Domain-Specific Physical Activity and Stroke in Sweden

This cohort study investigates the association of leisure time, work time, transport, and household physical activity with stroke incidence and outcomes in Sweden.


Introduction
Public strategies that promote physical activity may effectively reduce the burden of cardiovascular diseases. 1Objective measurements indicate a dose-response association between physical activity and stroke-free survival, 2,3 and modifiable factors, including physical inactivity, have been associated with a substantial proportion of the population-attributable stroke risk in all major regions of the world. 4Increasing physical activity is a sustainable concept that can improve overall health, 5 prevent other cardiovascular risk factors, 6 and stimulate several potentially neuroprotective mechanisms. 7though previous research has highlighted the beneficial association of overall physical activity with reduced risk of stroke, many details concerning the association between prestroke physical activity and stroke outcomes remain unclear. 8Most prior studies have not distinguished domainspecific physical activity, including leisure time, work time, transport, and household physical activity, when estimating the risk of stroke and poststroke outcomes. 9While leisure -time physical activity has consistently been associated with a reduced risk of stroke, [10][11][12] studies have yielded conflicting results regarding work time physical activity, indicating both a decreased risk 13 and an increased risk. 14Additionally, there is a lack of data on the associations of transport or household physical activity with stroke.Device-based measurements fail to capture information about physical activity domains, potentially favoring dynamic activities, such as walking or running, over resistance activities, such as weight lifting or isometric exercises.Conversely, self-reported measures are susceptible to social desirability and recall biases. 15better understanding of domain-specific associations of physical activity is crucial for tailoring stroke-prevention strategies.Moreover, prospective, population-based data on the association between domain-specific physical activities and stroke are scarce.In this study, we aimed to investigate associations of domain-specific physical activity with stroke incidence and poststroke death or activities of daily living (ADL) dependency using a prospective, population-based cohort.

Methods
This prospective cohort study was based on data from a randomly sampled population cohort established by the Interplay Between Genetic Susceptibility and External Factors (INTERGENE)   research program in 2001. 16Data collection for the INTERGENE cohort was approved by the Regional Ethics Committee in Gothenburg.Data linkage to national registries for this study was approved by the Swedish Ethical Review Authority.All participants provided written informed consent to take part in examinations, including linkage to future health end points.This study follows the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cohort studies.
An invitation to participate was sent out to 8625 peopled aged 24 to 74 years living in an urban-rural area covering western Sweden.The setting comprised Sweden's second-largest city, Gothenburg (urban), and smaller communities and dispersed settlement areas throughout the region (rural).Baseline data, including questionnaires and clinical measurements, from 3614 participants (41.9% acceptance rate) were collected from 2001 to 2004.Baseline assessments were performed at the Sahlgrenska University Hospital in Gothenburg and using a mobile research bus.
Reexaminations using the same metrics were performed from 2014 to 2016 in 1394 participants (of 2108 invited) who attended the baseline examination at Sahlgrenska University Hospital and were still alive.

Physical Activity Assessments
Self-reported physical activity levels were evaluated using questionnaires administered on the day of examination across 4 domains: leisure time, work time, transportation, and household physical activity.Participants were asked to estimate their mean physical activity level over the past year and consider significant variations that may have occurred between seasons.Leisure time physical Downloaded from jamanetwork.comby guest on 06/05/2024 activity was evaluated using the validated Saltin-Grimby Physical Activity Level Scale. 17,18Work time, transportation, and household physical activity questions were part of a validated, selfadministered physical activity questionnaire that was used in the Swedish National Public Health Survey. 19For each domain, physical activity levels were categorized as low, intermediate, or high (eTable 1 in Supplement 1).These assessments were repeated during reexaminations.

JAMA Network Open | Neurology
Objective evaluations of physical activity were performed in a subgroup of 496 participants at reexaminations using a sealed pedometer: Yamax CW 700 (Yamax Health and Sports).The Yamax pedometer has been found to have high step count accuracy when worn on the hip. 20Participants were instructed by a research nurse to wear the pedometer on the hip during all waking hours for 6 days.To account for activity during transportation, 160 steps were added to the total step count per reported minute of cycling.The mean daily step count was calculated for participants who had recorded at least 1000 steps per day over a period of 10 hours for 4 consecutive days.Investigators repeatedly tested the function of all pedometers before and throughout the study.between self-reported physical activity and pedometer-derived data.Follow-up time was calculated as the time in days from the date of cohort entry until the date of the first stroke diagnosis, death, or end of follow-up, whichever occurred first.There was no loss to follow-up in national registers.

