ScanDish
RU Open the bot

Home · Research

How not to read in extra results

The reading rule here is simple: if a document did not name a number, the article does not add it. That is how percents and hazard ratios stay out where a source itself asks that they not be given, and how new ones are not written in.

Only what is written · 9 min · a calm pattern

How not to read in extra results

Where the source itself closes a number

PREDIMED can be retold like this: adults who already had high cardiovascular risk, a Mediterranean pattern with extra-virgin olive oil or with nuts against advice to eat less fat, and fewer major cardiovascular events in the Mediterranean groups. The 2013 paper was retracted and republished in 2018 because of randomization problems, and the republication kept the finding of fewer events. It was not a weight-loss trial. A percent and a hazard ratio are not given, even if a reader remembers them from someone else’s retelling.

DASH is retold like this: a pattern higher in vegetables, fruit, and low-fat dairy and lower in saturated fat lowered blood pressure compared with a typical American control diet. Millimeters of mercury are not given. The word “lowered” already belongs to the source. A number on the cuff would be an addition; The reading method is to stop at the word that is allowed.

A number that can be repeated because it is written

The CDC page has numbers, and they can be repeated because they are on the page: about 1 to 2 pounds a week, a loss of 20 pounds in 2 weeks as an unrealistic goal, 5% of 200 pounds is 10 pounds. The caveat that this is not a personal guarantee is written too, and it is not cut off. The method does not forbid numbers. It forbids numbers that are not in the document, and it forbids moving a written number into a different document.

The same habit shows with salt and fiber. Salt under 5 g a day and fruit and vegetables from 400 g stand on the WHO fact sheet, and fiber of at least 25 g a day for adults stands in the 2023 guideline. Those numbers are cited with the name of their document. A risk percent from 25 g is not cited, because it is not in the guideline, and millimeters from 5 g of salt are not cited, because the fact sheet does not give them; A written limit and an invented outcome are different things.

How an addition usually disguises itself

An addition often looks like a translation: grams of salt into millimeters of pressure, PREDIMED events into percents of weight, a Mifflin resting estimate into a prescription of what was eaten. The 1990 equation estimates rest. For men it is (10 × kg) + (6.25 × cm) − (5 × age) + 5, and for women the same with −161 instead of +5. It does not measure what was eaten; Calling it the result of how much must be eaten in order to lose weight writes in a prescription the retelling does not have.

Another mask is gluing two real numbers into a third. Five percent in the CDC example and ten percent of sugar at WHO both exist and both concern different things: one is an example about weight, the other is the share of free sugars in energy. Their product or their joint effect is not published in the sources. The method says to leave them in different sentences; That way the article does not acquire a false precision.

Memory and someone else’s chart

Memory of a paper once read does not replace a line of the source. If the allowed retelling has no duration, no sample size, and no risk in percent, the encyclopedia does without them. The retraction and republication of PREDIMED teach exactly this: do not cite the convenient version of the paper. The living finding that can be repeated is fewer events, with no percent; A convenient figure from a withdrawn retelling does not come back here.

A bot follows the same rule. It can repeat a number that is in a document, for example 25 g of fiber or the salt limit, and it keeps the document’s name with it. It does not build a personal risk, personal millimeters, or a personal percent of weight. A food log shows what was eaten, not a missing cell in a results table; Reading stays reading, rather than becoming a new study.

What this text draws on

  • PREDIMED, Estruch et al., NEJM: adults who already had high cardiovascular risk. A Mediterranean pattern with extra-virgin olive oil or with nuts was compared with advice to eat less fat. The Mediterranean groups had fewer major cardiovascular events. The 2013 paper was retracted and republished in 2018 because of randomization problems; the republication kept the finding of fewer events. Not a weight-loss trial. Do not give a percent or a hazard ratio.
  • DASH, Appel et al., NEJM 1997: a pattern higher in vegetables, fruit, and low-fat dairy and lower in saturated fat lowered blood pressure compared with a typical American control diet. Do not give millimeters of mercury.
  • CDC, Steps for Losing Weight (January 17, 2025): a gradual pace of about 1 to 2 pounds a week is more likely to be kept than quicker loss; losing 20 pounds in 2 weeks is the example of an unrealistic goal. Modest loss, CDC’s example: 5% of 200 pounds is 10 pounds, can improve blood pressure, cholesterol, and blood sugar and could lower risk of heart disease, prediabetes, and type 2 diabetes — not a personal guarantee. Log food and drinks, activity, sleep, and stress. Goals are specific, two or three at a time; examples include water instead of a sugary drink, a 15-minute walk, a vegetable with supper. Setbacks are expected. Rewards are not food. Medicines, illness, stress, genes, hormones, environment, and age affect weight. Weight and medicine questions go to a clinician. Medicines and bariatric surgery are clinical options, not DIY.
  • WHO Healthy diet fact sheet: energy intake in balance with expenditure; fruit and vegetables ≥400 g/day; free sugars <10% of energy (below 5% for further benefit); fat <30%; saturated fat <10%; trans fat <1%; salt <5 g/day; unsaturated fats (fish, avocado, nuts, vegetable oils) preferred to saturated and industrial trans fats.
  • WHO carbohydrate guideline, 2023: at least 25 g of fiber a day for adults, from food.
  • Mifflin M. D. et al., Am J Clin Nutr, 1990; 51(2):241-247: a resting estimate. Men (10 × kg) + (6.25 × cm) − (5 × age) + 5. Women the same with −161 instead of +5. Not a measurement of what was eaten and not a prescription.

A short paraphrase of open guides and reviews (WHO, national recommendations, peer‑reviewed summaries). Not a quotation and not a treatment protocol.

Short answers

Why is there no percent in the PREDIMED retelling?

The source does not hand a percent or a hazard ratio to this retelling. It can be said that major cardiovascular events were fewer in the Mediterranean groups. It was not a weight-loss trial. The 2013 paper was retracted and republished in 2018 with the same finding about events.

Is every number in an article forbidden?

No. What is forbidden is an addition. Numbers the document itself wrote, such as 5% of 200 pounds at CDC or salt under 5 g at WHO, can be repeated, together with their caveats and the document’s name.