Article Text

Accelerometer-derived physical activity and the risk of death, heart failure, and stroke in patients with atrial fibrillation: a prospective study from UK Biobank
  1. Hyo-Jeong Ahn1,
  2. Eue-Keun Choi1,2,
  3. Tae-Min Rhee3,
  4. JungMin Choi1,
  5. Kyung-Yeon Lee1,
  6. Soonil Kwon1,
  7. So-Ryoung Lee1,2,
  8. Seil Oh1,2,
  9. Gregory Y H Lip4,5
  1. 1 Department of Internal Medicine, Seoul National University Hospital, Seoul, Republic of Korea
  2. 2 Department of Internal Medicine, Seoul National University College of Medicine, Seoul, Republic of Korea
  3. 3 Department of Internal Medicine, Seoul National University Hospital Healthcare System Gangnam Center, Seoul, Republic of Korea
  4. 4 Liverpool Centre for Cardiovascular Science, University of Liverpool, Liverpool, UK
  5. 5 Department of Clinical Medicine, Aalborg University, Aalborg, Denmark
  1. Correspondence to Dr Eue-Keun Choi, Seoul National University Hospital, Jongno-gu, Korea (the Republic of); choiek417{at}


Objective Data on cardiovascular outcomes according to objectively measured physical activity (PA) in patients with atrial fibrillation (AF) are scarce. This study explored the associations between PA derived from wrist-worn accelerometers and the risk of death, incident heart failure (HF), and incident stroke in patients with AF.

Methods From 37 990 patients with AF in UK Biobank, 2324 patients with accelerometer data were included. Weekly moderate-to-vigorous PA (MVPA) duration was computed from accelerometer data. The primary outcome was all-cause mortality. Secondary outcomes were cardiovascular mortality, incident HF, and incident stroke. Restricted cubic splines estimated the dose–response associations between MVPA duration and the outcomes. The adjusted HRs (aHRs) of the outcomes according to adherence to PA standard guidelines (performing MVPA≥150 min/week) were also evaluated.

Results The mean age was 66.9±6.2 years and 64.9% were male. During a median follow-up of 6.7 years, there were 181 all-cause deaths, 62 cardiovascular deaths, 225 cases of incident HF, and 91 cases of incident stroke; the overall incidence rate per 1000 patient-years was 11.76, 4.03, 15.16 and 5.99, respectively. There was a linear inverse dose–response relationship between MVPA (≥108 min/week) and all-cause mortality. Performing MVPA for 105–590 min/week was associated with a lower risk of HF than those with no measurable MVPA. The risk of stroke and cardiovascular mortality was not associated with MVPA. Performing guideline-adherent MVPA was related to a 30% lower risk of all-cause mortality (aHR: 0.70 (0.50–0.98), p=0.04) and 33% lower risk of HF (aHR 0.67 (0.49–0.93), p=0.02).

Conclusion In patients with AF, accelerometer-derived PA data supports lower risks of all-cause mortality and HF according to a greater level of MVPA and adherence to PA guidelines. Regular MVPA should be encouraged in patients with AF as a part of integrated management.

  • physical activity
  • death
  • heart disease

Data availability statement

Data are available upon reasonable request.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


  • The benefit of regular physical activity (PA) for improving cardiovascular health has been well established. Only limited data, the majority of which are based on self-reported questionnaires, are available about the impact of PA on cardiovascular outcomes in patients with atrial fibrillation (AF).


  • From accelerometer-derived PA data in patients with AF, performing moderate-to-vigorous PA (MVPA) for more than approximately 105 min/week was associated with a lower risk of all-cause mortality and incident heart failure (HF). Adherence to PA guidelines (performing ≥150 min/week of (accelerometer-measrued) MVPA) was related to 30% and 33% lower risks of all-cause mortality and incident HF, respectively.


  • Patients with AF participate in 29.2 min/week less MVPA than the adults without prevalent AF. Our findings demonstrate that undertaking at least 105 min/week MVPA should be encouraged as a means to reduce morbidity and mortality in patients with AF as a part of integrated management.


