Physical activity and advanced fibrosis in MASLD, MetALD, and ALD in a nationally representative cohort: NHANES 2017-2020.
Oct 14, 2025DOI: 10.1097/hc9.0000000000000797Journal ArticleHepatology Communications
Written by: Nicholas DunnWanyu ZhangHanna L BlaneyLuis Antonio DiazArpan PatelRobert J WongJuan Pablo ArabZobair YounossiCraig McClainMaiying KongAshwani K Singal
Physical activity (PA) has been widely recognized as a key component in the treatment of metabolic disorders, such as metabolic dysfunction–associated steatotic liver disease (MASLD), formerly known as NAFLD. Empirical evidence indicates that recreational or leisure-time PA (LTPA), particularly aerobic and resistance exercise, is associated with reduced hepatic steatosis, improved insulin sensitivity, and decreased risks of advanced fibrosis and death in individuals with MASLD.1–5 However, the effect of PA on the broader range of steatotic liver disease (SLD), such as metabolic dysfunction and alcohol-associated liver disease (MetALD) and alcohol-associated liver disease (ALD), is unknown. Furthermore, the contribution of work-related PA (WPA) to liver health has not been examined in a systematic way despite its interest in an occupational setting.PA regulates the major biologic metabolic pathways, such as hepatic lipid metabolism, inflammation, and mitochondrial function, thereby resulting in improvements in hepatic triglyceride content and liver histology independent of weight loss.1 In addition, PA exerts systemic influences through enhancing cardiorespiratory fitness, insulin resistance, and lipid homeostasis, all of which play an important role in the pathogenesis of SLD. Prospective cohort studies have demonstrated that LTPA is linked to reduced all-cause mortality in MASLD, with vigorous activity conferring additional survival benefit.2,5,6 Whether these results apply to MetALD and ALD, where alcohol-associated hepatotoxicity may neutralize some of the beneficial effects of PA, has yet to be determined.Aside from LTPA, the role of WPA in liver disease has not been reviewed. Whereas LTPA, which is structured, is usually followed by beneficial health effects, the effect of WPA is less certain. WPA can entail prolonged standing, frequent movement, and exertion. It is unclear if this type of PA has similar metabolic beneficial effects. Limited research indicates that WPA is neutral or even harmful to health because of heightened energy expenditure without adequate recovery, stress, or compensation mechanisms like elevated caloric intake. As such, it is important to consider whether WPA exerts similar preventive benefits on liver fibrosis and SLD spectrum disease progression.Our aim was to evaluate the association of both WPA and LTPA with prevalent advanced fibrosis in MASLD, MetALD, and ALD through the National Health and Nutrition Examination Survey (NHANES) 2017–2020 data. Specifically, we (1) examined if LTPA is related to lower fibrosis risk by SLD subtype; (2) examined WPA effects on the severity of liver disease; and (3) examined PA type, intensity, and SLD subtype interaction. By addressing these knowledge gaps, our findings will contribute to a fuller understanding of PA as a modifiable risk factor for liver fibrosis, informing targeted public health initiatives and clinical guidelines for patients across the whole SLD spectrum.
METHODS
This study used data from the 2017–March 2020 pre-pandemic NHANES database. We included adults (≥18 y) with reliable elastography assessments. Exclusion criteria included fasting duration of <3 hours, fewer than 10 valid liver stiffness measurements (LSMs), and an IQR/M ratio >30%. We included participants with liver steatosis (controlled attenuation parameter >248 dB/m).7 and at least 1 cardiometabolic risk factor (CMRF). Subjects with missing information on alcohol use or PA and those positive for active HBV or HCV infection were excluded (Figure 1). The SLD population was classified into 3 groups based on alcohol consumption.8 Participants who answered “No” to ALQ111 (“Ever had a drink of any kind of alcohol”) or “Never in the last year” to ALQ121 (“Past 12 mo, how often drink alcoholic beverages”) were classified as MASLD. For participants reporting alcohol consumption, average daily intake (in grams) was calculated by dividing the number of drinks reported per day (ALQ130) by the frequency of drinking (ALQ121), assuming 14 g of alcohol per drink. Men consuming 30–60 g/d and women consuming 20–50 g/d were classified as MetALD, while those below these thresholds were classified as MASLD and those exceeding these thresholds were classified as ALD. The link to the detailed questionnaire is https://wwwn.cdc.gov/Nchs/Data/Nhanes/Public/2017/DataFiles/ALQ_J.htm
Study population attrition diagram from NHANES 2017–2020.
