Nutrition: the evidence

NU-P-01Spreading protein across the day

PrescriptionProtein is distributed across the day's meals rather than concentrated, and a meal-prep container is portioned to carry its share. The per-meal dose is specified at BB-N-02 and PL-N-02 and is not restated here. Band: min "protein somewhere in the day", standard "spread across the day's meals", max "an even dose at every meal".

Evidence gradeB

EffectThe widely repeated ceiling of 20 to 25 g of protein per meal is specific to fast-digesting protein given without other macronutrients. Slower proteins, and protein consumed alongside other macronutrients, delay absorption and plausibly extend utilization. While doses above roughly 20 g do raise amino acid oxidation, that is not the fate of all of it, and the review's conclusion is that per-meal intakes above the commonly cited cap are usable [7].

PopulationReview of acute and longer-term studies in resistance-trained and untrained adults.

Sources[7]

CaveatsA narrative review rather than a meta-analysis, by authors who are prolific in this area and whose position it argues for. What it establishes is that the 20 to 25 g cap is not a hard ceiling, which is a smaller claim than "distribution does not matter". Distribution's effect on the total is what BB-N-02b covers, and it is graded B there too.

NU-P-02How many meals a day

PrescriptionMVIII lets the member choose 1 to 5 meals and does not recommend a number. That absence is the prescription. Band: min "no meal structure at all", standard "offer one to five, recommend none", max "a prescribed number of meals".

Evidence gradeC

EffectA meta-analysis with regression across 15 studies initially found feeding frequency positively associated with reductions in fat mass and body fat percentage, and with an increase in fat-free mass. Sensitivity analysis showed the positive findings were the product of a single study, and the authors conclude the results need interpreting with circumspection [6].

Population15 experimental studies.

Sources[6]

CaveatsThis is a deflator and should be read as one. It does not show that meal frequency is irrelevant; it shows that a meta-analysis pointing that way collapsed to one study when it was leaned on. BB-N-02b prescribes 3 to 6 meals for hypertrophy on separate grounds, which is a protein-distribution argument rather than a frequency-and-fat-loss argument. The two are not in conflict and are frequently confused.

NU-P-03Dietary fiber

PrescriptionNone, and that is the finding. MVIII does not track fiber, does not target it, and does not report it, while carrying strong evidence that it matters. Band: min "no mention of fiber", standard "not tracked", max "a daily target reported next to the macros".

Evidence gradeA

EffectA series of systematic reviews and meta-analyses drawing on just under 135 million person-years from 185 prospective studies and 58 clinical trials, assessed with GRADE, quantifying indicators of carbohydrate quality against non-communicable disease incidence, mortality and risk factors, and establishing an evidence base for quantitative recommendations on dietary fiber intake [8].

Population185 prospective studies and 58 randomized trials.

Sources[8]

CaveatsThe evidence is about long-run disease risk, not about body composition or training, which is why it has never bumped into anything MVIII computes. That is an explanation and not a defense. This is the largest and best-graded body of evidence in this document and the app is silent on it. Fiber is a field on almost every food database row the app already reads.

NU-P-04Daily water

PrescriptionMVIII computes a water goal of 35 mL per kilogram of bodyweight per day. Nothing establishes that number, or any number.

Evidence gradeC

EffectAlthough research teams and professional organizations have described total water intakes and Adequate Intakes for children, women and men, there is no widespread consensus regarding human water requirements, and they remain undefined, because of the complexity of the human water regulatory network and large inter-individual differences [11].

PopulationReview across sex and life-stage groups.

Sources[11]

CaveatsThe National Academies Adequate Intake values the app's sources sheet already cites are population survey figures, not requirements, and this review says so explicitly. A per-kilogram rule is a simplification of a figure that was never a requirement in the first place. It is not harmful and it is not evidenced, and both halves of that belong on screen.

NU-P-05Refrigerated storage of cooked food

PrescriptionA cooked meal-prep dish is stated to keep 3 to 4 days refrigerated, and never longer. The refrigerator is assumed at 40 °F or below, food is refrigerated within 2 hours of cooking, and reheating is to 165 °F.

Evidence gradeB

EffectUSDA Food Safety and Inspection Service guidance: use most cooked leftovers within 3 to 4 days; keep the refrigerator at 40 °F or below; refrigerate within 2 hours of cooking, or within 1 hour above 90 °F; reheat leftovers to 165 °F measured with a thermometer, and bring sauces, soups and gravies to a rolling boil. Bacteria grow most rapidly between 40 °F and 140 °F, doubling in as little as 20 minutes [12]. An FAO/WHO expert meeting reviewing Listeria monocytogenes in ready-to-eat foods reports that outbreaks and resultant deaths continue to occur globally and that new quantitative contamination data exists for nearly every factor in the earlier risk assessment [13].

PopulationNational guidance for domestic food handling [12]; international risk assessment [13].

Sources[12][13]

CaveatsThis is the one entry in the whole MVIII evidence system where being wrong makes somebody ill rather than merely mistaken, and it is the reason it exists. Guidance rather than a trial, which is the correct source type: no randomized trial of leftover storage times will ever be run, and the agency that sets the rule is the right authority for it. Three to four days is a floor across food types, not a per-dish figure. Some dishes keep less. The honest implementation states 3 to 4 days and does not let anything, model or catalog, claim longer without a specific reason. Freezing is a separate question this entry does not cover. # PART B — Findings

NU-D-01Energy density decides the calories, at a constant weight of food

FindingEighteen normal-weight women were given meals across three two-day sessions with free access to a main entree varying in energy density, matched for palatability. They ate a similar weight of food in all three conditions, taking in 1,376, 1,519 and 1,800 kcal respectively (P < 0.0001). There were no differences in hunger or fullness before meals, after meals, or across the two days. The authors conclude energy density affects intake independent of macronutrient content or palatability, and that overconsumption of high-fat foods may be due to their energy density rather than their fat [1].

