Physical Activity Frequency and the Risk of Stroke: A Nationwide Cohort Study in Korea.

BACKGROUND: The current guideline recommends moderate- to vigorous-intensity physical activity (PA) at least 40 min/day for 3 to 4 days/week. Although recent evidence has demonstrated that low-dose PA could reduce cardiovascular mortality, the relationship between low-dose PA and the risk of stroke remains uncertain. METHODS AND
RESULTS: Using data from a nation-wide sample cohort in Korea, we examined 336 326 individuals who received a general health examination between 2009 and 2010. Level of PA was assessed using a questionnaire for weekly PA frequencies regarding 3 intensity categories: light, moderate, and vigorous. Moderate- to vigorous-intensity PA (MVPA) was classified into 4 frequency categories: none, 1 to 2, 3 to 4, or ≥5 times/week. Cox proportional hazard models were constructed to estimate the risk of stroke. During the average follow-up of 3.6 years, 2213 stroke cases occurred. MVPA was none in 50%, 1 to 2 times/week in 20%, 3 to 4 times/week in 13%, and ≥5 times/week in 18% of the cohort. Individuals with MVPA 1 to 2 times/week had a 16% reduced risk of stroke (adjusted hazard ratio, 0.84; 95% confidence interval, 0.73-0.96) compared with those with no MVPA. The population attributable fraction of no MVPA was 12%, which was the second most important risk factor for a stroke after hypertension.
CONCLUSIONS: Even 1 to 2 times a week of MVPA might be beneficial to prevent a first-ever stroke in the general population, although a quantitative validation of the questionnaire is needed. From a public health perspective, we need to encourage inactive people to start exercising with more-achievable goals.

Clinical Perspective

What Is New?

Even 1 to 2 times a week of moderate‐ to vigorous‐intensity physical activity might be beneficial to prevent a first‐ever stroke in the general population.

What Are the Clinical Implications?

From a public health perspective, we need to encourage physically inactive people just to start any level of moderate‐ to vigorous‐intensity physical activity, considering that 50% of individuals in this cohort were not participating in moderate‐ to vigorous‐intensity physical activity.

Introduction

Physical activity (PA) is effective for preventing a stroke and reducing mortality.1, 2 However, approximately half of East Asians and 24% of US citizens still self‐reported as being physically inactive, suggesting that PA may not sufficiently work as a preventive strategy for strokes.3, 4 The reason many people are still physically inactive could simply be attributed to being unaware of the substantial benefit of PA. However, it is also possible that the recommended standards are too high for people to initiate or continue PA. The current stroke guidelines from the American Stroke Association recommend at least 40 min/day of moderate‐ to vigorous‐intensity aerobic PA 3 to 4 days/week.5 The World Health Organization also recommends 150 min/week of moderate‐intensity exercise or 75 min/week of vigorous‐intensity exercise.6 However, more than half the people in the United States and the United Kingdom and even up to 80% of all adults in Asian countries failed to comply with the guideline,7, 8, 9 implying it is not a practical or easily achievable goal for Western societies’ people. Recent studies showed a minimum of 90 min/week of moderate‐intensity exercise or 30 to 59 min/week of vigorous‐intensity exercise are markedly beneficial for all‐cause and cardiovascular mortality.3, 10 We hypothesized that PAs, even below the current recommended level, are effective for preventing a first‐ever stroke. To study this, we analyzed the database of general health examinations from a nation‐wide sample cohort in Korea.

Methods

National Health Insurance System‐National Sample Cohort of Korea

In 1989, universal healthcare coverage was enacted in Korea, and the national health insurance system, a single‐insurer system, began in 2000.11 By maintaining healthcare utilization and prescription data, the national health insurance system has created a sample cohort of the population that provides useful health insurance and health examination data to public health researchers and policy makers. Of the total eligible population, 2.2%, or a cohort of 1 025 340 participants, was randomly sampled from the 2002 Korean health insurance database to obtain baseline data. These cohort participants were followed for 11 years. During the follow‐up period, a representative sample of newborns (age 0) was added annually, and deceased or emigrated participants were excluded. After 11 years, in 2013, the database included 1 014 730 participants. Information regarding each participant's insurance eligibility, medical treatment history, healthcare provider's institution, and general health examination was included, because, in Korea, nation‐wide general health examinations are conducted biennially for citizens aged ≥40 years. An insured employee and a household member of the insured self‐employed can receive the examination regardless of his or her age.