The primary outcome was at-risk advanced fibrosis, defined as an Agile 3+ score ≥0.45. The Agile 3+ score is a probability-based model that incorporates vibration-controlled transient elastography LSM along with common laboratory parameters, including age, AST, ALT, platelet count, and a history of type 2 diabetes mellitus.9–11 The Agile 3+ score has a similar sensitivity and specificity versus LSM (C-statistics of 0.78) but with lower (8.3 vs. 13.1%) indeterminate results.10
PA was quantified as metabolic equivalent (MET)-minutes per week based on the NHANES Physical Activity Questionnaire, adapted from the Global Physical Activity Questionnaire (GPAQ). Participants reported the number of days per week and average duration (in minutes) per day engaged in vigorous-intensity and moderate-intensity activities in 3 domains: occupational (work-related tasks including paid/unpaid work and household chores), transportation (walking or bicycling), and leisure-time recreational activities. Vigorous-intensity activities were assigned a MET score of 8.0, and moderate-intensity activities, including walking or bicycling for transportation, were assigned a MET score of 4.0. Total MET-minutes per week for each activity were computed by multiplying days per week by duration and corresponding MET scores, and then summing across domains. Participants were then categorized as achieving either high (≥600 MET-min/wk) or low (<600 MET-min/wk) PA. Further details regarding NHANES questionnaire administration and MET assignment can be found on the NHANES website (https://wwwn.cdc.gov/Nchs/Data/Nhanes/Public/2017/DataFiles/PAQ_J.htm).
Participants were categorized into either high physical activity (≥600 MET-min/wk) or low physical activity (<600 MET-min/wk) groups based on established international guidelines. Specifically, the threshold of 600 MET-min/wk aligns with the World Health Organization’s (WHO) recommendations that adults should engage in at least 150 minutes of moderate-intensity or 75 minutes of vigorous-intensity PA weekly, equivalent to roughly 600 MET-minutes per week (150 min × 4 METs for moderate intensity or 75 minutes × 8 METs for vigorous intensity).12,13 This categorization has been widely adopted in epidemiological research examining associations between PA and health outcomes.14,15 Notably, approximately half of the participants in NHANES legitimately reported zero minutes of both LTPA and WPA, resulting in a zero-inflated distribution that cannot be adequately addressed through traditional data transformations such as log transformation.16 Therefore, the simplest and most statistically appropriate approach is to analyze PA as a binary variable.
NHANES employs a complex, multistage, stratified sampling design to ensure nationally representative estimates. Survey-specific methods were used to account for sampling weights, strata, and primary sampling units. A survey design object was created in R using the svydesign function. Logistic regression models (svyglm) with a quasibinomial family examined associations between work-related and leisure-related MET-minutes per week and advanced fibrosis.
Among the confounders considered, dietary intake is crucial yet challenging to accurately adjust for using standard statistical methods due to compensatory dietary responses. Therefore, we applied unsupervised clustering analysis based on 2-day dietary recall data (including percentage calories from protein, carbohydrates, sugar, fiber, alcohol, saturated fats, monounsaturated fats, and polyunsaturated fats) to categorize NHANES participants into 4 distinct dietary patterns: general diet, high-fat low-carbohydrate diet, low-fat high-carbohydrate diet, and high-alcohol diet. While detailed analyses of dietary patterns in relation to SLD subtypes will be reported separately, the derived dietary clusters were utilized as confounding variables in this analysis. Other confounder variables considered include age (rounded to the nearest decade to avoid collinearity with the outcome variable), sex, race/ethnicity, education level, income-to-poverty ratio, SLD etiology, the number of metabolic risk factors, caffeine intake, and smoking. Metabolic risk factors were defined as the presence of any of the following: overweight or obesity, defined as a body mass index ≥25 kg/m2; impaired glucose metabolism, indicated by self-reported diabetes, a fasting glucose level ≥100 mg/dL, or an HbA1c ≥5.7%; low HDL cholesterol, defined as <40 mg/dL for men or <50 mg/dL for women; elevated triglycerides, defined as ≥150 mg/dL; and hypertension, defined as either a self-reported diagnosis, systolic blood pressure ≥130 mm Hg, diastolic blood pressure ≥85 mm Hg, or having been told by a health professional that one has high blood pressure.
