You order the grilled salmon with greens and skipped the fries. At home there's a "high protein" ready meal in the fridge for Tuesday, and the bread in the cupboard says wholegrain on the front.
None of it feels like junk food, the problem is most of it counts as ultra-processed anyway. The salmon came out of a kitchen that cooks the way a factory does.
Ultra-processed has become the most argued-over phrase in nutrition. The World Health Organization put out a draft model for defining it last week, and more than 100 scientists wrote to the White House budget office in October asking it to release a stalled US definition. While that gets argued over, here's what the numbers say.
An average restaurant main carries a day and a half's worth of salt and about 1,500 calories, and the people who eat most of their meals this way die earlier.
The ready meal in the fridge is the same food in a tray, sometimes worse.
This week is about where the salt, sugar and oil you don't see come from.
Today: what ultra-processed means, what the research found, and why a restaurant meal belongs in the same bucket as a frozen one.
Wednesday: not all of it is equal.
Friday: the swaps, shelf by shelf, on one page.
The like-for-like swaps for the packaged foods most of us buy every week are already written up in the Whole Food Swaps guide, if you'd rather not wait for Friday.
The Evidence
The research is clear on three of these and still arguing about the fourth.
What counts. The term comes from a Brazilian research group whose definition, laid out in The Lancet last year, is about how a food is made rather than how many calories it has. Ultra-processed means an industrial formulation built from extracted ingredients, with additives to make it taste, look and last the way the maker wants.
The working test is the label: 5 or more ingredients and at least 1 you couldn't buy at a supermarket (protein isolate, modified starch, emulsifier, "natural flavour"). Most readers here cook the majority of the time from scratch and already know the obvious offenders. These are the ones that surprise people:
Most packaged bread, breakfast cereal and granola, including the wholegrain ones.
Flavoured yogurt, plant milks, protein bars and shakes.
Ready meals, meal-kit sauces, stock cubes, salad dressings, most things labelled "light".
A large share of what comes out of a restaurant kitchen, which buys the same pre-made sauces, bases and dressings you'd find in the supermarket, in bigger tubs.
The restaurant meal counts, whatever the menu says. When researchers analysed full-service chain restaurant meals in the US, the average adult order, a main, a side and half a starter, came to:
About 1,500 calories, before a drink. With a drink and half a dessert, 2,020.
3,512mg of sodium, which is a day and a half's worth of salt in one sitting. 98% of American men and 80% of women are already over the daily limit before a meal like this, and because women's blood pressure climbs more steeply than men's from the 30s onward, the salt matters more the older you get.
28g of saturated fat and 11g of fiber, so most of a day's saturated fat and under half a day's fiber.
A 2022 check of 40,902 menu items from large US and UK chains found 96% over the recommended level for at least 1 of salt, fat, saturated fat or sugar, and that includes the salads.
Eating this way often shows up in the death statistics. In a 17-year follow-up of 35,084 Americans, people eating 2 or more meals a day prepared away from home had a 49% higher risk of dying than people eating out less than once a week. That lines up with the wider ultra-processed picture, from a BMJ review that pooled 45 analyses covering 9.9 million people:
The heaviest ultra-processed eaters had a 21% higher risk of dying over the studies' follow-up than the lightest. If 100 of every 1,000 people in the light group died, about 121 did in the heavy group.
50% higher risk of dying from heart disease, and 12% higher risk of type 2 diabetes for every extra 10% of the diet that was ultra-processed.
The heart-death and diabetes links were rated convincing, the strongest grade the review gives. These are observational studies, so people who eat this way differ in other ways too. The restaurant numbers above, though, were measured in a lab, and they don't depend on anyone's memory.
The processing itself seems to do something, and this is the part still being argued. In October, Virginia Tech researchers published a small study in Nature Metabolism: 32 adults ate 2 meals, 3 days apart, matched almost exactly for calories, carbs, fat, protein, fiber and salt. One was ultra-processed, one wasn't. After the ultra-processed meal, insulin went higher and the brain's reward areas responded differently to pictures of the food. It's 32 people and 2 meals, so don't build anything on it yet. What it points at is that the salt, sugar and oil are only part of the story.
Back to the grilled fish. The fish itself is fine, and the numbers above come from what it sits in. Restaurant kitchens run on salt, sugar and fat because that's what makes food taste better than yours, and because the sauce, the dressing, the stock and the marinade usually arrive in a tub rather than being made that morning. The oil is bulk refined seed oil, kept hot for the whole service and used in amounts you'd never pour at home, so a sautéed main can carry 2 or 3 tablespoons, about 300 calories, before anything else is counted. The research on seed oils themselves at home quantities is mixed, and we'll come back to it. At restaurant quantities you're simply eating a lot of oil, on everything.
What This Means For You
Here's how to use this before Friday's swaps arrive.
Run the label test on the things you don't usually check. You already read the front of the packet, so turn over the jar of sauce, the stock, the dressing and the snack bar in your bag. 5 or more ingredients plus 1 you couldn't buy means it's ultra-processed. Nothing has to change yet, tonight is only about finding out.
At a restaurant, sauce on the side and keep the dish. Dressing and sauce on the side, grilled over fried, skip the bread, and treat the side of fries as the dessert. Those 4 moves take most of the hidden salt, sugar and oil out of the meal without you ordering the sad option.
Treat ready meals and meal kits as what they are. A ready meal is a restaurant meal in a plastic tray with the same tubs behind it, so the label test applies there too. Meal kits are better only if the sauce comes as ingredients rather than a sachet.
Halve it before it arrives. If the average order is 1,500 calories, asking for a box when you order and putting half in it first is the simplest fix there is. Tomorrow's lunch is sorted and tonight's meal is a normal size.
Wednesday is about which foods inside it actually carry the risk, because the answer changes what you'd swap.
Start with the label test tonight. It takes 30 seconds and most people are surprised by what turns up.
If you want the swaps now, the Whole Food Swaps guide has a like-for-like replacement for each of the common packaged foods, with what to look for on the label.
Which packaged food failed the label test and surprised you? Hit reply and tell us. We read every one.
Wednesday Preview
Wednesday: the ultra-processed foods that look fine in the data, the ones that don't.
Until Wednesday.
Longevity Daily / The Building Decades
P.S. Eating out this week? Forward this to whoever's picking the restaurant. The Whole Food Swaps guide has the packet-by-packet version for home.