Association between fast eating speed and metabolic dysfunction-associated steatotic liver disease: a multicenter cross-sectional study and meta-analysis

Metabolic dysfunction-associated steatotic liver disease (MASLD), formerly termed nonalcoholic fatty liver disease (NAFLD), is highly prevalent and linked to obesity, metabolic syndrome, hypertension, and diabetes. The nomenclature changed in 2023 to emphasize metabolic dysfunction, while clinical criteria and natural histories remain essentially identical between MASLD and NAFLD. Lifestyle modification with weight loss is currently the only recommended therapy. Eating speed has emerged as a potential lifestyle factor affecting metabolic health. Prior studies report that fast eating is associated with overweight/obesity, glycemic excursions, incident diabetes, and elevated liver enzymes, yet evidence linking fast eating to MASLD is mixed across populations and analytic adjustments. This study aimed to evaluate whether frequent fast eating is associated with higher MASLD risk in a large, multicenter Chinese cohort and to synthesize existing evidence via meta-analysis.

Previous observational studies suggested that MASLD patients may eat faster than those without MASLD, and some cohorts (e.g., Korean adults) found higher MASLD risk among fast eaters. Associations have also been reported in male patients with type 2 diabetes. However, other studies found no independent association after adjusting for adiposity (BMI, waist circumference) and other confounders, including a prior Chinese cohort. Across seven reviewed studies (one retrospective cohort, six cross-sectional) conducted in Asian populations (Japan, China, South Korea, Iran), eating speed categories varied (2–4 levels), and MASLD definitions ranged from ultrasound diagnosis to indices (FLI) or CT liver/spleen attenuation ratio. Adjustment strategies differed, with some studies using limited covariate control. These mixed findings motivated a meta-analysis using adjusted estimates to clarify whether fast eating independently associates with MASLD beyond adiposity and energy intake.

Design and participants: Sub-analysis of a multicenter cross-sectional study recruiting 2704 adults from 10 clinics of obesity, diabetes, and metabolic diseases across six provinces/municipalities in China (January–December 2011). After exclusions for missing data, excess alcohol intake (>20 g/day men, >10 g/day women), viral hepatitis, other liver diseases, or endocrine disorders, 1965 participants (977 men, 988 women) were included. Data collection: Standardized, face-to-face questionnaires recorded demographics, smoking, alcohol use, medical history, medications, and eating habits (skipping breakfast, late eating, snacking, fast eating). Geographic region categorized by the Qinling–Huaihe line (South: Shanghai, Jiangsu, Henan; North: Heilongjiang, Tianjin, Shandong). Anthropometrics: height, weight, waist circumference (WC), blood pressure; BMI calculated as kg/m². Laboratory measures: fasting serum TG, TC, ALT, AST, GGT, LDL-c, HDL-c, uric acid using Hitachi 7600 analyzer; HbA1c via HPLC (BIO-RAD II TURBO). Oral glucose tolerance test (75 g) measured fasting glucose (FBG) and 2-hour postload glucose (2hPBG); glucose status classified as NGT, IGR, or T2DM. MASLD assessment: Hepatic steatosis diagnosed by standardized ultrasonography (increased hepatic echogenicity vs renal cortex and portal vein blurring) performed by experienced sonographers on calibrated machines. Exposure: Fast eating defined as meal time <5 minutes; frequency self-reported as ≤1 time/month, ≤1 time/week, or ≥2 times/week. Statistical analysis: Continuous variables summarized as mean±SD or median (IQR); categorical as counts (%). Group comparisons via one-way ANOVA or Wilcoxon tests; categorical via chi-squared tests. Multivariable logistic regression modeled MASLD odds by fast eating categories (reference: ≤1 time/month) and as a continuous frequency variable. Model 1 adjusted for age, sex, smoking, drinking, BMI, and region. Model 2 further adjusted for WC, FBG, HbA1c, TG, TC, HDL-c, LDL-c, ALT, AST, GGT, and uric acid. Interactions tested for sex, region, smoking, glucose status, and BMI via likelihood ratio tests. Meta-analysis: Systematic searches in PubMed and Web of Science up to Nov 2023 using terms for eating speed and fatty liver/MASLD. Inclusion: cross-sectional/cohort/case-control studies reporting adjusted OR/RR for non-slow eaters vs slow eaters. After screening 25 records (PRISMA), 4 eligible studies were combined with current study for meta-analysis; quality assessed by NOS and AHRQ scales by two independent reviewers. Random-effects model computed pooled ORs with 95% CIs; heterogeneity assessed by I² and τ²; leave-one-out sensitivity analyses performed. Software: R 4.2.3 with rms, epiDisplay, dplyr, ggplot2, PMCMRplus, metafor, forestploter; two-sided P<0.05 considered significant.

