Abstract
Background
Ulcerative colitis (UC) lacks established primary prevention strategies despite its economic burden. Physical activity (PA) is promising, yet studies using self-reported data show inconsistent results. We examined the association between accelerometer-measured physical activity and UC risk in a large prospective cohort.
Methods
This cohort study analyzed data from 95,197 UK Biobank participants without baseline inflammatory bowel disease and with valid accelerometer data, with follow-up from 2013 to 2015 through 2023. Machine learning models estimated time spent in sedentary behavior, light physical activity (LPA), and moderate-to-vigorous physical activity (MVPA). MVPA patterns were categorized as active weekend warriors (≥231.5 min/week, ≥50% in 1–2 days), active regular (≥231.5 min/week, not meeting weekend warrior criteria), and inactive (<231.5 min/week). Cox proportional hazards regression assessed associations between PA and UC risk.
Results
Over 8.86 years median follow-up, 247 incident UC cases were recorded. MVPA showed a dose–response relationship with UC risk (trend p = 0.004). Compared to the lowest MVPA quartile, the third (231.5 to <402.3 min/week) and highest (>402.3 min/week) quartiles showed multivariable adjusted hazard ratios (aHRs) of 0.59 and 0.63 (95% CI: 0.40–0.85 and 0.43–0.91). Neither LPA nor sedentary time showed significant associations with UC risk. Both weekend warrior (aHR: 0.72, 95% CI: 0.53–0.99) and regular MVPA patterns (aHR: 0.57, 95% CI: 0.39–0.82) correlated with reduced UC risk. These associations remained robust across various subgroups and sensitivity analyses.
Conclusion
Accelerometer-measured MVPA is associated with reduced UC risk. Both weekend warrior and regular MVPA patterns conferred protection, suggesting that flexibility in exercise scheduling may be feasible.
Keywords: accelerometer, IBD, physical activity, primary prevention, ulcerative colitis
1. Introduction
Ulcerative colitis (UC), a lifelong inflammatory bowel disease (IBD), poses a growing global health challenge (1–3). With a worldwide prevalence of 5 million cases as of 2023 (1) and rising incidence rates in newly industrialized and select high-income nations (2, 3), UC imposes substantial economic burdens on healthcare systems (2, 3). The absence of established preventive strategies underscores the critical need for research into potential risk-reduction measures.
The etiology of UC involves both genetic and environmental factors, including sedentary lifestyle (4). Physical activity (PA), the energy-expending movement of skeletal muscles (5), has shown beneficial effects on various non-communicable diseases (6). Recent guidelines encourage regular exercise for IBD patients (7). However, Prospective studies on the PA-UC risk relationship have produced mixed findings. The Nurses’ Health Study, involving 194,711 women, found no significant link between high PA levels and UC risk (8). Intriguingly, a follow-up analysis of the same cohort suggested that 42% of UC cases might be preventable through a healthy lifestyle, with adequate PA as a key component (9). Conversely, the European Prospective Investigation into Cancer and Nutrition (EPIC) and a Danish cohort study reported no relationship between PA levels and IBD risk (10, 11). Notably, questionnaire-based PA assessments in these studies face limitations in accurately capturing intensity, duration, and energy expenditure (12), while being prone to recall bias and measurement errors.
While prior studies largely relied on self-reported physical activity (PA) with inherent recall and misclassification biases (13, 14), this study leverages wrist-worn accelerometry in approximately 90,000 UK Biobank participants to objectively quantify the PA-ulcerative colitis (UC) association. Specifically, we aimed to: (1) delineate the precise dose–response relationship and identify a reproducible MVPA threshold (i.e., 231.5 min/week) for UC primary prevention; and (2) compare the protective efficacy of the “weekend warrior” pattern (concentrated 1–2 days) against regular daily MVPA, thereby assessing the feasibility of flexible exercise scheduling. Additionally, we evaluated the independent effects of light PA and sedentary time, and tested the robustness of our findings across diverse subgroups. By providing these accelerometer-derived, quantifiable targets and pattern-specific risk estimates, this study moves beyond methodological refinement to offer evidence-based, temporally flexible strategies for UC prevention, potentially alleviating the growing burden of this disease.
2. Methods
2.1. Data source and study design
Drawing from the UK Biobank, a large-scale prospective cohort study, this research analyzed data from about 500,000 participants. Recruitment occurred between 2006 and 2010, encompassing individuals aged 40–69 from 22 assessment centers across England, Wales, and Scotland (15). Ethical approval for the UK Biobank study was granted by the North West Research Ethics Committee, with all participants providing written informed consent. The study design adhered to STROBE guidelines for cohort studies. Participants completed a comprehensive baseline assessment, including a touch-screen questionnaire and computer-assisted interview, capturing detailed sociodemographic, lifestyle, and health data. Physical examinations were conducted, and biospecimens were collected.
The analysis focused on a subsample of 103,662 participants who received random invitations via email to wear wrist-worn accelerometers (AX3; Axivity Ltd) continuously for 7 days between 2013 and 2015 (16, 17). The participation rate was 45%, with no significant baseline differences between respondents and non-respondents (17). Accelerometers captured triaxial movement at 100 Hz, with a sensitivity range of ± 8 g. Data were processed following established protocols (17), including gravity calibration. The raw data were condensed into 5-s intervals and quantified using average vector magnitude. Non-wear periods (stationary intervals exceeding 60 min with standard deviation below 13.0 milli-gravity across all axes (17)) were imputed using the average vector magnitude and intensity distribution from corresponding time points on other days.
Exclusion criteria were applied to participants failing accelerometer data quality control (n = 7,052), including insufficient wearing time (n = 6,984), calibration failure (n = 17), insufficient data (n = 38), and implausibly high acceleration values (n = 13). Participants with prevalent inflammatory bowel disease before wearing the accelerometer were also excluded (n = 1,433). The final analytic sample comprised 95,197 participants (Figure 1).
Figure 1.
The flow chart of participants through the study.
