Abstract
INTRODUCTION
There is a lack of studies on the association between specific physical activity (PA) types and dementia. We examined the association of leisure time physical activity (LTPA), occupational physical activity (OPA), and sedentary lifestyle with dementia risk and cognitive decline using the UK‐Biobank study.
METHODS
Baseline PA was collected using questionnaires. A total of 502,481 dementia‐free participants were recruited in 2006–2010 and followed for 10 years until the end of 2020 for the ascertainment of dementia. Associations of PA with incident dementia and cognitive decline were examined.
RESULTS
Higher levels of LTPA and OPA and lower levels of sedentary hours were associated with lower dementia risk. The fifth quintiles of LTPA (hazard ratio [HR] = 0.53, 95% confidence interval [CI]: 0.43‐0.67) and OPA (HR = 0.68, 95% CI:0.51‐0.90) had lower dementia risk, whereas the fifth quintile of sedentary lifestyle had higher dementia risk (HR = 1.23, 95% CI:1.08‐1.41).
DISCUSSION
Our findings suggest the promotion of an active lifestyle suggested to be preventive of dementia risk. This research has been conducted using the UK Biobank Resource under Application Number 71022.
Keywords: cognition, dementia, lifestyle, physical activity
1. INTRODUCTION
With the world population ageing rapidly, the number of dementia cases is expected to rise dramatically. 1 The 2020 Report of the Lancet Commission identified 12 modifiable risk factors of dementia across the life span, including lower education, smoking, and physical inactivity, and it is estimated that 40% of all dementia cases can be attributed to these risk factors. 2 Among the modifiable factors, physical inactivity is an important contributor to the risk of dementia 3 .
Several studies have shown that physical activity (PA) has a protective effect against dementia. 4 , 5 , 6 , 7 , 8 , 9 , 10 A meta‐analysis of 58 longitudinal studies found a significant association between high PA and lower risk of dementia, and this association remained significant after excluding studies with follow‐up periods < 20 years or those of low to moderate quality. 11 A recent UK‐Biobank study has reported that engaging in vigorous and other exercises (such as swimming and cycling) was significantly associated with lower risk of dementia. 9 , 12 In addition, an international consortium study involving multiple populations worldwide, reported a significant dose‐response relationship of higher duration and amount of PA with lower risk of dementia 10 .
Studies have also investigated the association between leisure time physical activity (LTPA) and the risk of dementia. 13 , 14 , 15 , 16 Most of the studies have shown that engaging in LTPA is associated with a significantly lower risk of dementia, and a meta‐regression found a significant linear relationship of higher LTPA with lower risk of dementia. 17 However, there have only been a few studies examining the association between occupational physical activity (OPA) and incident dementia. A prospective Danish cohort study involving approximately 5000 middle‐aged participants followed for 30 years found that participants with higher OPA had a significantly increased risk of dementia compared to those with sedentary jobs. 15 However, the association between LTPA and dementia risk was not significant in this study.
There have been several studies reporting significant associations between higher PA and a lower risk of cognitive decline. 18 , 19 One of the largest studies was a UK cohort of approximately 10,000 participants, which found that the low PA group performed worse in memory, fluid intelligence, and verbal fluency, and these relationships were significant at baseline and follow‐ups, except for a non‐significant difference in memory at baseline. 20 In a Japanese study, higher quartiles of light‐intensity PA were associated with significantly lower odds of decline in the Mini‐Mental Status Examination (MMSE) score. 21 Similarly, physically active participants had significantly higher MMSE score after 4 years of follow‐up in a Korean study. By contrast, a study conducted in Mexico showed no significant associations with MMSE score after 3 years of follow‐up 22 .
Thus, the available data on specific PA domains, particularly from occupation and sedentary lifestyle, are limited. This study aims to examine the association between specific types of PA and cognitive health in a large sample size. We hypothesize that higher levels of PA, including LTPA and OPA, as well as lower levels of a sedentary lifestyle, are associated with a reduced risk of dementia and slower cognitive decline. We investigate these associations using the UK‐Biobank cohort, which provides a unique opportunity to study a diverse and sizable population.
RESEARCH IN CONTEXT
Systematic review: There is a growing consensus that certain lifestyle and psychosocial factors can contribute to cognitive impairment and dementia. However, little is known about the relative contribution of specific physical activity (PA) domains, particularly from occupation and sedentary lifestyle on risk of dementia and cognitive decline. We systematically examined the association of leisure time physical activity (LTPA), occupational physical activity (OPA) and sedentary lifestyle with the risk of dementia and cognitive decline in the UK‐Biobank study.
Interpretation: Our findings confirmed that the higher LTPA and OPA, and lower sedentary hours were associated with reduced risk of dementia at follow‐up and can potentially be used to support better targeting of modifiable risk factors of dementia.
Future direction: Promotion of active lifestyle is needed to better preserve cognition among elderly population. Future work is needed to investigate if PA as effective intervention can be used in the primary prevention of dementia.
2. METHODS
2.1. Study population
This study was based on the UK Biobank, a large‐scale population‐based prospective cohort study. 23 Between 2006 and 2010, over 500,000 participants aged between 40 and 69 from the United Kingdom were enrolled in the cohort and followed up at three time points. The study was conducted at 22 assessment centers located in both rural and urban areas in the United Kingdom . At the baseline survey, participants completed a PA questionnaire and underwent cognitive performance assessments. Health records were retrieved through linkage to inpatient records. The UK‐Biobank study was approved by the North West Multi‐Center Research Ethics Committee (MREC, https://www.ukbiobank.ac.uk/learn‐more‐about‐uk‐biobank/about‐us/ethics).
A total of 502,481 participants were recruited into the UK‐Biobank cohort between 2006 and 2010 (Figure 1). Participants were excluded from the analysis if they withdrew consent (n = 67), had prevalent dementia at baseline (n = 120), had missing age information or were below 40 years of age (n = 4), had missing PA measurements (n = 215,215), or had missing information on any of the covariates (n = 85,451), resulting in a sample size of 201,642 participants for the analysis of incident dementia. Of these, a sub‐group analysis was conducted to examine the association between PA and cognitive decline using 1,671 participants with cognition assessments at both baseline and second follow‐up,
FIGURE 1.