- In 1965 participants (mean age 53.6 years; mean BMI 24.80±3.61 kg/m²), overall MASLD prevalence was 48.8%. - MASLD prevalence by fast eating frequency: ≤1 time/month 46.2%, ≤1 time/week 50.5%, ≥2 times/week 59.3% (P for trend <0.001). - Unadjusted ORs vs ≤1 time/month: ≤1 time/week OR 1.19 (0.93–1.52), ≥2 times/week OR 1.70 (1.31–2.20); fast eating frequency as continuous: OR 1.28 (1.14–1.45), all P<0.001 for trend. - Model 1 (adjusted for age, sex, smoking, drinking, BMI, region): ≤1 time/week OR 1.20 (0.91–1.60), ≥2 times/week OR 1.45 (1.08–1.96); frequency as continuous OR 1.20 (1.04–1.39), P=0.011. - Model 2 (additional adjustment for WC, FBG, HbA1c, lipids, liver enzymes, UA): ≤1 time/week OR 1.08 (0.78–1.49), ≥2 times/week OR 1.81 (1.26–2.59); frequency as continuous OR 1.29 (1.09–1.53), P=0.003. - Subgroups (fully adjusted, frequency as continuous) showed significant trends in women (P=0.007), North China (P=0.004), non-smokers (P=0.014), IGR (P=0.001), and BMI <25 kg/m² (P=0.002); no significant interactions by sex, region, smoking, glucose status, or BMI (all P for interaction >0.19). - Meta-analysis (four eligible studies plus current study): pooled OR 1.22 (95% CI 1.07–1.39) for frequent fast eating and MASLD; moderate heterogeneity (I²=46%, τ²=0.0192, P=0.05). Leave-one-out sensitivity analyses preserved significance. - Fast eaters (≥2 times/week) had 81% higher adjusted MASLD risk vs ≤1 time/month group (Model 2, P=0.011).

The study demonstrates that frequent fast eating is independently associated with higher MASLD risk after comprehensive adjustment for demographic, behavioral, adiposity, metabolic, and hepatic enzyme covariates. These findings, supported by a meta-analysis of adjusted estimates from multiple studies, address prior inconsistencies and suggest the association is not fully mediated by obesity or calorie intake. Potential mechanisms include increased energy intake due to diminished satiety signaling with rapid eating, altered postprandial gut hormone responses (reduced ghrelin suppression; lower PYY/GLP-1 responses), and downstream effects on the gut–liver axis and hepatic lipid metabolism. Modulation of liver lipid handling via gut peptide signaling pathways and microbiota-derived metabolites is hypothesized but requires further study. The results underscore eating speed as a modifiable behavior potentially relevant to MASLD risk reduction, particularly among women, individuals in Northern China, non-smokers, those with impaired glucose regulation, and those with BMI <25 kg/m².

Frequently eating fast is positively associated with an increased risk of MASLD in a large multicenter Chinese sample, and meta-analytic evidence corroborates this association. Future research should elucidate causal mechanisms, assess impacts on liver fibrosis, and test whether slowing eating speed reduces MASLD risk or progression.

- Cross-sectional design precludes causal inference. - Lack of data on total energy intake and dietary composition; residual confounding by diet cannot be excluded. - All included studies in meta-analysis were from Asian populations, limiting generalizability to other ethnicities. - Non-uniform definitions and categorization of fast eating across studies. - Self-reported eating speed may be subject to recall or reporting bias.