2.2. Exposure assessment
Total mean PA volume was calculated as the average vector magnitude in milligravity units (17). PA intensities were categorized using machine learning models (16) trained on free-living data from time-use diaries and wearable cameras. PA was categorized based on metabolic equivalent of tasks (METs): sedentary time (≤1.5 METs, e.g., driving or TV watching), light physical activity (LPA) (<3 METs, excluding sedentary activities, e.g., slow walking or self-care), and moderate-to-vigorous physical activity (MVPA) (≥3 METs, e.g., swimming, jogging, or cycling). For each participant, total wear duration and time allocated to different activity intensities were computed.
The MVPA proportions were converted to minutes per day, facilitating participant categorization into Inactive, Active-Weekend Warrior, and Active-Regular groups. The primary analysis employed two thresholds: the guideline-based ≥150 min/week (5, 18) and the sample median of ≥230.4 min/week, reflecting UK national health surveys indicating approximately half of the population as physically active (19). Secondary analyses explored additional thresholds. Participants were categorized as active weekend warriors if they met or exceeded the MVPA threshold with at least half of their total MVPA occurring over 1–2 days (6). Those meeting the MVPA threshold without the weekend warrior pattern were classified as active regular, while those below the threshold were deemed inactive.
The UK Biobank Data Showcase provides access to all study variables for registered researchers (Overall acceleration average: Field 90,012, PA intensities: Fields 40,047–40,049, Moderate-Vigorous Day average: Field 40,045).
2.3. Ascertainment of UC
The IBD cases were identified using 3-digit ICD-10 code (K51 for UC, K50 for CD) and UK Biobank self-reported data (IBD: Field 1,461, CD: Field 1,462, UC: Field 1,463). Participant hospitalization data were updated until October 1, 2023. Participants were followed until UC diagnosis, death, or the end of follow-up, whichever occurred first; those without UC were censored at their last follow-up date.
2.4. Covariates assessment
Potential confounding factors were assessed using data from touch-screen questionnaires, anthropometry, biochemical markers, accelerometer outputs, and medical histories. Confounders examined included sex, age, Townsend deprivation index (TDI), education level, Body mass index (BMI), alcohol consumption frequency, smoking status, non-steroidal anti-inflammatory drug (NSAID) and antibiotic use, and pre-existing cardiovascular disease (CVD), diabetes, and hypertension (identified by self-report or health records pre-accelerometer). NSAID use was included as a covariate given its previously reported potential association with UC risk (9, 20). Additional variables for females were hormone therapy, oral contraception, menopausal status, and parity. All covariates were derived from the baseline assessment (Instance 0, 2006–2010), as it provided complete data for all participants; Instance 1 (2012–2013) was not used due to its limited sample size (n = 20,312).
2.5. Statistical analysis
Participant characteristics at study entry were summarized by total PA volume quartiles, presenting means with SDs for continuous measures and percentages for categorical variables.
Cox proportional hazards regression models assessed the association between exposure quartiles (sedentary time, LPA, and MVPA) and UC incidence risk. Follow-up started at accelerometer wear completion. Models were adjusted for sex, age, BMI, alcohol consumption, smoking status, socioeconomic status, education level, medical history (including cardiovascular disease, hypertension, and diabetes), medication use (NSAIDs and antibiotics), and female-specific variables (menopausal status, parity, contraceptive use, and hormone replacement therapy). For female-specific variables, we used interaction terms with sex.
Missing covariate data were handled via complete-case analysis, yielding 94,676 participants with complete data for all covariates (99.5% of the original cohort) in the final multivariate models. Supplementary Table 1 details the missing covariates among participants. To evaluate independent associations, we additionally constructed a mutually adjusted Cox model that simultaneously included LPA, MVPA, and sedentary time as quartile-based exposures, with results presented in Supplementary Table 2. Schoenfeld residuals analysis confirmed the proportional hazards assumption, showing no violations. A correlation matrix was generated to evaluate potential collinearity between covariates. No substantial multicollinearity issues were identified, except for the expected high correlations between sex variables and their interactions with female-specific variables.
Restricted cubic spline Cox models evaluated the dose–response relationship between PA, sedentary time, and UC incidence risk. Subgroup analyses assessed robustness and explored variations by stratifying for sex, age (<65, ≥65), BMI (<25, ≥25), TDI (< median, ≥ median), smoking status (never, current or former), or cardiovascular disease status (yes, no). Potential interactions between these risk factors and PA were assessed using cross-classification. Interaction significance was assessed via log-likelihood ratio tests, comparing models with cross-classified categories against those treating factors independently.
Robustness checks were performed to validate the findings. These included excluding events within 2 years post-accelerometer wear to mitigate reverse causality concerns, and employing competing risk models using Fine-Gray subdistribution hazards to assess potential bias, with death as a competing risk for UC incidence.
Analyses were conducted in R 4.4.1, employing two-sided tests with significance threshold p < 0.05.
3. Results
3.1. Characteristics of the study population
The study included 95,197 participants (56.3% women) with a mean age of 62.31 years (SD 7.85) and BMI of 26.7 kg/m2 (SD 4.5). Average weekly PA durations were: LPA 2121.90 min (SD 690.21), MVPA 290.06 min (SD 243.30), and sedentary behavior 3949.80 min (SD 763.81) (Figure 1 and Table 1). After excluding participants with missing covariate data, a total of 94,676 participants (99.5% of the original cohort) with complete data were included in the final multivariate regression analyses.
Table 1.