Mounting evidence supports the benefit of regular physical activity (PA) for improvement of cardiovascular health and reduction in risk of all-cause mortality.1 2 Recent guidelines developed by WHO also recommend regular PA in adults living with chronic conditions or disabilities.1

Atrial fibrillation (AF) is a disease of ageing and is associated with high multimorbidities.3 Given its collateral health conditions, patients with AF are prone to impaired quality of life and worse clinical outcomes, including higher risks of stroke, heart failure (HF), and death, resulting in excess healthcare costs.4 5 Meanwhile, patients with AF engaged in less daily PA than adults without AF.6

For patients with AF, several studies supported the protective effect of PA on reducing symptoms, incidence, recurrence and burden of AF, and improving quality of life.7 8 However, relatively little is known about the potential associations between PA and risk of cardiovascular outcomes such as ischaemic stroke, HF, or mortality, specifically for patients with AF. As such, there are limited recommendations for the optimal intensity and duration of PA in patients with AF in the current European Society of Cardiology (ESC) and American Heart Association (AHA) guidelines of AF management,9 despite the clinical importance of PA as a part of holistic or integrated care in AF.10

Although we previously demonstrated that regular PA after AF diagnosis was related to lower risks of HF and mortality from a nationwide population cohort,11 data were derived from self-reported questionnaires, which might limit the accurate estimation of PA frequency and volume. Current PA guidelines are established based on the studies in which PA data are mainly reported from questionnaires rather than device-measured PA.2 12 Considering the limitation of self-reported data (ie, recall bias), objectively measured PA data—of which the validity has been supported by the strong correlations between wrist acceleration and PA energy expenditure or oxygen uptake in adults13 14—are needed to evaluate the association with cardiovascular health in patients with AF.

In this study, we analysed the PA data of patients with AF measured by a wrist-worn accelerometer from the UK Biobank. We investigated the association between PA, or more specifically, adherence to PA guidelines and the risk of death, incident HF, and incident stroke in patients with AF.


Study cohort

The UK Biobank is a large-scale cohort of 502 413 adults aged 40–69 years recruited from 22 assessment centres in the UK (response rate: 5.4%) in 2006–2010.15 16 With written informed consent, data of physical measurement, biosamples and detailed questionnaires about lifestyle were collected, and all participants were prospectively followed by linked health-related records including the National Health Service (NHS). The database secures the accuracy of the participants’ diagnosis and follow-up health information by the expert-led adjudication and ascertainment.15 17 It incorporates data from national death registries (including information on the date and causes of death), cancer registries, hospital inpatient records and primary care records which are identified based on the International Classification of Diseases (ICD)-9 and ICD-10 codes to produce reliable and valid information.15 16 The study is reported according to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guideline (STROBE checklist) (online supplemental file 2).

Supplemental material

Assessment of PA from the wrist-worn accelerometer

Among the total population, 236 519 UK Biobank participants were invited to wear a wrist-worn accelerometer for 1 week in February 2013–December 2015, of whom 106 053 (44.8%) agreed to participate and 103 695 (43.8%) submitted data.18 Individuals who accepted the invitation were found to have similar characteristics to those who declined it.18

The volume of PA was measured using Axivity AX3 (Newcastle upon Tyne, UK) wrist-worn triaxial accelerometer for a week. The sensor captured acceleration for 7 days at a frequency of 100 Hz with a dynamic range of ±8 g, and the signals were calibrated to gravity.18 The average vector magnitude was processed by combining sampling data into 5 s epochs, and overall mean acceleration represents the global PA.13 19 When all three axes had a SD <13.0 milligravity (mg) for≥60 min, it was considered as non-wear time.19

Inclusion and exclusion of the study population

This study includes participants diagnosed with AF before or at the time of accelerometer data collection. Only a subset of participants with AF who agreed and provided their accelerometer-measured PA data were included in the analysis. Patients with AF who had a prior diagnosis of stroke and HF were excluded. Also, we excluded those with insufficient wear time (wear time <72 hours)20 21 and poor calibration (an average vector magnitude acceleration >100 mg or <10 mg).20