Flow chart of the study population *Covariates: age, sex, qualification, household income, loneliness, Townsend Index, parental dementia, alcohol drinker, ever smoked, sleep duration, BMI, hearing problem, hypertension, diabetes, and IPAQ level. **Domains: reaction time, visual memory, numeric memory, prospective memory, reasoning
2.2. Physical Activity
Participants’ PA was collected at baseline, which included LTPA, OPA, and sedentary lifestyle. Participants were asked about their participation in five leisure time activities in the last 4 weeks, which were walking for pleasure, strenuous sports, other exercises like swimming and cycling, light do‐it‐yourself (DIY), and heavy DIY. Each LTPA amount was calculated as the product of frequency, duration, and metabolic equivalent of task (MET) value (with MET values of 3.3, 8.0, 4.5, 2.25 and 4.5 assigned to walking for pleasure, strenuous sports, other exercises, light DIY, and heavy DIY respectively) 24 and the total LTPA was calculated as the sum of individual amounts. The frequency (once in the last 4 weeks, 2‐3 times in the last 4 weeks, once a week, 2‐3 times a week, 4‐5 times a week or every day) and duration (< 15 minutes, between 15 and 30 minutes, between 30 minutes and 1 hour, between 1 and 1.5 hours, between 1.5 and 2 hours, between 2 and 3 hours, or > 3 hours) of LTPA were recorded, and each category was assigned to specific value to facilitate calculation (see Table S1).
The study collected information on the working status of participants. Those who were in paid employment or self‐employed were asked about the frequency of their job involving heavy manual or PA and, mainly walking or standing. Four levels of frequency (0%, 25%, 75%, and 100%) were assigned to the responses of never, sometimes, usually, and always respectively. The MET values of 4.25 and 2.25 were assigned to the two categories of heavy manual or PA and mainly walking or standing respectively. The total OPA level was calculated as the sum of the product of the weekly working hours, frequency, and MET value. People who were not working were assigned a value of 0 in the OPA calculation.
The sedentary lifestyle of participants was measured by calculating the daily hours spent watching TV, using a computer (non‐occupational), and driving over the last 4 weeks. The daily sedentary hours were calculated as the sum of these activities, with any time exceeding 10 hours being truncated to 10 hours 24 .
The levels of LTPA, OPA, and sedentary hours were categorized into quintiles. In cases where > 5% of the participants had a value of zero, we followed a specific approach for categorization. Only the participants with non‐zero values were categorized into quintiles, while those with a value of zero were grouped together to create an additional category. This additional category served as the reference level for comparison. The specific value ranges for each level of the categorized variable can be found in Table S2.
2.3. Cognitive function assessment
At both baseline and second follow‐up, cognitive function assessments were administered on a computer, which included reaction time, visual memory, numeric memory, prospective memory, and reasoning. The reaction time was a continuous measurement, while prospective memory was a binary measurement (right or wrong). The other three measurements were count variables, and a higher value indicated better performance. The details of the assessments have been described elsewhere. 25
The data collected from the five domains of cognitive assessments at baseline were subjected to dimension reduction to generate a summary score for each participant's overall cognitive performance. Since prospective memory was a binary measurement, the Factor Analysis of Mixed Data (FAMD) provided in an R package called FactoMineR was used. 26 The reaction time and visual memory score were log‐transformed due to their right skewed distribution. The first dimension (31.8% variance explained) was used as the summary score of cognitive performance. A higher value of this score was correlated with better results in the cognitive assessments. The same transformation matrix was applied to cognitive assessments at the second follow‐up, and cognitive decline was calculated as the cognitive score at second follow‐up minus that at baseline, so a negative value indicates faster cognitive decline over time.
2.4. Incident dementia and censoring
Inpatient and outpatient records were periodically retrieved through linkage and included diagnoses and corresponding dates for diseases coded using the International Classification of Diseases, 9th Revision (ICD‐09) or 10th Revision (ICD‐10). The diagnosis of all‐cause dementia and subtypes (vascular dementia and Alzheimer's disease) were determined as described in a previous study (Table S3), 9 but no dementia diagnosis under ICD‐09 was identified. The duration from baseline to diagnosis was calculated in days. For participants who were either: (i) dead before diagnosis of all‐cause dementia (or its subtype), or (ii) alive at the end of study but not diagnosed with all‐cause dementia (or its subtype), were treated as censoring observations. The duration of the study period from baseline to either event (death or dementia) or end of study was calculated in days. The date of death was recorded, with the cause of death confirmed using death certificates from either the National Health Service (NHS) Information Centre or NHS Central Register, depending on the administrative area of the participants.
2.5. Covariates
The participants reported their demographic information, which included sex, date of birth, educational qualification, and household income. The age at baseline was calculated as the date of recruitment minus the date of birth expressed in years. The Townsend deprivation index was assigned to each participant based on their postcode and was determined using UK national census output area data.
The participants reported their alcohol consumption status as never, previous, or current drinker. In this analysis, ‘never drinkers’ were merged with ‘previous drinkers’. Similarly, the participants reported their smoking status as never, previous, or current smoker with ‘previous smokers’ and ‘current smokers’ merged together to form ‘ever smokers’. Sleep duration was collected as the number of hours they slept in a 24‐hour time period in the last 4 weeks, which was then classified into low (< 6 hours), optimal (6–8 hours) and high (> 8 hours) . 27 Participants were classified as isolated if they responded ‘yes’ to the question, “Do you often feel lonely?”.
The body mass index (BMI) was calculated using the weight (kg) and height (m) measured at the first visit to the assessment center. The BMI was further classified into underweight (< 18.50), normal weight (18.50‐24.99), overweight (25.00‐29.99), and obese (≥30) according to WHO criteria. 28 The participants reported their family's medical history, and a family history of dementia was defined as having at least one biological parent suffering from Alzheimer's disease or dementia.