Evidence gradeB

Population18 normal-weight women, within-subjects.

Sources[1]

CaveatsOne study, 18 participants, one sex, two-day sessions. It is graded B rather than C because the design isolates the variable unusually cleanly and the result is large. The finding is about a mechanism, not a prescription: nothing here says a member should eat to a density target.

NU-D-02Volume matters independently of density, and can be made of air

FindingTwenty-eight lean men were given isoenergetic yogurt-based milkshakes of identical ingredients and identical weight, varying only in volume (300, 450, 600 mL) through incorporated air. Energy intake at lunch 30 minutes later was 12% lower after the 600 mL preload than after the 300 mL one, with greater reductions in hunger and greater increases in fullness [2].

Evidence gradeC

Population28 lean men, within-subjects.

Sources[2]

CaveatsA single study in one sex on a laboratory preload, and the effect is on the next meal rather than on a day or a week. It is included because it separates volume from density, which nothing else does, and because the implication is unusual: the signal deciding fullness can be moved by something with no mass and no calories.

NU-D-03Fat and carbohydrate together, and the judgment problem

FindingIn an auction task during fMRI, participants were willing to pay more for foods combining fat and carbohydrate than for equally familiar, equally liked and equally caloric foods containing only fat or only carbohydrate, with the potentiated reward associated with response in the dorsal striatum and mediodorsal thalamus. The same study found participants were better able to estimate the energy density of fat-only foods than of fat-plus-carbohydrate foods [3].

Evidence gradeC

PopulationAdult participants in a neuroimaging auction paradigm.

Sources[3]

CaveatsA laboratory valuation task, not eating behavior, and reward response is not intake. The second finding is the one that matters operationally and is the one almost never quoted: the combination people value most is also the one their estimation is worst at. That is the overlap where "I hardly ate anything" is honestly believed and factually wrong.

NU-D-04Density is a large part of it and not all of it

FindingTwenty weight-stable adults were admitted as inpatients and randomized to ultra-processed or unprocessed diets for two weeks each, eating ad libitum. Meals were matched for presented calories, energy density, macronutrients, sugar, sodium and fiber. Energy intake was 508 ± 106 kcal/day greater on the ultra-processed diet (p = 0.0001), from carbohydrate (+280 kcal/d) and fat (+230 kcal/d) but not protein. Participants gained 0.9 ± 0.3 kg on the ultra-processed diet and lost 0.9 ± 0.3 kg on the unprocessed one, with weight change correlated to intake at r = 0.8 [4].

Evidence gradeB

Population20 adults, inpatient, four weeks total.

Sources[4]

CaveatsEnergy density was matched and the overeating happened anyway, which is why this entry sits next to NU-D-01 rather than under it. Density is a real and large lever and it is not the whole mechanism; food form, texture and eating rate remain candidates and this trial cannot separate them. Twenty participants at one site.

NU-D-05Hyper-palatable foods have a definition now

FindingResearch on hyper-palatable foods has used descriptive terms such as "fast foods" and "sweets" which are not standardized and lack specificity. A quantitative definition was developed and applied to the Food and Nutrient Database for Dietary Studies to determine the prevalence of such foods in the US food system [5].

Evidence gradeC

PopulationUS food system database.

Sources[5]

CaveatsA definitional and prevalence paper, carrying no outcome. It is cited so that "hyper-palatable" can be used in MVIII's copy as something countable rather than as an adjective.

NU-F-01The named diet is not what decides the outcome

FindingA Bayesian network meta-analysis of 48 randomized trials including 7,286 people, comparing popular named diets by macronutrient class, adjusted for behavioral support and exercise. Weight loss against no diet at six months was largest for low-carbohydrate diets at 8.73 kg (95% CrI 7.27 to 10.20) and remained substantial at twelve months, with low-fat diets close behind and differences between named programs small [9].

Evidence gradeA

Population7,286 overweight or obese adults across 48 randomized trials.

Sources[9]

CaveatsDifferences between classes were small relative to the difference between dieting and not dieting, which is the finding, and it is routinely quoted as whichever diet came out fractionally ahead. Weight loss is the outcome, not body composition and not adherence beyond twelve months.

NU-F-02And neither genotype nor insulin secretion predicts who should pick which

FindingDIETFITS randomized 609 adults without diabetes to a healthy low-fat or a healthy low-carbohydrate diet for 12 months, delivered through 22 diet-specific group sessions, and additionally tested whether three single-nucleotide-polymorphism genotype patterns or insulin secretion (INS-30) predicted which diet somebody would respond to [10].

Evidence gradeA

Population609 adults aged 18 to 50, BMI 28 to 40, twelve months.

Sources[10]

CaveatsBoth arms emphasized diet quality rather than a macronutrient number, so this compares two well-constructed diets rather than a good diet against a bad one. Read with NU-F-01: the pattern somebody picks matters far less than the fact that they picked one and stayed on it. An app offering six macro splits and nine food patterns is offering preference, and should describe them that way rather than as competing strategies. # Where the app and this document meet An earlier draft of this section claimed the app contradicts three of its own published bands. Checked against the entries, two of the three were wrong, and both were wrong in the specific way the entries themselves warn about. What is left is smaller, truer, and more useful.