8.8 million people, 51 cohorts, one direction. And what 'associated' means in numbers you can use.

The largest pooled analysis of ultra-processed food and chronic disease to date was released today in Family Medicine and Community Health: 51 prospective cohorts, 8,819,894 adults, followed for two to thirty-two years across four continents. Comparing the highest intake with the lowest, the hazard of cardiovascular events was 24% higher, of metabolic syndrome or diabetes 24% higher, of obesity 23% higher, of cancer 12% higher, and of death from any cause 18% higher. The dose-response version is the useful one: each additional 100 grams a day — roughly a bag of chips or a bowl of sweetened cereal — was associated with 3% higher mortality and 14% higher cardiovascular risk. The authors' recommendation is replacement, not restriction: swap minimally processed food in, rather than simply eat less.

Turning relative risk into something you can feel. Every number above is relative. An 18% higher hazard of death means 18% more than whatever your baseline is, and baselines differ. Suppose a 60-year-old's chance of dying in the next ten years is about one in ten. An 18% relative increase moves that to a little under twelve in a hundred: two more deaths per hundred people over a decade. That is a real public-health number when it is applied to a population eating half its calories this way, and a modest one for any individual — which is why the authors frame the finding at the population level and why you should not panic over a single bag. The limits are the standard ones, stated by the authors: self-reported diet in every included study, and lifestyle patterns that travel with processed food and are hard to fully adjust for. Publicly funded, no financial disclosures.

Loree’s take

What I take from the dose-response curve is permission to be ordinary. Every 100 grams replaced with real food moves the needle a little, and the needle moves in the same direction in every cohort. Nobody needs to reach zero. They need to swap the breakfast cereal for eggs and oats, the afternoon bar for fruit and nuts, and let the arithmetic run for thirty years. The study says the thirty years are where the 18% lives.

Ben’s take

Eight point eight million people is an awkward number for the packaged-food industry, and the response will be the one it has used for a decade: the category is too broad, the studies are self-reported, processing is not the problem. Some of that is fair and the authors say so. Watch, over the next weeks, who makes the argument and who paid for the making. We have a post coming on exactly that.

Why it matters to a recovering optimizer. This is the nutrition chapter of our book in a single table, and the table says what we said: the specific diet matters far less than whether the food is real. You do not need a macro, a window or a tribe. You need the 100 grams, replaced. Then the next 100.