UK nutrient rules flag about half the ultraprocessed foods Eatwell-based profiles call unhealthy

Research published in BMJ Medicine compared how UK food-policy instruments classify processed and ultraprocessed products already eaten by the public. Using National Diet and Nutrition Survey Year 12 (2019-20) food codes, 262 processed and 1,636 ultraprocessed items, investigators tested the Food Standards Agency nutrient profiling model and voluntary front-of-pack traffic-light rules against three new profiles built from Eatwell Guide energy and salt ceilings. The design is a classification comparison, not a trial of health outcomes. Current tools marked only about half of these products unhealthy; guidance-based profiles marked more than 90 percent.

The FSA model, which defines products high in fat, salt or sugar for advertising rules, classified nearly 58 percent of ultraprocessed items and 50 percent of processed items as unhealthy. Traffic-light criteria classified just over 48 percent and 46 percent. NPM1, flagging excess saturated fat, free sugar or sodium, marked 91.5 percent of ultraprocessed items unhealthy. Adding total fat raised that to almost 95 percent; adding energy density raised it to almost 98 percent. Guidance ceilings used were 35, 10 and 5 percent of energy from total fat, saturated fat and free sugars, plus 6 grams of salt daily.

The gap is partly by design. The 2004-05 FSA-Ofcom model was built to restrict HFSS advertising, not to copy Eatwell ceilings. Traffic lights are voluntary and nutrient-specific; collapsing them into one unhealthy label is an analytical choice shops do not make. News shorthand of nearly twice overstates the jump: 58 percent to 91.5 percent is about 1.6-fold. A profile that fails nearly all ultraprocessed items also loses ranking power for taxes and ad bans. Minimally processed foods were omitted, so oils, cheese and spreads were never tested.

In a linked BMJ Medicine editorial, Franco Sassi and Shashika Rathnayaka note that product scores are not diet scores. A larger share of ultraprocessed food is associated with poorer nutrient profiles and lower intake of fruit, vegetables and legumes, but rules can shift purchases toward other scored products without improving the whole diet. Funding was not stated in the news report. This design can expose instrument discordance. It cannot show that a unified profile would reduce obesity or cardiovascular events. What is needed next is modeled substitution in NDNS under alternative thresholds, including minimally processed foods.