Abstract

Background:

Physical activity volume (PAV) has been linked to a wide range of health outcomes; however, its association with incident nervous system diseases remains incompletely understood. This study explored their relationship using data from UK Biobank.

Methods:

A retrospective cohort study was conducted involving 278,306 participants from the UK Biobank. PAV was quantified as metabolic equivalent of task (MET) minutes per week, derived from self-reported physical activity levels, and categorized into three groups. Incident cases of nervous system diseases were identified through ICD-10 codes obtained from hospital inpatient records, death registries, and self-reports. Cox proportional hazards models were employed to estimate hazard ratios (HRs) and 95% confidence intervals (CIs), adjusting for a series of covariates. Restricted cubic splines were applied to assess potential non-linear associations.

Results:

Women and individuals aged 60 years or older exhibited higher incidence rates of nervous system diseases. After multivariable adjustment, higher PAV was significantly associated with a lower risk of overall nervous system diseases (HR < 1). A non-linear dose–response relationship was observed, with the lowest risk occurring at a PAV level of 1,356 MET-min/week. Subgroup analyses indicated that elevated PAV conferred protective effects against several specific conditions. Conversely, higher PAV was associated with an increased risk of meningitis.

Conclusion:

Increased levels of physical activity are associated with a reduced risk of numerous nervous system diseases, with optimal protection observed at approximately 1,356 MET-min/week. These findings support the promotion of moderate-to-vigorous physical activity as a preventive strategy for neurological disorders, particularly among high-risk populations.

1 Introduction

The increasing global burden of nervous system diseases is notable, driven by aging populations. Nervous system diseases, including neurodegenerative diseases (1, 2), cerebrovascular diseases (e.g., stroke) (3), and other neurological disorders (4), have become leading causes of disability and death worldwide (5, 6). With the development of medical technique and pharmacological treatments, the treatments of nervous system diseases are continually evolving (7–9). While clinical treatment often only symptoms management, and some conditions need long-term rehabilitation, the role of exercise in the recovery of the nervous system remains significant (10). As we all know, exercise is a well-established protective factor for some diseases, such as diabetes (11), gastroesophageal reflux disease (12), and even some cancers. Besides, a growing body of evidence suggests that physical activity also contribute to brain health (10). Proposed mechanisms include enhanced cerebral blood flow, reduction of inflammation, improved vascular function, and the promotion of neurotrophic factors. Numerous studies have linked exercise to a reduced risk of individual nervous system diseases, particularly dementia and stroke (13). However, notable gaps in knowledge remain. Current studies either have focused on a single disease outcome or simply illustrate exercise condition using active vs. inactive. The quantitative relationship between the total volume of activity and nervous system disease is unclear (14). Understanding this dose–response relationship is crucial for informing precise public health recommendations.

Physical activity volume(PAV), the total amount of physical activity one performs, is regarded as the comprehensive measure of total exercise “dose” (15). It is a crucial concept in public health and the key factor linked to most health benefit, calculated by manipulating frequency, intensity, and time (16). In order to make sense of association between the volume of physical activity and the subsequent incidence of nervous system disease, we conducted a large-scale retrospective cohort study using date from UK Biobank to explore their relationships.

2 Materials and methods2.1 Date source

The UK Biobank is a long-term prospective cohort study established to identify the determinants of disabling conditions and data collected from 22 health centers across the United Kingdom (17). At recruitment, participants completed a detailed touch-screen questionnaire which mainly focus on demographic, socioeconomic information, health behaviors and living environment situations were collected by specific measurements. More details about the UK Biobank can be accessed online (http://www.ukbio bank.ac.uk/). The ethical approval was granted by the research ethics committee (11/ NW/0382). Before the research, all participants provided written informed consent and we were permitted to use UK Biobank data under study number 99709. At the beginning, data from 502,428 participants were potentially available. After excluding UK Biobank removed (n = 430), missing data on key covariates (n = 90,935), with no PAV data (n = 86,107), previous diagnosis of Nervous System Diseases at baseline (n = 46,650), data from a total of 278,306 eligible participants, aged from 40 to 69 at recruit in UK Biobank was used in our study. The participants were enrolled between 2006 and 2010, and the follow-up completed on March 31, 2025.