The Modified International Physical Activity Questionnaire (IPAQ) was administered to measure frequency and duration of walking, as well as moderate and vigorous activity in the last 7 days. 29 Each individual activity level was calculated as the product of frequency, duration, and MET value. The MET values of 3.3, 4.0, and 8.0 were respectively assigned to walking, moderate and vigorous activity. 30 The total PA level MET value was defined as the sum of all three types, and the activity group (low, moderate, and high) was determined according to the IPAQ Guidelines 30 .
The presence of diabetes and hypertension at baseline were determined by medical records. The codes used to identify diabetes included ICD‐9: 250, 6480; and ICD‐10: E10, E11, E12, E13, E14, O24. 31 The codes for hypertension included ICD‐9: 401, 402, 403, 404, 405; and ICD‐10: I10, I11, I12, I13, I15, O10. 32 Participants also reported their hearing status, with responses of “No”, “Yes”, and “Completely deaf”. The last two responses were merged into one category of “Yes or Completely deaf”.
2.6. Statistical analysis
Cox regression with hazard ratios (HRs) and 95% confidence intervals (CIs) was used to examine the association of PA with incident dementia. The demographics (age at baseline, sex, education qualification, household income, and Townsend deprivation index), lifestyle (alcohol consumption status, smoking status, level of sleep duration, loneliness, and IPAQ group), anthropometrics (BMI), and health conditions (presence of diabetes, hypertension, and hearing problem) and family history of dementia were considered as potential covariates in the analysis. From the list of potential covariates, backward stepwise variable selection was used to determine the covariates to include into the model for adjustment by identifying the model with the smallest Akaike Information Criterion (AIC) value, and individual physical activities were then included in three separate models (Model 1). Different types of PA were also combined into one model to investigate potential confounding effects among them (Model 2). To address the non‐proportionality of hazard for loneliness and sleep duration, these covariates were specified as the strata in the stratified Cox regression model. Multiple linear regressions were applied to examine the association of PA with cognitive decline, and the regression coefficient was reported with its 95% CI and adjustment for covariates was determined using the same variable selection procedure as described above. The linear trend was also examined by treating PA levels as continuous. All the statistical analyses were conducted using R Statistical Software 4.1.2 (R Foundation for Statistical Computing, Vienna, Austria).
3. RESULTS
The mean (SD) age of participants included in the dementia analysis was 56.1 (8.1) years, and 95,919 (47.6%) were female (Table 1). Over half of the study participants had an education qualification at college or A/AS level or equivalent (55.3%), while 17.4% and 25.0% came from low and low‐middle income households respectively. The mean (SD) of Townsend deprivation index was −1.5 (3.0) and 17.5% reported feeling lonely. Most participants were current alcohol drinkers (93.5%) and approximately half (55.0%) were current or previous smokers. 88.3% reported optimal sleeping length of 6‐8 hours. The proportion of obesity was 24.2and 13.2% participants had a family history of dementia, 25.0% had a hearing problem, 6.8% had hypertension and 1.6% had diabetes. Participants in the cognitive decline analysis had higher proportions of college education (55.8%) and family history of dementia (26.3%) but lower proportions of low household income (9.5%) and obesity (19.7%).
TABLE 1.
Characteristics of participants included in analyses for dementia and cognitive decline
| Characteristics | Dementia analysis (N = 201,624) | Cognitive decline analysis (N = 1671) |
|---|---|---|
| Age, mean (SD), yr | 56.1 (8.1) | 55.2 (7.6) |
| Female, n (%) | 95 919 (47.6) | 740 (44.3) |
| Education qualification | ||
| College | 84 189 (41.8) | 932 (55.8) |
| A/AS level or equivalent | 27 209 (13.5) | 210 (12.6) |
| O level/GCSEs or equivalent | 50 023 (24.8) | 326 (19.5) |
| CSE or equivalent | 12 448 (6.2) | 70 (4.2) |
| NVQ/HND/NHC or equivalent | 16 386 (8.1) | 85 (5.1) |
| Others | 11 369 (5.6) | 48 (2.9) |
| Household income, n (%) | ||
| Low | 35 014 (17.4) | 158 (9.5) |
| Low‐Mid | 50 335 (25.0) | 350 (20.9) |
| Mid | 57 302 (28.4) | 543 (32.5) |
| Mid‐High | 46 547 (23.1) | 483 (28.9) |
| High | 12 426 (6.2) | 137 (8.2) |
| Townsend Index, mean (SD) | −1.5 (3.0) | −1.9 (2.4) |
| Loneliness, n (%) | 35 299 (17.5) | 239 (14.3) |
| Current alcohol drinker, n (%) | 188 441 (93.5) | 1596 (95.5) |
| Ever smoked, n (%) | 110 861 (55.0) | 985 (58.9) |
| Sleep duration | ||
| Less than 6 hr | 9 611 (4.8) | 69 (4.1) |
| 6–8 hr | 178 052 (88.3) | 1498 (89.6) |
| More than 8 hr | 13 961 (6.9) | 104 (6.2) |
| BMI category, n (%) | ||
| Underweight | 896 (0.4) | 3 (0.2) |
| Normal weight | 64 066 (31.8) | 568 (34.0) |
| Overweight | 87 950 (43.6) | 770 (46.1) |
| Obesity | 48 712 (24.2) | 330 (19.7) |
| Parental dementia, n (%) | 26 575 (13.2) | 440 (26.3) |
| Hearing problem, n(%) | 50 362 (25.0) | 416 (24.9) |
| Hypertension, n(%) | 13 744 (6.8) | 89 (5.3) |
| Diabetes, n(%) | 3 164 (1.6) | 13 (0.8) |
Abbreviations: AS, Advanced Subsidiary, BMI, body mass index; CSE, certificate of secondary education; GCSE, general certificate of secondary education; HNC, higher national certificate; HND, higher national diploma.