Participant characteristics by fourths of total mean accelerometer-measured physical activity volume levels.
| Characteristics | Overall | Q1 | Q2 | Q3 | Q4 |
|---|---|---|---|---|---|
| n | 95,197 | 23,824 | 23,821 | 23,772 | 23,780 |
| Sex | |||||
| Female | 53,632 (56.3) | 11,896 (49.9) | 13,550 (56.9) | 14,049 (59.1) | 14,137 (59.4) |
| Male | 41,565 (43.7) | 11,928 (50.1) | 10,271 (43.1) | 9,723 (40.9) | 9,643 (40.6) |
| Age at accelerometer assessment | 62.31 (7.85) | 65.07 (7.31) | 63.08 (7.63) | 61.58 (7.73) | 59.50 (7.65) |
| BMI (kg/m2) | 26.71 (4.53) | 28.45 (5.15) | 26.96 (4.40) | 26.25 (4.09) | 25.17 (3.70) |
| Townsend Index scorea | −1.73 (2.82) | −1.59 (2.90) | −1.78 (2.80) | −1.81 (2.77) | −1.74 (2.80) |
| College degree or equivalent | 42,471 (44.6) | 9,820 (41.2) | 10,782 (45.3) | 10,856 (45.7) | 11,013 (46.3) |
| Cigarette smoking | |||||
| Never | 54,326 (57.1) | 12,613 (52.9) | 13,565 (56.9) | 13,908 (58.5) | 14,240 (59.9) |
| Previous | 34,248 (36.0) | 9,071 (38.1) | 8,649 (36.3) | 8,387 (35.3) | 8,141 (34.2) |
| Current | 6,623 (7.0) | 2,140 (9.0) | 1,607 (6.7) | 1,477 (6.2) | 1,399 (5.9) |
| Alcohol consumption | |||||
| Never | 5,381 (5.7) | 1,657 (7.0) | 1,277 (5.4) | 1,215 (5.1) | 1,232 (5.2) |
| <1 time/week | 19,350 (20.3) | 5,660 (23.8) | 4,744 (19.9) | 4,595 (19.3) | 4,351 (18.3) |
| 1–2 time/week | 23,887 (25.1) | 5,751 (24.1) | 6,028 (25.3) | 5,997 (25.2) | 6,111 (25.7) |
| 3–4 time/week | 24,809 (26.1) | 5,469 (23.0) | 6,153 (25.8) | 6,465 (27.2) | 6,722 (28.3) |
| >4 time/week | 21,770 (22.9) | 5,287 (22.2) | 5,619 (23.6) | 5,500 (23.1) | 5,364 (22.6) |
| Pre-existing chronic conditions | |||||
| Cardiovascular disease | 5,498 (5.8) | 2,451 (10.3) | 1,402 (5.9) | 1,034 (4.3) | 611 (2.6) |
| Hypertension | 25,782 (27.1) | 9,415 (39.5) | 6,832 (28.7) | 5,488 (23.1) | 4,047 (17.0) |
| Diabetes | 4,408 (4.6) | 2,216 (9.3) | 1,030 (4.3) | 721 (3.0) | 441 (1.9) |
| Regular use of NSAIDsb | 23,785 (25.0) | 6,987 (29.3) | 6,035 (25.3) | 5,614 (23.6) | 5,149 (21.7) |
| Antibiotic use in past 3 months | 1,289 (1.4) | 310 (1.3) | 313 (1.3) | 335 (1.4) | 331 (1.4) |
| Female-specific variables | |||||
| Postmenopause | 36,279 (67.6) | 8,552 (71.9) | 9,446 (69.7) | 9,422 (67.1) | 8,859 (62.7) |
| Parity | |||||
| 0 | 11,504 (21.4) | 2,509 (21.1) | 2,850 (21.0) | 2,988 (21.3) | 3,157 (22.3) |
| 1–2 | 30,601 (57.1) | 6,765 (56.9) | 7,765 (57.3) | 8,033 (57.2) | 8,038 (56.9) |
| 3+ | 11,527 (21.5) | 2,622 (22.0) | 2,935 (21.7) | 3,028 (21.6) | 2,942 (20.8) |
| Ever use of oral contraceptives | 45,778 (85.4) | 9,767 (82.1) | 11,553 (85.3) | 12,131 (86.3) | 12,327 (87.2) |
| Ever use of HRT | 20,852 (38.9) | 5,639 (47.4) | 5,684 (41.9) | 5,216 (37.1) | 4,313 (30.5) |
| LPA (min/week) | 2121.9 (690.2) | 1526.9 (464.3) | 2005.8 (478.1) | 2296.8 (549.9) | 2659.6 (687.0) |
| MVPA (min/week) | 290.1 (243.3) | 145.5 (127.5) | 233.0 (167.7) | 310.2(207.0) | 471.9 (303.3) |
| Sedentary time (min/week) | 3949.8 (763.8) | 4447.8 (732.1) | 4094.6 (633.9) | 3834.5 (622.0) | 3421.2 (668.1) |
Continuous variables are presented as mean (SD); categorical variables are presented as n (%).
aTownsend deprivation index = a composite area-level measure of deprivation based on unemployment, non-car ownership, non-home ownership, and household overcrowding; a higher score indicates higher deprivation.
bRegular NSAID use was assessed through a questionnaire. Participants were asked about their regular use (defined as most days of the week for the last 4 weeks) of common over-the-counter medications, including various brand names for aspirin, ibuprofen, paracetamol, and codeine. The questionnaire provided a comprehensive list of brand names to ensure accurate reporting of NSAID use, including combinations and less commonly recognized trade names.
BMI, Body Mass Index; NSAID, nonsteroidal anti-inflammatory drug; HRT, hormone replacement therapy; LPA, light physical activity; MVPA, moderate-to-vigorous physical activity.
Over 8.86 years median follow-up (interquartile range, 8.30–9.37), 247 incident UC cases were recorded. Participants were stratified into quartiles (Q1–Q4) based on accelerometer-measured Total mean PA volume levels (Table 1). Compared to the least active quartile (Q1), the most active quartile (Q4) comprised younger individuals with lower BMI and a higher proportion of females. An inverse trend in chronic disease (including cardiovascular disease, hypertension, and diabetes) prevalence was observed from Q1 to Q4. Q4 reported less NSAIDs and hormone replacement therapy use but higher oral contraceptive use. While differences in education level, material deprivation, smoking status, and alcohol consumption were noted across quartiles, these socioeconomic and lifestyle factors showed less pronounced trends compared to the health indicators.
