An evidence-focused briefing, current to 5 August 2026
The newest diet research does not identify a single winning macronutrient split. It makes a more practical argument: dietary quality, degree of processing, sodium exposure, and sustained energy change matter more than whether a plan is called low-carbohydrate, plant-based, Mediterranean, or time-restricted. The strongest new evidence is not evenly distributed: a large randomized salt-substitution study is unusually actionable, whereas several prominent 2025–2026 findings remain observational, short-term, or dependent on adherence.[1–3]
1. Diet pattern still beats nutrient tribalism
The durable signal is a pattern built mainly from vegetables, fruit, pulses, whole grains, nuts, and unsaturated-fat sources, with protein selected to fit the person and context. The World Health Organization’s current framing is deliberately broad: adequacy, balance, moderation, and diversity, with minimally processed foods low in unhealthy fats, free sugars, and sodium as the foundation. It recommends at least 400 g of fruit and vegetables and 25 g of naturally occurring fibre daily for people older than 10 years, while emphasizing that the exact diet must vary with culture, availability, age, and activity.[4]
A 2025 synthesis focused on secondary cardiovascular prevention adds useful restraint. Across 14 randomized trials enrolling 14,065 patients with established cardiovascular disease, dietary interventions as a whole did not significantly reduce all-cause mortality, cardiovascular mortality, recurrent events, or revascularisation. Yet the Mediterranean-diet subgroup did show lower cardiovascular mortality. The 22 prospective cohorts in the review also associated high adherence to a healthy pattern with lower all-cause and cardiovascular mortality; the authors judged much of the evidence at risk of bias or heterogeneous. That is a more defensible conclusion than treating “Mediterranean” as a guarantee: it is a strong template, but the trial base in people who already have cardiovascular disease is thinner and less uniform than popular summaries imply.[5]
A 20-year Greek cohort moved in the same direction for diets that are both plant-forward and environmentally oriented: higher EAT–Lancet and Mediterranean scores were associated with fewer cardiovascular events. Because this was observational, it strengthens plausibility rather than proving that the diet caused the difference.[6]
2. Ultra-processing has become a serious public-health signal—but not a complete measure of food quality
The 2025 Lancet series argues that diets based on ultra-processed foods displace meals centred on whole foods and cooking, worsen nutrient balance, and promote overeating through properties such as energy density, hyper-palatability, soft texture, and altered food matrices. Its authors describe convergent evidence from prospective studies, meta-analyses, interventions, and mechanistic work across multiple chronic-disease outcomes.[7]
The important qualification is that the NOVA processing category cannot replace nutrition assessment. A proof-of-concept menu derived from foods classified as ultra-processed supplied 91% of calories from those foods yet achieved a Healthy Eating Index score of 86/100 and met most macro- and micronutrient targets; sodium, whole grains, vitamin D, vitamin E, and choline remained problems. This does not show that an ultra-processed diet is equivalent to a minimally processed one in long-term trials. It does show why public advice should target the practical cluster—sugary drinks, salty snacks, refined starches, low-fibre packaged foods, and displacement of home-prepared meals—rather than assume that a processing label alone captures all nutritional value.[8]
The new U.S. Dietary Guidelines reflect the first half of that message, recommending whole, nutrient-dense foods and substantial cuts in highly processed foods, refined carbohydrates, and sugar-sweetened beverages. They retain a sodium ceiling of 2,300 mg/day for the general population and advise less alcohol.[9]
3. Replacing some sodium with potassium is one of the clearest scalable interventions
The 2025 prespecified stroke subgroup analysis from the Salt Substitute and Stroke Study deserves attention because it tested a concrete household change at scale. In 15,249 Chinese adults with prior stroke, a salt substitute containing 75% sodium chloride and 25% potassium chloride lowered systolic blood pressure by about 2 mm Hg over a median 61 months. It reduced recurrent stroke by 14% and deaths by 12%; there was no statistically detectable excess of hyperkalaemia.[1]
This is stronger evidence than most dietary headlines: it is a large cluster-randomized trial with hard clinical endpoints, not just a biomarker study. It is not a universal self-treatment recommendation. WHO notes that lower-sodium, potassium-containing salts are most suitable where discretionary salt is an important sodium source and where health systems can identify and manage people at risk of hyperkalaemia, particularly those with kidney disease.[4]
4. “Plant-based” is a direction, not a result
A 2026 meta-analysis restricted to people with overweight or obesity found that plant-based diets produced little or no average difference from omnivorous comparison diets for weight, blood pressure, glucose, insulin, total cholesterol, triglycerides, HDL cholesterol, or body fat. LDL cholesterol fell slightly; possible BMI and HbA1c improvements were supported by very low-certainty evidence, and longer interventions looked more favourable.[2]
A separate 2026 meta-analysis of 17 observational studies, covering 456,783 adults, linked greater plant-based adherence with lower odds of hypertension, but found only borderline associations with type 2 diabetes and dyslipidaemia and no significant association with obesity.[10]
Together, these studies argue against two common overstatements: a plant-based label does not automatically produce weight loss, and plant-forward eating is not nutritionally trivial. The likely advantages depend on what replaces animal foods—fibre-rich pulses, vegetables, nuts, and whole grains are a different intervention from refined starches and highly processed vegan substitutes. Nutrient planning remains material for fully vegan diets, particularly for vitamin B12 and several minerals; the U.S. guidelines explicitly advise monitoring and targeted supplementation where appropriate.[9]
5. For diabetes and weight loss, intensity and maintenance matter more than the label
The best 2025 evidence does not support a simple claim that low-carbohydrate digital programmes outperform ordinary primary care. In a 12-month randomized trial of 115 adults with type 2 diabetes, a 12-week digitally supported low-carbohydrate programme produced modestly greater weight loss at three months, but the difference was gone at 12 months and there was no between-group improvement in HbA1c or cardiovascular risk factors.[3]
At the other extreme, a 2025 preprint reported diabetes remission in 34 of 47 completers after three months of a 600–800 kcal/day formula diet, with roughly 14–15 kg of weight loss. Remission was less frequent in participants with longer-standing diabetes. It is a compelling signal about the metabolic effect of substantial weight loss, but it is small, short, medication-intensive, and not yet peer reviewed; it should not be treated as evidence for unsupervised very-low-calorie dieting.[11]
Time-restricted eating remains promising rather than settled. A 2025 review reports improvements in weight, insulin sensitivity, and glucose metabolism across clinical trials, while also identifying the need for longer-term studies and individualized implementation. The credible inference is modest: timing can be a useful adherence tool for some people, not a demonstrated replacement for dietary quality or sustained energy change.[12]
What should change in practice?
The evidence supports a short hierarchy. First, improve the default food pattern: more minimally processed plant foods, fibre, and unsaturated-fat sources; fewer sugary drinks, refined snacks, and sodium-heavy packaged foods. Second, treat sodium reduction—and, for appropriate patients, potassium-enriched salt—as a major cardiovascular intervention rather than a minor dietary detail. Third, choose a weight-management structure that a person can sustain; lower-carbohydrate, Mediterranean-style, plant-forward, and time-restricted approaches are tools, not identities. Finally, be cautious with alcohol narratives. A 2025 analysis of three long-running U.S. cohorts associated regular light-to-moderate drinking with lower diabetes incidence, but this is observational and cannot establish a preventive benefit; it should not be read as a reason for non-drinkers to begin drinking.[13]
The practical consensus is therefore less glamorous than any diet trend: eat a varied diet centred on intact or minimally processed foods, control sodium, preserve dietary quality during weight loss, and judge a diet by outcomes and adherence rather than its name.[4]