Inclusion and Exclusion of Participants

From this cohort, we selected the participants who took the examination at least once between 2009 and 2010 because major changes were made in 2009 to the content of the examination and the questionnaire, especially related to PAs. Of the 352 896 individuals who received the general health examination between 2009 and 2010, we excluded participants who (1) had a previous stroke (N=5917), (2) had incomplete data on PA (N=5155), and (3) were disqualified because of death or emigration during the same year of the examination (N=5498). Finally, data for 336 326 participants (171 681 men and 164 645 women) were available for analyses. The present study was approved by the Institutional Review Board at the Seoul National University Hospital (Approval No. E‐1610‐092‐799) and national health insurance system review committee of research support (NHIS‐2017‐2‐337). Data were fully anonymized and de‐identified for the analysis. Written informed consent was waived.

Assessment of PA and Potential Covariates at the Baseline

Level of PA was assessed using a questionnaire (Figure S1) for frequencies of weekly PA in each of the 3 intensity categories: vigorous (≥20 min/day; eg, running); moderate (≥30 min/day; eg, brisk walking); or light (≥30 min/day; eg, walking). Vigorous‐intensity PA is defined as intense exercise that causes shortness of breath such as running, aerobic dancing, bicycling at high speeds, or uphill walking. Moderate‐intensity PA is defined as exercises that cause mild shortness of breath such as brisk walking, tennis (doubles), or bicycling at a usual speed. Light‐intensity PA is defined as walking at a usual pace. Frequencies of moderate‐ to vigorous‐intensity PA (MVPA) were further classified into 4 categories: none, 1 to 2 times, 3 to 4 times, or ≥5 times/week. To reduce the possible confounding effect, the frequencies of light‐intensity PA were adjusted in all multivariable regression models.

Age strata were defined by 19 age groups (Table S1; Figure S2). Income was categorized into 3 groups (lower 30%, middle 40%, and upper 30%) based on 10 strata. Cerebrovascular disease risk factors included the following: body mass index; smoking status (never versus past versus current use of cigarettes); alcohol use (heavy, >7 drinks/week for women, >14 per week for men; moderate, 1–7 for women, 1–14 for men; and none); diabetes mellitus (self‐reported history of diabetes mellitus or fasting glucose ≥126 mL/dL); hypertension (self‐reported history of hypertension, systolic blood pressure ≥140 mm Hg, or diastolic blood pressure ≥90 mm Hg); and hyperlipidemia (self‐reported history of hyperlipidemia or total cholesterol >240 mg/dL).

Follow‐up and Stroke Assessment

Participants were followed for incident stroke from the baseline examination through the date of hospitalization due to stroke or December 31, 2013, using claims data. The diagnosis of a stroke was coded by the physicians in charge during hospitalization upon discharge as indicated by the International Classification of Diseases, Tenth Revision codes I63.X for ischemic stroke and I61.X for intracerebral hemorrhage. Health Insurance Review & Assessment Service in Korea has been conducting quality assessment program dedicated for acute stroke, which reported 92.6% accuracy of coding for acute stroke diagnosis in 2010.12

Statistical Analyses

We compared baseline characteristics of the patients within each categories of MVPA based on the chi‐squared and 1‐way ANOVAs for categorical and continuous variables, respectively. The values are presented as the means±SDs or as percentages, as appropriate. Multivariable Cox proportional hazards models were used to determine hazard ratios (HRs) and 95% confidence intervals (CIs) of incident stroke (total stroke, ischemic stroke, and intracerebral hemorrhage) across PA categories (any/no MVPA as well as the 4 MVPA categories). In the analyses for each stroke subtype, individuals were censored when the corresponding type of stroke occurred. The model was adjusted for age, sex, income, smoking, alcohol consumption, body mass index, hypertension, diabetes mellitus, hyperlipidemia, and frequencies of light‐intensity PA. Population attributable fractions for physical inactivity and other stroke predictors (hypertension, diabetes mellitus, hypercholesterolemia, current smoker, and heavy drinker) determined by the baseline assessment were estimated by using adjusted HR and prevalence. We tested effect modification by each of the covariates on the associations between PA and incident stroke using interaction terms in the regression. Proportional hazard assumptions were examined and met by using scaled Schoenfeld residuals. Sensitivity analysis for residual confounding was performed using the R “obsSens” package, which evaluates the effect of hypothetical unmeasured confounding factor in the model.13 Significance levels were set at a P value of <0.05 for 2‐tailed tests. Statistical analyses were performed using R statistical software (R, version 3.3.0; R Project).

Results

Baseline characteristics according to MVPA are described in Table 1. More physically active participants tended to be men, younger, have higher incomes, more likely to smoke, use alcohol, and had a higher body mass index. Of the 1 198 033 person‐years of follow‐up (average, 3.6 years), 2213 incident stroke cases occurred (1866 ischemic strokes and 347 intracerebral hemorrhages).