A total of 1,841 (0.91%) incident dementia cases were identified since recruitment and before the end of the study date, December 31, 2020. The total follow‐up duration was 2,335,381 person‐years, with a median (first quartile, third quartile) of 11.8 (11.1, 12.5) years. Among all‐cause dementia cases, 391 (21.2%) were first diagnosed with vascular dementia and 718 (39.0%) were first diagnosed with Alzheimer's disease. The rest (n = 732) were other causes of dementia.
Association of PA with incident dementia is shown in Table 2. LTPA had a significant linear trend with dementia risk (p < 0.001). Compared to zero LTPA, the fifth quintile of LTPA had a significant reduction in the risk of dementia (HR 0.53, 95% CI 0.43, 0.67). Linear association between OPA and dementia risk was also significant (p < 0.001) with the fifth quintile of OPA having a significant reduction in hazard when compared to zero OPA (HR 0.68, 95% CI 0.51, 0.90). Similarly, the fifth quintile of sedentary hours had significantly higher dementia risk (HR 1.23, 95% CI 1.08, 1.41) than the first quintile, with significant linear trend between sedentary lifestyle and dementia risk (p < 0.001). Similar trends of association were observed for vascular dementia and Alzheimer's disease (Table 3 ). Although most of the associations were non‐significant, third (HR 0.60, 95% CI 0.37, 0.97) and fourth (HR 0.61, 95% CI 0.38, 0.99) quintiles of LTPA and second quintile of OPA (HR 0.39, 95% CI 0.16, 0.96) were associated with significantly lower risk of vascular dementia compared with zero level, and second quintile of sedentary lifestyle was associated with significantly lower risk of Alzheimer's disease than first quintile (HR 0.77, 95% CI 0.62, 0.97). Furthermore, we performed an additional Cox regression analysis specifically targeting dementia of other causes. As shown in Table S4, higher levels of LTPA and OPA were associated with a reduced risk of dementia of other causes.
TABLE 2.
Association of physical activity with dementia
| Model 1 | Model 2 | |||||||
|---|---|---|---|---|---|---|---|---|
| Exposure a | Levels | Hazard ratio (95% CI) separate model | p Value | p for trend | Hazard ratio (95% CI) common model | p Value | p for trend | Category size |
| Leisure time physical activity | 0 | Ref. | <0.001 | Ref. | <0.001 | 10847 | ||
| 1 | 0.70 (0.58, 0.85) | <0.001 | 0.72 (0.59, 0.87) | <0.001 | 38351 | |||
| 2 | 0.66 (0.54, 0.81) | <0.001 | 0.67 (0.55, 0.82) | <0.001 | 39029 | |||
| 3 | 0.55 (0.45, 0.69) | <0.001 | 0.55 (0.45, 0.68) | <0.001 | 37296 | |||
| 4 | 0.58 (0.47, 0.72) | <0.001 | 0.57 (0.46, 0.71) | <0.001 | 38103 | |||
| 5 | 0.55 (0.44, 0.68) | <0.001 | 0.53 (0.43, 0.67) | <0.001 | 37998 | |||
| Occupational physical activity | 0 | Ref. | <0.001 | Ref. | <0.001 | 120980 | ||
| 1 | 0.98 (0.80, 1.20) | 0.83 | 0.98 (0.80, 1.20) | 0.82 | 17285 | |||
| 2 | 0.76 (0.57, 1.02) | 0.063 | 0.76 (0.57, 1.01) | 0.062 | 15187 | |||
| 3 | 0.66 (0.51, 0.87) | 0.003 | 0.65 (0.50, 0.85) | 0.002 | 16553 | |||
| 4 | 0.80 (0.62, 1.05) | 0.11 | 0.78 (0.60, 1.02) | 0.069 | 15647 | |||
| 5 | 0.72 (0.54, 0.96) | 0.024 | 0.68 (0.51, 0.90) | 0.008 | 15972 | |||
| Sedentary hours | 1 | Ref. | <0.001 | Ref. | <0.001 | 70273 | ||
| 2 | 0.94 (0.81, 1.08) | 0.38 | 0.93 (0.80, 1.07) | 0.29 | 41711 | |||
| 3 | 1.09 (0.94, 1.25) | 0.24 | 1.07 (0.93, 1.23) | 0.38 | 33407 | |||
| 4 | 1.07 (0.92, 1.25) | 0.39 | 1.04 (0.89, 1.21) | 0.64 | 22508 | |||
| 5 | 1.30 (1.13, 1.48) | <0.001 | 1.23 (1.08, 1.41) | 0.003 | 33725 | |||
Note: The hazard ratios (95% CIs) were derived from stratified cox regression model with adjustment for age, sex, household income, loneliness, Townsend index, family history of dementia, current drinker, BMI level, hearing problem, hypertension, diabetes, loneliness, and sleep duration by including them as predictors in the model except for loneliness and sleep duration, which were specified as the strata to address their non‐proportionality of hazard.
LTPA: zero as reference, OPA: zero as reference, Sedentary lifestyle: first quintile as reference.
Abbreviations: BMI, body mass index; CI, confidence interval; LTPA, leisure time physical activity; OPA, occupational physical activity.
TABLE 3.
