Accelerometer-derived physical activity patterns and risk of incident gout: a prospective cohort study of 97,387 UK Biobank participants Authors: Qiang Wang, Yueming Liu, Bin Zhu Published: 2025 ABSTRACT BackgroundThe influence of physical activity (PA) patterns on gout risk is unclear. We investigated the associations of accelerometer-derived PA patterns, specifically regularly active (RA) and weekend warrior (WW), with incident gout.MethodsWe analyzed 97,387 gout-free UK Biobank participants with 7-day wrist-worn accelerometer data. Participants were categorized as inactive (< 150 min/week moderate-to-vigorous PA [MVPA]), WW (≥ 150 min MVPA with ≥ 50% concentrated in 1–2 days), or RA (≥ 150 min MVPA, not meeting WW criteria). Multivariable Cox models estimated hazard ratios (HRs) for incident gout.ResultsDuring a median 8.0 years’ follow-up, 833 gout cases occurred. In minimally adjusted models, both RA (HR = 0.67, 95% CI: 0.55–0.83, P < 0.001) and WW (HR = 0.76, 95% CI: 0.65–0.89, P < 0.001) patterns showed reduced risk compared to inactive pattern. However, results diverged after further adjustment for baseline sUA: the RA pattern maintained a robust protective association (HR = 0.79, 95% CI: 0.65–0.98, P = 0.029), whereas the association for the WW pattern was attenuated and lost statistical significance (HR = 0.93, 95% CI: 0.79–1.08, P = 0.331). Subgroup analyses revealed significant interactions with comorbidities (P for interaction < 0.01), where the WW pattern showed no inverse association in participants with hypertension or chronic kidney disease, while RA remained protective. These findings were robust across sensitivity analyses and genetic risk strata.ConclusionWhile any moderate-to-vigorous physical activity is beneficial, the regularly active pattern demonstrates a more robust inverse association with gout than the weekend warrior pattern. Regularity in physical activity appears crucial for gout risk reduction, particularly for metabolically vulnerable individuals, as the sporadic nature of the weekend warrior pattern may be insufficient to counteract established metabolic risks.Supplementary InformationThe online version contains supplementary material available at 10.1186/s13075-025-03716-3. Introduction Gout is a prevalent metabolic arthritis characterized by the deposition of monosodium urate crystals within joints and surrounding tissues, triggering intense inflammation. Fueled by rising obesity rates, dietary shifts, and an aging population, the global prevalence of gout has surged in recent years, establishing it as a major chronic disease posing a substantial public health burden [1]. Beyond causing debilitating joint pain and functional impairment, gout is strongly associated with an elevated risk of cardiovascular disease (CVD), chronic kidney disease (CKD), and all-cause mortality. Despite a well-understood pathophysiology and the availability of cost-effective treatments, gout management remains suboptimal for many patients [2]. Physical activity, a crucial modifiable lifestyle factor, has been extensively linked to lower risks of various chronic diseases, including CVD [3], type 2 diabetes [4], and certain cancers [5]. Proposed mechanisms include improved insulin sensitivity, weight control, and reduced systemic inflammation [6]. However, the specific role of physical activity in gout risk reduction remains elusive due to a lack of high-quality prospective evidence [7, 8]. Consequently, while the 2020 American College of Rheumatology (ACR) guidelines recommend weight loss, they notably omit explicit recommendations regarding physical activity [9]. A critical unresolved question is whether physical activity is effectively associated with reduced gout risk, and if so, whether this protection depends solely on total activity volume or if the temporal distribution (regularity) also matters. The weekend warrior pattern—compressing 50% recommended weekly activity volume into 1–2 days—has been shown to confer cardiovascular benefits comparable to regular activity in recent studies [10, 11]. However, it remains unknown whether this sporadic pattern is sufficient to mitigate the risk of gout. Addressing this gap has been hampered by the limitations of conventional epidemiological studies, which rely primarily on self-reported questionnaires. These subjective methods suffer from inherent limitations, including recall bias and reporting errors, and struggle to accurately capture key dimensions of daily activity such as intensity, frequency, duration, and temporal patterns [12–15]. Accelerometers represent the gold standard for objective physical activity measurement, enabling the continuous and precise recording of 24-h activity intensity, sedentary time, and detailed activity patterns [16]. The UK Biobank, a large-scale prospective cohort, has compiled extensive health information, biological samples, and genetic data. Critically, it also includes wrist-worn accelerometer data collected over 7 days in a large subset of participants, providing vast, objective physical activity data [17]. This resource offers an unprecedented opportunity to investigate the complex relationship between physical activity and gout risk comprehensively. Therefore, leveraging this rich accelerometer data, this study aims to prospectively investigate the association between different physical activity patterns, specifically regularly active (RA) versus weekend warrior (WW), and incident gout. Such findings will provide high-level evidence to guide the development of precise, targeted physical activity prescriptions for gout risk reduction. Methods Study population This study utilized data from the UK Biobank, a large-scale prospective cohort that recruited over 500,000 participants aged 37 to 73 years between 2006 and 2010 [18]. To assess the association between physical activity and the risk of incident gout, this study initially recruited 502,006 individuals from the UK Biobank. Participants without valid wrist-worn accelerometer data were excluded (N = 401,972), leaving 100,034 participants. Then, 2,560 individuals with existing gout at baseline (time of accelerometer wearing) were excluded, resulting in 97,474 participants. Missing covariates were imputed using the “mice” R package. Following further exclusions for self-reported incident gout (N = 72) and withdrawn consent (N = 15), the final analytical cohort comprised 97,387 individuals (Fig. S1). Accelerometer-derived physical activity and pattern definition Physical activity was quantified using