Nutrition & Dietary Patterns: What the Primary Evidence Supports

The unit of analysis is the pattern, not the nutrient. The strongest nutrition evidence now rests on whole dietary patterns and food groups rather than isolated micronutrients, and on this ground several signals are settled.

The Mediterranean pattern is the one dietary approach with hard-outcome RCT backing. In PREDIMED, the largest dietary RCT for cardiovascular endpoints, a Mediterranean diet supplemented with extra-virgin olive oil or nuts cut major cardiovascular events (MI, stroke, CV death) by roughly 30% versus a low-fat control in high-risk adults [1]. The original 2013 report was retracted and republished in 2018 after a randomization error; the corrected analysis upheld the conclusion, and that history should be disclosed. In secondary prevention, the CORDIOPREV trial found a Mediterranean diet slowed carotid atherosclerosis progression over seven years versus a low-fat diet [2], though this endpoint is carotid intima-media thickness, a surrogate, not a hard event count. Observationally, each 2-point increase in Mediterranean adherence tracks with ~8% lower all-cause mortality [3], a cohort estimate that complements but cannot replace the RCT.

Fiber and whole grains are the most robustly supported "eat more of this" message in the literature. Higher dietary fiber intake is associated dose-dependently with lower all-cause and cardiovascular mortality, coronary disease, type 2 diabetes, and colorectal cancer, roughly 15-30% reductions at the highest versus lowest intake, with the authors judging the relationship possibly causal [4]. Independently, each 90 g/day of whole grains tracks with ~17% lower all-cause mortality, ~22% lower CVD, and ~15% lower cancer mortality, with a clear dose-response [5]. These two converge through overlapping mechanisms while capturing distinct effects, isolated fiber versus the whole-grain food matrix.

Ultra-processed food proportion is independently associated with worse outcomes. In the NutriNet-Santé cohort, each 10% increment in the ultra-processed share of the diet tracked with a 12% higher risk of cardiovascular disease [6] and roughly 14% higher all-cause mortality [7]. Both associations persisted after adjustment for nutrient composition and dietary-quality scores, pointing to processing-specific factors beyond simple nutrients. Both studies draw on the same volunteer cohort, so this is association, not proof of causation.

For dietary fat, the direction is settled and the replacement nutrient is decisive. Reducing saturated fat lowers composite cardiovascular events by ~17% (RR 0.83), an effect driven by trials that replaced saturated fat with polyunsaturated fat, with no significant reduction in all-cause mortality and benefit attenuated when the replacement is refined carbohydrate [8]. On the companion omega-6 question, higher circulating linoleic acid, a biomarker, not a food-frequency estimate, is associated with lower cardiovascular events and CV mortality across 30 cohorts [9], which points the opposite way from the "seed oils are uniquely toxic" narrative. The magnitude of the saturated-fat effect, and whether seed oils are harmful, both remain contested (see Contested).

Protein source differentiates mortality more cleanly than protein quantity. Substituting 3% of energy from plant protein for animal protein is associated with lower all-cause mortality, with a steep source gradient: replacing processed red meat yields the largest reduction (HR 0.66, ~34% lower), unprocessed red meat HR 0.88, egg HR 0.81, while fish and poultry show no significant mortality association [10]. In mid-life (ages 50, 65), high total protein intake (≥20% of calories) is associated with improved cancer and all-cause mortality, an effect that is absent when the protein is plant-sourced and that reverses after age 65, where low protein intake becomes harmful [11]. Both are observational, with health-conscious cohorts and self-reported diet; the muscle-performance camp contests the implications (see Blockers).

A weight-loss drug became a cardiovascular drug, and it is a diet-adjacent lever worth noting. In SELECT, semaglutide cut major cardiovascular events by ~20% (HR 0.80) in overweight or obese adults with established CVD and no diabetes [12], an outcome, not a surrogate.