2.2 Calculation of PAV

PAV is the representation of the metabolic equivalent of tasks, and calculated by each participant’s weekly physical activity volume. In this study, the PAV was only used to evaluated physical exercise activity and calculated based on the 2011 Compendium of Physical Activities (18). As the physical activity intensity is usually divided into vigorous, moderate, and light, and each category is assigned eight, five, and three METs. According to the minimum time consumed for each category and its frequencies, the PAV was finally calculated as follows: (8 METs×20 min × times) + (5 METs×30 min × times) + (3 METs×30 min × times). Then, PAV was divided into the following three groups, which, respectively, is <500 group, 500–1,000 group, and ≥1,000 group.

2.3 Ascertainment of nervous system disease incidence

The incident Nervous System disease events was identified in accordance with the ‘first occurrence’ data fields that the UK biobank generated that map the clinical codes from primary care, hospital inpatient admissions, death records, and self-reported medical conditions to 3-digit code of International Classification of Disease (ICD-10) and provide, for each participant, the date that code first occurred in any source.

2.4 Covariates

The selection of covariates in our study reference to a review of several published studies and Global Burden Disease report (19). To mitigate the influence of potential confounding effects, the following five types covariates were considered in multivariable models: Demographic information (Sex, Age, Ethnic background, BMI and MET score), Socioeconomic Status (Employment status and University education), Medical History (Diabetes, Hypertension and Cancer), Health Behaviors (Alcohol drinker status, Smoking status, Sleep duration, Time watch TV, Time use computer, Coffee intake, Tea intake and IPAQ activity group) and Living Environment (PM2.5, PM10, Greenspace percentage in 300 m, Nitrogen dioxide and Nitrogen oxides). Among these, Sleep duration, Time watch TV and Time use compute were self-reported as the number of hours per day or the times per day. BMI or Body Mass Index was calculated using the internationally recognized formula: weight (kg) divided by height (cm) squared. IPAQ (International Physical Activity Questionnaire) is a self-administered questionnaire that was used to assess the physical activity in the past 7 days. As regular PA was defined as performing at least 30 min of moderate PA at least five times a week, we categorized each participant’s weekly PA levels based on that. In this study, we used IPAQ to evaluate the physical activity intensity. Furthermore, considering diabetes, hypertension and cancer are major components of the global burden of disease, this study included these three conditions based on the first reported date in the first occurrence dataset and the date of attending the assessment center.

2.5 Statistical analysis

Statistical analyses used SAS (version 9.4, SAS Institute Inc., Cary, NC, United States) and R (version 4.4.1), with statistical significance at p < 0.05 using t-tests for continuous variables and chi-square for categorical variables. Cox proportional hazards regression models were used to explore the associations between PAV score and total nervous system disorder incidence, and hazard ratios (HRs) and 95% confidence intervals (CIs) were shown. The four models were tested: (1) Model 1 was adjusted for demographic information; (2) Model 2 was adjusted for Model 1 plus socioeconomic status; (3) Model 3 was adjusted for model 2 plus medical history factors; (4) Model 4 further adjusted for health behaviors factors; and (4) Model 5 finally adjusted for living environment factors.

We used restricted cubic spline (RCS) regression to explore potential non-linear associations between PAV score total nervous system disorder incidence. The subgroup analysis was used to further explore associations between PAV score and specific nervous system disorder incidence.

3 Results3.1 Characteristics of the study population

Data from 502,428 participants were available at beginning, after excluding UK Biobank removed (n = 430), missing data on key covariates (n = 90,935), with no PAV data (n = 86,107), previous diagnosis of Nervous System Diseases at baseline (n = 46,650), data from a total of 278,306 eligible participants in UK Biobank was used in our study.