Axivity AX3 accelerometers worn on the dominant wrist for 7 consecutive days (100 Hz sampling; ± 8 g range) between 2013 and 2015 [17]. Raw accelerometer data were calibrated to local gravity, filtered for noise, and processed to identify non-wear periods (defined as unbroken episodes of at least 60 min with SD of acceleration < 13.0 mg). To operationalize MVPA episodes, a validated two-stage machine-learning architecture was employed [19]. First, a balanced Random Forest classifier assigned behavior labels to non-overlapping 30-s time windows based on acceleration features. Second, a Hidden Markov Model smoothed the activity sequence to correct unrealistic interruptions and define coherent episodes using transition probabilities. MVPA was defined as waking behavior ≥ 3 metabolic equivalent of tasks (METs). Daily MVPA duration (Field ID 40045) was aggregated from these smoothed windows without a minimum bout duration requirement, consistent with current guidelines [20–22]. Participants were categorized into three mutually exclusive groups based on their weekly MVPA volume and temporal distribution, adopting the accelerometer-derived definitions validated by Khurshid et al. [23] and consistent with World Health Organization guidelines [24]: (1) inactive: < 150 min/week of MVPA; (2) weekend warrior (WW): ≥ 150 min/week of MVPA, with ≥ 50% of the total weekly MVPA accumulated within the most active 1–2 days; (3) regularly active (RA): ≥ 150 min/week of MVPA, but not meeting the specific criteria for the WW pattern. Gout Gout was identified using data from “First Occurrence” provided by UK Biobank (Field ID 131258, International Classification of Diseases (ICD), Tenth Revision, code M10). This dataset integrates diagnoses from hospital inpatient records, death registries, primary care, and self-reports [25]. To improve diagnostic specificity, participants whose gout status was based solely on self-report were excluded. Due to differences in follow-up frequency and intervals among participants, the study defined the first-recorded gout event during follow-up as the endpoint. Therefore, the observation was considered to end at the earliest of the following: occurrence of gout, death, or the last known follow-up date (October 31, 2022). Covariates During enrollment in the UK Biobank study, participants’ characteristics including sex, age, ethnicity, smoking status, alcohol consumption, educational attainment, and income level were collected via touchscreen questionnaires. Additionally, trained staff measured participants' height and weight during their assessment visit, with body mass index (BMI) subsequently calculated using the standard formula (classified as: underweight, <18.5 ; normal weight, 18.5–24.9 ; overweight, 25.0–29.9 ; and obese, 30.0 ).Biochemical covariates, specifically low-density lipoprotein (LDL) cholesterol, triglycerides, and total cholesterol, were measured from baseline blood samples. The Townsend Deprivation Index (TDI), derived from participants' residential postcodes, integrates area-level aggregate data on car ownership, unemployment, homeownership, and household overcrowding [26]. A healthy diet score (HDS, range: 0–5) was calculated based on self-reported weekly intake frequencies of vegetables, fruit, fish, and meat, with higher scores indicating healthier dietary patterns [27]. History of CVD, CKD, diabetes, and hypertension was ascertained from corresponding fields as defined in a previously published study [28]. The selection of these covariates for adjustment was guided by a Directed Acyclic Graph (DAG) constructed based on a priori knowledge of temporal and biological relationships (Fig. S2). Statistical analysis Primary analysis:​​ Multivariable Cox proportional hazards models estimated hazard ratios (HRs) for incident gout across physical activity patterns, using the inactive group as reference. ​​Model 1 was adjusted for age, sex, and BMI. Model 2 was further adjusted for ethnicity, smoking status, drinking status, HDS, TDI, household income, education level, lipid profiles (LDL, triglycerides, total cholesterol), and baseline history of CVD, CKD, diabetes, and hypertension. Model 3 included additional adjustment for baseline serum uric acid (sUA) to examine whether the association was independent of participant’s baseline urate burden. We also performed direct comparisons between the WW and RA groups. In these models, total MVPA duration was included as an additional covariate in all models to isolate the effect of temporal distribution independent of physical activity volume. Sensitivity analyses: To evaluate robustness, we conducted following supplementary analyses: (1) varying the definition of the active group using quartile-based MVPA duration thresholds (25th: 95.0 min; 50th: 211.6 min; 75th: 374.4 min). (2) applying a stricter WW criterion requiring ≥ 75% of total MVPA within 1–2 days. (3) excluding participants diagnosed with gout within the first 2 years of wearing the monitoring device. (4) assessing the influence of medications by excluding participants taking Urate-Lowering Therapy (ULT) or colchicine at baseline, and subsequently adjusting for urate-elevating medications (Model 4) or other non-ULT urate-lowering drugs (Model 5). (5) redefining baseline CKD history based on estimated glomerular filtration rate (eGFR) (Table S1). (6) including self-reported incident gout cases which were excluded in the primary analysis; (7) conducting falsification tests using conductive and sensorineural hearing loss (ICD-10: H90) and acute appendicitis (ICD-10: K35) as negative control outcomes to assess unmeasured confounding; and (8) testing interactions between physical activity patterns and prespecified subgroup variables. Restricted cubic splines (RCS) RCS with 3 knots (placed at default quantile positions: 0.10, 0.50, and 0.90) were employed to model the dose–response relationship between MVPA and gout risk. Cox proportional hazards models were used to estimate HRs and 95% confidence intervals (CIs). Likelihood ratio tests were performed to evaluate nonlinearity. The median MVPA level served as the reference point (HR = 1). Observations with MVPA exceeding the 90th percentile were further excluded to mitigate the influence of extreme values. Inverse probability of treatment weighting (IPTW) To evaluate the independent association of different physical activity patterns on the risk of gout, we employed the IPTW method within the framework of the Average Treatment effect on the Treated (ATT). We compared each