In the gut, the actionable levers are well-supported; the supplement layer is not. Fecal microbiota transplant resolves recurrent Clostridioides difficile infection, ~81% after a first infusion and ~94% cumulatively, versus ~23-31% for vancomycin, in a trial stopped early for overwhelming superiority [13]. That is the one microbiome intervention with strong primary outcome support; FMT for metabolic or other conditions, personalized-testing claims, and probiotic-supplement benefits remain weaker and are not published here as fact.

Where the Evidence Is Contested

The direction of the relationship between saturated fat and heart disease is broadly agreed: saturated fat raises apoB and LDL, and replacing it with polyunsaturated fat cuts cardiovascular events, with Cochrane estimating roughly 17% [8] and the AHA roughly 30% [14]. What is disputed is the magnitude, and whether any of this touches mortality. Cochrane finds no mortality signal at all. The Minnesota RCT lowered cholesterol but did not lower mortality [15], and the PURE observational data linked saturated fat to lower mortality, though that finding is confounded [16]. A related argument concerns seed oils and omega-6 fats. Ramsden advances the view that they are uniquely harmful through oxidized linoleic acid [15]. Against this, higher circulating linoleic acid tracks with lower cardiovascular disease across 30 cohorts [9], and critics note that the trials showing harm are confounded by trans-fat margarine and calorie mismatch [17]. The quality of the opposing evidence keeps both questions open.

Whether the timing of eating matters beyond the calories themselves is contested. One camp argues from mechanism, holding that meal timing flips autophagy and circadian switches; this work is mouse-weighted [18] and associated with Panda, Mattson, and Longo. The other camp points to isocaloric human trials: alternate-day fasting performs no better than daily calorie restriction [19], adding an 8-hour eating window shows no benefit [20], and 16:8 was null with roughly 65% of the weight lost coming from lean mass [21]. Calorie restriction itself benefits cardiometabolic risk and the aging clock [22]. A parallel dispute surrounds the MIND diet. Observational work reports a benefit of about a 7.5-year cognitive-age advantage, with a dose-response pattern [23]. The randomized trial was null, finding no between-group difference in cognition at 3 years, with a difference of 0.035 and a p-value of 0.23 [24]. Healthy-user confounding is the expected explanation for the gap.

Protein sits at the center of two competing priorities. The muscle camp holds that at least 1.6 g/kg builds muscle and that strength predicts survival [25, 26]. The longevity camp counters that high animal protein in mid-life raises IGF-1 and mortality [11], though this is confounded and reverses after 65; exercise dissolves most of the conflict. The blue zones are similarly split. Newman's record-keeping and fraud analyses, in a preprint with an active rebuttal, undermine the demography of the longevity hotspots. At the same time, the plant-forward lifestyle those zones prescribe is independently well-supported, as in PREDIMED [1] and the Adventists [27].

The gut microbiome field divides what is real from what is hype. The real side includes fecal transplant for C. diff [13] and associations involving fiber, fermented foods, and microbial richness [28, 29, 4]. The hype side covers probiotic supplements, personalized tests, and the promise to fix your gut to cure almost anything, all resting on weak, observational, or expert-opinion evidence. Finally, the GLP-1 drugs are contested less on data than on framing. The outcome case is strong: roughly 15 to 21% weight loss [30, 31] and a 20% reduction in cardiovascular events [12]. The cost case is also real: about 40% of the loss is lean mass [32], and about two-thirds of the weight returns after stopping [33].