Assessment of moderate-to-vigorous PA

The current literature lacks a defined cut-off for moderate-to-vigorous physical activity (MVPA) using wrist-worn accelerometers in adults. We determined a threshold that corresponds to the moderate-intensity PA from triaxial wrist-worn accelerometer data as the sum of 5 s epochs with mean acceleration ≥100 mg,14 22 based on the study of 30 adults evaluating raw triaxial accelerometer output to develop regression equations for estimating energy expenditure; the study found that moderate-intensity PA corresponds to the derived acceleration intensity thresholds of 93–101 mg.14 Indeed, several studies successfully computed the moderate PA with the threshold of 100 mg from the accelerometer raw data provided by UK Biobank and estimated the associations with cardiovascular outcomes.21 23 To avoid misclassifying the artefact as MVPA, we further identified MVPA in bouts (≥80% of the activity meets the MVPA threshold, ≥100 mg, for at least a bout of 5 min).21 24 Based on the computed MVPA levels, we determined the individual’s adherence to the PA guidelines according to the recommendations of the ESC,25 AHA26 and WHO1: MVPA level meeting the standard recommendation of ESC/AHA/WHO, ≥150 min/week; meeting the extended recommendation for additional health benefit, ≥300 min/week.

Covariates and study outcomes

Demographic data, including age, sex, ethnicity and other lifestyle behaviours (smoking and alcohol consumption) were collected based on self-reported assessment. The presence of comorbidities (hypertension, diabetes mellitus, dyslipidaemia and myocardial infarction) was identified based on the ICD-9 and ICD-10 coded registry of hospital or primary care records. The frailty of the individual was quantified as Charlson Comorbidity Index (CCI). The Townsend Deprivation Index (TDI) indicates the level of deprivation among individuals within a population and represents an overall socioeconomic status.

We defined all-cause mortality as a primary outcome. The secondary outcomes are cardiovascular mortality, incident HF and incident stroke that occurred during follow-up after the collection of the accelerometer data. Online supplemental table 1 provides a detailed explanation of the covariates and outcomes defined by operational definitions incorporating admission/outpatient use data and diagnostic/procedural codes from the UK Biobank.16 27

Supplemental material

Participants were followed up from the measurement of accelerometer data to 30 September 2021, or the occurrence of primary and secondary outcomes, whichever came first. We evaluated the risks of the primary and secondary outcomes according to the continuous variable of MVPA (min/week). Also, we divided the study population into two groups according to the adherence to the PA guidelines (those meeting ESC/AHA/WHO standard or extended guidelines and the others) and compared the risks of the primary and secondary outcomes.

Sensitivity and subgroup analyses

We performed sensitivity analyses by excluding the primary and secondary outcomes that occurred during the first 90 days of follow-up. We conducted additional sensitivity analyses since there has yet to be a consensus on the cut-off value to define MVPA using wrist-worn accelerometer data. These analyses involved using alternative cut-off points for MVPA, specifically setting thresholds at ≥125 mg and ≥150 mg. The aim was to replicate the observed associations between MVPA and the outcomes that demonstrated significant relationships in the initial analysis, where an MVPA cut-off of 100 mg was used.

Subgroup analyses were performed for the associations between the adherence to PA standard guidelines and the risks of primary and secondary outcomes: according to age (<65 vs ≥65 years), sex (male vs female), CCI (<3 vs ≥3), and congestive HF, hypertension, age ≥75 years (doubled), diabetes mellitus, stroke (doubled), vascular disease, age 65–74 years, sex category (female) (CHA2DS2-VASc) score (<2 vs ≥2).

Statistical analysis

Continuous variables are described as mean±SD or median with IQR, and categorical variables as numbers (%). Student’s t-test or Mann-Whitney test compared continuous variables of baseline characteristics between study groups and Pearson’s χ2 test or Fisher’s exact test for categorical variables as required. The Cox proportional hazards regression models evaluated the associations between MVPA and the risks of outcomes and represented HRs with 95% CIs. Model 1 showed an unadjusted risk, and Model 2 was adjusted for age and sex. Model 3 was adjusted for age, sex, body mass index (BMI), smoking status (current smoking), alcohol consumption (daily drinking), CCI and TDI. We analysed adjusted HRs (aHRs) of the primary outcomes (adjusted by Model 3) according to the continuous range of MVPA from 0 min/week to the maximum value and illustrated as restricted cubic splines (RCS). The aHRs were represented as a reference to the risk of MVPA 0 min/week. The comparison of the cumulative risks of primary and secondary outcomes between patients with AF who meet the ESC/AHA/WHO standard (MVPA ≥150 min/week) or extended guideline (≥300 min/week) and those who are below the recommended MVPA levels were shown as Kaplan-Meier survival curves with log-rank p values. All analyses were performed using Stata (V.17, StataCorp LLC, College Station, Texas, USA). A value of two-sided p<0.05 was considered statistically significant.