active group (WW and RA) separately with the inactive group. Propensity scores (PS) were estimated using a multivariable logistic regression model that included covariates potentially associated with both physical activity and the risk of gout. Under the ATT framework, individuals in the exposed (active) group were assigned a weight of 1, while individuals in the control (inactive) group were weighted by PS/(1 − PS). Covariate balance before and after weighting was assessed using the standardized mean difference (SMD), with an SMD < 0.1 indicating adequate balance. Based on the weighted sample, we applied Cox proportional hazards models to estimate the HRs and 95% CIs for gout incidence across groups. In addition, we calculated the weighted incidence rates of gout (per 1,000 person-years) for each group to reflect the absolute risk. Genetic susceptibility analysis​ The gout polygenic risk score (PRS) was obtained from the Polygenic Score Catalog (https://www.pgscatalog.org/) as PGS000199. This score comprises 29 single nucleotide polymorphisms (SNPs) identified through genome-wide association studies (GWAS) in individuals of European ancestry [29, 30]. Full details of the selected SNPs are provided in Table S2. Using the Swiss Army Knife tool on the UK Biobank Research Analysis Platform, the individual PRS was calculated as follows: For each SNP, the number of effect alleles (0, 1, or 2) carried by each participant was multiplied by the SNP's published effect size (β coefficient) derived from GWAS. The resulting weighted values for all 29 SNPs were then summed to yield the final PRS for each participant. Based on the calculated PRS distribution, participants were stratified into tertiles, resulting in three genetic risk groups: low genetic risk (lowest tertile), intermediate genetic risk (middle tertile), and high genetic risk (highest tertile). Software implementation All analyses were performed using R (Version 4.3.0). Primary packages included “survival” (for Cox models), “mice” (for multiple imputation), “cobalt” (for IPTW), “rms” (for RCS), and “forestploter” (for forest plot). Statistical significance was defined as two-sided P < 0.05. Results Baseline characteristics of the study population​ A total of 97,387 participants were included in the final analysis. Participants were classified into inactive (N = 36,757), weekend warrior (WW; N = 40,352), and regularly active (RA; N = 20,278) groups based on the criteria defined above (Table 1). Statistically significant differences (p < 0.001) were observed across multiple baseline characteristics. Compared to the inactive group, both RA and WW participants were younger, had a lower proportion of females, lower obesity prevalence, healthier lifestyle and socioeconomic profiles, higher proportions of ideal Healthy Diet Scores, higher levels of college education, and lower prevalence of current smoking. Additionally, the active groups exhibited a lower prevalence of comorbidities, including diabetes, hypertension, CVD, and CKD. Median total MVPA time differed significantly (P < 0.001), with the RA group reporting the highest (417.60 min/week), followed by the WW group (286.58 min/week), and the inactive group the lowest (69.48 min/week) .Table 1Baseline characteristicsCharacteristicInactive (N = 36,757)Regularly Active (N = 20,278)Weekend Warrior (N = 40,352)PAge (years)< 0.001 Mean (SD)56.60 (7.89)54.98 (7.88)55.93 (7.74)Gender< 0.001 Female24,439 (66.49%)10,530 (51.93%)20,931 (51.87%) Male12,318 (33.51%)9,748 (48.07%)19,421 (48.13%)BMI (kg/m2)< 0.001 Underweight144 (0.39%)188 (0.93%)230 (0.57%) Normal11,461 (31.18%)9,551 (47.10%)17,268 (42.79%) Overweight14,929 (40.62%)7,928 (39.10%)17,109 (42.40%) Obese10,223 (27.81%)2,611 (12.88%)5,745 (14.24%)Ethnicity< 0.001 Non White1,420 (3.86%)683 (3.37%)1,085 (2.69%) White35,337 (96.14%)19,595 (96.63%)39,267 (97.31%)TDI< 0.001 Median (Q1, Q3)−2.45 (−3.80, −0.19)−2.06 (−3.61, 0.54)−2.57 (−3.90, −0.47)Education level< 0.001 UnKnown3,996 (10.87%)1,356 (6.69%)2,906 (7.20%) College13,120 (35.69%)10,072 (49.67%)18,964 (47.00%) Other levels19,641 (53.43%)8,850 (43.64%)18,482 (45.80%)Drinking status< 0.001 Never1,447 (3.94%)543 (2.68%)904 (2.24%) Previous1,192 (3.24%)565 (2.79%)909 (2.25%) Current34,118 (92.82%)19,170 (94.54%)38,539 (95.51%)Smoking status< 0.001 Never20,324 (55.29%)11,791 (58.15%)23,831 (59.06%) Previous13,205 (35.93%)7,250 (35.75%)14,160 (35.09%) Current3,228 (8.78%)1,237 (6.10%)2,361 (5.85%)Household income< 0.001 Less than 18,0006,859 (18.66%)2,866 (14.13%)5,315 (13.17%) 18,000 to 30,9999,668 (26.30%)4,593 (22.65%)9,452 (23.42%) 31,000 to 51,99910,312 (28.05%)5,739 (28.30%)11,574 (28.68%) 52,000 to 100,0007,889 (21.46%)5,322 (26.25%)10,734 (26.60%) Greater than 100,0002,029 (5.52%)1,758 (8.67%)3,277 (8.12%)HDS category< 0.001 Ideal14,768 (40.18%)8,800 (43.40%)16,959 (42.03%) Medium17,783 (48.38%)9,395 (46.33%)19,256 (47.72%) Poor4,206 (11.44%)2,083 (10.27%)4,137 (10.25%)LDL Cholesterol< 0.001 Median (Q1, Q3)3.55 (2.97, 4.16)3.49 (2.95, 4.05)3.54 (3.00, 4.11)Triglycerides< 0.001 Median (Q1, Q3)1.47 (1.04, 2.12)1.30 (0.96, 1.93)1.35 (0.98, 1.99)Total Cholesterol< 0.001 Median (Q1, Q3)5.69 (4.98, 6.52)5.63 (4.96, 6.41)5.68 (4.99, 6.47)Serum UA< 0.001 Median (Q1, Q3)294.30 (244.50, 350.30)292.50 (242.70, 348.80)296.90 (246.20, 352.10)Total MVPA< 0.001 Median (Q1, Q3)69.48 (28.01, 109.32)417.60 (296.45, 590.40)286.58 (209.96, 417.60)History of Diabetes< 0.001 No34,942 (95.06%)19,745 (97.37%)39,374 (97.58%) Yes1,815 (4.94%)533 (2.63%)978 (2.42%)History of Hypertension< 0.001 No26,712 (72.67%)16,353 (80.64%)32,154 (79.68%) Yes10,045 (27.33%)3,925 (19.36%)8,198 (20.32%)History of CVD< 0.001 No30,975 (84.27%)18,089 (89.21%)35,915 (89.00%) Yes5,782 (15.73%)2,189 (10.79%)4,437 (11.00%)History of CKD< 0.001 No34,592 (94.11%)19,644 (96.87%)39,135 (96.98%) Yes2,165 (5.89%)634 (3.13%)1,217 (3.02%)Table 1: Baseline characteristics of the study population according to physical activity pattern. Participants were classified as Inactive (< 150 min/week MVPA), Weekend Warrior (≥ 150 min/week MVPA with ≥ 50% concentrated within 1–2 days), or Regularly Active (≥ 150 min/week MVPA not meeting WW criteria). Values are presented as mean (SD) for age, median (Q1, Q3) for skewed continuous variables (TDI, lipid profiles, serum UA, and total MVPA), and number (percentage) for categorical variables. P-values assess the differences across the three physical activity patterns. Continuous variables were compared using the Kruskal–Wallis rank sum test, and categorical variables