The substitution of minimally processed food and culinary home preparations for ready-to-eat products is increasing worldwide, which is overlooked as a cause of concern. The technological developments and the rise in highly processed food availability have introduced the concept of ultra-processed food (UPF). Food classification systems based on processing are now a new basis for epidemiological research. Different results from these classifications might influence conclusions on the population's consumption of UPF or its association with health outcomes. The aim of this study was to compare classification systems and to find out if their results are comparable when evaluating the extent of high/UPF on the overall diet. Portuguese data from the year 2000 was extracted from the DAFNE-AnemosSoft, and 556 food/beverages items were classified according to five systems. The contribution of UPF was calculated as a percentage of total available amount and discrepancy ranges used for comparisons. Results of UPF availability contributions were: NOVA 10.2%; UNC 15.2%; IFPRI 16.7%; IFIC 17.7%; IARC 47.4%. The highest discrepancy ranges were from alcoholic beverages (97.4%), milk/milk products (94.2%), sugar/sugar products (90.1%), added lipids (74.9%), and cereals/cereal products (71.3%). Inconsistencies among classifications were huge and the contribution from highly/UPF presented high discrepancies. Caution must be taken when comparing and interpreting such data.
Keywords: food classification systems; food processing; household surveys; ultra-processed food.