Equity, diversity and inclusion statement

The author group is gender balanced and consists of junior, mid-career and senior researchers; however, all members of the author group are in one discipline, cardiology. Our study population included both male and female patients with AF from different socioeconomic backgrounds, but the majority were white. The limited generalisability is considered in the discussion.


Baseline characteristics

Among 37 990 patients with AF in the UK Biobank, only 3002 participants had available accelerometer data after their AF diagnosis. Of them, we excluded those with insufficient wear time (n=189), poor data calibration (n=25), prior history of stroke (n=214) and prior history of HF (n=250). Finally, 2324 patients with AF who have reliable accelerometer data without stroke and HF were included in the analyses (figure 1). The differences in baseline characteristics of patients with AF between excluded (who do not have accelerometer data or have accelerometer data before their AF diagnosis) and included (who agreed and provided their accelerometer-measured PA data) participants are presented in online supplemental table 2. The mean age did not significantly differ between excluded and included patients with AF (67.2 years vs 67.1 years; p=0.74). Included patients were mostly male (67.4% vs 62.9%; p<0.001), had lower BMI (28.1 kg/m2 vs 29.1 kg/m2, p<0.001) and had a lower rate of smoking (10.7% vs 5.8%; p<0.001), but they presented a higher CHADS-VASc score (2.2 vs 2.4; p<0.001) and CCI (3.0 vs 3.5; p<0.001) than excluded patients.

Figure 1

Selection of the study population. AF, atrial fibrillation.

The baseline characteristics of the study population are described in table 1. The mean age was 66.9±6.2 years, and 64.9% (n=1508) were male. Most patients were white (98.5%), and the common comorbidities were hypertension (n=61.9%) and dyslipidaemia (47.4%). The mean CHADS-VASc score, CCI, and TDI were 2.2±1.1, 3.2±1.7, and −1.9±2.7, respectively.

Table 1

Baseline characteristics of the study population

Overall mean acceleration was 24.9±7.4 mg, and the MVPA duration was estimated as 105.8 (IQR: 37.4–229.4) min/week. Patients with AF who meet the PA standard guideline (n=896, 38.6%) were younger and more likely to be male (both p<0.001). They also had lower CHADS-VASc scores and CCI (p<0.001). There was a significant difference in the median MVPA duration between those who met the standard PA guideline and the others (280.4 vs 51.3 min/week, p<0.001). Less than one-fifth of the total population (n=415, 17.9%) was performing PA adhering to the PA extended guideline, and their MVPA duration was measured as 420.1 (IQR: 352.5–550.1) min/week (online supplemental table 3).

Mortality, HF and stroke according to MVPA

During a median follow-up of 6.7 years, there were 181 all-cause deaths, 62 cardiovascular deaths, 225 incident HF and 91 incident stroke; overall incidence rate per 1000 patient-year (PY) was 11.76, 4.03, 15.16 and 5.99, respectively. The continuous dose–response associations between accelerometer-derived MVPA and primary and secondary outcomes are represented in figure 2. The risks of all-cause mortality, cardiovascular mortality, incident HF and incident stroke are computed according to the duration of MVPA, using MVPA of 0 min/week as a reference point. After adjusting by model 3, these risks were plotted continuously as HRs along with their corresponding 95% CIs. A higher level of MVPA was consistently associated with a lower risk of all-cause mortality. Specifically, performing MVPA ≥108.4 min/week was significantly associated with a lower risk of all-cause mortality (figure 2A). Although any amount of MVPA was continuously associated with lower estimates of HRs for cardiovascular mortality, statistical significance was not observed for all ranges of MVPA (figure 2B). Of note, there was a specific volume of MVPA that is related to a lower risk of HF; the risk of HF was lower when MVPA was achieved between 105.0 and 590.2 min/week than the minimum amount of MVPA (0 min/week). The trend—lower aHR of HF according to the increment of MVPA—was maintained for MVPA ≥590.2 min/week, but no further significance was observed over this volume (figure 2C). The correlation between MVPA and the benefit on the risk of stroke was uncertain (figure 2D).