Table 1

Baseline Characteristics According to the Frequencies of Moderate to Vigorous PA Per Week

	Frequencies of Moderate‐to‐Vigorous PA (Per Week)				P Value
	None	1 to 2	3 to 4	≥5
	n=166 681	n=65 813	n=44 915	n=58 917
Sex					<0.001
Male	44.4%	57.9%	57.8%	57.1%
Female	55.6%	42.1%	42.2%	42.9%
Age, y					<0.001
<40	25.4%	38.0%	29.8%	20.3%
40 to 49	24.8%	29.3%	29.1%	26.0%
50 to 59	22.4%	19.6%	23.1%	26.9%
60 to 69	16.1%	8.9%	12.6%	18.7%
≥70	11.2%	4.1%	5.4%	8.1%
Income					<0.001
Lower 30%	23.8%	18.8%	19.1%	20.4%
Mid 40%	40.8%	40.8%	37.4%	35.5%
Upper 30%	35.4%	40.4%	43.5%	44.1%
Smoking status					<0.001
Never	68.2%	56.8%	57.9%	60.0%
Former	9.9%	15.0%	17.0%	18.8%
Current	21.9%	28.2%	25.1%	21.2%
Alcohol drinking					<0.001
None	61.8%	43.5%	45.6%	50.5%
Moderate	20.4%	33.1%	30.8%	27.4%
Heavy	17.8%	23.4%	23.7%	22.2%
Body mass index, kg/m2	23.6±3.3	23.6±3.3	23.9±3.2	24.0±3.0	<0.001
<25.0	69.1%	68.7%	66.2%	65.6%
25 to 29.9	27.4%	27.8%	30.1%	31.0%
≥30.0	3.5%	3.5%	3.7%	3.4%
Frequency of light‐intensity PA, per week	2.0±2.5	2.5±2.1	3.1±2.1	4.3±2.2	<0.001
Systolic BP, mm Hg	122.6±15.6	121.3±14.5	122.2±14.6	123.6±14.9	<0.001
Diastolic BP, mm Hg	76.1±10.2	75.9±10.0	76.3±10.0	76.7±10.0	<0.001
Hypertension	26.2%	20.0%	23.6%	28.3%	<0.001
Fasting glucose, mg/dL	97.9±24.7	96.1±22.1	97.5±23.4	98.9±24.3	<0.001
Diabetes mellitus	9.2%	6.7%	8.4%	10.8%	<0.001
Total cholesterol, mg/dL	195.8±37.6	194.7±36.4	195.4±36.6	196.0±37.0	<0.001
Hypercholesterolemia	14.2%	12.6%	13.7%	14.5%	<0.001

Values are mean±SD or percentage of frequency. BP indicates blood pressure; PA, physical activity.

Any MVPA was associated with a lower risk of total stroke (adjusted HR, 0.78; 95% CI, 0.71–0.86) and ischemic stroke (HR, 0.77; 95% CI, 0.70–0.85), but it was not significantly associated with the risk of an intracerebral hemorrhage (HR, 0.84; 95% CI, 0.67–1.04; Figure 1). Dose‐response analysis showed even the individuals in the MVPA 1 to 2 times/week group had a 16% reduced risk of incident stroke (HR, 0.84; 95% CI, 0.73–0.96) compared with those in the no MVPA group. MVPAs 3 to 4 times/week and ≥5 times/week were also associated with a lower risk of stroke (HR, 0.79; 95% CI, 0.68–0.91 in 3–4 times/week; HR, 0.78; 95% CI, 0.70–0.89 in ≥5 times/week). Similar results were observed for ischemic stroke, but not for intracerebral hemorrhage (Figure 1; Table S2).

Figure 1

Frequency of moderate to vigorous PA and risk of incident stroke. CI indicates confidence interval; HR, hazard ratio; PA, physical activity.

No significant interactions were observed between MVPA (any/no or the 4 categories) and each of the confounders (Figure 2; Table S3). Sensitivity analysis to examine the trend of estimates of hazard on model with add on of an unmeasured confounder with relative hazard of 0.7 to 0.9 (Table S4). In most situations, patients who did MVPA had lower risk of stroke even if a favorable unmeasured confounder exists. Comparison of the effect of different International Classification of Diseases, Tenth Revision codes for defining outcome was provided in Table S5.

Figure 2

Forest plot of the effect of interaction between confounders and moderate‐to‐vigorous physical activity on risk of stroke. The reference group for all analyses is the group of not having moderate‐to‐vigorous physical activity. Please refer to full detail of the analysis in Table S3. CI indicates confidence interval; MVPA, moderate‐to‐physical activity.