were compared using Pearson's chi-squared testAbbreviations: BMI body mass index, CKD chronic kidney disease, CVD cardiovascular disease, HDS healthy diet score, LDL low-density lipoprotein, MVPA moderate-to-vigorous physical activity, SD standard deviation, TDI Townsend deprivation index, UA uric acid, Q1 first quartile (25th percentile), Q3 third quartile (75th percentile) Table 1Baseline characteristicsCharacteristicInactive (N = 36,757)Regularly Active (N = 20,278)Weekend Warrior (N = 40,352)PAge (years)< 0.001 Mean (SD)56.60 (7.89)54.98 (7.88)55.93 (7.74)Gender< 0.001 Female24,439 (66.49%)10,530 (51.93%)20,931 (51.87%) Male12,318 (33.51%)9,748 (48.07%)19,421 (48.13%)BMI (kg/m2)< 0.001 Underweight144 (0.39%)188 (0.93%)230 (0.57%) Normal11,461 (31.18%)9,551 (47.10%)17,268 (42.79%) Overweight14,929 (40.62%)7,928 (39.10%)17,109 (42.40%) Obese10,223 (27.81%)2,611 (12.88%)5,745 (14.24%)Ethnicity< 0.001 Non White1,420 (3.86%)683 (3.37%)1,085 (2.69%) White35,337 (96.14%)19,595 (96.63%)39,267 (97.31%)TDI< 0.001 Median (Q1, Q3)−2.45 (−3.80, −0.19)−2.06 (−3.61, 0.54)−2.57 (−3.90, −0.47)Education level< 0.001 UnKnown3,996 (10.87%)1,356 (6.69%)2,906 (7.20%) College13,120 (35.69%)10,072 (49.67%)18,964 (47.00%) Other levels19,641 (53.43%)8,850 (43.64%)18,482 (45.80%)Drinking status< 0.001 Never1,447 (3.94%)543 (2.68%)904 (2.24%) Previous1,192 (3.24%)565 (2.79%)909 (2.25%) Current34,118 (92.82%)19,170 (94.54%)38,539 (95.51%)Smoking status< 0.001 Never20,324 (55.29%)11,791 (58.15%)23,831 (59.06%) Previous13,205 (35.93%)7,250 (35.75%)14,160 (35.09%) Current3,228 (8.78%)1,237 (6.10%)2,361 (5.85%)Household income< 0.001 Less than 18,0006,859 (18.66%)2,866 (14.13%)5,315 (13.17%) 18,000 to 30,9999,668 (26.30%)4,593 (22.65%)9,452 (23.42%) 31,000 to 51,99910,312 (28.05%)5,739 (28.30%)11,574 (28.68%) 52,000 to 100,0007,889 (21.46%)5,322 (26.25%)10,734 (26.60%) Greater than 100,0002,029 (5.52%)1,758 (8.67%)3,277 (8.12%)HDS category< 0.001 Ideal14,768 (40.18%)8,800 (43.40%)16,959 (42.03%) Medium17,783 (48.38%)9,395 (46.33%)19,256 (47.72%) Poor4,206 (11.44%)2,083 (10.27%)4,137 (10.25%)LDL Cholesterol< 0.001 Median (Q1, Q3)3.55 (2.97, 4.16)3.49 (2.95, 4.05)3.54 (3.00, 4.11)Triglycerides< 0.001 Median (Q1, Q3)1.47 (1.04, 2.12)1.30 (0.96, 1.93)1.35 (0.98, 1.99)Total Cholesterol< 0.001 Median (Q1, Q3)5.69 (4.98, 6.52)5.63 (4.96, 6.41)5.68 (4.99, 6.47)Serum UA< 0.001 Median (Q1, Q3)294.30 (244.50, 350.30)292.50 (242.70, 348.80)296.90 (246.20, 352.10)Total MVPA< 0.001 Median (Q1, Q3)69.48 (28.01, 109.32)417.60 (296.45, 590.40)286.58 (209.96, 417.60)History of Diabetes< 0.001 No34,942 (95.06%)19,745 (97.37%)39,374 (97.58%) Yes1,815 (4.94%)533 (2.63%)978 (2.42%)History of Hypertension< 0.001 No26,712 (72.67%)16,353 (80.64%)32,154 (79.68%) Yes10,045 (27.33%)3,925 (19.36%)8,198 (20.32%)History of CVD< 0.001 No30,975 (84.27%)18,089 (89.21%)35,915 (89.00%) Yes5,782 (15.73%)2,189 (10.79%)4,437 (11.00%)History of CKD< 0.001 No34,592 (94.11%)19,644 (96.87%)39,135 (96.98%) Yes2,165 (5.89%)634 (3.13%)1,217 (3.02%)Table 1: Baseline characteristics of the study population according to physical activity pattern. Participants were classified as Inactive (< 150 min/week MVPA), Weekend Warrior (≥ 150 min/week MVPA with ≥ 50% concentrated within 1–2 days), or Regularly Active (≥ 150 min/week MVPA not meeting WW criteria). Values are presented as mean (SD) for age, median (Q1, Q3) for skewed continuous variables (TDI, lipid profiles, serum UA, and total MVPA), and number (percentage) for categorical variables. P-values assess the differences across the three physical activity patterns. Continuous variables were compared using the Kruskal–Wallis rank sum test, and categorical variables were compared using Pearson's chi-squared testAbbreviations: BMI body mass index, CKD chronic kidney disease, CVD cardiovascular disease, HDS healthy diet score, LDL low-density lipoprotein, MVPA moderate-to-vigorous physical activity, SD standard deviation, TDI Townsend deprivation index, UA uric acid, Q1 first quartile (25th percentile), Q3 third quartile (75th percentile) Baseline characteristics Table 1: Baseline characteristics of the study population according to physical activity pattern. Participants were classified as Inactive (< 150 min/week MVPA), Weekend Warrior (≥ 150 min/week MVPA with ≥ 50% concentrated within 1–2 days), or Regularly Active (≥ 150 min/week MVPA not meeting WW criteria). Values are presented as mean (SD) for age, median (Q1, Q3) for skewed continuous variables (TDI, lipid profiles, serum UA, and total MVPA), and number (percentage) for categorical variables. P-values assess the differences across the three physical activity patterns. Continuous variables were compared using the Kruskal–Wallis rank sum test, and categorical variables were compared using Pearson's chi-squared test Abbreviations: BMI body mass index, CKD chronic kidney disease, CVD cardiovascular disease, HDS healthy diet score, LDL low-density lipoprotein, MVPA moderate-to-vigorous physical activity, SD standard deviation, TDI Townsend deprivation index, UA uric acid, Q1 first quartile (25th percentile), Q3 third quartile (75th percentile) Density plots (Fig. S3) revealed distinct weekly MVPA distribution patterns: RA participants maintained consistent activity throughout the week, whereas WW participants concentrated the majority of MVPA within 1–2 days, with substantially lower activity on remaining days. Association between physical activity patterns and gout risk​ Tests based on Schoenfeld residuals showed that the proportional hazards assumption was not violated for the physical activity patterns (Model 1 P = 0.068, Model 2 P = 0.169, Model 3 P = 0.074). During a median follow-up of 8.0 years, 833 incident gout cases were documented. In terms of absolute risk, the incidence rate (IR) was highest in the inactive group (1.35 per 1,000 person-years), followed by the WW group (1.00 per 1,000 person-years), and lowest in the RA group (0.81 per 1,000 person-years). Compared with the inactive group, the RA and WW patterns were associated with an absolute risk reduction (ARR) of 0.54 and 0.35 cases per 1,000 person-years, respectively (Fig. 1A). In the multivariable Cox regression analysis, both RA (HR = 0.67, 95% CI: 0.55–0.83; P < 0.001) and WW (HR = 0.76, 95% CI: 0.65–0.89; P < 0.001) showed significantly reduced gout risk in Model 1 compared with the inactive group. These associations persisted in Model 2 after further adjustment for sociodemographic factors, lifestyle habits, and medical history. However, after