Ways of eating, compared honestly

"Which diet is best for weight loss?" has an answer, and it is not the one the question expects.

The answer first

When 121 randomised trials of fourteen named diet programmes were compared in a network meta-analysis, most patterns produced modest weight loss and better blood pressure at six months — and at twelve months the differences between them had largely disappeared. The current clinical guidelines say the same in their own language: lower-carbohydrate, lower-fat and mixed patterns can achieve similar weight reduction, and the choice should fit the person.

So the question this site asks of every pattern is not "does it work?" but "what does it ask of you, what does it give back, and what does the evidence actually show — for weight and for everything else?"

How to read a pattern page

Every way-of-eating page has the same structure, so they can be compared:

• What it is, in one line and without a verdict.
• The pattern itself — carbohydrate, fat and protein where the pattern sets them, timing where it does, what it excludes or limits. Every number carries a source; where a pattern sets no number, the page says so instead of inventing one.
• What the evidence shows — an evidence grade from A to D, the weight-loss result at six and twelve months where trials exist, what the pattern is most strongly supported for (often not weight), and, deliberately, what it claims but has not shown.
• Practicality — cost, cooking effort, eating out, social friction and how well it fits the local kitchen. This is editorial judgement, scored from one to three with a stated reason, and it is the part most sites skip.
• Safety — who should talk to a doctor first, and what to monitor.
• In the app — whether the app can verify the pattern as a rule, treats it as soft context, or does not support it. Only the balanced policy has a reviewed, verifiable rule today; the pages say so.

The patterns

Dietary patterns

• Mediterranean pattern — grade B for weight, A for cardiovascular outcomes. The pattern with the strongest evidence for something other than the scale.
• DASH — grade B. Designed for blood pressure; about 1.4 kg more than control diets over 8–24 weeks, through the deficit inside it.
• Nordic pattern — grade C. The regional counterpart of the Mediterranean pattern; recommended for health and the environment, with no weight trial in the registry.
• Volumetrics — grade C. Eating by energy density; the mechanism is well shown in short studies, the long-term weight evidence is thin.
• Flexitarian — grade D. Mostly plants, meat sometimes, no definition — and so nothing to test.

Lower-carbohydrate

• Lower-carbohydrate eating — grade A for "not better than a balanced diet at a year", B for a faster start.
• Ketogenic diet — grade B. Under about 50 g of carbohydrate a day; a fast start, an adaptation week, no advantage at a year.

Lower-fat

• Low-fat eating — grade A. The historical standard; equivalent to low-carbohydrate at twelve months in the head-to-head trials.

Method

• Higher-protein eating — grade A. The one macronutrient lever with consistent evidence — for what the loss is made of, not how much.
• Low glycaemic index — grade B. Sensible swaps; little or no difference for weight in the Cochrane review.
• Portion control — grade B. The same food, less of it; a solid effect on intake, a weight effect inferred rather than measured.

Exclusion

• Vegetarian eating — grade B. About 2 kg more than comparison diets over four months, through energy density and fibre.
• Vegan eating — grade B. The largest effect among the vegetarian patterns; B12 is not optional.
• Paleo — grade C. No grains, legumes or dairy; a few kilograms more in small, short trials of people with chronic disease.

Timing

• Time-restricted eating (16:8) — grade B. A simpler way to hold a deficit for some people; no advantage over an equal deficit when calories are matched.
• 5:2 fasting — grade B. Two low days a week; sustained loss at 6–12 months in the umbrella review, as a way of arranging a weekly deficit.
• Alternate-day fasting — grade B. The strictest form; the same loss as a daily deficit over a year, with more dropout.

Commercial programmes

• Atkins — grade B. Phased low-carbohydrate; among the best at six months, differences gone by a year.
• Zone — grade B. A fixed 40/30/30 split; modest loss, no advantage, the hormonal theory unsupported.
• Points-based programmes — grade B. The evidence is for the structure and the support, not the points.

Very-low-calorie programmes, medically supervised diets and weight-loss medication are a separate matter and will get their own pages only after expert review.

When a "diet" is medicine

Very-low-calorie programmes of 800 to 1200 kilocalories a day are reserved by the guidelines for specialist services with medical supervision. A gluten-free diet is a treatment for coeliac disease, not a weight-loss pattern. Prescription weight-loss medication is a conversation with a doctor. None of those belongs on a self-service page, and this site will not pretend otherwise.