We estimated population attributable fractions for physical inactivity and other stroke risk factors such as smoking, heavy drinking, hypertension, diabetes mellitus, and hypercholesterolemia. No MVPA was the second most important risk factor for a stroke after hypertension, which accounted for 12% of the total strokes and 13% of ischemic strokes found in the participants (Table 2).

Table 2

Population Attributable Fraction and Adjusted HRs of Physical Inactivity and Other Stroke Predictors

Stroke Predictor	Total Stroke		Ischemic Stroke		Intracerebral Hemorrhage
	Adjusted HR (95% CI)	PAFa	Adjusted HR (95% CI)	PAF	Adjusted HR (95% CI)	PAF
Moderate‐to‐vigorous PA: none (vs any moderate‐to‐vigorous PA)	1.24 (1.14–1.36)	12%	1.26 (1.14–1.39)	13%	1.18 (0.94–1.47)	9%
Current smoker (vs nonsmoker)	1.53 (1.36–1.73)	9%	1.61 (1.41–1.83)	9%	1.13 (0.83–1.55)	3%
Heavy drinker (vs nondrinker)	1.11 (0.98–1.26)	2%	1.02 (0.89–1.18)	0%	1.76 (1.32–2.36)	12%
Hypertension	1.63 (1.49–1.78)	22%	1.53 (1.39–1.69)	20%	2.29 (1.82–2.90)	33%
Diabetes mellitus	1.52 (1.38–1.68)	8%	1.68 (1.52–1.87)	11%	0.85 (0.62–1.14)	−3%
Hypercholesterolemia	1.05 (0.94–1.17)	1%	1.04 (0.93–1.18)	1%	1.04 (0.78–1.37)	1%

Each model was adjusted for baseline age group, sex, income, body mass index, frequency of light‐intensity PA, and all other stroke predictors in the table. CI indicates confidence interval; HR, hazard ratio; PA, physical activity; PAF, population attributable fraction.

PAF was calculated as Pc (1−1/HRadj). Pc (ordered as listed in the table), the prevalence of the stroke predictor among stroke cases, was 62.1, 24.9, 18.8, 57.7, 24.2, and 17.8 for total stroke; 63.1, 25.1, 17.4, 57.4, 26.2, and 17.7 for ischemic stroke; and 57.3, 24.0, 27.4, 59.0, 14.4, and 17.5 for intracerebral hemorrhage.

Discussion

In this nation‐wide and standardized Korean cohort, we found that MVPA is associated with an ≈20% lower risk of stroke compared with physical inactivity, which has a similar magnitude of effect compared with previous studies.14 Interestingly, our study demonstrated, for the first time, that a level of MVPA below the recommendation (1–2 times/week) is also protective against a stroke, independent of other stroke risk factors.

The strengths of this study include the following: a very large sample size from a representative nation‐wide cohort; the controlling of potential confounding factors, including dose of light‐intensity PA; and the comparability with the current stroke guidelines in terms of PA dose. However, several limitations are also noted. Because our questionnaire is composed of a minimal duration of PA for each intensity and weekly frequencies (eg, How many times a week do you participate in ≥30 minutes of moderate‐intensity PA?), we could not calculate the exact metabolic equivalent times for additional analysis. Thus, someone in the low‐frequency group may actually have performed the recommended amount of PA. Validation of our questionnaire against a more‐quantitative method is needed to confirm the protective effect of low‐dose MVPA for stroke prevention. Changing behavior during long‐term periods could not be assessed in this study because of the relatively short‐term study period. Diet, a possible important confounding factor, was not considered in the model because the questionnaire did not include it. Because the outcomes were defined with only hospitalized stroke with the administrative health data, nonhospitalized stroke, such as severe stroke with death at home, may be underestimated.