Participant characteristics were stratified based on whether their work-related and recreational exercise exceeded 600 MET-minutes. Continuous variables were compared using svymean() to estimate survey-weighted means and svyvar() to estimate survey-weighted variances. Group comparisons for continuous variables were conducted using svyttest() to account for the complex survey design. Categorical variables were summarized as survey-weighted proportions using svyby() and compared between groups with svychisq() to assess differences across categories. All analyses incorporated sample weights, clustering, and stratification to ensure nationally representative estimates. A 2-tailed p value of <0.05 was considered statistically significant. Logistic regression analysis was used for the primary outcome, at-risk advanced fibrosis (defined as an Agile 3+ score ≥0.45), and conducted accounting for the NHANES complex survey design, including stratification, clustering, and sampling weights.
RESULTS
A total of 4050 participants with SLD were included in the analysis (Figure 1). The subtypes of SLD were MASLD (n=2236), MetALD (n=1355), and ALD (n=457). Subjects performing ≥600 versus <600 METs/wk for both LTPA and WPA were younger, males, with college or higher education, and had higher income above the poverty threshold. In contrast, subjects with a higher number of CMRF more often performed <600 METs/wk LTPA and WPA. Regarding the subtype of SLD, subjects with MetALD and ALD more often and those with MASLD less often performed ≥600 METs/wk LTPA and WPA (Table 1). Further, the noninvasive test scores like LSM on transient elastography, Agile 3+ score, and controlled attenuation parameter were lower in subjects with ≥600 versus <600 METs/wk of both LTPA and WPA (Table 1).
Baseline characteristics of study participants stratified by physical activity level (<600 vs. ≥600 METs/wk)
A total of 1867 (47.7%) subjects performed ≥600 METs of WPA. The proportion of subjects performing ≥600 of WPA was 41.6% in MASLD, 51.0% in MetALD, and 62.2% in ALD. Similarly, a total of 1176 (32.8%) subjects performed ≥600 METs of LTPA. The proportion of subjects performing ≥600 minutes of LTPA was 26.5% in MASLD, 38.2% in MetALD, and 41.5% in ALD (Figure 2).
Proportion of individuals engaging in ≥600 METs/wk of physical activity across each SLD group. The number of patients does not match the percentage due to the NHANES survey design. Abbreviations: MET, metabolic equivalent; NHANES, National Health and Nutrition Examination Survey; SLD, steatotic liver disease.
The prevalence of at-risk advanced fibrosis was lower among subjects performing ≥600 versus <600 METs/wk WPA (19.6% vs. 22.1%, p<0.001), with 14% reduced odds (OR: 0.86, 95% CI: 0.71–1.03) of advanced fibrosis (Figures 3A, 4A). In a multivariable logistic regression model controlled for age, sex, race/ethnicity, education level, income-to-poverty ratio, SLD etiology, and the number of CMRFs, the odds for advanced fibrosis with ≥600 versus <600 METs/wk of WPA was similar (OR: 1.02, 95% CI: 0.77–1.35) (Figure 4B). Similarly, the prevalence of at-risk advanced fibrosis was lower among subjects performing ≥600 versus <600 METs/wk LTPA (13.1% vs. 24.7%, p<0.0001) with 54% reduced odds (OR: 0.46, 95% CI: 0.34–0.62) of advanced fibrosis in unadjusted analysis (Figures 3B, 4C) and 43% reduced odds (OR: 0.57, 95% CI: 0.39–0.82) in adjusted analysis (Figure 4D).
(A) Prevalence of advanced fibrosis by WPA and (B) LTPA across steatotic liver disease groups. Abbreviations: LTPA, leisure-time physical activity; MASLD, metabolic dysfunction–associated steatotic liver disease; MetALD, metabolic dysfunction and alcohol-associated liver disease; SLD, steatotic liver disease; WPA, work physical activity.
Association between WPA and advanced fibrosis across steatotic liver disease groups for unadjusted (A) and adjusted analyses (B). (C, D) Association between LTPA and advanced fibrosis across steatotic liver disease groups for unadjusted and adjusted analyses, respectively. The adjusted analyses controlled for demographics (age, sex, income-to-poverty ratio, education), number of cardiometabolic risk factors, caffeine intake, diet cluster, and smoking. Abbreviations: ALD, alcohol-associated liver disease; LTPA, leisure-time physical activity; MASLD, metabolic dysfunction–associated steatotic liver disease; MetALD, metabolic dysfunction and alcohol-associated liver disease; SLD, steatotic liver disease; WPA, work physical activity.
On a univariate analysis, age and number of CMRF were associated with increased odds for advanced fibrosis, with respective OR (95% CI) of 1.07 (1.06–1.08) and 1.88 (1.72–1.94), respectively. In contrast, MetALD and ALD versus MASLD, Hispanic and Asian versus White race, and education level were associated with lower odds for advanced fibrosis (Supplemental Figure S1, Supplemental Digital Content 1, http://links.lww.com/HC9/C109).