further adjustment for baseline sUA in Model 3, the results diverged: the RA pattern maintained a significant protective association (HR = 0.79, 95% CI: 0.65–0.98; P = 0.029), whereas the association for the WW pattern was attenuated and no longer statistically significant (HR = 0.93, 95% CI: 0.79–1.08; P = 0.331). Given that the median total MVPA duration was significantly higher in the RA group (417.6 min) compared with the WW group (286.6 min), this suggests that the lower activity volume in the WW group may not be sufficient to provide independent protection after accounting for baseline uric acid levels (Fig. 1A). When directly comparing the two active groups with adjustment for total MVPA duration (Fig. 1B), the WW group exhibited a trend toward higher risk compared to the RA group (HR = 1.21, 95% CI: 0.98–1.51) in the fully adjusted model. Although this difference was marginally significant (P = 0.079), it indicates that the regularly active pattern may confer superior protection compared to the weekend warrior pattern, potentially due to the sustained physiological benefits of consistent physical activity.Fig. 1Associations of physical activity patterns with incident gout risk. A Forest plot showing HRs and 95% CIs for the RA and WW patterns compared with the inactive group (reference). Model 1 was adjusted for age, sex, and BMI. Model 2 was further adjusted for ethnicity, Townsend Deprivation Index, education level, smoking status, drinking status, household income, Healthy Diet Score, lipid profiles (LDL, triglycerides, total cholesterol), and baseline comorbidities (diabetes, hypertension, CVD, and CKD). Model 3 included all covariates in Model 2 plus baseline sUA. B Forest plot of the direct comparison between the active groups, with the RA pattern serving as the reference; total MVPA volume was additionally adjusted for in all models for this comparison. The absolute IR is expressed per 1,000 person-years. Abbreviations: HR, hazard ratio; CI, confidence interval; RA, regularly active; WW, weekend warrior; IR, incidence rate; ARR, absolute risk reduction; BMI, body mass index; CVD, cardiovascular disease; CKD, chronic kidney disease; sUA, serum uric acid Associations of physical activity patterns with incident gout risk. A Forest plot showing HRs and 95% CIs for the RA and WW patterns compared with the inactive group (reference). Model 1 was adjusted for age, sex, and BMI. Model 2 was further adjusted for ethnicity, Townsend Deprivation Index, education level, smoking status, drinking status, household income, Healthy Diet Score, lipid profiles (LDL, triglycerides, total cholesterol), and baseline comorbidities (diabetes, hypertension, CVD, and CKD). Model 3 included all covariates in Model 2 plus baseline sUA. B Forest plot of the direct comparison between the active groups, with the RA pattern serving as the reference; total MVPA volume was additionally adjusted for in all models for this comparison. The absolute IR is expressed per 1,000 person-years. Abbreviations: HR, hazard ratio; CI, confidence interval; RA, regularly active; WW, weekend warrior; IR, incidence rate; ARR, absolute risk reduction; BMI, body mass index; CVD, cardiovascular disease; CKD, chronic kidney disease; sUA, serum uric acid Kaplan–Meier analyses corroborated these findings, showing significantly divergent cumulative gout incidence curves (overall log-rank P < 0.001; Fig. 2). The inactive group exhibited the highest cumulative incidence, while the RA group demonstrated the lowest. The WW curve occupied an intermediate position, consistently situated below the inactive curve but above the RA curve. Pairwise comparisons (Bonferroni-corrected) confirmed significant risk reductions for both RA (vs. inactive: P < 0.001) and WW (vs. inactive: P < 0.001). Although the RA group showed a lower cumulative incidence compared with the WW group, the difference did not reach statistical significance (P = 0.108).Fig. 2Kaplan–Meier estimates of cumulative gout incidence by physical activity pattern. Cumulative incidence curves for inactive (blue), regularly active (green), and weekend warrior (orange) groups over a median follow-up of 8.0 years. Shaded areas represent 95% CIs. The overall difference was tested using the log-rank test. Pairwise comparisons were adjusted using the Bonferroni correction (P.adj) Kaplan–Meier estimates of cumulative gout incidence by physical activity pattern. Cumulative incidence curves for inactive (blue), regularly active (green), and weekend warrior (orange) groups over a median follow-up of 8.0 years. Shaded areas represent 95% CIs. The overall difference was tested using the log-rank test. Pairwise comparisons were adjusted using the Bonferroni correction (P.adj) Subgroup analyses of the association between physical activity patterns and incident gout Subgroup analyses were conducted to evaluate the consistency of the associations across various population characteristics (Fig. 3 and Table S3). Generally, the inverse association between the RA pattern and gout risk remained robust across most strata, particularly among high-risk subgroups. Notably, RA showed a significant protective association specifically in participants with obesity (HR = 0.66, 95% CI: 0.45–0.96) and those with hyperuricemia (sUA ≥ 420 µmol/L; HR = 0.73, 95% CI: 0.55–0.97), whereas the WW pattern did not reach statistical significance in these metabolically vulnerable groups.Fig. 3Subgroup analyses of the association between physical activity patterns and incident gout. Forest plots displaying the HRs for (A) demographic factors and (B) clinical/metabolic factors. The inactive group served as the reference for all comparisons. HRs were derived from the fully adjusted model (Model 3, including adjustment for sUA), with the specific stratification variable excluded from the covariates to avoid collinearity. P for interaction was calculated using likelihood ratio tests. Abbreviations: HR, hazard ratio; sUA, serum uric acid; HDS, healthy diet score; CVD, cardiovascular disease; CKD, chronic kidney disease Subgroup analyses of the association between physical activity patterns and incident gout. Forest plots displaying the HRs for (A) demographic factors and (B) clinical/metabolic factors. The inactive group served as the reference for all comparisons. HRs were derived from the fully adjusted model (Model 3, including adjustment for sUA), with the specific stratification variable excluded from the covariates to avoid collinearity. P for interaction was calculated using likelihood