;12(8):e0182709. pmid:28817676
View Article
PubMed/NCBI
Google Scholar

[193] View Article

[194] PubMed/NCBI

[195] Google Scholar

[ref57] 57. Garcia L, Mendonça G, Benedetti TRB, Borges LJ, Streit IA, Christofoletti M, et al. Barriers and facilitators of domain-specific physical activity: a systematic review of reviews. BMC Public Health. 2022;22(1):1964. pmid:36289461
View Article
PubMed/NCBI
Google Scholar

[197] View Article

[198] PubMed/NCBI

[199] Google Scholar

[ref58] 58. Zhang Y, Koene M, Reijneveld SA, Tuinstra J, Broekhuis M, van der Spek S, et al. The impact of interventions in the built environment on physical activity levels: a systematic umbrella review. Int J Behav Nutr Phys Act. 2022;19(1):156. pmid:36550583
View Article
PubMed/NCBI
Google Scholar

[201] View Article

[202] PubMed/NCBI

[203] Google Scholar

Figures

Abstract

Introduction

Materials and methods

Study population

Variables

Terminology.

Correlates of physical activity behaviour.

Outcome variables.

Data cleaning

Statistical analysis

Bivariate analysis.

Parametric regression analysis.

Non-parametric machine learning analysis.

Results

Primary physical activity outcome

Secondary physical activity outcomes

Discussion

Retirement

Past physical activity behaviour

Mediterranean diet

Gender

Strengths and limitations

Conclusions

Supporting information

S1 File. Regression models and machine learning models for the secondary physical activity outcomes.

S2 File. Characteristics of Paracelsus 10.000 participants with increased cardiovascular risk and established cardiovascular disease: sample included in the analysis (n = 1173) versus excluded participants (n = 3064).

References