Figure 2

The risk of (A) all-cause mortality, (B) cardiovascular mortality, (C) heart failure and (D) stroke according to the accelerometer-derived moderate to vigorous physical activity. HRs were adjusted by age, sex, body mass index, current smoking, daily drinking, Charlson Comorbidity Index and Townsend Deprivation Index.

Adherence to guideline recommendations for PA and cardiovascular outcomes

Among total patients with AF, 38.6% (n=896) adhere to the ESC/AHA/WHO standard (MVPA ≥150 min/week) PA guideline and 17.9% (n=415) adhere to the extended (MVPA ≥300 min/week) PA guideline. The cumulative risks of primary and secondary outcomes in patients with AF stratified by adherence to PA standard guidelines are shown in figure 3.

Figure 3

Cumulative risks of (A) all-cause mortality, (B) cardiovascular mortality, (C) heart failure and (D) stroke, stratified by the adherence to PA standard guidelines according to the accelerometer-derived moderate to vigorous physical activity. HRs were adjusted by age, sex, body mass index, current smoking, daily drinking, Charlson Comorbidity Index and Townsend Deprivation Index. PA, physical activity

Patients with AF meeting the PA standard guidelines exhibited a higher survival rate (93.1% vs 86.8%, log-rank p=0.001) over the 8-year follow-up period following the accelerometer PA measurement. Also, they were associated with a 30% lower risk of all-cause mortality compared with those performing MVPA below the standard recommendations (aHR: 0.70 (0.50–0.98), p=0.039) (figure 3A).

A similar trend was confirmed for HF; those adhering to standard PA guidelines had a higher HF-free survival rate (92.5% vs 84.1%, log-rank p=0.015), and aHR was estimated as 0.67 (0.49–0.93), p=0.015 (figure 3C). There were no significant differences in the risk of cardiovascular mortality and stroke according to adherence to the standard PA guidelines (figure 3B, D).

Primary and secondary outcomes stratified by the MVPA duration of standard and extended cut-offs are described in online supplemental table 4. In line with figure 2 of dose–response relationships between the MVPA time and the outcomes, the risk of all-cause mortality was lower when higher MVPA was achieved, and the lower risk of HF attenuated above MVPA over 300 min/week. The risks of the outcomes according to the adherence to extended PA guidelines are shown in online supplemental figure 1, showing reduced benefits in all-cause mortality and HF compared with the analyses of standard recommendations.

Sensitivity and subgroup analyses

The sensitivity analysis performed by excluding the primary and secondary outcomes that occurred during the first 90 days of the follow-up is presented in online supplemental table 5. The lower risks of all-cause mortality and HF consistently remained in patients with AF who meet the standard PA guidelines. The additional sensitivity analysis of all-cause mortality and HF, which presented significantly lower risks associated with the increment of MVPA, using alternative cut-off points for MVPA (mean acceleration ≥125 mg and ≥150 mg) are presented in online supplemental figures 2 and 3. In line with the results using MVPA cut-off of mean acceleration ≥100 mg, a higher level of MVPA was consistently associated with a lower risk of all-cause mortality across the different MVPA cut-off points, namely ≥125 mg and ≥150 mg (online supplemental figure 2A, B). For HF, the same trend—a specific range of MVPA that is related to a lower risk of HF—was also observed for the different MVPA cut-offs; performing MVPA 33.0–554.7 min/week (MVPA cut-off ≥125 mg) or 37.4–152.9 min/week (MVPA cut-off ≥150 mg) is associated with a lower risk of HF (online supplemental figure 2C, D). According to the adherence to PA standard guidelines, the lower cumulative risks of all-cause mortality (by 34%–43%) and HF (by 34%–41%) were also maintained when analysed by the alternative MVPA cut-offs of 125 mg and 150 mg (online supplemental figure 3).

Several subgroup analyses performed by age (<65 vs ≥65 years), sex (male vs female), CCI (<3 vs ≥3) and CHADS-VASc score (<2 vs ≥2) revealed no interactions between groups for all primary and secondary outcomes, except for the accentuated lower risk of all-cause mortality for those with age <65 years and meeting standard PA guidelines (aHR: 0.271 for age <65 years vs 0.847 for age ≥65 years, p-for-interaction=0.018) (online supplemental table 6).