ratio tests. Abbreviations: HR, hazard ratio; sUA, serum uric acid; HDS, healthy diet score; CVD, cardiovascular disease; CKD, chronic kidney disease Formal tests for interaction revealed significant effect modification by history of comorbidities. Significant interactions were observed for history of hypertension (P-interaction = 0.009) and history of CKD (P-interaction = 0.008). Stratified analyses indicated that the protective benefit of the WW pattern was largely confined to participants without these comorbidities (No Hypertension: HR = 0.76, P = 0.015; No CKD: HR = 0.84, P = 0.044). In contrast, among participants with a history of hypertension or CKD, the WW pattern lost its protective association (Hypertension: HR = 1.08; CKD: HR = 1.29), showing a trend towards null or increased risk. Conversely, the RA pattern maintained a favorable trend towards risk reduction even among those with comorbidities (Hypertension: HR = 0.75; CKD: HR = 0.59), suggesting that regular physical activity may be necessary to be associated with lower risk in populations with compromised metabolic or renal health. Dose–response relationships between MVPA volume and gout risk RCS was utilized to further explore the dose–response relationship between physical activity volume and incident gout. To mitigate the potential bias introduced by extreme outliers, the primary analysis focused on participants within the 0–90th percentile of MVPA distribution (Fig. 4), while the analysis of the full population is presented in Fig. S4. In the broad population (Q90), we observed a linear inverse association between total MVPA volume and gout risk across all models (P for nonlinearity > 0.05). In Model 1 and Model 2, increasing MVPA duration was strongly associated with reduced risk (P for overall < 0.001 and P = 0.001, respectively). Importantly, this protective linear trend persisted in the fully adjusted model (Model 3, adjusted for sUA), although the strength of the association was attenuated (P for overall = 0.046; Fig. 4A). Stratified RCS analyses by activity pattern provided deeper insights into this association. For the RA group, the inverse dose–response relationship remained statistically significant even after adjusting for sUA in Model 3 (P for overall = 0.019; Fig. 4B). In contrast, for the WW group, the dose–response association was not statistically significant in the fully adjusted model (P for overall = 0.354; Fig. 4C). Analysis of the full dataset (Q100; Fig. S4) revealed wider confidence intervals. specifically, the statistical significance for Total MVPA was lost in the fully adjusted model (P for overall = 0.177), likely driven by the extreme variability at the highest spectrum of physical activity. Similarly, the association in the WW group remained non-significant (P for overall = 0.187), whereas the RA group maintained a significant inverse association (P for overall = 0.026).Fig. 4Dose–response relationships between MVPA volume and gout risk. estricted cubic splines modeling the association between MVPA volume and incident gout for Total MVPA, regularly active (RA), and weekend warrior (WW) groups. Analyses were restricted to participants within the 0–90th percentile (Q90) of MVPA distribution to mitigate the influence of extreme outliers. Solid lines represent HRs, and shaded areas indicate 95% confidence intervals. The reference point was set at the median MVPA value. Model 1, 2, and 3 correspond to the adjustment levels described in Fig. 1. P for overall indicates the significance of the overall association; P for nonlinearity tests for deviation from linearity. Abbreviations: MVPA, moderate-to-vigorous physical activity; HR, hazard ratio Dose–response relationships between MVPA volume and gout risk. estricted cubic splines modeling the association between MVPA volume and incident gout for Total MVPA, regularly active (RA), and weekend warrior (WW) groups. Analyses were restricted to participants within the 0–90th percentile (Q90) of MVPA distribution to mitigate the influence of extreme outliers. Solid lines represent HRs, and shaded areas indicate 95% confidence intervals. The reference point was set at the median MVPA value. Model 1, 2, and 3 correspond to the adjustment levels described in Fig. 1. P for overall indicates the significance of the overall association; P for nonlinearity tests for deviation from linearity. Abbreviations: MVPA, moderate-to-vigorous physical activity; HR, hazard ratio IPTW analysis To corroborate the robustness of our findings and mitigate potential confounding bias, we performed an IPTW analysis. Covariate balance was effectively achieved, with absolute standardized mean differences for all covariates reduced to < 0.10 after weighting for both comparisons (Fig. 5A–B, Fig. S5A, Table S4-6).Fig. 5IPTW analysis. A–B Love plots showing the absolute standardized mean differences (SMD) of covariates before (orange) and after (blue) IPTW adjustment for the comparison of (A) regularly active vs. inactive and (B) weekend warrior vs. inactive. An absolute SMD < 0.10 indicates adequate covariate balance. C Weighted HRs for gout risk derived from IPTW-adjusted Cox proportional hazards models. Model 1, 2, and 3 correspond to the adjustment levels described in Fig. 1. Abbreviations: IPTW, Inverse probability of treatment weighting; HR, hazard ratio; IR/1000PY, incidence rate (per 1,000 person-years); sUA, serum uric acid; TDI, Townsend Deprivation Index IPTW analysis. A–B Love plots showing the absolute standardized mean differences (SMD) of covariates before (orange) and after (blue) IPTW adjustment for the comparison of (A) regularly active vs. inactive and (B) weekend warrior vs. inactive. An absolute SMD < 0.10 indicates adequate covariate balance. C Weighted HRs for gout risk derived from IPTW-adjusted Cox proportional hazards models. Model 1, 2, and 3 correspond to the adjustment levels described in Fig. 1. Abbreviations: IPTW, Inverse probability of treatment weighting; HR, hazard ratio; IR/1000PY, incidence rate (per 1,000 person-years); sUA, serum uric acid; TDI, Townsend Deprivation Index In the IPTW-adjusted logistic regression analysis (Model 3), the RA pattern maintained a robust and statistically significant inverse association with gout risk compared to the inactive group (HR = 0.78, 95% CI: 0.63–0.98; P = 0.031; Fig. 5C). In contrast, the association for the WW pattern was attenuated and did not reach statistical significance (HR = 0.92, 95% CI: 0.78–1.09; P = 0.319; Fig. 5C). These findings align with the multivariable