3.2. Association of PA intensity and sedentary time with incident UC
Restricted cubic spline Cox regression revealed a U-shaped association between TPA and UC risk (P for non-linearity = 0.0018, Figure 2), with the nadir at approximately 30.9 milligravity (aHR: 0.54). LPA showed no significant association with UC risk, while MVPA demonstrated a significant dose–response relationship (P for trend = 0.004). Compared to the lowest MVPA quartile, the third (231.5 to <402.3 min/week) and highest quartiles (> 402.3 min/week) had aHRs of 0.53 and 0.57 (95% CI: 0.35–0.82 and 0.37–0.87). The magnitude of risk reduction was not substantially different between these two highest quartiles, suggest that a potential plateauing of the protective effect at higher activity levels. Sedentary time showed no significant link to UC risk (all in Table 2). These associations remained materially unchanged in mutually adjusted models simultaneously accounting for LPA, MVPA, and sedentary time (Supplementary Table 2).
Figure 2.
Restricted cubic spline for the association between accelerometer-measured average total volume, and moderate-to-vigorous-intensity physical activities and the risk of UC incidence. All adjusted for age, sex, body mass index, Townsend Deprivation Index, educational level, smoking status, alcohol consumption, and pre-existing chronic conditions (hypertension, diabetes and cardiovascular disease), non-steroid anti-inflammatory drug use, antibiotics use, and, among female participants, use of oral contraception, menopausal status, hormone therapy and parity. The solid line is the estimate (HR) and the shaded area is the 95% CI. HR indicates hazard ratio; CI, confidence interval; UC, ulcerative colitis.
Table 2.
Association of light, moderate-to-vigorous physical activity and sedentary time with the risk of incident ulcerative colitis.
| Exposure | Total | Events (ratea) | Crude Model | Adjusted Modelb |
|---|---|---|---|---|
| HR (95%CI), p-value | HR (95%CI), P-value | |||
| Light physical activity, min/week | ||||
| Q1 (<1634.9) | 23,675 | 66 (3.2) | Ref | Ref |
| Q2 (1634.9- to <2068.7) | 23,668 | 64 (3.1) | 0.97 (0.66, 1.43), 0.873 | 1.06 (0.72, 1.58), 0.758 |
| Q3 (2068.7- to <2552.1) | 23,677 | 48 (2.3) | 0.68 (0.44, 1.04), 0.0736 | 0.77 (0.49, 1.19), 0.241 |
| Q4 (>2552.1) | 23,656 | 65 (3.1) | 0.94 (0.64, 1.40), 0.774 | 1.10 (0.73, 1.66), 0.658 |
| P for trend | 0.437 | 0.986 | ||
| Moderate to vigorous physical activity, min/week | ||||
| Q1 (<112.5) | 23,743 | 81 (4.0) | Ref | Ref |
| Q2 (112.5- to <231.5) | 23,597 | 65 (3.2) | 0.75 (0.52, 1.08), 0.125 | 0.78 (0.53, 1.13), 0.188 |
| Q3 (231.5- to <402.3) | 23,689 | 46 (2.2) | 0.52 (0.34, 0.78), 0.00174 | 0.53 (0.35, 0.82), 0.00426 |
| Q4 (>402.3) | 23,647 | 51 (2.5) | 0.56 (0.37, 0.84), 0.00486 | 0.57 (0.37, 0.87), 0.00991 |
| P for trend | 0.0009 | 0.00262 | ||
| Sedentary time, min/week | ||||
| Q1 (<3437.9) | 23,673 | 58 (2.8) | Ref | Ref |
| Q2 (3437.9- to <3950.9) | 23,672 | 53 (2.6) | 1.00 (0.65, 1.54), 0.992 | 0.99 (0.64, 1.52), 0.964 |
| Q3 (3950.9- to <4457.8) | 23,668 | 66 (3.2) | 1.17 (0.78, 1.77), 0.448 | 1.13 (0.75, 1.72), 0.555 |
| Q4 (>4457.8) | 23,663 | 66 (3.2) | 1.26 (0.84, 1.89), 0.269 | 1.11 (0.73, 1.70), 0.621 |
| P for trend | 0.196 | 0.508 | ||
aIncidence rates per 10,000 person-years.
bAdjusted for age, sex, body mass index, Townsend Deprivation Index, educational level, smoking status, alcohol consumption, and pre-existing chronic conditions (hypertension, diabetes and cardiovascular disease), non-steroid anti-inflammatory drug use, antibiotics use, and, among female participants, use of oral contraception, menopausal status, hormone therapy and parity.
HR, hazard ratio; CI, confidence interval; Ref, reference.
These findings remained consistent across a 2-year blanking period (Supplementary Table 3) and in competing risk analyses considering death (Supplementary Table 4). The null association for sedentary time was also confirmed in the mutually adjusted model (Supplementary Table 2). The PA-UC relationship was not significantly modified by sex, age (<65, ≥65), BMI (<25, ≥25), TDI (<median, ≥median), smoking status (never, current or former), or CVD status (yes, no) (all P for interaction >0.1) (Supplementary Tables 5–10).
3.3. Association of MVPA pattern with incident UC
After identifying a significant link between MVPA and UC risk, we investigated optimal MVPA patterns. The UK National Health Service advises distributing MVPA throughout the week (18). However, the benefits of concentrating exercise into 1–2 days [“weekend warrior” pattern (21)] compared to more consistent activity remain unclear. We analyzed 94,099 participants with complete daily MVPA data, excluding 1,089 individuals from the initial cohort of 95,197 due to incomplete records (Supplementary Tables 11, 12). Based on a median MVPA threshold of ≥231.5 min/week, we classified 26,235 (27.9%) participants as active weekend warriors, 20,832 (22.1%) as active regular, and 47,032 (50.0%) as inactive (Supplementary Table 12). Active weekend warriors demonstrated substantially higher MVPA on their two most active days versus the other five (Figure 3 and Supplementary Figure 1). In contrast, the active regular group exhibited more consistent MVPA distribution throughout the week.
