Diet

The feed will hand you a new food rule before lunch.

About 20% of that advice is worth a month of practice. The other 80% is a swap you will forget by Friday.

US death data points at a short food list, not at a perfect plate. This page names the list, a calorie finding, and a way to see your own week.

Ten foods, almost half of a certain kind of death

Renata Micha and colleagues estimated how ten dietary factors lined up with US deaths from heart disease, stroke, and type 2 diabetes in 2012. Suboptimal intake was associated with 318,656 of 702,308 cardiometabolic deaths, or 45.4% (Micha et al., JAMA, 2017).

The largest pieces, as a share of all cardiometabolic deaths, were high sodium 9.5%, low nuts and seeds 8.5%, high processed meats 8.2%, low seafood omega-3 fats 7.8%, low vegetables 7.6%, low fruits 7.5%, and high sugar-sweetened drinks 7.4%. Low polyunsaturated fats and high unprocessed red meat sat much lower. These shares overlap, so they are not a stack you add to 80%.

That is the field pattern. A minority of food gaps, not every ingredient you have heard of, carried the estimated deaths. It is a model from surveys and meta-analyses, with uncertainty intervals in the paper. It is not a law that 20% of your meals create 80% of your health.

The extra calories had an address

Kevin Hall and colleagues housed adults and served two menus, matched for the calories offered, sugar, fat, fiber, and other macros as designed. People ate about 500 kilocalories a day more on the ultra-processed menu than on the unprocessed one (Hall et al., Cell Metabolism, 2019). They were not told to overeat. The menu did the work.

So the ignored majority in a kitchen is often the hunt for a rare superfood, while the weekday menu is whatever is fastest and salted. If weight is the reason you are here, the same calorie finding is used on weight loss. The week-by-week version is on eating.

A pattern beat a lecture

In PREDIMED, a Mediterranean pattern with extra-virgin olive oil or nuts was associated with about 30% fewer major cardiovascular events than the control diet (Estruch et al., New England Journal of Medicine, 2018 republication). Olive oil, nuts, vegetables, fish, and less processed meat. Not a 40-item rulebook.

The same diet-quality idea is one of five habits tied to extra years of life in a separate cohort paper. Women at age 50 with all five habits had about 14 extra years versus none. Men had about 12.2 (Li et al., Circulation, 2018). The other four habits are not foods. They are laid out on longevity.

A target you choose, and a week you count

One chosen target, not a study: aim for about 80% of your meals to be built from foods you already trust, and leave about 20% for meals you simply want. If that split makes you swing harder, drop the target. The Hall study is about the menu in front of you, not about earning dessert.

For seven days, write four lines: sugary drinks, late-night eating, restaurant or takeout meals, and the breakfast or lunch you repeat without thinking. At the end, mark which line held the most calories or the most "I did not mean to" moments.

Illustrative: one sample week had three restaurant dinners, and those dinners held most of the calories above the person's usual plan. Your week will not match this row. Count yours. Do not publish their 20% as your 20%.

What the log usually exposes, and what generic diet plans bury, is repetition. The same drink, the same late bowl, the same lunch. That repeat is the concentration. The variety you worry about on Saturday is often the small share.

Guardrails

Heart risk in the population, and what to do with a diagnosis, is a clinical question. The spending side of that system is healthcare. Cooking the repeat meals is cooking.

Misreads

"Everyone eats 20% of their favorite foods 80% of the time."
Many kitchens repeat a short list. Prove it with your seven lines. It is not a census.

"A woman's recipes were 20% of her list and 80% of her calories."
That story had no source. It is gone. Use the illustrative week only as a format.

8020 move: Pick the single log line that dominated the week. Change that line for the next seven days. Leave the rest of the food rules unread.

How to read 45% without making it 80

45.4% means the model tied ten dietary gaps to almost half of US cardiometabolic deaths in 2012. It does not mean half of all deaths, and it does not mean your personal risk is 45%. The uncertainty interval in the paper is wide enough that you should quote 45.4% as an estimate, then look at which factor is largest, not at a blended "food causes everything" line.

The shares also overlap. A person can have high sodium and low vegetables at once. Adding 9.5 and 8.5 and 8.2 until you reach a fake 80 is the mistake the old page made in a different costume. Pick the biggest factor you actually eat. For many logs that is sodium, processed meat, or sugary drinks. Then stop.

Hall's about 500 kilocalories is a daily difference between two menus in a small housed study, not a license to call every packaged food equal. If your default lunch is built from those packages, change the lunch. If your default lunch is already simple and the log says restaurant dinners, change the dinners. The study tells you the menu can move intake. The log tells you which menu is yours.

PREDIMED's about 30% was fewer major cardiovascular events in a trial of a Mediterranean pattern, later republished after a methods correction. It is a reason to keep olive oil, nuts, vegetables, and fish in the 80% target if you use that target. It is not a reason to buy a 12-step Mediterranean plan and ignore the sodium you already measured.

Sources

• Micha et al., JAMA, 2017. 45.4% of US cardiometabolic deaths in 2012. Ranked factor shares as cited.
• Hall et al., Cell Metabolism, 2019. About 500 kilocalories a day more on the ultra-processed menu.
• Estruch et al., NEJM, PREDIMED, republished 2018. About 30% fewer major cardiovascular events.
• Li et al., Circulation, 2018. Diet quality as one of five habits. About 14.0 and 12.2 extra years at age 50.
• The sample week is marked Illustrative. It is not a study.