Association of physical activity with vascular dementia and Alzheimer's disease.
| Vascular dementia | Alzheimer's disease | |||||||
|---|---|---|---|---|---|---|---|---|
| Exposure a | Levels | Hazard ratio (95% CI) separate model | p Value | p for trend | Hazard ratio (95% CI) common model | p Value | p for trend | Category Size |
| Leisure time physical activity | 0 | Ref. | 0.01 | Ref. | 0.17 | 10847 | ||
| 1 | 0.97 (0.64, 1.47) | 0.89 | 0.92 (0.63, 1.32) | 0.64 | 38351 | |||
| 2 | 0.93 (0.60, 1.43) | 0.73 | 0.99 (0.68, 1.44) | 0.97 | 39029 | |||
| 3 | 0.60 (0.37, 0.97) | 0.038 | 0.89 (0.61, 1.31) | 0.55 | 37296 | |||
| 4 | 0.61 (0.38, 0.99) | 0.047 | 0.88 (0.60, 1.30) | 0.52 | 38103 | |||
| 5 | 0.69 (0.43, 1.11) | 0.13 | 0.74 (0.50, 1.11) | 0.14 | 37998 | |||
| Occupational physical activity | 0 | Ref. | 0.07 | Ref. | 0.20 | 120980 | ||
| 1 | 0.74 (0.44, 1.25) | 0.26 | 1.06 (0.77, 1.47) | 0.72 | 17285 | |||
| 2 | 0.39 (0.16, 0.96) | 0.041 | 0.93 (0.59, 1.45) | 0.75 | 15187 | |||
| 3 | 0.77 (0.45, 1.34) | 0.36 | 0.64 (0.41, 1.01) | 0.056 | 16553 | |||
| 4 | 0.75 (0.41, 1.35) | 0.33 | 1.00 (0.66, 1.49) | 0.98 | 15647 | |||
| 5 | 0.68 (0.37, 1.25) | 0.21 | 0.77 (0.48, 1.23) | 0.28 | 15972 | |||
| Sedentary hours | 1 | Ref. | 0.30 | Ref. | 0.46 | 70273 | ||
| 2 | 0.97 (0.70, 1.32) | 0.83 | 0.77 (0.62, 0.97) | 0.026 | 41711 | |||
| 3 | 1.10 (0.81, 1.50) | 0.54 | 0.93 (0.74, 1.15) | 0.49 | 33407 | |||
| 4 | 0.92 (0.64, 1.30) | 0.63 | 0.92 (0.72, 1.18) | 0.53 | 22508 | |||
| 5 | 1.20 (0.89, 1.62) | 0.23 | 1.06 (0.85, 1.32) | 0.61 | 33725 | |||
Note: Vascular dementia adjustment: age, sex, education qualification, household income, loneliness, Townsend index, family history of dementia, current drinker, ever smoked, BMI level, hypertension, diabetes, LTPA, OPA and sedentary hours.
Alzheimer's disease adjustment: age, sex, household income, Townsend index, family history of dementia, current drinker, BMI level, hearing problem, hypertension, diabetes, LTPA, OPA and sedentary hours.
LTPA: zero as reference, OPA: zero as reference, Sedentary lifestyle: first quintile as reference. The hazard ratios (95% CIs) were derived from Cox regression model.
Abbreviations: BMI, body mass index; CI, confidence interval; LTPA, leisure time physical activity; OPA, occupational physical activity.
As in Table 4, linear trends of LTPA (p = 0.35) and sedentary hours (p = 0.21) with cognitive decline were not significant. Compared to zero LTPA, the fifth quintile of LTPA had non‐significantly slower cognitive decline (β = 0.05, 95% CI −0.26, 0.36). For sedentary hours, when compared to its first quintile, the fifth quintile of sedentary hours had faster cognitive decline (β = −0.14, 95% CI −0.28, 0.01). Although the linear trend between OPA and cognitive decline was significant (p = 0.02), no significant differences between zero OPA level and quintiles of non‐zero OPA levels were found (e.g., fifth quintile: β = −0.13, 95% CI −0.34, 0.08).
TABLE 4.
Association of physical activity with cognitive decline
| Model 1 | Model 2 | |||||||
|---|---|---|---|---|---|---|---|---|
| Exposure a | Level | Regression coefficient (95% CI) | p Value | p for trend | Regression coefficient (95% CI) | p Value | p for trend | Category Size |
| Leisure time physical activity | 0 | Ref. | 0.29 | Ref. | 0.35 | 48 | ||
| 1 | 0.02 (−0.29, 0.33) | 0.90 | 0.02 (−0.28, 0.33) | 0.88 | 240 | |||
| 2 | −0.05 (−0.36, 0.25) | 0.73 | −0.05 (−0.36, 0.25) | 0.73 | 287 | |||
| 3 | 0.03 (−0.27, 0.33) | 0.84 | 0.03 (−0.28, 0.33) | 0.87 | 359 | |||
| 4 | 0.10 (−0.20, 0.39) | 0.53 | 0.10 (−0.21, 0.40) | 0.54 | 372 | |||
| 5 | 0.04 (−0.26, 0.34) | 0.81 | 0.05 (−0.26, 0.36) | 0.75 | 365 | |||
| Occupational physical activity | 0 | Ref. | 0.03 | Ref. | 0.02 | 1013 | ||
| 1 | 0.01 (−0.16, 0.18) | 0.93 | 0.00 (−0.16, 0.17) | 0.96 | 154 | |||
| 2 | −0.13 (−0.30, 0.04) | 0.15 | −0.14 (−0.31, 0.04) | 0.13 | 146 | |||
| 3 | −0.15 (−0.33, 0.03) | 0.11 | −0.15 (−0.33, 0.04) | 0.11 | 134 | |||
| 4 | −0.14 (−0.33, 0.05) | 0.16 | −0.14 (−0.34, 0.05) | 0.16 | 117 | |||
| 5 | −0.13 (−0.33, 0.08) | 0.22 | −0.13 (−0.34, 0.08) | 0.22 | 107 | |||
| Sedentary hours | 1 | Ref. | 0.24 | Ref. | 0.21 | 563 | ||
| 2 | −0.03 (−0.16, 0.09) | 0.60 | −0.03 (−0.16, 0.10) | 0.62 | 380 | |||
| 3 | −0.08 (−0.23, 0.06) | 0.25 | −0.09 (−0.23, 0.05) | 0.22 | 282 | |||
| 4 | 0.08 (−0.10, 0.25) | 0.39 | 0.08 (−0.09, 0.25) | 0.37 | 168 | |||
| 5 | −0.12 (−0.27, 0.02) | 0.095 | −0.14 (−0.28, 0.01) | 0.066 | 278 | |||
Note: The regression coefficients (95% CIs) were derived from multiple linear regression model with adjustment for age, sex, education qualification, family history of dementia, current drinker.