Figure 3.
Weekly MVPA distribution: top 2 days vs. remaining 5 days in active individuals (active: ≥231.5 min/week MVPA, sample median). MVPA distribution for active individuals (≥231.5 min/week, n = 47,067): 2 most active days (blue) vs. remaining 5 (yellow). (A) Weekend warriors (≥50% MVPA in 1–2 days). (B) Regular actives. Note: 147 weekend warriors (0.6%) with zero MVPA on 5 days assigned 1 min for log scale. MVPA indicates moderate to vigorous physical activity.
Multivariable-adjusted models revealed that compared to inactive individuals, both weekend warrior and regular activity patterns correlated with reduced risk of UC (Table 3). This association remained consistent across various MVPA thresholds (Table 3). At the 25th percentile threshold (≥112.5 min/week), both weekend warriors (aHR 0.74, 95% CI 0.55–0.99) and regular activity participants (aHR 0.59, 95% CI 0.41–0.84) showed significantly reduced UC risk. Similarly, at the 50th percentile (≥231.5 min/week), weekend warriors (aHR 0.72, 95% CI 0.53–0.99) and regular activity (aHR 0.57, 95% CI 0.39–0.82) were protective. For the guideline-based threshold (≥150 min/week), only regular activity was significantly protective (aHR 0.63, 95% CI 0.44–0.90). At the highest threshold (≥402.0 min/week), trends persisted but lost statistical significance. Event rates were comparable between the active weekend warrior and active regular groups (Figures 4A,B). When stratified by MVPA deciles, the weekend warrior pattern did not significantly differ from the regular activity pattern in terms of UC risk (aHR 1.08, 95% CI 0.78–1.48). Analyses using a 2-year blanking period and competing risk models (accounting for death) further supported the main findings (Supplementary Tables 13, 14). Notably, analyses with the 2-year blanking period demonstrated more pronounced risk reduction at lower thresholds (Supplementary Table 13). Overall, regular activity consistently demonstrated the lowest aHRs across all thresholds, suggesting more consistent protection against UC risk.
Table 3.
Associations between physical activity pattern and incident ulcerative colitis across varying MVPA thresholda.
| Activity threshold | Activity Patternb | Event (ratec) | Hazard ratio (95% CI) | p_value |
|---|---|---|---|---|
| 25th percentile (≥112.5 min MVPA/week) | Inactive (n = 23,480) | 81 (4.01) | Ref | Ref |
| Weekend warrior (n = 45,559) | 111 (2.78) | 0.74 (0.55–0.99) | 0.046 | |
| Regular activity (n = 25,060) | 51 (2.32) | 0.59 (0.41–0.84) | 0.004 | |
| Guideline-based (≥150 min MVPA/week) | Inactive (n = 31,414) | 101 (3.73) | Ref | Ref |
| Weekend warrior (n = 38,772) | 93 (2.74) | 0.78 (0.58–1.04) | 0.091 | |
| Regular activity (n = 23,913) | 49 (2.34) | 0.63 (0.44–0.90) | 0.011 | |
| 50th percentile (≥231.5 min MVPA/week) | Inactive (n = 47,032) | 147 (3.61) | Ref | Ref |
| Weekend warrior (n = 26,235) | 58 (2.52) | 0.72 (0.53–0.99) | 0.043 | |
| Regular activity (n = 20,832) | 38 (2.08) | 0.57 (0.39–0.82) | 0.003 | |
| 75th percentile (≥402.0 min MVPA/week) | Inactive (n = 70,568) | 194 (3.16) | Ref | Ref |
| Weekend warrior (n = 11,044) | 25 (2.58) | 0.85 (0.55–1.30) | 0.449 | |
| Regular activity (n = 12,487) | 24 (2.18) | 0.69 (0.45–1.06) | 0.088 |
aAdjusted for age, sex, body mass index, townsend deprivation index, educational level, smoking status, alcohol consumption, and pre-existing chronic conditions (hypertension, diabetes and cardiovascular disease), non-steroid anti-inflammatory drug use, antibiotics use, and, among female participants, use of oral contraception, menopausal status, hormone therapy and parity.
bInactive defined as weekly MVPA below the specified threshold.
cIncidence rates per 10,000 person-years.
MVPA, moderate-to-vigorous physical activity; CI, confidence interval; Ref, reference.
Figure 4.
Cumulative risk of incident UC events stratified by activity pattern. Crude cumulative risk of incident UC events across accelerometer-derived activity patterns: Inactive (orange), Active-Weekend Warrior (blue), and Active-Regular (gray). Panel (A) uses the sample median activity threshold (≥231.5 min MVPA/week), while Panel (B) uses the guideline-based threshold (≥150 min MVPA/week). At-risk numbers are shown below each plot. Shaded areas represent 95% confidence intervals. UC, ulcerative colitis; MVPA, moderate-to-vigorous physical activity.
4. Discussion
4.1. Main findings
This large-scale cohort study found MVPA significantly lowered UC risk after adjusting for potential confounders. A dose–response relationship was observed, with 231.5 to over 402.3 min of weekly MVPA (such as swimming, jogging, stationary cycling, or other aerobic activities) associated with a 37–41% lower adjusted UC risk compared to the lowest quartile. Notably, this protective effect plateaued at higher activity levels, suggesting that moderate MVPA may suffice for optimal UC risk reduction. Adjusted analyses showed no significant associations between UC risk and either LPA or sedentary time. The protective effect of MVPA remained significant after mutual adjustment for LPA and sedentary time, confirming its independent role in UC risk reduction. Both consistent weekly MVPA and the “weekend warrior” approach (concentrating MVPA into 1–2 days per week) were associated with reduced UC risk in multivariable-adjusted models. These findings remained robust across various sensitivity analyses and subgroup examinations.
