Logistic regression analysis linking individual food groups or nutrients and the presence of NAFLD (n = 797).

<p>¹ Tertiles based on the distribution of all subjects, except two categories for the fast food intake (intake = 0 vs. intake > 0 g per day)</p><p>² Further adjusted for BMI, current smoker status (yes/no), current drinker status (yes/no), central obesity (yes/no), triglyceride >1.7 mmol/l (yes/no), reduced HDL-cholesterol (yes/no), hypertension (yes/no), impaired fasting glucose or diabetes (yes/no), and the <i>PNPLA3</i> genotypes (CC vs. CG vs. GG genotypes). Energy intake was further adjusted in the multivariate model of food group intake analysis.</p><p>³<i>P</i>_<i>trend</i> was examined by entering tertiles of food group intake as a fixed factor for the models of food group intake analysis, whereas <i>P</i> value was used for the models of nutrient intake analysis.</p><p>⁴ Included mainly sweetened beverages or juice, carbonated drinks and sports drinks</p><p>NAFLD, non-alcoholic fatty liver disease</p><p>Logistic regression analysis linking individual food groups or nutrients and the presence of NAFLD (n = 797).