Among UK biobank participants, a nearly equal gender distribution was shown, participants were almost white (94.8%), and age below 60 were162000 (58.2%). According to each participant’s PAV score, the number of participant in each group were, respectively, 40,780 (MTE < 500 group), 41,976 (500–1,000 group), 195,550 (MTE ≥ 1,000 group). Compared with those whose PAV score was above 1,000, participants in PAV < 500 group tended to be younger, possess a higher BMI, exhibit higher levels of education, and sleep less, smoke more. Additionally, participants in this group spent more time in watching TV and using computer. Furthermore, they also had a higher proportion of Diabetes and Hypertension than the other two groups. While no difference was shown when compared the ratio of Cancer. The characteristics of study participants in the primary overall and by the divided group were presented in Table 1.

CharacteristicsPAV <500 group (40780) n (%)PAV 500–1,000 group (41976) n (%)PAV ≥1,000 group (195550) n (%)Total (278306) n (%)p-valueSex<0.0001Male20,502 (50.3%)19,716 (47.0%)96,406 (49.3%)136,624 (49.1%)Female20,278 (49.7%)22,260 (53.0%)99,144 (50.7%)141,682 (50.9%)Age at recruitment<0.0001<6025,799 (63.3%)25,068 (59.7%)111,133 (56.8%)162,000 (58.2%)≥6014,981 (36.7%)16,908 (40.3%)84,417 (43.2%)116,306 (41.8%)BMI(kg/m2)<0.0001<18.5404 (1.0%)346 (0.8%)1,600 (0.8%)2,350 (0.8%)18.5–20593 (1.5%)698 (1.7%)3,864 (2.0%)5,155 (1.9%)20–259,925 (24.3%)12,607 (30.0%)66,937 (34.2%)89,469 (32.1%)25–3016,801 (41.2%)18,037 (43.0%)85,241 (43.6%)120,079 (43.1%)≥3013,057 (32.0%)10,288 (24.5%)37,908 (19.4%)61,253 (22.0%)Ethnic background<0.0001White38,142 (93.5%)39,599 (94.3%)186,182 (95.2%)263,923 (94.8%)African359 (0.9%)331 (0.8%)1,226 (0.6%)1916 (0.7%)Asian1,213 (3.0%)1,046 (2.5%)3,697 (1.9%)5,956 (2.1%)Mixed252 (0.6%)252 (0.6%)1,174 (0.6%)1,678 (0.6%)Other ethnic group814 (2.0%)748 (1.8%)3,271 (1.7%)4,833 (1.7%)Current employment status<0.0001Employment27,312 (67.0%)26,821 (63.9%)116,083 (59.4%)170,216 (61.2%)Unemployment3,574 (8.8%)2,674 (6.4%)12,566 (6.4%)18,814 (6.8%)Retired9,894 (24.3%)12,481 (29.7%)66,901 (34.2%)89,276 (32.1%)University education<0.0001No26,121 (64.1%)25,455 (60.6%)127,080 (65.0%)178,656 (64.2%)Yes14,659 (35.9%)16,521 (39.4%)68,470 (35.0%)99,650 (35.8%)Alcohol drinker status<0.0001Current37,521 (92.0%)39,307 (93.6%)183,436 (93.8%)260,264 (93.5%)Previous1,421 (3.5%)1,149 (2.7%)5,523 (2.8%)8,093 (2.9%)Never1838 (4.5%)1,520 (3.6%)6,591 (3.4%)9,949 (3.6%)Smoking status<0.0001Current4,941 (12.1%)4,066 (9.7%)18,037 (9.2%)27,044 (9.7%)Previous13,804 (33.9%)14,340 (34.2%)69,671 (35.6%)97,815 (35.1%)Never22,035 (54.0%)23,570 (56.2%)107,842 (55.1%)153,447 (55.1%)Sleep duration (Hours/day)<0.0001<710,391 (25.5%)9,643 (23.0%)44,471 (22.7%)64,505 (23.2%)≥730,389 (74.5%)32,333 (77.0%)151,079 (77.3%)213,801 (76.8%)Diabetes<0.0001No35,839 (87.9%)38,098 (90.8%)181,052 (92.6%)254,989 (91.6%)Yes4,941 (12.1%)3,878 (9.2%)14,498 (7.4%)23,317 (8.4%)Hypertension<0.0001No24,156 (59.2%)26,149 (62.3%)126,220 (64.5%)176,525 (63.4%)Yes16,624 (40.8%)15,827 (37.7%)69,330 (35.5%)101,781 (36.6%)Cancer0.5145No31,203 (76.5%)32,225 (76.8%)150,139 (76.8%)213,567 (76.7%)Yes9,577 (23.5%)9,751 (23.2%)45,411 (23.2%)64,739 (23.3%)Coffee intake (Cups/day)<0.0001Not intake12,213 (29.9%)11,671 (27.8%)54,193 (27.7%)78,077 (28.1%)17,473 (18.3%)8,583 (20.4%)41,369 (21.2%)57,425 (20.6%)27,033 (17.2%)8,062 (19.2%)38,618 (19.7%)53,713 (19.3%)35,052 (12.4%)5,315 (12.7%)24,446 (12.5%)34,813 (12.5%)43,586 (8.8%)3,761 (9.0%)16,411 (8.4%)23,758 (8.5%)52,411 (5.9%)2,106 (5.0%)9,492 (4.9%)14,009 (5.0%)≥63,012 (7.4%)2,478 (5.9%)11,021 (5.6%)16,511 (5.9%)Tea intake (Cups/day)<0.0001Not intake8,078 (19.8%)7,322 (17.4%)32,769 (16.8%)48,169 (17.3%)13,798 (9.3%)3,724 (8.9%)16,776 (8.6%)24,298 (8.7%)25,750 (14.1%)6,385 (15.2%)28,768 (14.7%)40,903 (14.7%)35,912 (14.5%)6,499 (15.5%)29,870 (15.3%)42,281 (15.2%)45,486 (13.5%)6,005 (14.3%)28,458 (14.6%)39,949 (14.4%)54,335 (10.6%)4,599 (11.0%)22,483 (11.5%)31,417 (11.3%)≥67,421 (18.2%)7,442 (17.7%)36,426 (18.6%)51,289 (18.4%)Time watch TV (Hours/day)<0.0001<13,140 (7.7%)3,476 (8.3%)16,691 (8.5%)23,307 (8.4%)14,978 (12.2%)5,772 (13.8%)27,764 (14.2%)38,514 (13.8%)210,549 (25.9%)11,718 (27.9%)55,760 (28.5%)78,027 (28.0%)39,503 (23.3%)9,795 (23.3%)45,848 (23.4%)65,146 (23.4%)46,331 (15.5%)6,307 (15.0%)29,465 (15.1%)42,103 (15.1%)≥56,279 (15.4%)4,908 (11.7%)20,022 (10.2%)31,209 (11.2%)Time use computer (Hours/day)<0.0001<117,276 (42.4%)18,425 (43.9%)91,329 (46.7%)127,030 (45.6%)111,633 (28.5%)12,818 (30.5%)61,116 (31.3%)85,567 (30.7%)25,889 (14.4%)5,870 (14.0%)25,118 (12.8%)36,877 (13.3%)32,178 (5.3%)1912 (4.6%)7,607 (3.9%)11,697 (4.2%)41,259 (3.1%)1,025 (2.4%)3,942 (2.0%)6,226 (2.2%)≥52,545 (6.2%)1926 (4.6%)6,438 (3.3%)10,909 (3.9%)IPAQ activity group<0.0001Low39,524 (96.9%)8,807 (21.0%)1,222 (0.6%)49,553 (17.8%)Moderate1,256 (3.1%)33,169 (79.0%)78,983 (40.4%)113,408 (40.7%)High0 (0%)0 (0%)115,345 (59.0%)115,345 (41.4%)PM2.5*9.92 [8.2, 21.3]9.91 [8.17, 20.1]9.90 [8.17, 20.2]9.91 [8.2, 21.3]<0.0001PM10*16.0 [11.8, 30.7]16.0 [11.8, 26.3]16.0 [11.8, 31.4]16.0 [11.8, 31.4]<0.0001Greenspace percentage in 300m*29.8 [0.3, 99.2]29.3 [0.4, 99.1]29.8 [0.2, 99.2]29.7 [0.2, 99.2]<0.0001Nitrogen dioxide*26.2 [12.9, 97.1]26.2 [12.9, 108]26.0 [12.9, 108]26.0 [12.9, 108]<0.0001Nitrogen oxides*42.3 [19.7, 247]42.0 [19.7, 265]41.9 [19.7, 266]41.9 [19.7, 266]<0.0001