The testing for interaction between WPA and SLD subtypes (p=0.98) and LTPA and SLD subtypes (p=0.55) was negative, indicating there is no evidence of difference among the SLD subtypes. For comprehensiveness, we proceeded to explore the point estimate of each subtype.
The prevalence of at-risk advanced fibrosis comparing ≥600 versus <600 METs/wk WPA was 25.2% versus 26.8% (Figure 3A), with no difference (OR: 0.86, 95% CI: 0.71–1.03) of advanced fibrosis in unadjusted analysis (Figure 4A) and in adjusted analysis, (OR: 1.03, 95% CI: 0.75–1.72) (Figure 4B). The prevalence of at-risk advanced fibrosis comparing ≥600 versus <600 METs/wk LTPA was 29.5% versus 16.8% (Figure 3B), with (OR: 0.48, 95% CI: 0.32–0.72) in unadjusted analysis (Figure 4C) and (OR: 0.56, 95% CI: 0.34–0.93) in adjusted analysis (Figure 4D).
The prevalence of at-risk advanced fibrosis comparing ≥600 versus <600 METs/wk WPA was 16.0% versus 16.6% (Figure 3A), with no difference (OR: 0.92, 95% CI: 0.72–1.18) of advanced fibrosis in unadjusted analysis (Figure 4A) and in adjusted analysis, (OR: 1.21, 95% CI: 0.69–2.11) (Figure 4B). The prevalence of at-risk advanced fibrosis comparing ≥600 versus <600 METs/wk LTPA was 11.9% versus 19.0% (Figure 3B), with (OR: 0.57, 95% CI: 0.32–1.03) in unadjusted analysis (Figure 4C) as well as in adjusted analysis (OR: 0.57, 95% CI: 0.68–1.09) (Figure 4D).
The prevalence of at-risk advanced fibrosis comparing ≥600 versus <600 METs/wk WPA was 13.3% versus 15.6% (Figure 3A), with no difference (OR: 0.95, 95% CI: 0.61–1.50) of advanced fibrosis in unadjusted analysis (Figure 4A) and in adjusted analysis, (OR: 1.14, 95% CI: 0.37–3.53) (Figure 4B). The prevalence of at-risk advanced fibrosis comparing ≥600 versus <600 METs/wk LTPA was 7.0% versus 19.1% (Figure 3B), with (OR: 0.32, 95% CI: 0.15–0.69) in unadjusted analysis (Figure 4C) as well as in adjusted analysis (OR: 0.33, 95% CI: 0.09–1.17) (Figure 4D).
The adjusted OR for WPA ≥600 METs/wk was 1.22 (95% CI: 0.72–2.06). The adjusted OR for LTPA ≥600 METs/wk was 0.55 (95% CI: 0.34–0.89).
In the adjusted multivariable regression models for all subtypes of SLD in both WPA and LTPA analyses, consistent variables associated with increased odds for advanced fibrosis were age and number of CMRFs. In the LTPA analysis, Hispanic versus Caucasian race was associated with lower odds for advanced fibrosis for all individuals with any SLD and for MetALD subtype (Supplemental Table S1, Supplemental Digital Content 2, http://links.lww.com/HC9/C110). Similarly, in the WPA analysis, Hispanic versus Caucasian race and education level were associated with lower odds for advanced fibrosis for individuals with MetALD subtype (Supplemental Table S2, Supplemental Digital Content 2, http://links.lww.com/HC9/C110).
DISCUSSION
Within this NHANES cohort, we identified substantial cross-sectional associations between PA and advanced fibrosis among individuals with SLD. Further, these associations differed by type of activity, with LTPA but not WPA independently associated with the reduced odds of advanced fibrosis in MASLD and MetALD. Although our sample was too small to demonstrate an independent effect of LTPA in individuals with ALD, numerical trends suggest that LTPA may confer similar protective benefits on advanced fibrosis. The results demonstrate that LTPA is a modifiable risk factor in treating advanced fibrosis in SLD.