LTPA: zero as reference, OPA: zero as reference, Sedentary lifestyle: first quintile as reference.
Abbreviations: CI, confidence interval; LTPA, leisure time physical activity; OPA, occupational physical activity.
4. DISCUSSION
Using the nation‐wide prospective cohort of UK Biobank, this study found that the higher LTPA and OPA, and lower sedentary hours were associated with reduced risk of dementia at follow‐up. Although the linear trend between OPA and cognitive decline was significant, the differences between the quintiles of non‐zero OPA levels and zero OPA level were non‐significant. For LTPA and sedentary lifestyle, although their linear trends with cognitive decline were non‐significant, the directions of their linear trends with an increased risk of dementia were consistent.
Multiple longitudinal studies have shown that higher LTPA are associated with a lower risk of dementia. 14 , 33 , 34 However, most of these studies involve older participants, and only a few studies have participants with a mean age below 60 years (mean age 48.1 to 56 years). 17 A Finnish cohort study of 3559 participants followed for an average of 24.4 years found that lower frequency of participating in LTPA at baseline was significantly associated with a higher risk of dementia. 14 Our findings are consistent with previous literature where we showed that LTPA was associated with reduced risk of dementia.
A few studies have so far examined OPA. In the prospective Danish cohort study followed for 30 years in the middle aged group (40‐59 years), higher level of OPA was significantly associated with an increased incidence of dementia. 15 However, result from our study showed lower risk of dementia among people with higher OPA. This disparity might be due to differences in study population as the Danish study only included working males or by different methods of assessing OPA or longer follow‐up in the Danish study, where the participant's job was only classified into groups of sedentary, light physical work, and heavy physical work.
In this study, we found that the association between OPA and dementia risk and that with cognitive decline contradicts. This might be due to the less detailed cognitive assessment in the UK Biobank which may not be sensitive to changes in PA. Moreover, this might also be due to selection bias as analysis of cognitive decline only included 1,671 participants, who had higher proportions of college education and family history of dementia but lower proportions of low household income and obesity. Further investigation is warranted to better understand the contribution of OPA in other domains of cognitive function.
In our analyses, higher levels of LTPA and OPA were associated with lower dementia risk, although most of the associations became non‐significant when specifically examining Alzheimer's disease. We further observed that the association between OPA and all‐cause dementia was primarily driven by dementia cases arising from other causes. A linear trend between LTPA and vascular dementia was observed suggesting the beneficial effect of PA on vascular health and stroke risk and thus reducing the risk of vascular dementia. In contrast, the relationship between PA and Alzheimer's disease, which is characterized by the accumulation of amyloid plaques and tau tangles in the brain, may be related to its effects on reducing inflammation, oxidative stress, and improving synaptic plasticity and brain‐derived neurotrophic factor (BDNF) levels. However, given that the numbers are rather small, these results should be interpreted with caution.
The association between PA and the risk of dementia was also examined in a previous study using UK‐Biobank data. 9 However, frequency and duration of each activity item were modeled separately with incident dementia, providing only partial association between different activity aspects of duration or frequency of PA with incident dementia. Additionally, the use of principal component analysis in the previous study may not have fully reflected the potential associations between specific types of PA and the risk of dementia. By contrast, our study examined the association of PA with both incident dementia and cognitive decline. The direction of the linear trend for both outcomes were consistent when comparing the fifth quintile of LTPA with zero value and when comparing fifth and first quintiles of sedentary lifestyle, although the trends were only significant in the dementia analysis, non‐significant trends for cognitive decline analysis could partly be explained by the much smaller sample size.
This study has several strengths including its large sample size, and long follow‐up period to result in > 1800 dementia cases at the end of follow‐up. Considering that different types (LTPA, OPA and sedentary lifestyle) of physical activities might be correlated with each other, they were mutually adjusted to elucidate independent association of each category with dementia risk and cognitive decline. Also, two outcomes (dementia and cognitive decline) were included in this analysis, and a summary score was generated using dimension reduction for cognitive performance, where FAMD was used to accommodate for the binary variable in the cognitive data.
The limitations of the study include; first, compared to a single time point measurement, using multiple time points to assess PA provides a more comprehensive understanding of individuals' activity levels, as demonstrated in studies like the ARIC cohort. 35 This approach is particularly valuable for individuals approximately the age of retirement who may undergo lifestyle changes. However, in our large prospective cohort, only 14.2% of subjects (∼71,000) were surveyed during the second follow‐up, and a limited number of dementia cases occurred thereafter. As a result, we were unable to examine the PA patterns over time in this study. Nevertheless, with a longer follow‐up period, more dementia cases may arise among these participants, providing an opportunity for future studies to investigate longitudinal PA and its association with dementia risk. Second, while the participants were followed for a median of 11.8 years, more than half of the dementia cases were diagnosed within 10 years since recruitment, and the small number of cases diagnosed beyond 10 years is insufficient to rule out the possibility of prodromal dementia influencing the results. Third, the self‐reporting of PA in this study introduces the possibility of recall bias, which may compromise the reliability of the findings. Furthermore, the use of MET values to estimate activity levels may not accurately reflect the actual amount of PA undertaken by participants. Additionally, the semi‐quantitative nature of the questions regarding frequency and duration of activities may introduce misclassification of participants into higher or lower activity categories, potentially affecting the accuracy of the associations observed. To further strengthen the evidence base, it is important to incorporate more objective measurements of PA, such as activity diaries and accelerometers, 36 to corroborate the results of our study. Fourth, measurement of sedentary behavior in this study only included watching TV, playing video game, and driving a car, which cover 47%−51% of total sedentary hours, 35 while other activities like listening to music were not measured which may also introduce misclassification bias in the study. Finally, we are unable to distinguish between aerobic and anaerobic activities due to the lack of specific information collected in the questionnaire. This limitation prevents us from examining the potential differential effects of aerobic exercise specifically on the risk of dementia and cognitive decline.