Characteristics of study population in UK Biobank.

3.2 Relationships of PAV score and total nervous system disorder incidence

As shown in Table 2, negative associations between PAV score and the incidence of nervous system diseases were shown (HR < 1). Consistently, after adjusted the potential confounders of Socioeconomic Status (Employment status and University education) and Health Behaviors factors (Alcohol drinker status, Smoking status, Sleep duration, Time watch TV, Time use computer, Coffee intake, Tea intake and IPAQ activity group), the association seemed somewhat strengthened in Grouped PAV score. Whereas, after the adjustment of Socioeconomic Status (Employment status and University education) in Continuous PAV score, the association was insignificant, which might be caused by statistical difference. Additionally, adjusting the potential confounders of Medical History (Diabetes, Hypertension and Cancer) and Living Environment (PM2.5, PM10, Greenspace percentage in 300 m, Nitrogen dioxide and Nitrogen oxides), the incidence of total nervous system disorder seemed decrease. The p values for trend were all below 0.05 in above situations except in model 2 for Continuous PAV score.

ModelGrouped PAV scoreContinuous PAV scoreHR (95% CI)PCox-regressionPCox-regressionModel 10.942 (0.929,0.954)<0.0010.003Model 20.931 (0.919,0.944)<0.0010.497Model 30.953 (0.941,0.967)<0.001<0.001Model 40.945 (0.925,0.966)<0.001<0.001Model 50.946 (0.926,0.967)<0.001<0.001

Associations between PAV score and total nervous system disorder incidence.

Model 1: Adjust Sex, Age at recruitment, Ethnic background, BMI and MET score.

Model 2: Adjust Variables in Model 1 plus Employment status and University education.

Model 3: Adjust Variables in Model 2 plus Diabetes, Hypertension and Cancer.

Model 4: Adjust Variables in Model 3 plus Alcohol drinker status, Smoking status, Sleep duration, Time watch TV, Time use computer, Coffee intake, Tea intake and IPAQ activity group.

Model 5: Adjust Variables in Model 4 plus PM2.5, PM10, Greenspace percentage in 300 m, Nitrogen dioxide and Nitrogen oxides.

Non-linear relationship between the score of PAV and incidence of nervous system disorder after adjusting potential confounding factors. The risk of nervous system disorder was lowest when the score of PAV was 1,356. Considering the influence of gender and sex, female had higher incidence of Nervous System Diseases and population age above 60 were more likely had Nervous Diseases. More details were presented in Figure 1.

Restricted cubic spline for testing the hypothesis of nonlinear correlation between MET score and total nervous system disorder incidence in (A). Total population, (B). Population grouped by sex, (C). Population grouped by age.

3.3 Relationships of PAV score and specific nervous system disorder incidence

Subgroup analysis was used to explored the relationships between PAV and some specific nervous system diseases. As shown in Figure 2, PAV had no influence on most nervous system diseases. While it was a risk factor to Meningitis in bacterial diseases classified elsewhere. Additionally, PAV played an protective effect in the following nervous system diseases: other degenerative disease of basal ganglia, other extrapyramidal and movement disorders, multiple sclerosis, transient cerebral ischaemic attacks and related syndromes, sleep disorder, nerve root and plexus compressions in diseases classified elsewhere, mononeuropathies of upper limb, other polyneuropathies, primary disorders of muscles, other myopathies, hemiplegia, paraplegia and tetraplegia, other paralytic syndromes, other disease of spinal cord.