References

1. Estruch et al. 2018 (PREDIMED), NEJM — Mediterranean diet & CVD · PubMed 29897866
2. CORDIOPREV — Mediterranean diet 7-yr RCT · PubMed 34372670
3. Sofi F, Abbate R, Gensini GF, Casini A. 2014 (Am J Clin Nutr) — Updated meta-analysis Mediterranean diet adherence; 16 prospective studies;… · PubMed 24476641
4. Reynolds et al. 2019, Lancet — dietary fibre & health outcomes · PubMed 30638909
5. Aune D, Keum N, Giovannucci E et al. 2016 (BMJ) — Dose-response meta-analysis of whole grain intake; 45 prospective cohorts; n=778,076; 90g/day… · PubMed 27301975
6. Srour B, Fezeu LK, Kesse-Guyot E et al. 2019 (BMJ) — NutriNet-Santé; n=105,159; 5yr follow-up; UPF as % of dietary weight; HR for CVD events 1.12… · PubMed 31142457
7. Schnabel L, Kesse-Guyot E, Allès B et al. 2019 (JAMA Intern Med) — NutriNet-Santé; n=44,551; UPF % → ACM ~14% higher per 10% increment (HR ~1.14);… · PubMed 30742202
8. Hooper L, Martin N, Abdelhamid A, Davey Smith G. 2020 (Cochrane Database Syst Rev) — 15 RCTs, n=56,675; reduced/modified dietary fat; saturated fat… · PubMed 32827219
9. Marklund et al. 2019, Circulation — linoleic acid biomarker & CVD · PubMed 30971107
10. Song M, Fung TT, Hu FB et al. 2016 (JAMA Intern Med) — NHS + HPFS; n=131,342; animal protein 10% kcal → plant protein: HR 0.66 ACM (0.59–0.75) in… · PubMed 27479196
11. Levine ME, Suarez JA, Brandhorst S et al. 2014 (Cell Metab) — NHANES I; n=6,381, 18-yr follow-up; moderate/high protein (≥20% kcal) in ages 50–65:… · PubMed 24606898
12. Lincoff et al. 2023 (SELECT), NEJM — semaglutide & CV events · PubMed 37952131
13. van Nood et al. 2013 NEJM — FMT vs vancomycin for recurrent C. difficile: 81%/94% vs ~23–31% · PubMed 23323867
14. Sacks et al. 2017, Circulation — AHA dietary fats advisory · PubMed 28620111
15. Ramsden et al. 2016, BMJ — Minnesota Coronary Experiment · PubMed 27071971
16. Dehghan et al. 2017, Lancet (PURE) — fats, carbs & mortality · PubMed 28864332
17. Physionic (Verhoeven) — seed oils: the raging health debate
18. de Cabo & Mattson 2019, NEJM — IF review · PubMed 31881139
19. Templeman et al. 2021, Sci Transl Med — isocaloric ADF vs CR · PubMed 34135111
20. Liu et al. 2022, NEJM — TRE + CR vs CR alone · PubMed 35443107
21. Lowe et al. 2020, JAMA Intern Med (TREAT) — 16:8 TRE · PubMed 32986097
22. Kraus et al. 2019, Lancet D&E (CALERIE) — 2yr caloric restriction · PubMed 31303390
23. Morris MC, Tangney CC, Wang Y et al. 2015 (Alzheimer's & Dementia) — MIND diet score (0–15); n=923, age 58–98y, mean 4.7yr follow-up; top tertile… · PubMed 26086182
24. Barnes LL, Dhana K, Liu X et al. 2023 (NEJM) — MIND diet RCT; n=604 (301 MIND, 303 healthy control); adults ≥65, BMI ≥25, family history of AD,… · PubMed 37466280
25. Morton et al. 2018, BJSM — protein + RT meta (1.6 g/kg) · PubMed 28698222
26. Leong et al. 2015, Lancet (PURE) — grip strength & mortality · PubMed 25982160
27. Orlich et al. 2013, JAMA IM — Adventist vegetarian mortality · PubMed 23836264
28. Wastyk et al. 2021, Cell — fermented foods vs fiber (diversity) · PubMed 34256014
29. Le Chatelier et al. 2013, Nature — microbiome richness & metabolism · PubMed 23985870
30. Wilding et al. 2021 (STEP 1), NEJM — semaglutide weight loss · PubMed 33567185
31. Jastreboff et al. 2022 (SURMOUNT-1), NEJM — tirzepatide · PubMed 35658024
32. GLP-1 body composition — ~40% lean-mass loss (STEP/SURMOUNT DXA) · PubMed 39996356
33. Wilding et al. 2022 (STEP 1 extension), DOM — weight regain · PubMed 35441470