4.2. Clinical implications
These results offer significant implications for UC prevention strategies. Firstly, the study revealed an inverse dose–response link between MVPA and UC incidence risk, with 231.5–402.3 min of MVPA per week significantly reducing UC incidence risk by 41% (aHR: 0.59, 95% CI: 0.40–0.85). It is suggested that adults engage in 231.5–402.3 min of MVPA weekly. These accelerometer-based measurements provide valuable reference for developing UC prevention guidelines. Notably, a potential plateau effect in protective benefits was observed at higher activity levels (>402.3 min/week), suggesting there may be an optimal range of activity (Table 2). Secondly, this study is the first to compare the impact of the “weekend warrior” pattern and regular exercise patterns on UC risk. Results indicated that both patterns were associated with reduced UC risk across various MVPA thresholds. At the 231.5 min per week threshold, both weekend warriors and regular activity demonstrated significant protective effects. Direct comparison between the two active patterns showed no significant difference (aHR 1.08, 95% CI: 0.78–1.48), suggesting that the protective effects are broadly comparable despite the slightly lower point estimate for regular activity. For those with hectic weekdays, weekend-concentrated exercise may still offer health benefits. However, the regular activity pattern consistently showed the lowest aHRs across all thresholds, indicating it may be the most ideal exercise pattern, aligning with existing public health recommendations (5). However, given the observational nature of our study and the relatively small number of UC events in subgroup analyses, this finding should be interpreted with caution and warrants confirmation in future prospective studies with larger case numbers. Overall, this study strongly supports MVPA as a potential UC prevention strategy, with both weekend warrior and regular exercise regimens showing protective effects.
4.3. Comparison with other studies and implications
Our findings align with a recent meta-analysis (3 cohort, 7 case–control studies) showing high PA significantly lowered UC risk versus low activity (relative risk [RR]: 0.62 [95% CI: 0.43, 0.88], p = 0.008) (22). Similarly, a study objectively measuring physical fitness in over 240,000 Swedish adolescent males found an inverse correlation between physical fitness and UC risk (23). However, these findings diverge from several questionnaire-based prospective cohort studies that found no association between PA and UC risk (8, 10, 11). Indeed, the variability in PA definitions across different studies makes comparisons challenging (24). Questionnaire-based surveys not only suffer from recall bias but also have limited reliability (12). For instance, the frequent, yet irregular occurrence of LPA challenges its precise measurement through self-reported surveys.
The protective effect of MVPA (such as swimming, jogging, stationary cycling, or other aerobic activities) on UC risk likely results from multiple interacting factors. Animal studies have shown that MVPA reduces chronic bowel inflammation by inhibiting local proinflammatory responses in DSS-induced colitis (25, 26). Additionally, MVPA may modulate the gut microbiome, which is pivotal in regulating exercise-induced oxidative stress, inflammation, and metabolism (27). Notably, the conversion of exercise-induced lactate to propionate by specific microbes may be protective (28), given that lower propionate levels have been observed in UC patients (29). Furthermore, β-hydroxybutyrate, produced during aerobic exercise, contributes to this effect by promoting regulatory T cell differentiation, thus reducing the Th17/Treg ratio (30). PA may also activate autophagy, a process known to prevent various inflammatory conditions (31). These multifaceted pathways likely work synergistically to confer protection against UC development.
4.4. Limitations
This study has several limitations. Firstly, the UK Biobank cohort’s older, predominantly European demographic may bias findings toward late-onset IBD and limit generalizability. Secondly, the optimal MVPA threshold for wrist-worn accelerometers remains uncertain. Thirdly, as an observational study, we cannot definitively rule out residual confounding or reverse causality, despite adjusting for numerous confounders. Notably, our analytical model did not incorporate all potential dietary factors (e.g., fiber, red or processed meat intake), which may have contributed to residual confounding. Additionally, temporal gap between accelerometer data and covariates may affect baseline accuracy, despite using most temporally proximate covariate data and known variable stability (13). Fifth, the one-week measurement period may not fully represent long-term activity patterns, although studies have shown moderate to high correlations in repeated assessments over extended periods (32–34). Sixth, our MVPA classification method, while validated (16), may underestimate non-ambulatory activities. Lastly, we did not account for the potential impact of daily timing of PA on health outcomes, an important consideration for future research.
5. Conclusion
In this large prospective cohort of approximately 95,000 UK Biobank participants, accelerometer-measured MVPA was associated with a significantly reduced risk of incident UC over a median follow-up of 8.86 years. The protective association followed a non-linear dose–response pattern, with risk reduction appearing to plateau at higher activity levels. Notably, both the ‘weekend warrior’ pattern (concentrated activity within 1–2 days) and regular daily MVPA were associated with comparable risk reductions, suggesting that accumulated MVPA—rather than its weekly distribution—may be the key determinant of UC prevention. While MVPA represents a promising modifiable target for UC primary prevention, further studies are warranted to validate optimal thresholds in diverse populations.
Funding Statement
The author(s) declared that financial support was received for this work and/or its publication. This study was funded by Shenzhen Science and Technology Program (No. JCYJ20240813145408012), National Natural Science Foundation of China (Nos. 82601067 and 82270581), China Postdoctoral Science Foundation (Nos. 2025M782046 and 2024M751322).
Edited by: Bowen Li, Shanxi Normal University, China
Reviewed by: Haiyang Wang, Anhui Normal University, China
Haoting Shi, Shanghai Jiao Tong University, China
Ogail Dawod, Jazan University, Saudi Arabia
Abbreviations: AHR, adjusted hazard ratio; BMI, body mass index; CI, confidence interval; CVD, cardiovascular disease; EPIC, European Prospective Investigation into Cancer and Nutrition; IBD, inflammatory bowel disease; LPA, light physical activity; METs, metabolic equivalent of tasks; MVPA, moderate-to-vigorous physical activity; NSAID, non-steroidal anti-inflammatory drug; PA, Physical activity; SD, standard deviation; STROBE, Strengthening Reporting of Observational Studies in Epidemiology; TDI, Townsend deprivation index; TPA, Total mean physical activity volume; UC, Ulcerative colitis.