This is the first study using accelerometer-derived PA data to explore the association between MVPA and cardiovascular outcomes in patients with AF. Our principal findings are as follows: (1) patients with AF consistently showed a lower risk of all-cause mortality when MVPA was ≥108 min/week; (2) performing MVPA between 105 and 590 min/week was associated with lower risk of HF; and (3) patients with AF who adhered to ESC/AHA/WHO standard PA guideline (performing MVPA ≥150 min/week) experienced 30% and 33% lower risks of all-cause mortality and HF compared with those with PA below the recommendations. We provide evidence that PA should be encouraged for patients with AF, as in the general population. This observation is important since patients with AF tend to be older, multimorbid and frail, and therefore, may be wary of increasing their PA due to perceived risk of unfavourable episodes such as musculoskeletal injuries or sudden cardiac adverse events.28 29

Our study has several novel approaches over previous studies evaluating cardiovascular outcomes according to PA in AF. We analysed the PA intensity and volume based on an objective measurement using wrist-worn accelerometers, which captures actual PA undertaken more closely than self-report. Although the assessment of the PA via self-report has well-established strengths, it is subject to recall bias and correlates modestly with measured energy expenditure.30 31 Indeed, several studies noted that the quantification of PA using the self-report method can either underestimate or overestimate the strength of some relationships with risk factors.21 32 Therefore, evidence from device-measured data is prioritised, and this is in line with the recent expert consensus that standardisation of PA data collection, processing and analytical procedures is required.33 Given that most of the evidence on PA largely depends on self-reported questionnaires,11 31 34 35 a more objective method that accurately reflects the PA amount and the benefits of PA on cardiovascular outcomes is needed. Our findings from accelerometer-derived PA data demonstrate the potential benefit of PA in AF, enabling precise estimation of PA. In addition, we focused on MVPA, which is in line with the recent finding emphasising the particularly important role of moderately intense PA in reducing cardiovascular risk.2 23 Furthermore, we demonstrate the benefit of MVPA on cardiovascular morbidity and mortality in multifaceted analyses by analysing continuous dose–response associations which would have been only approximated if relying solely on self-reported data, and further, by categorising patients with AF according to PA guideline adherence.

Using a different methodology, we have now confirmed and strengthened the prior reports on PA and the health benefits in AF. The benefit of PA in AF on all-cause mortality has been reported in several studies, including the HUNT3 and EORP-AF registry studies.34 35 Our findings on RCS curves reaffirmed ‘any attempt of MVPA is better than no activity at all’,20 showing continuously lower estimates of HRs for all-cause mortality according to the increment of MVPA. This is also consistent with the contemporary emphasis on any short bouts of a moderate or relatively modest amount of vigorous PA linking to lower mortality and incident cardiovascular disease, calling for the promotion of a small amount of MVPA.23 36 Of note, we suggested a specific amount of MVPA (108 min/week) for the apparent benefit on all-cause mortality and there was no upper threshold limiting the benefit. A more accessible interpretation of our results that can be practically applied in the clinical setting is that the benefit on all-cause mortality can be achieved through the daily equivalent of 15 min of MVPA, such as brisk walking, aerobics or running for individuals with normal fitness levels. In our study, cardiovascular mortality accounts for 34.3% of all-cause mortality, a consistent proportion with previous findings, which range from one-third to nearly half.37 38 Although statistical significance between MVPA and decreased cardiovascular mortality was not secured, which could be partly explained by the small sample size of our cohort, we found that any amount of MVPA was continuously associated with lower estimates of HRs. While cardiovascular mortality was the predominant single disease entity explaining all-cause mortality, a significant majority of non-cardiovascular causes, such as cancer, infections or respiratory system diseases, also contributed to the overall mortality. The substantial benefit of PA on all-cause mortality might be attributed to the decreased risk of non-cardiovascular mortality as an outcome of the cumulative positive impact of PA on individuals' general well-being, such as enhancing immune function or reducing systemic inflammation.39