In this study, we found that higher LTPA and OPA, and lower sedentary hours were associated with a reduced risk of dementia but not with cognitive decline. Our findings highlight the importance of regular PA, including both leisure‐time and occupational activities, as a potential preventive measure against dementia.
CONFLICT OF INTEREST STATEMENT
Authors have no conflict of interest to declare. Author disclosures are available in the supporting information.
CONSENT STATEMENT
All human subjects provided informed consent.
Supporting information
Supporting information
Supporting information
Supporting information
Supporting information
Supporting information
ACKNOWLEDGMENTS
The authors have nothing to report. This research was conducted under UK Biobank Project #71022. S. Hilal was supported by NUS start‐up grant [A‐0006088‐00‐00], National Medical Research Council Singapore, Transition Award [A‐0006310‐00‐00], Ministry of Education, Academic Research Fund Tier 1 [A‐0006106‐00‐00] and Absence Leave Grant [A‐8000336‐00‐00].
Huang X, Tan CS, Kandiah N, Hilal S. Association of physical activity with dementia and cognitive decline in UK Biobank. Alzheimer's Dement. 2023;15:e12476. 10.1002/dad2.12476
DATA AVAILABILITY STATEMENT
De‐identified data used in this study are publicly available upon request to the UKB.
REFERENCES
- 1. GBD 2019 Dementia Forecasting Collaborators. Dementia Forecasting Collaborators . Estimation of the global prevalence of dementia in 2019 and forecasted prevalence in 2050: an analysis for the Global Burden of Disease Study 2019. Lancet Public Health. 2022;7(2):e105‐e125. doi: 10.1016/S2468-2667(21)00249-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Livingston G, Huntley J, Sommerlad A, et al. Dementia prevention, intervention, and care: 2020 report of the Lancet Commission. Lancet. 2020;396(10248):413‐446. doi: 10.1016/S0140-6736(20)30367-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Nikitara K, Odani S, Demenagas N, Rachiotis G, Symvoulakis E, Vardavas C. Prevalence and correlates of physical inactivity in adults across 28 European countries. Eur J Public Health. 2021;31(4):840‐845. doi: 10.1093/eurpub/ckab067 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Kivimaki M, Singh‐Manoux A, Pentti J, et al. Physical inactivity, cardiometabolic disease, and risk of dementia: an individual‐participant meta‐analysis. BMJ. 2019;365:l1495. doi: 10.1136/bmj.l1495 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Podewils LJ, Guallar E, Kuller LH, et al. Physical activity, APOE genotype, and dementia risk: findings from the Cardiovascular Health Cognition Study. Am J Epidemiol. 2005;161(7):639‐651. doi: 10.1093/aje/kwi092 [DOI] [PubMed] [Google Scholar]
- 6. Wu W, Ding D, Zhao Q, et al. Medium‐to‐high late‐life physical activity is associated with lower risk of incident dementia: The Shanghai Aging Study. J Alzheimers Dis. 2020;73(2):751‐758. doi: 10.3233/JAD-190937 [DOI] [PubMed] [Google Scholar]
- 7. Sabia S, Dugravot A, Dartigues JF, et al. Physical activity, cognitive decline, and risk of dementia: 28 year follow‐up of Whitehall II cohort study. BMJ. 2017;357:j2709. doi: 10.1136/bmj.j2709 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Llamas‐Velasco S, Contador I, Villarejo‐Galende A, Lora‐Pablos D, Bermejo‐Pareja F. Physical activity as protective factor against dementia: A prospective population‐based study (NEDICES). J Int Neuropsychol Soc. 2015;21(10):861‐867. doi: 10.1017/S1355617715000831 [DOI] [PubMed] [Google Scholar]
- 9. Zhu J, Ge F, Zheng Y, et al. Neurology . 2022. doi: 10.1212/WNL.0000000000200701 [DOI]
- 10. Wu W, Ding D, Zhao Q, et al. Dose‐response relationship between late‐life physical activity and incident dementia: A pooled analysis of 10 cohort studies of memory in an international consortium. Alzheimers Dement. 2022. doi: 10.1002/alz.12628 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Iso‐Markku P, Kujala UM, Knittle K, Polet J, Vuoksimaa E, Waller K. Physical activity as a protective factor for dementia and Alzheimer's disease: Systematic review, meta‐analysis and quality assessment of cohort and case‐control studies. Br J Sports Med. 2022;56(12):701‐709. doi: 10.1136/bjsports-2021-104981 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Lourida I, Hannon E, Littlejohns TJ, et al. Association of lifestyle and genetic risk with incidence of dementia. JAMA. 2019;322(5):430‐437. doi: 10.1001/jama.2019.9879 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Hansson O, Svensson M, Gustavsson AM, et al. Midlife physical activity is associated with lower incidence of vascular dementia but not Alzheimer's disease. Alzheimers Res Ther. 2019;11(1):87. doi: 10.1186/s13195-019-0538-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Tolppanen AM, Solomon A, Kulmala J, et al. Leisure‐time physical activity from mid‐ to late life, body mass index, and risk of dementia. Alzheimers Dement. 2015;11(4):434‐443. doi: 10.1016/j.jalz.2014.01.008 e6 [DOI] [PubMed] [Google Scholar]
- 15. Nabe‐Nielsen K, Holtermann A, Gyntelberg F, et al. The effect of occupational physical activity on dementia: results from the Copenhagen Male Study. Scand J Med Sci Sports. 2021;31(2):446‐455. doi: 10.1111/sms.13846 [DOI] [PubMed] [Google Scholar]
- 16. Juul Rasmussen I, Rasmussen KL, Thomassen JQ, et al. Physical activity in leisure time and at work and risk of dementia: A prospective cohort study of 117,616 individuals. Atherosclerosis. 2022;360:53‐60. doi: 10.1016/j.atherosclerosis.2022.08.004 [DOI] [PubMed] [Google Scholar]