Participant Characteristics
However, emigration history was not known.Age-adjusted cumulative stroke hazards stratified by domain-specific physical activity level and accounting for the competing risk of death were calculated using hazard regression models. 22Crude cumulative incidences of stroke were also calculated using Kaplan-Meier survival curves.
Multiple imputation by chained equations was used to handle missing observations in regression analyses.We used the classification and regression trees method, which imputes missing values iteratively by constructing estimating models for each variable with missing data, leveraging all other variables in the dataset. 23Adjusted hazard ratios (aHRs) for the association between physical activity and first stroke incidence were calculated using Cox proportional hazard regression models.Schoenfeld residuals were used to test the proportional hazard assumption.Adjusted odds ratios for the association between physical activity and death or dependency 3 months after stroke were calculated using binary logistic regression models.For each physical activity domain, we constructed regression models adjusted for age and sex.In addition, a fully adjusted Cox regression model was fitted for each physical activity domain, including all baseline characteristics.
Mixed-effects Cox regression models adjusted for age and sex were used to assess the association between 2 repeated measurements of domain-specific physical activity and stroke incidence (at baseline in 2001-2004 and at reexaminations in 2014-2016).In these analyses, the follow-up was split at the time of reexamination and physical activity was treated as a timedependent categorical variable.Subgroup analyses adjusted for age and sex were conducted by exploring the interaction between baseline characteristics and leisure time physical activity using Cox proportional hazard models.
All statistical analyses were 2-tailed, and significance was interpreted based on 95% CIs not including 1 or P < .05.All analyses were conducted using R statistical software version 4.3.2(R Project for Statistical Computing).Data were analyzed from September through October 2023.

Results
There were 3614 participants (aged between 24 and 77 years at baseline; mean [SD] age, 51. 4 [13.1]years; 1910 female [52.9%]) (Table 1).The numbers of participants in each physical activity category at baseline and at reexaminations with corresponding pedometer-derived data are presented in eTable 1 in Supplement 1.The mean number of steps recorded over a 6-day period exhibited positive correlations with self-reported leisure time physical activity (r = 0.24; P < .001)and transport physical activity (r = 0.17; P < .001).In contrast, there was a negative correlation between the mean number of steps and household physical activity (r = −0.11;P = .02)and no correlation with work time physical activity (r = 0.01; P = .86).
Age a There were 54 participants with incomplete ADL assessments.
b Smoking was defined as current smoking or smoking in the past year.
c High alcohol intake was defined as more than 10 g/d for females and more than 20 g/d for males.
d Financial stability was defined as managing monthly expenses without difficulty.
e Presence of a social network was defined as communication or engagement with 3 or more familiar individuals per week.

Discussion
In this 20-year prospective, population-based cohort study, we observed that intermediate (eg, bicycling, walking outdoors, or playing table tennis Ն4 hr/wk) and high (eg, running, swimming, or playing tennis 2-3 hr/wk or engaging in competitive sports) levels of leisure time physical activity were associated with a reduction in stroke incidence.Similar benefits were found for an intermediate physical activity level in transportation (eg, walking or biking 20-40 min/d).A high level of leisure time physical activity was also associated with a lower risk of poststroke death or ADL dependency, and there was a potentially mitigating interaction between physical activity and the stroke risk associated with smoking or having a family history of stroke.
1][12] In contrast, some studies have found that high levels of work time physical activity were associated with increased risk of stroke 14 and overall cardiovascular disease and mortality, 24,25 whereas other studies 13,26 found a decreased stroke incidence with increased work time physical activity.Few prior studies have investigated the association of transport or household physical activity with stroke.However, a 2022 retrospective study 26 on the US population found that individuals who walked, bicycled, or did house or yard work had a significant reduction in stroke prevalence.Daily walking or cycling to and from work has also been shown to be associated with a reduced risk of stroke. 11erging evidence suggests that prestroke physical activity is associated with improved stroke outcomes. 8However, most prior studies have used retrospective assessments of prestroke physical activity, limiting the interpretability of their findings.Using prospective assessments, Mediano et al 27 reported that higher levels of total, leisure time, and work time physical activity were associated with a lower risk of mortality.9][30][31] Similar to our results, those of Rist et al 32 demonstrated that increased levels of prestroke physical activity were associated with greater independence in instrumental ADL tasks assessed 3 years after a stroke.
Self-reported assessments of physical activity typically exhibit limited association with objective data, 33 with the latter being more consistently associated with cardiovascular outcomes. 34Leisure time and transport physical activity were positively correlated with pedometer-derived measurements in this study, whereas work time and household activities were not.Considering that leisure time and transport physical activities were also the only domains associated with stroke incidence, aerobic activities (ie, a higher daily step count) may be particularly beneficial for stroke prevention. 18,19However, observed correlations were weak.Given that pedometers measured total physical activity over a few days while questionnaires measured domain-specific physical activity over a year, discrepancies were expected.Still, the absence of stronger correlation suggests that there is limited validity in capturing the true physical activity level through domain-specific questions b Smoking was defined as current smoking or smoking in the past year.
c High alcohol intake was defined as more than 10 g/d for females and more than 20 g/d for males.
d Financial stability was defined as managing monthly expenses without difficulty.
e Presence of a social network was defined as communication or engagement with 3 or more familiar individuals per week.

