Among 11,168 adults in the 2016-2019 Korea National Health and Nutrition Examination Survey, those in the highest quintile of ultra-processed food (UPF) intake had 28% higher odds of NAFLD than the lowest quintile (OR 1.28, 95% CI 1.06-1.56; p-trend <0.001). Higher UPF intake tracked with greater adiposity, higher LDL-c and non-HDL-c, and elevated ALT or ALT/AST ratio. Processed meat and fish drove the subgroup signal, and substituting 100 g/day with minimally processed alternatives was associated with 11% lower odds (OR 0.89, 0.80-0.98).
The direction matches Western cohorts, and this adds evidence from an Asian population, where UPF research is thin and diets differ. The effect size is modest, and the design limits what it proves. A single 24-hour recall is a noisy measure of habitual intake, and cross-sectional data cannot establish temporality or rule out reverse causation. NAFLD was defined by a composite score rather than imaging or biopsy, which adds misclassification. The processed meat and fish finding (OR 1.04 per unit) is small and fragile. The substitution result is the most practical element, since it suggests swapping specific foods rather than avoiding a vague category. UPF intake also travels with total calories, adiposity, and lower diet quality, so independent liver effects are hard to isolate despite multivariable adjustment. I rate this as confirmatory and incremental. It supports favoring minimally processed foods, especially in place of processed meats, for liver health, but prospective cohorts are needed before treating the association as causal.