Cox regression results, reinforcing the observation that while regular activity offers independent protection, the protective association of the irregular WW pattern may be limited after fully accounting for baseline metabolic profiles. Furthermore, when directly comparing the two active groups using IPTW (Fig. S5B), no statistically significant difference was observed (HR = 1.12, 95% CI: 0.89–1.42; P = 0.337), although the hazard ratio greater than 1 continues to suggest a trend toward lower risk in the RA group compared to the WW group. Joint associations of physical activity patterns and PRS with gout risk Formal tests for interaction revealed no statistically significant interaction between physical activity patterns and PRS categories across all models (all P for interaction > 0.10) (Fig. 6). This lack of interaction suggests that the association between physical activity and gout risk is consistent regardless of an individual's genetic predisposition. In the stratified analysis, the RA pattern demonstrated a robust and significant protective association across all models, maintaining a 41% risk reduction even in the fully adjusted model including sUA (Model 3: HR = 0.59, 95% CI: 0.37–0.92; P = 0.020). Conversely, while the WW pattern showed a protective association in the minimally adjusted model (Model 1: P = 0.010), this association was attenuated and lost statistical significance after adjusting for sociodemographic and lifestyle factors (Model 2: P = 0.072) and sUA (Model 3: HR = 0.77, 95% CI: 0.56–1.05; P = 0.093).Fig. 6Joint associations of physical activity patterns and PRS with gout risk. Forest plot showing the HRs for physical activity patterns stratified by genetic risk categories (Low, Middle, High) based on a gout PRS. Tests for interaction evaluated whether the association of physical activity varied across genetic risk strata. Models correspond to those described in Fig. 1. Note that no statistically significant interaction was observed (all P > 0.10), and the RA pattern maintained a significant protective association in Model 3 independent of genetic risk. Abbreviations: HR, hazard ratio; PA, physical activity; PRS, polygenic risk score Joint associations of physical activity patterns and PRS with gout risk. Forest plot showing the HRs for physical activity patterns stratified by genetic risk categories (Low, Middle, High) based on a gout PRS. Tests for interaction evaluated whether the association of physical activity varied across genetic risk strata. Models correspond to those described in Fig. 1. Note that no statistically significant interaction was observed (all P > 0.10), and the RA pattern maintained a significant protective association in Model 3 independent of genetic risk. Abbreviations: HR, hazard ratio; PA, physical activity; PRS, polygenic risk score Collectively, these results indicate that while high genetic risk is a strong determinant of gout susceptibility, RA pattern confers a significant protective benefit independent of genetic background. The benefit of the WW pattern, however, appears less robust when accounting for covariates and genetic risk. Sensitivity analyses The results of the sensitivity analyses were largely consistent with our primary findings, reinforcing the distinction between the two activity patterns (Fig. S6-14). In the fully adjusted models (Model 3), the RA pattern maintained a robust and statistically significant (or marginally significant) protective association with gout risk across most scenarios. In contrast, the association for the WW pattern was consistently attenuated and failed to reach statistical significance in Model 3 across these sensitivity analyses (all P > 0.05). Finally, the falsification tests showed no significant associations between physical activity patterns and the negative control outcomes (hearing loss or acute appendicitis; both P > 0.05), suggesting that our observed results were not driven by residual confounding or systematic bias. Discussion Based on a large prospective cohort from the UK Biobank, this study systematically investigated the association between different physical activity patterns and the risk of incident gout. Leveraging objective accelerometer data, our findings provide a nuanced understanding of the "volume vs. regularity" debate in gout prevention. While both the RA and WW patterns were associated with a reduced risk of gout compared to inactivity in minimally adjusted models, our results reveal a critical distinction between the two. The protective association of the RA pattern remained robust and statistically significant even after fully adjusting for baseline sUA and in specific high-risk subgroups. In contrast, the benefit of the WW pattern was largely attenuated and lost statistical significance after accounting for baseline metabolic profiles and comorbidities. This suggests that while total MVPA volume is a driver of risk reduction, the regularity of activity serves as a vital determinant of risk reduction potential, particularly for individuals with established metabolic risk factors. Previous studies focusing on the weekend warrior phenomenon have generally reported that for outcomes such as cardiovascular mortality, cancer, and all-cause mortality, the WW pattern confers benefits comparable to regular activity [10, 23, 31]. However, our study indicates that this equivalence might not fully extend to gout. Unlike cardiovascular risk reduction, which can benefit from cumulative effects of sporadic physical activity, gout is fundamentally driven by the crystallization of monosodium urate in the presence of hyperuricemia [23]. The attenuation of the protective association for the WW pattern after adjusting for baseline sUA suggests that the benefit of this sporadic activity pattern is highly dependent on metabolic control, specifically serum urate levels. Conversely, the RA pattern maintained a robust risk reduction even after strictly adjusting for baseline sUA. This finding implies that regular physical activity serves as a potent factor capable of modifying the subsequent clinical trajectory—potentially by limiting the progression of hyperuricemia or enhancing anti-inflammatory defenses—regardless of the initial baseline urate burden. The divergence between RA and WW patterns after adjusting for baseline sUA highlights the importance of the frequency and regularity of the physiological stimulus. We propose that the protective association of physical activity on gout is likely cumulative and requires sustained engagement