The beneficial impact of PA on liver health is well established in mitigating steatosis in those at risk for MASLD and in reducing the risk of advanced fibrosis in those with established MASLD. LTPA has been shown to reduce liver fat content, improve biochemical markers, and risk of advanced fibrosis in patients with MASLD. This beneficial effect is independent of weight loss, with aerobic exercise showing the potential for rapid improvement in hepatic steatosis and biochemical markers even with short-term intervention.1 Longer programs of 4 weeks to 6 months duration are associated with even more benefits, like a decrease in intrahepatic triglyceride content, insulin resistance, and histological improvement in MASLD.17–24 In another report, ≥150 min/wk of moderate-intensity LTPA prevented MASLD with additive effects from resistance exercise.3 In an Israeli National Health and Nutrition Survey, it was also observed that regular LTPA, especially anaerobic, was associated with less severe steatosis per SteatoTest.25
Besides improving liver outcomes, observational studies support the benefit of LTPA in improving patient survival. For example, the long-term follow-up of NHANES cohorts in one study showed that LTPA was associated with a lower risk of mortality among patients with MASLD.2,6 In addition, vigorous LTPA also seems to have other protection against all-cause mortality, as shown in a previous NHANES cohort with longer mortality follow-up.5 Despite strong findings in MASLD, the impact of LTPA on MetALD and ALD is unknown. Our study fills the knowledge gap of similar benefits in individuals with MetALD or ALD. In our study, LTPA was associated with a beneficial effect of reducing odds of advanced fibrosis across the whole SLD spectrum, with a very significant risk reduction in MASLD and MetALD and a trend toward benefit in ALD. The absence of a statistically significant correlation in ALD can be due to a smaller number of observations or different injury mechanisms in ALD.
Our study demonstrates that LTPA but not WPA was linked with an independent effect to benefit in reducing the odds of advanced fibrosis in any SLD subgroup. Several large-scale studies across diverse cardiometabolic outcomes have consistently demonstrated that LTPA confers protective health benefits, whereas WPA does not—a phenomenon known as the “physical activity paradox.” For example, the Copenhagen General Population Study revealed that high LTPA significantly reduces cardiovascular and all-cause mortality, whereas elevated WPA is linked to increased risks.26 Similarly, UK Biobank and NANES analyses have also shown mortality benefits from LTPA but minimal or no protective effects from occupational activity.27–29 This paradox has also been observed for type 2 diabetes incidence and hypertension, where LTPA consistently correlates with reduced risk, while WPA does not confer similar protection.30,31 Potential mechanisms underlying this discrepancy include that LTPA typically involves structured intensity, adequate recovery periods, and voluntary engagement, which collectively promote beneficial metabolic adaptations, improved mitochondrial function, reduced oxidative stress, and lower systemic inflammation. In contrast, WPA is often characterized by prolonged exertion with inadequate recovery and sustained physiological stress, leading to elevated chronic inflammation, increased oxidative stress, and impaired cardiovascular and metabolic health.32,33 Our results in liver fibrosis are consistent with current literature in cardiometabolic ourcomes.34
Our study provides novel data on the benefits of LTPA across the SLD spectrum. Other strengths are the use of a database representative of the US population and controlling for socioeconomic status, which can be associated with the type and intensity of work and related stress.34 We used education and household income as surrogate variables of socioeconomic status, and measured the outcome of fibrosis using a more accurate noninvasive test of AGIL3+ score.34 However, we recognize certain limitations of our study. First, a cross-sectional observational design limits the determination of causality and generalizability of our findings. Further, the PA was self-reported without detailed information on the quality and intensity of PA, which could have resulted in misclassification of PA levels. The database was also unable to provide information on the total energy expenditure in the 2 different types of PA. In addition, SLD subtypes were categorized by self-reported alcohol use, which can be affected by underreporting of alcohol use due to recall bias, the sick-quitter effect, and stigma.35 Despite these limitations, our findings encourage the prescription of PA in MASLD and MetALD, as well as the implementation of sustained, population-wide, best-practice communication campaigns to promote PA among the overall population.36
In conclusion, this study clearly demonstrates the differential impact of PA types on advanced liver fibrosis in patients across all subtypes of SLD, including MASLD, MetALD, and ALD. LTPA consistently showed a significant protective effect against advanced fibrosis in patients with MASLD and MetALD, and a similar beneficial trend in ALD. In contrast, WPA did not confer protection against advanced fibrosis after adjusting for relevant confounders, underscoring a critical distinction between structured leisure exercise and occupational PA. These findings highlight the importance of promoting structured leisure-time exercise rather than relying solely on occupational activity to mitigate fibrosis risk in patients with SLD. Further prospective research and interventional studies are necessary to confirm these findings and optimize PA recommendations tailored specifically to the diverse SLD patient population.
HB
Dr. Hanna Blaney
Gastroenterology · Hepatology & Liver TransplantationVirginia Commonwealth University Medical Centerhannab