Data availability statement
Publicly available datasets were analyzed in this study. The data that support the findings of this study are available from UK Biobank (https://www.ukbiobank.ac.uk/enable-your-research/register) but restrictions apply to availability. Data access requires application and approval through UK Biobank’s Access Management System.
Ethics statement
The studies involving humans were approved by Ethical approval for the UK Biobank study was granted by the North West Research Ethics Committee (ref 11/NW/0382), with all participants providing written informed consent. The studies were conducted in accordance with the local legislation and institutional requirements. Written informed consent for participation was not required from the participants or the participants’ legal guardians/next of kin in accordance with the national legislation and institutional requirements.
Author contributions
ZL: Conceptualization, Writing – original draft, Writing – review & editing, Data curation, Formal analysis. HK: Conceptualization, Data curation, Formal analysis, Funding acquisition, Supervision, Writing – review & editing. HF: Data curation, Formal analysis, Writing – review & editing. QL: Data curation, Formal analysis, Writing – review & editing. YC: Conceptualization, Funding acquisition, Supervision, Writing – review & editing.
Conflict of interest
The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Generative AI statement
The author(s) declared that Generative AI was not used in the creation of this manuscript.
Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.
Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
Supplementary material
The Supplementary material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpubh.2026.1955306/full#supplementary-material
References
- 1.Le Berre C, Honap S, Peyrin-Biroulet L. Ulcerative colitis. Lancet. (2023) 402:571–84. doi: 10.1016/S0140-6736(23)00966-2, [DOI] [PubMed] [Google Scholar]
- 2.Ng SC, Shi HY, Hamidi N, Underwood FE, Tang W, Benchimol EI, et al. Worldwide incidence and prevalence of inflammatory bowel disease in the 21st century: a systematic review of population-based studies. Lancet. (2017) 390:2769–78. doi: 10.1016/S0140-6736(17)32448-0, [DOI] [PubMed] [Google Scholar]
- 3.Buie MJ, Quan J, Windsor JW, Coward S, Hansen TM, King JA, et al. Global hospitalization trends for Crohn’s disease and ulcerative colitis in the 21st century: a systematic review with temporal analyses. Clin Gastroenterol Hepatol. (2023) 21:2211–21. doi: 10.1016/j.cgh.2022.06.030, [DOI] [PubMed] [Google Scholar]
- 4.Piovani D, Danese S, Peyrin-Biroulet L, Bonovas S. Environmental, nutritional, and socioeconomic determinants of IBD incidence: a global ecological study. J Crohns Colitis. (2020) 14:323–31. doi: 10.1093/ecco-jcc/jjz150, [DOI] [PubMed] [Google Scholar]
- 5.Bull FC, Al-Ansari SS, Biddle S, Borodulin K, Buman MP, Cardon G, et al. World Health Organization 2020 guidelines on physical activity and sedentary behaviour. Br J Sports Med. (2020) 54:1451–62. doi: 10.1136/bjsports-2020-102955, [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Khurshid S, Al-Alusi MA, Churchill TW, Guseh JS, Ellinor PT. Accelerometer-derived “weekend warrior” physical activity and incident cardiovascular disease. JAMA. (2023) 330:247–52. doi: 10.1001/jama.2023.10875, [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Ananthakrishnan AN, Kaplan GG, Bernstein CN, Burke KE, Lochhead PJ, Sasson AN, et al. Lifestyle, behaviour, and environmental modification for the management of patients with inflammatory bowel diseases: an International Organization for Study of inflammatory bowel diseases consensus. Lancet Gastroenterol Hepatol. (2022) 7:666–78. doi: 10.1016/S2468-1253(22)00021-8, [DOI] [PubMed] [Google Scholar]
- 8.Khalili H, Ananthakrishnan AN, Konijeti GG, Liao X, Higuchi LM, Fuchs CS, et al. Physical activity and risk of inflammatory bowel disease: prospective study from the nurses’ health study cohorts. BMJ. (2013) 347:f6633. doi: 10.1136/bmj.f6633, [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Lopes EW, Chan SSM, Song M, Ludvigsson JF, Håkansson N, Lochhead P, et al. Lifestyle factors for the prevention of inflammatory bowel disease. Gut. (2022):gutjnl-2022-328174. doi: 10.1136/gutjnl-2022-328174, [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Chan SSM, Luben R, Olsen A, Tjonneland A, Kaaks R, Teucher B, et al. Body mass index and the risk for Crohn’s disease and ulcerative colitis: data from a European prospective cohort study (the IBD in EPIC study). Am J Gastroenterol. (2013) 108:575–82. doi: 10.1038/ajg.2012.453, [DOI] [PubMed] [Google Scholar]
- 11.Rasmussen NF, Bech BH, Rubin KH, Andersen V. Associations between participation in, intensity of, and time spent on leisure time physical activity and risk of inflammatory bowel disease among older adults (PA-IBD): a prospective cohort study. BMC Public Health. (2021) 21:634. doi: 10.1186/s12889-021-10492-7, [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Shephard RJ. Limits to the measurement of habitual physical activity by questionnaires. Br J Sports Med. (2003) 37:197–206; discussion 206. doi: 10.1136/bjsm.37.3.197, [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Strain T, Wijndaele K, Dempsey PC, Sharp SJ, Pearce M, Jeon J, et al. Wearable-device-measured physical activity and future health risk. Nat Med. (2020) 26:1385–91. doi: 10.1038/s41591-020-1012-3, [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Bassett DR, Troiano RP, McClain JJ, Wolff DL. Accelerometer-based physical activity: total volume per day and standardized measures. Med Sci Sports Exerc. (2015) 47:833–8. doi: 10.1249/MSS.0000000000000468, [DOI] [PubMed] [Google Scholar]