to counteract metabolic risks. First, although we controlled for baseline sUA, this snapshot does not capture changes over time. Regular exercise is known to improve insulin sensitivity and enhance renal urate excretion [32]. The consistent exercise in the RA group may be more effective at maintaining lower sUA levels or preventing the progression of hyperuricemia over the follow-up period compared to the infrequent bouts performed by weekend warriors. Consequently, RA participants starting with the same baseline sUA as inactive individuals may end up with a lower cumulative urate burden over 8 years. Second, regular exercise induces long-term adaptations in the innate immune system, potentially raising the threshold for crystal-induced inflammation via downregulation of the NLRP3 inflammasome [33]. This physiological adaptation likely requires frequent reinforcement. Thus, RA participants may achieve a modified trajectory through better longitudinal metabolic control and immune conditioning, whereas the sporadic nature of the WW pattern may fail to provide this continuous protection once the baseline advantage is removed. Our subgroup and interaction analyses further reinforce the superiority of the RA pattern for vulnerable populations. We observed significant interactions where the protective association of the WW pattern was attenuated to null in participants with a history of hypertension or CKD. In contrast, the RA pattern showed consistent protective trends even in these high-risk groups. This carries profound clinical implications for precision exercise prescription. For the general healthy population, the WW pattern remains a viable option to reduce gout risk, primarily by helping to maintain a healthier metabolic profile. However, for high-risk individuals—such as those with obesity, hyperuricemia, or existing comorbidities—our data suggests that WW pattern is insufficient to offset the elevated risk. For these patients, regularity is non-negotiable. The RA pattern should be explicitly recommended as the standard of care, as it demonstrates the capacity to reduce risk independent of the patient's starting metabolic profile. A key strength of our study is the integration of gout PRS. We found no significant interaction between genetic risk and physical activity patterns, indicating that the benefits of MVPA are universal. However, consistent with our main findings, the RA pattern provided a robust risk reduction (HR ~ 0.59 in Model 3) across all genetic strata, whereas the WW pattern was less consistent. This implies that individuals with a high genetic predisposition to gout can significantly offset their risk through physical activity, but adherence to a regular activity routine yields the most reliable protection. Our study findings have significant public health and clinical implications, advocating for a nuanced, stratified approach to gout prevention. They strongly support the inclusion of physical activity as a cornerstone for gout risk reduction strategies. Crucially, our results suggest a paradigm shift towards precision exercise prescription. For the general population with healthy metabolic profiles, the weekend warrior pattern remains a feasible and beneficial option to combat physical inactivity and maintain metabolic health, particularly for those with time constraints. However, for metabolically vulnerable individuals—such as those with hyperuricemia, obesity, or comorbidities—our findings indicate that flexibility should not come at the expense of regularity. For these high-risk groups, achieving the recommended MVPA volume through consistent, regular sessions appears necessary to secure the robust, independent protection against gout that sporadic exercise fails to provide. While weight loss remains paramount, this study provides robust evidence to warrant explicit recommendations regarding not just the volume, but also the frequency of physical activity in future gout management and prevention guidelines. This addresses a notable gap in the current 2020 ACR guidelines [9], suggesting that clinicians should emphasize regular engagement in physical activity as a standard of care for patients prone to gout. Limitations First, accelerometer data were collected during a single 7-day period. While this provides an objective snapshot, it may not perfectly reflect an individual's typical long-term activity pattern. Second, despite adjusting for numerous potential confounders and employing IPTW, residual confounding from unmeasured factors (e.g., specific dietary components like purine intake not captured by HDS) cannot be ruled out. Third, we lacked longitudinal sUA measurements to definitively track urate trajectories over the follow-up period. Fourth, the study population consisted predominantly of individuals of White European ancestry, potentially limiting generalizability to other racial or ethnic groups where gout epidemiology and associated factors may differ. Fifth, we acknowledge the healthy volunteer selection bias inherent in the UK Biobank cohort [34]. Consequently, the absolute incidence rates of gout observed in our study may be lower than those in the general population. However, previous methodological studies indicate that while this non-representativeness affects prevalence estimates, exposure–disease associations remain widely generalizable [35]. Since our primary aim was to compare the relative risk between different physical activity patterns rather than estimate gout prevalence, this bias is unlikely to compromise the internal validity of our findings. Finally, as an observational study, causality cannot be definitively established, although the prospective design and comprehensive sensitivity analyses support the robustness of the observed associations. association of physical activity varied across genetic risk strata. Models correspond to Conclusions In conclusion, while engaging in any moderate-to-vigorous physical activity is associated with a lower risk of gout compared to inactivity, the regularly active pattern demonstrates a more robust inverse association than the weekend warrior pattern. Future clinical guidelines might consider highlighting the potential importance of not just the volume, but the regularity of physical activity, particularly for individuals with compromised metabolic health. Code availability Analytic code supporting this research will be available upon reasonable request to the corresponding author. Supplementary Information Supplementary Material 1. Supplementary Material 1.