Strengths and Limitations
Our findings should be interpreted considering several strengths and limitations.Strengths include the prospective population-based design, a long-term follow-up, and robust measures of stroke incidence, mortality, and poststroke ADL dependency through extensive data collection.The possibility to link the INTERGENE cohort with data from the Swedish national registers is a particular strength of this study.We were also able to adjust for a wide range of important risk factors and potential confounders, for which data were not available in prior studies.Furthermore, our study benefited from the high validity and completeness of Swedish register data, minimizing data quality concerns and missing observations. 36,37mitations include the study participation rate of 41.9% at baseline, which may have introduced participation bias.Additionally, most participants were assessed only once, which limits our understanding of longitudinal changes in physical activity behaviors.Second, self-reported measures of physical activity are susceptible to social desirability and recall bias. 15Physical activity assessments referred to the year mean, which may increase the uncertainty of estimates.However, 1-year estimates ensure a consistent recall period across participants without effects from seasonal variations.Furthermore, the objective evaluation of physical activity was conducted in a subset of participants, and the representativeness of pedometer data in reflecting usual physical activity levels is limited.Although Swedish government registers have excellent coverage, they do not capture stroke events or deaths if a person has emigrated, and we were not able to control for this factor.
Additionally, despite adjustment for multiple baseline characteristics, the potential for unmeasured confounding and the influence of time-varying factors cannot be ruled out.The observational design of our study also renders it susceptible to reverse causality, wherein individuals with specific health conditions may have altered their physical activity levels due to their health status.

Conclusions
In this cohort study of 3614 individuals from the general population in western Sweden, intermediate and high levels of leisure time physical activity and an intermediate level of transport physical activity were associated with a decreased risk of stroke.Higher levels of leisure time physical activity at baseline were also associated with a lower risk of death or ADL dependency 3 months after stroke.
These findings suggest that integrating leisure time physical activity and active transport into public health policies may be a sustainable intervention to reduce the burden of stroke.

Stroke Incidence and Outcomes Using Swedish personal identification numbers, data collected in the INTERGENE cohort were linked to data from national Swedish registries. Stroke events were identified in the National Patient Register held by the National Board of Health and Welfare in Sweden using International Statistical
Baseline characteristics were presented as numbers with percentages for categorical variables and means with SDs for continuous variables.Spearman rank correlation was used to test the correlation JAMA Network Open | Neurology Domain-Specific Physical Activity and Stroke in Sweden JAMA Network Open.2024;7(5):e2413453.doi:10.1001/jamanetworkopen.2024.13453(Reprinted) May 29, 2024 3/12 Downloaded from jamanetwork.comby guest on 06/05/2024

Table 1 .
-adjusted cumulative stroke hazards stratified by domain-specific physical activity levels are Baseline Characteristics of Study Participants Adjusted associations of domain-specific physical activity levels with stroke incidence and poststroke death or ADL dependency are presented in Table 2. Models adjusted separately for lifestyle factors, socioeconomic factors, comorbid conditions, and genetic factors are presented in eTable 2 in Supplement 1. Intermediate and high levels of leisure time physical activity were associated with a decreased incidence of stroke compared with low levels across all models; for JAMA Network Open | Neurology Domain-Specific Physical Activity and Stroke in Sweden Domain-Specific Physical Activity and Stroke in Sweden example, aHRs in model 1 were 0.54 (95% CI, 0.38-0.77)for intermediate and (aHR, 0.47; 95% CI, 0.31-0.73)for high levels.An intermediate compared with a low level of physical activity in transportation was also associated with a reduced stroke hazard in all models (eg, model 1: aHR, 0.69; 95% CI, 0.52-0.93).However, there was no association for a high level of transport physical