In the previous reports which investigated the dose‐response relationship between MVPA frequency and risk of stroke, the multivariable analysis adjusted with risk factors was inconclusive, and a different cutoff from the current stroke guidelines made it difficult to compare with. The current stroke guidelines recommend MVPA at least 40 min/day for 3 to 4 days/week for primary stroke prevention.5 We found that 31% of our cohort met the recommended level of PA and had approximately a 20% lower risk of stroke. The World Health Organization recommends ≥150 min/week of moderate‐intensity or 75 min/week of vigorous‐intensity aerobic physical activity or an equivalent combination.6 In our cohort, only ≈18% of participants met this guideline and had a 21% reduction of risk. Similarly, ≈80% of the adult population of East Asian countries, such as Taiwan, China, or Japan, did not meet this recommendation.3 The far more serious problem is that currently over half of our cohort did not participate in any level of MVPA and were in a physically inactive state. In this study, we found reduction of stroke risk with even 1 to 2 times/week of MVPA, which is slightly less than the recommendation but significant. Although there was a statistically significant trend for dose‐response relationship, the difference of point estimates was modest between each group. This finding has importance from a clinical and public health perspective, because time is one of the larger hurdles when physically inactive modern people begin to exercise.15, 16 There is an opinion that the current recommendation may not be perceived as a practical, achievable, and sustainable goal for physically inactive people.17 If all physically inactive people start to perform any level of MVPA, in this population, 12% of total strokes and 13% of ischemic strokes would be prevented based on the estimation of population attributable fractions.

Although the efficacy of PA for preventing a stroke is supported by many studies, there were still some discrepancies found in previous evidence. It may be necessary to confirm intensity and volume of PA, the different effects by sex and ethnic groups, and the different effects on ischemic stroke and intracerebral hemorrhage. Overall, our study showed that MVPA is effective for preventing a stroke even after adjusting traditional risk factors, and MVPA is effective in most subgroups except for the current smoker and low‐income‐status groups. This finding supports the results of previous studies that it is suitable to recommend moderate‐ to vigorous‐intensity PA rather than light‐intensity PA in the guideline for preventing a stroke.18 There were disparities of effects according to sex with some studies, suggesting that PA is only effective in men and others only in women.1 However, our study from representative population‐based cohort data showed no significant effect modification by sex, which is consistent with the recent meta‐analysis.19 Another interesting point was that participants with older age or more vascular risk factors performed the highest frequency of MVPA, which is different from the previous results of Western countries.20, 21 Although MVPA had a similar degree of effect on the risk of an ischemic stroke and an intracerebral hemorrhage (20% versus 15% lower risk, respectively), the association was significant in an ischemic stroke, but not in an intracerebral hemorrhage. Hypertension and heavy drinking was significantly associated with an increased risk of an intracerebral hemorrhage with population attributable fractions of 33% and 12%, respectively. Whether the benefit of MVPA extends to intracerebral hemorrhage is less certain, possibly attributed to true biological difference compared to ischemic stroke or attributed to insufficient power to detect a significant difference. A study from China showed a similar magnitude of effect in intracerebral hemorrhages, but no statistical significance. Thus, for the association of PA and risk of intracerebral hemorrhage, data are needed from a longer follow‐up period or a pooled analysis of recent studies.

The minimal goals such as 1 to 2 times/week or 10 minutes daily of MVPA may be perceived easier to start and sustain exercise, because lack of time or low self‐efficacy are important barriers to PA.22 Furthermore, once an individual regularly completes the minimum exercise goals, they may be more likely to increase their amount of exercise to achieve more benefits. Recent data showed that a minimal amount of MVPA might be effective to reduce all‐cause and cardiovascular mortality. Based on these results, the Japanese government is encouraging people to conduct “+10 min/day of moderate to vigorous PA” as a minimal starting dose of PA to encourage both sedentary and active individuals to become progressively more active.23 We agree that “Let's start with +10 min/day of activity” is an appropriate and more‐attainable public health message, although the effectiveness of which needs to be validated.

In conclusion, even 1 to 2 times a week of MVPA might be beneficial to prevent a first‐ever stroke in the general population compared with physical inactivity. From a public health perspective, we need to encourage physically inactive people just to start any level of MVPA, considering that 50% of individuals in this cohort were not participating in MVPA. Future research is needed to validate how effectively such a policy would work to compel more people to start increasing their PA and thus prevent stroke.

Sources of Funding

This research was supported by a grant of the Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health & Welfare, Republic of Korea (Grant No.: HI17C0076), and also supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT & Future Planning (NRF‐2015R1A2A2A01007770).

Disclosures

None.

Figure S1. Questionnaire used in general health examination in Korea (English version).

Figure S2. Histogram of the number of study participants within age strata.

Table S1. Age Strata and Frequencies of the Participants in the Analyzed Data Set

Table S2. Full Multivariable Models for Dose‐Response Analysis

Table S3. Analysis of the Effect of Interaction Between Confounders and Moderate‐to‐Vigorous Physical Activity on Risks of Stroke (Total Stroke, Ischemic Stroke, and Intracerebral Hemorrhage)

Table S4. Simple Bias Analysis for Unmeasured Residual Confounders

Table S5. Comparison of Effects of Different ICD‐10 Codes for Defining Acute Stroke

Click here for additional data file.