- 15.Sudlow C, Gallacher J, Allen N, Beral V, Burton P, Danesh J, 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:e1001779. doi: 10.1371/journal.pmed.1001779, [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Walmsley R, Chan S, Smith-Byrne K, Ramakrishnan R, Woodward M, Rahimi K, et al. Reallocation of time between device-measured movement behaviours and risk of incident cardiovascular disease. Br J Sports Med. (2021) 56:1008–17. doi: 10.1136/bjsports-2021-104050, [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Doherty A, Jackson D, Hammerla N, Plötz T, Olivier P, Granat MH, et al. Large scale population assessment of physical activity using wrist worn accelerometers: the UK biobank study. PLoS One. (2017) 12:e0169649. doi: 10.1371/journal.pone.0169649, [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.NHS Physical activity guidelines for adults aged 19 to 64. nhs.uk. (2022). Available online at: https://www.nhs.uk/live-well/exercise/physical-activity-guidelines-for-adults-aged-19-to-64/ (accessed September 14, 2024)
- 19.NHS England. Health survey for England, 2021 part 2. NHS England Digital. (2023). Available online at: https://digital.nhs.uk/data-and-information/publications/statistical/health-survey-for-england/2021-part-2 (accessed September 14, 2024)
- 20.Ananthakrishnan AN, Higuchi LM, Huang ES, Khalili H, Richter JM, Fuchs CS, et al. Aspirin, nonsteroidal anti-inflammatory drug use, and risk for Crohn disease and ulcerative colitis: a cohort study. Ann Intern Med. (2012) 156:350–9. doi: 10.7326/0003-4819-156-5-201203060-00007, [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Lee I-M, Sesso HD, Oguma Y, Paffenbarger RS. The “weekend warrior” and risk of mortality. Am J Epidemiol. (2004) 160:636–41. doi: 10.1093/aje/kwh274, [DOI] [PubMed] [Google Scholar]
- 22.Tiong HT, Fan D, Frampton C, Ananthakrishnan AN, Gearry RB. Physical activity is associated with a decreased risk of developing inflammatory bowel disease: a systematic review and Meta-analysis. J Crohn's Colitis. (2024) 18:1476–85. doi: 10.1093/ecco-jcc/jjae053, [DOI] [PubMed] [Google Scholar]
- 23.Melinder C, Hiyoshi A, Hussein O, Halfvarson J, Ekbom A, Montgomery S. Physical fitness in adolescence and subsequent inflammatory bowel disease risk. Clin Transl Gastroenterol. (2015) 6:e121. doi: 10.1038/ctg.2015.49, [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Wang Q, Xu K-Q, Qin X-R, Wen-Lu n, Yan-Liu n, Wang X-Y. Association between physical activity and inflammatory bowel disease risk: a meta-analysis. Dig Liver Dis. (2016) 48:1425–31. doi: 10.1016/j.dld.2016.08.129 [DOI] [PubMed] [Google Scholar]
- 25.Saxena A, Fletcher E, Larsen B, Baliga MS, Durstine JL, Fayad R. Effect of exercise on chemically-induced colitis in adiponectin deficient mice. J Inflamm. (2012) 9:30. doi: 10.1186/1476-9255-9-30, [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Packer N, Hoffman-Goetz L. Exercise training reduces inflammatory mediators in the intestinal tract of healthy older adult mice. Can J Aging. (2012) 31:161–71. doi: 10.1017/S0714980812000104, [DOI] [PubMed] [Google Scholar]
- 27.Mach N, Fuster-Botella D. Endurance exercise and gut microbiota: a review. J Sport Health Sci. (2017) 6:179–97. doi: 10.1016/j.jshs.2016.05.001, [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Scheiman J, Luber JM, Chavkin TA, MacDonald T, Tung A, Pham L-D, et al. Meta-omics analysis of elite athletes identifies a performance-enhancing microbe that functions via lactate metabolism. Nat Med. (2019) 25:1104–9. doi: 10.1038/s41591-019-0485-4, [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Zhuang X, Li T, Li M, Huang S, Qiu Y, Feng R, et al. Systematic review and Meta-analysis: short-chain fatty acid characterization in patients with inflammatory bowel disease. Inflamm Bowel Dis. (2019) 25:1751–63. doi: 10.1093/ibd/izz188, [DOI] [PubMed] [Google Scholar]
- 30.Hao F, Tian M, Wang H, Li S, Wang X, Jin X, et al. Exercise-induced β-hydroxybutyrate promotes Treg cell differentiation to ameliorate colitis in mice. FASEB J. (2024) 38:e23487. doi: 10.1096/fj.202301686RR, [DOI] [PubMed] [Google Scholar]
- 31.He C, Bassik MC, Moresi V, Sun K, Wei Y, Zou Z, et al. Exercise-induced BCL2-regulated autophagy is required for muscle glucose homeostasis. Nature. (2012) 481:511–5. doi: 10.1038/nature10758, [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Keadle SK, Shiroma EJ, Kamada M, Matthews CE, Harris TB, Lee I-M. Reproducibility of accelerometer-assessed physical activity and sedentary time. Am J Prev Med. (2017) 52:541–8. doi: 10.1016/j.amepre.2016.11.010, [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Saint-Maurice PF, Sampson JN, Keadle SK, Willis EA, Troiano RP, Matthews CE. Reproducibility of accelerometer and posture-derived measures of physical activity. Med Sci Sports Exerc. (2020) 52:876–83. doi: 10.1249/MSS.0000000000002206, [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Dwyer T, Pezic A, Sun C, Cochrane J, Venn A, Srikanth V, et al. Objectively measured daily steps and subsequent long term all-cause mortality: the Tasped prospective cohort study. PLoS One. (2015) 10:e0141274. doi: 10.1371/journal.pone.0141274, [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Publicly available datasets were analyzed in this study. The data that support the findings of this study are available from UK Biobank (https://www.ukbiobank.ac.uk/enable-your-research/register) but restrictions apply to availability. Data access requires application and approval through UK Biobank’s Access Management System.