PA is known to have a protective effect in preventing HF and not only for healthy adults but also for patients with established HF.40 We confirmed MVPA is also related to a lower risk of HF for patients with AF in line with our previous report derived from a nationwide population-based cohort, denoting regular moderate exercise after AF diagnosis is associated with a 5%–8% lower risk of incident HF.11 Preventing HF in patients with AF is an important healthcare strategy as HF and AF are closely intertwined with each other, sharing common predisposing factors,41 and their coexistence is related to adverse prognoses such as a higher risk of all-cause mortality and hospitalisation rates.42 Indeed, a recent randomised clinical trial of HF with reduced ejection fraction and permanent AF suggested that PA training can improve exercise capacity and cardiac function.43 In the present study, we show that a wide range of MVPA (105–590 min/week) is associated with a lower risk of HF. While the high volume of MVPA could not show definitive benefits in HF, we expect the loss of statistical significance might originate from the small sample and event numbers at MVPA ≥590 min/week. The minimum threshold of HF benefits is probably similar to those of all-cause mortality, and it would be reasonable to recommend that patients with AF invest 15 min of daily MVPA for their general well-being.

While PA is associated with a lower risk of stroke in the general population,44 relatively little is known about whether PA is related to the decreased risk of stroke in AF, as well.34 45 Although there have been signals that PA could be related to a lower risk of stroke, statistical significance was not found in most studies.11 34 Conflicting results might come from the small number of events in each study requiring long-term follow-up. The overall stroke rate varies between 7.87 and 12.07 per 1000 PY in other studies,11 45 while in our research, it is even lower at 5.99 per 1000 PY, primarily because we included a younger cohort. Stroke in AF has an apparent culprit pathology—remodelling and reduced function of the left atrium and left atrial appendage46—and the risk is primarily modified by anticoagulation use.47 Hence, we hypothesise that the accompanying overall health benefit of PA may have less of an impact on stroke than the other clinical outcomes.

Clinical implications

AF is a chronic condition with multimorbidity,3 and data from electronic wearable activity trackers recently demonstrated that individuals with AF engage in significantly less daily PA.6 The lower PA levels in AF are also evident in our study. For example, the overall acceleration (global PA level) and median MVPA duration of patients with AF in our study were lower than the previously reported MVPA of 93 669 non-AF adults of UK Biobank applying the same processing methods of MVPA from the wrist-worn accelerometer; 24.9 vs 28.1 mg for overall acceleration and 106 vs 135 min/week.21 We confirmed that patients with AF participate in 29.2 min/week lower MVPA (ie, 22% lower MVPA) than the adults without prevalent AF, implying the chronic conditions hampering active movement in AF and a possible gap that could be improved for cardiovascular health.

Notably, updated WHO PA guidelines suggest that even adults with a chronic condition should perform at least 150–300 min of moderate intensity PA or 75–150 min of vigorous PA, or an equivalent combination. We show that adherence to PA guidelines would be related to a 30%–33% lower risk of all-cause mortality and HF. As AF is associated with a 3.5-fold increase in mortality48 and HF is one of the most common causes of death among AF (14.5%),49 our findings demonstrate that MVPA must be encouraged as a means to reduce AF-related morbidity and mortality. Potential risks associated with performing PA in patients with AF (ie, injuries or sudden cardiac adverse events) should be managed carefully by the assessment of individual fitness levels and related medical conditions.

There could be a concern that it might not be suitable to apply recommendations of PA guidelines, required minimum cut-offs of PA, which is almost entirely established based on self-reported data to categorise the accelerometer-derived PA of our cohort; the primary methods for the assessment of PA have been by a self-report format until recently,1 and the majority of the studies which established the recommendations of guidelines quantified PA using self-reported questionnaires. There lacks standardisation of PA data collection and the comparison of PA-associated outcomes between studies using device-based measures (ie, accelerometer) and questionnaire-based measures is a clinical challenge. Therefore, comparing the clinical outcomes of our cohort who have accelerometer-derived PA according to recommended values of PA amount on guidelines of which constructed based groundworks of self-reported PA data require careful interpretation. Nonetheless, we believe that our study adds value to the otherwise limited interpretation provided by self-reported data in other studies.

