tree-level 2; and 'cooking methods' such as boiling, frying, formed tree-level 3. The associations between particular food intakes and health outcomes (presence of NAFLD and significant fibrosis, diabetes mellitus, hypertension, dyslipidaemia, and cardiovascular events) were determined by logistic regression. Results Individuals with NAFLD consumed significantly higher amount of cereals (as refined rice), fat and edible oils (as animal fat), meat (as red meat), sugar (as refined sugar) and fried foods, and lower amount of vegetables, pulses, nuts, seeds, and milk compared to controls (all P<0.05). Consumption of meat (as red meat), fats (as animal fat), nuts and refined rice was positively associated with both the presence of NAFLD and its severity (significant fibrosis), whereas, consumption of vegetables (as leafy vegetables), fruits, oily seeds, spices, and dried pulses was negatively associated with NAFLD. Fried and boiled food consumption were positively and negatively associated with NAFLD, respectively. Increased consumption of animal fats was associated with diabetes, hypertension, and cardiovascular outcomes, whereas consumption of wholegrain rice was negatively associated with these health-related outcomes. Conclusions Comprehensive evaluation of food intakes using validated FFQ and tree-based approach in a large, well-characterised population-based cases and controls has enabled the identification of specific dietary indicators associated with NAFLD, its severity and the co-morbidity cluster. These findings provide a basis for culturally sensitive advice to prevent the development of NAFLD as well as the design of individualised intervention in those with NAFLD.