Table 2 .
Associations of Baseline Domain-Specific Physical Activity With Stroke Incidence and Outcomes a Work time and household physical activity showed no association with stroke incidence.A high compared with a low level of leisure time physical activity level at baseline was associated with a lower risk of death or ADL dependency 3 months after stroke in all models (eg, model 1: adjusted odds ratio, 0.34; 95% CI, 0.16-0.71).
a Associations are shown with incidence of first stroke during 20-year follow-up and death or ADL dependency at 3 months after stroke.bModel 1 was adjusted for age and sex.Model 2 was adjusted for all covariates (age, sex, smoking, alcohol intake, education, economy, marital status, social network, living area, obesity, diabetes, hyperlipidemia, hypertension, atrial fibrillation, family history c The aHRs were calculated using Cox proportional hazard models.dTheaORswerecalculated using binary logistic regression models.Model 2 was not applied to death or ADL dependency to prevent overfitting given the limited number of outcome events.JAMA Network Open | NeurologyDomain-Specific Physical Activity and Stroke in Sweden JAMA Network Open.2024;7(5):e2413453.doi:10.1001/jamanetworkopen.2024.13453(Reprinted) May 29, 2024 6/12 Downloaded from jamanetwork.comby guest on 06/05/2024 activity.

Table 4 .
An interaction was found between leisure time physical activity and smoking, such that smoking (current smoking or smoking in the past year vs no smoking) was associated with an increased stroke hazard in participants with low or intermediate physical activity levels (aHR, 2.33; 95% CI, 1.72-3.15)but not participants with high physical activity levels (aHR, 1.09; 95% CI, 0.63-1.87).Another interaction was found between physical activity and family history of stroke, in which having a first-degree relative with a history of stroke was associated with increased stroke hazard in participants with low or intermediate physical activity levels (aHR, 1.73; 95% CI, 1.27-2.38)but not participants with high physical activity levels (aHR, 0.79; 95% CI, 0.38-1.62).

Table 3 .
Associations Between 2 Repeated Measurements of Physical Activity and Incidence of First Stroke Abbreviation: aHR, adjusted hazard ratio.aParticipantsincluded in these analyses had available data from baseline and the reexamination.The aHRs were computed using mixed-effects Cox regression models, with physical activity level included as a time-dependent categorical variable measured at 2 times during a 20-year follow-up(2001-2004 and again at 2014-2016).All associations are adjusted for age and sex.

Table 4 .
Association of Baseline Characteristics With First Stroke Incidence and Interaction With Leisure Time Physical Activity Incidence was measured during a 20-year follow-up, and aHRs for baseline characteristics were computed for low or intermediate and high leisure time physical activity using Cox proportional hazard regression.All associations are adjusted for age and sex. a Domain-Specific Physical Activity and Stroke in Sweden Some physical activity behaviors may also exhibit less stability over time, whereas self-reported leisure time physical activity may represent a consistent, often long-term behavior.35Insupport of this conclusion, leisure time physical activity exhibited the smallest intraindividual change over time.
Banda JA, Hutto B, Feeney A, et al.Comparing physical activity measures in a diverse group of midlife and older adults.Med Sci Sports Exerc.2010;42(12):2251-2257.doi:10.1249/MSS.0b013e3181e32e9a34.Perry AS, Dooley EE, Master H, Spartano NL, Brittain EL, Pettee Gabriel K. Physical activity over the lifecourse and cardiovascular disease.Circ Res.2023;132(12):1725-1740. doi:10.1161/CIRCRESAHA.123.32212135.Armstrong GK, Morgan K. Stability and change in levels of habitual physical activity in later life.Age Ageing.1998;27(suppl3):17-23.doi:10.1093/ageing/27.suppl_3.1736.Ludvigsson JF, Andersson E, Ekbom A, et al.External review and validation of the Swedish national inpatient register.BMC Public Health.2011;11:450.doi:10.1186/1471-2458-11-45037. Brooke HL, Talbäck M, Hörnblad J, et al.The Swedish cause of death register.Eur J Epidemiol.2017;32(9): 765-773.doi:10.1007/s10654-017-0316-1Classification of physical activity domains at baseline in 2001 to 2004 and at the reexamination in 2014 to 2016 with pedometer-derived data eTable 2. Adjusted associations between levels of baseline domain-specific physical activity with the incidence of first stroke during a 20-year follow-up and with death or ADL dependency at 3 months after stroke eFigure 1.Crude cumulative incidence of first stroke during a 20-year follow-up stratified by domain-specific physical activity eFigure 2. Intraindividual change in leisure time physical activity between baseline at 2001 to 2004 and the reexamination 2014 to 2016 in 1379 participants with available data eFigure 3. Intraindividual change in work time physical activity between baseline at 2001 to 2004 and the reexamination 2014 to 2016 in 1371 participants with available data eFigure 4. Intraindividual change in transport physical activity between baseline at 2001 to 2004 and the reexamination 2014 to 2016 in 1377 participants with available data eFigure 5. Intraindividual change in household physical activity between baseline at 2001 to 2004 and the reexamination 2014 to 2016 in 1378 participants with available data 33.