Finally, our observations on PA are aligned with the contemporary approach to AF management, which recommends a holistic or integrated care approach based on stroke prevention, symptom management with rate or rhythm control and comorbidities and lifestyle optimisation.50 Importantly, such an integrated care approach is associated with improved clinical outcomes in AF, and is currently recommended in guidelines.10


Our study has several limitations. First, the small numbers of the study cohort might not represent the overall general AF population. A selection bias might be introduced due to the very first age criteria (adults aged 40–69 years were recruited in the UK Biobank) and the modest rate of accelerometer data collection. Second, the majority of the patients with AF were white and male with greater socioeconomic accessibility to voluntarily engage in the UK Biobank study, and external validation in other groups will be required. Third, the processing of accelerometer data and a threshold of MVPA could vary across studies. Also, an accelerometer might underdetect stationary MVPA, such as cycling, resulting in misclassification bias. A wrist-worn accelerometer may not fully capture all MVPA, and translating the recorded acceleration values from other accelerometer placements (ie, hip) would require further validation. Fourth, there might be unmeasured confounding that cannot adjust more fragile and chronic patients with AF being less engaged in MVPA, thereby contributing to reverse causation. However, our maximal effort to adjust the HRs, including CCI and TDI, would minimise the residual confounding. Additionally, we validated our primary findings by excluding outcomes that occurred within the first 90 days of follow-up to reduce reverse causality. Fifth, categorising the accelerometer-derived PA amount of our study participants to define adherence to the PA guidelines might require cautious interpretation since cut-offs suggested by PA guidelines are based on self-reported data. A consistent effort to standardise the method of PA quantification and analytical strategy to incorporate the PA data collected in various ways would be required. Sixth, as data on the fitness level of the participants were not provided, a more personalised recommendation of MVPA based on their fitness levels would be needed. Lastly, the individuals’ longitudinal change in PA behaviour after the collection of the accelerometer data may not be taken into account in the association between PA and cardiovascular outcomes.


From the prospective analysis of patients with AF evaluating the associations between accelerometer-derived PA and cardiovascular outcomes, PA is associated with a lower risk of all-cause mortality and HF. There is an inverse dose–response relationship between MVPA and the risk of all-cause mortality and HF. In AF, adherence to standard PA guidelines (MVPA ≥150 min/week) is associated with a 30% and 33% lower risk of all-cause mortality and HF. Indeed, performing MVPA daily for at least 15 min is sufficient to achieve a significant health benefit. As patients with AF participate in lower PA than non-AF adults, this gap should be addressed by the encouragement of increasing PA levels, particularly moderate to vigorous intensity, as a part of integrated care AF management.

Data availability statement

Data are available upon reasonable request.

Ethics statements

Patient consent for publication

Ethics approval

The UK Biobank received ethical approval from the UK National Health Service’s National Research Ethics Service (ref 11/NW/0382). This study was conducted under UK Biobank application number 76593. Participants gave informed consent to participate in the study before taking part.


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.


  • Contributors H-JA had primary responsibility for writing the article. H-JA and T-MR conceived the idea and initiated the analysis plan. H-JA carried out data analyses for the current study. H-JA and T-MR contributed to the data collection and design of the work. E-KC supervised the current study. E-KC, JMC, K-YL, SK, S-RL, SO and GYHL reviewed the draft. All authors coordinated the study, provided comments on manuscript drafts and revised and approved the manuscript. E-KC is the guarantor of this work and takes responsibility for the integrity and accuracy of the data analysis. All authors read and approved the final manuscript.

  • Funding This research was supported by a grant from the Patient-Centered Clinical Research Coordinating Center (PACEN) funded by the Ministry of Health & Welfare, Republic of Korea (grant number: HC21C0028), and by the Korea Medical Device Development Fund grant funded by the Korea government (the Ministry of Science and ICT, the Ministry of Trade, Industry and Energy, the Ministry of Health & Welfare, the Ministry of Food and Drug Safety) (Project Number: HI20C1662, 1711138358, KMDF_PR_20200901_0173).

  • Competing interests E-KC: Research grants or speaking fees from Abbott, Bayer, BMS/Pfizer, Biosense Webster, Chong Kun Dang, Daewoong Pharmaceutical Co, Daiichi-Sankyo, DeepQure, Dreamtech Co, Ltd, Jeil Pharmaceutical Co Ltd, Medtronic, Samjinpharm, Seers Technology, and Skylabs. GYHL: Consultant and speaker for BMS/Pfizer, Boehringer Ingelheim, Anthos and Daiichi-Sankyo. No fees are received personally. GYHL is coprincipal investigator of the AFFIRMO project on multimorbidity in AF, which has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 899871.

  • Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting or dissemination plans of this research.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.