- 17. Xu W, Wang HF, Wan Y, Tan CC, Yu JT, Tan L. Leisure time physical activity and dementia risk: a dose‐response meta‐analysis of prospective studies. BMJ Open. 2017;7(10):e014706. doi: 10.1136/bmjopen-2016-014706 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Blondell SJ, Hammersley‐Mather R, Veerman JL. Does physical activity prevent cognitive decline and dementia?: A systematic review and meta‐analysis of longitudinal studies. BMC Public Health. 2014;14:510. doi: 10.1186/1471-2458-14-510 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Greene C, Lee H, In ThuretS. The long run: Physical activity in early life and cognitive aging. Front Neurosci. 2019;13:884. doi: 10.3389/fnins.2019.00884 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Singh‐Manoux A, Hillsdon M, Brunner E, Marmot M. Effects of physical activity on cognitive functioning in middle age: evidence from the Whitehall II prospective cohort study. Am J Public Health. 2005;95(12):2252‐2258. doi: 10.2105/AJPH.2004.055574 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Lee S, Yuki A, Nishita Y, et al. Research relationship between light‐intensity physical activity and cognitive function in a community‐dwelling elderly population‐an 8‐year longitudinal study. J Am Geriatr Soc. 2013;61(3):452‐453. doi: 10.1111/jgs.12119 [DOI] [PubMed] [Google Scholar]
- 22. Aarsland V, Borda MG, Aarsland D, et al. Association between physical activity and cognition in Mexican and Korean older adults. Arch Gerontol Geriatr. 2020;89:104047. doi: 10.1016/j.archger.2020.104047 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Sudlow C, Gallacher J, Allen N, et al. UK biobank: An open access resource for identifying the causes of a wide range of complex diseases of middle and old age. PLoS Med. 2015;12(3):e1001779. doi: 10.1371/journal.pmed.1001779 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Pearce M, Strain T, Kim Y, et al. Estimating physical activity from self‐reported behaviours in large‐scale population studies using network harmonisation: Findings from UK Biobank and associations with disease outcomes. Int J Behav Nutr Phys Act. 2020;17(1):40. doi: 10.1186/s12966-020-00937-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Lyall DM, Cullen B, Allerhand M, et al. Cognitive test scores in UK Biobank: Data rReduction in 480,416 participants and longitudinal stability in 20,346 participants. PLoS One. 2016;11(4):e0154222. doi: 10.1371/journal.pone.0154222 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Lê S, Josse J, Husson F. FactoMineR: An R Package for multivariate analysis. Journal of Statistical Software 03/18. 2008;25(1):1‐18. doi: 10.18637/jss.v025.i01 [DOI] [Google Scholar]
- 27. Tai XY, Chen C, Manohar S, Husain M. Impact of sleep duration on executive function and brain structure. Commun Biol. 2022;5(1):201. doi: 10.1038/s42003-022-03123-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. World Health Organization . Obesity: Preventing and managing the global epidemic. Report of a WHO consultation. 2000. World Health Organ Tech Rep Ser [PubMed]
- 29. Cassidy S, Chau JY, Catt M, Bauman A, Trenell MI. Cross‐sectional study of diet, physical activity, television viewing and sleep duration in 233,110 adults from the UK Biobank; the behavioural phenotype of cardiovascular disease and type 2 diabetes. BMJ Open. 2016;6(3):e010038. doi: 10.1136/bmjopen-2015-010038 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. IPAQ. Guidelines for data processing and analysis of the International Physical Activity Questionnaire (IPAQ)—Short and Long Forms. 2005:1‐15.
- 31. Ma H, Li X, Zhou T, et al. Glucosamine use, inflammation, and genetic susceptibility, and incidence of type 2 diabetes: A Prospective Study in UK Biobank. Diabetes Care. 2020;43(4):719‐725. doi: 10.2337/dc19-1836 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Shang X, Hill E, Zhu Z, et al. The association of age at diagnosis of hypertension with brain structure and incident dementia in the UK Biobank. Hypertension. 2021;78(5):1463‐1474. doi: 10.1161/HYPERTENSIONAHA.121.17608 [DOI] [PubMed] [Google Scholar]
- 33. Neergaard JS, Dragsbaek K, Hansen HB, Henriksen K, Christiansen C, Karsdal MA. Late‐life risk factors for all‐cause dementia and differential dementia diagnoses in women: A prospective cohort study. Medicine (Baltimore). 2016;95(11):e3112. doi: 10.1097/MD.0000000000003112 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Morgan GS, Gallacher J, Bayer A, Fish M, Ebrahim S, Ben‐Shlomo Y. Physical activity in middle‐age and dementia in later life: findings from a prospective cohort of men in Caerphilly, South Wales and a meta‐analysis. J Alzheimers Dis. 2012;31(3):569‐580. doi: 10.3233/JAD-2012-112171 [DOI] [PubMed] [Google Scholar]
- 35. Palta P, Sharrett AR, Deal JA, et al. Leisure‐time physical activity sustained since midlife and preservation of cognitive function: The atherosclerosis risk in communities study. Alzheimers Dement. 2019;15(2):273‐281. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Del Pozo Cruz B, Ahmadi M, Naismith SL, Stamatakis E. Association of daily step count and intensity with incident dementia in 78430 adults living in the UK. JAMA Neurol. 2022;79(10):1059‐1063. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Rosenberg DE, Norman GJ, Wagner N, Patrick K, Calfas KJ, Sallis JF. Reliability and validity of the Sedentary Behavior Questionnaire (SBQ) for adults. J Phys Act Health. 2010;7(6):697‐705. doi: 10.1123/jpah.7.6.697 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supporting information
Supporting information
Supporting information
Supporting information
Supporting information
Data Availability Statement
De‐identified data used in this study are publicly available upon request to the UKB.