The prognostic role of diet quality in patients with MAFLD and physical activity: data from NHANES

MAFLD, formerly NAFLD, involves hepatic triglyceride accumulation with metabolic dysfunction and is linked to increased mortality. With no approved pharmacologic therapies, lifestyle modification—dietary control and increased physical activity—remains first-line. Many patients face barriers to maintaining exercise routines, and low diet quality is an independent risk factor for adverse liver outcomes. Using the Healthy Eating Index (HEI) to quantify diet quality, this study investigates how diet quality interacts with physical activity levels to influence mortality among individuals with MAFLD, with a focus on those who are physically inactive.

Prior guidelines endorse combined diet and exercise to improve hepatic steatosis and histology in fatty liver disease. However, patients with fatty liver often have lower adherence to national PA guidelines and face physiological, psychological, and socio-environmental barriers to exercise. Low dietary quality has been associated with NAFLD-related liver mortality, and HEI is a validated measure of overall diet quality. Evidence suggests exercise can mitigate harmful effects of short-term poor diet on vascular function and metabolic health, but the interaction between long-term diet quality and PA level on MAFLD outcomes has been unclear. This study addresses that gap using NHANES III with long-term mortality follow-up.

Design and data source: Analysis of NHANES III (1988–1994) with linked mortality through December 31, 2015 (linked file released 2019). Population: Adults in NHANES III; 19,599 screened. Exclusions: missing hepatic steatosis data (n=6157), missing other key data (n=3030), and those without MAFLD (n=6703), yielding 3709 MAFLD participants. Outcomes: All-cause and cause-specific (cardiovascular-related, cancer-related) mortality. Exposure measures: - Diet quality: Healthy Eating Index (HEI) derived from 24-hour dietary recall provided by NHANES. - Physical activity (PA): Self-reported frequency and type of past-month activities (walking, jogging/running, cycling, swimming, dance, calisthenics, gardening/yard work, weightlifting, others). Each activity had an assigned MET value. Because session duration was not available, an activity score was computed as frequency × MET per activity and summed; participants were dichotomized into PA inactive vs PA active by the median total activity score. Other variables: Demographics (age, sex, race), socioeconomic indicators (education years; low education <12 years; poverty income ratio with low income <1), behavioral factor (overdrinking), comorbidities (type 2 diabetes, hypertension), anthropometrics (BMI, waist-to-hip ratio), labs (HbA1c, ALT, AST, cholesterol, triglycerides), renal function (eGFR by 2009 CKD-EPI), and liver fibrosis indices (FIB-4, NFS). Hepatic steatosis and MAFLD definition: Hepatic steatosis assessed via ultrasound (none, mild, moderate, severe; any of mild–severe counted as steatosis). MAFLD defined as ultrasound-confirmed steatosis plus either overweight/obesity, type 2 diabetes, or ≥2 metabolic risk factors in non-obese individuals per international consensus. Statistical analysis: Continuous variables summarized as mean±SD or median (IQR) and compared by t-test or Mann–Whitney U; categorical variables as n (%) and compared by χ². Cox proportional hazards models assessed associations of HEI and PA with mortality, adjusting for race, age, sex, socioeconomic factors, diabetes, hypertension, metabolic profiles, renal function, liver enzymes, and fibrosis scores to avoid Simpson’s paradox. Hazard ratios (HR) with 95% CI reported. Restricted cubic splines visualized continuous HEI–mortality relationships; contour plots illustrated HEI–PA interactions. Two-sided P<0.05 defined significance. Analyses performed in R.

- Cohort: 3709 MAFLD participants; mean age 46.6±15.4 years; 49.7% male; median follow-up 26.2 years (IQR 19.3–28.1); deaths: 1549 (41.8%). - Overall Cox multivariable model for all-cause mortality: • HEI score: HR 0.995 (95% CI 0.992–0.999), P=0.017 (inverse association). • PA level: HR 0.999 (95% CI 0.999–1.000), P=0.029 (inverse association; small effect due to scale). • Other significant covariates (direction): male sex (HR 1.156), older age (HR 1.067 per year), race (HR 1.228), overdrinking (HR 1.323), type 2 diabetes (HR 1.150), hypertension (HR 1.207), higher WHR (HR 4.790), higher HbA1c (HR 1.103), higher triglycerides (HR 1.038), higher AST (HR 1.010), higher ALT (protective; HR 0.991), higher eGFR (protective; HR 0.994); cholesterol, FIB-4, NFS not significant after adjustment. - HEI spline: Mortality risk decreased with higher HEI; inflection at HEI ≈62.8. - PA subgroup analyses (multivariable): • PA inactive: HEI inversely associated with all-cause mortality, HR 0.992 (95% CI 0.986–0.997), P=0.002. • PA active: No significant association, HR 0.999 (95% CI 0.994–1.004), P=0.672. • Contour plots and RCS confirmed these patterns. - Cause-specific mortality: In overall cohort, HEI and PA not significantly associated with cardiovascular- or cancer-related mortality after adjustment. In PA inactive subgroup, higher HEI associated with lower cardiovascular mortality (HR 0.988, 95% CI 0.978–0.999, P=0.027) and lower cancer mortality (HR 0.986, 95% CI 0.975–0.998, P=0.018); no significant associations in PA active subgroup. - Baseline differences by PA: PA active vs inactive had higher HEI (median 64.4 vs 61.9; P<0.001), lower BMI (28.9 vs 30.2 kg/m²; P<0.001), lower diabetes prevalence (20.8% vs 26.8%; P<0.001), lower WHR and HbA1c, and better survival by Kaplan–Meier (log-rank P=0.024).

Findings suggest differential impacts of lifestyle components on prognosis in MAFLD: higher diet quality reduces long-term mortality primarily among individuals with low physical activity, while in physically active individuals, the mortality impact of diet quality appears attenuated, consistent with literature showing exercise can offset adverse metabolic and vascular effects of poor diet. Thus, for patients unable to sustain exercise, improving diet quality may be crucial to mitigate mortality risk. Results align with evidence that higher diet quality is associated with lower systemic inflammation, oxidative stress, frailty, and mortality, and may explain inconsistencies in prior NAFLD studies that did not account for PA or used different disease definitions (NAFLD vs MAFLD). Clinically, encouraging adherence to at least one lifestyle pillar—either being physically active or maintaining a high-quality diet—may confer meaningful survival benefits in MAFLD.

A high-quality diet was independently associated with reduced all-cause mortality among MAFLD patients with low physical activity, but not among those with active PA levels. These findings highlight the importance of emphasizing diet quality for sedentary individuals with MAFLD. Future research should incorporate longitudinal, repeated and objective assessments of diet and physical activity (e.g., detailed food records, wearable trackers), quantify energy expenditure (MET-min/week), and verify these interactions in prospective cohorts and intervention trials.

- PA and diet were self-reported, subject to recall and reporting bias. - HEI based on a single 24-hour dietary recall, which may not fully capture usual intake, though adequate for group stratification. - Lacked exercise session duration; unable to compute MET-min/week, so PA was approximated via frequency×MET and dichotomized at the median, potentially causing misclassification. - PA and diet were assessed only at baseline; lifestyle changes over time were not captured. - Despite comprehensive adjustment, residual confounding is possible.