A reference class, not a person — and not everyone. This compares two animal-based patterns for one common profile: the metabolically-impaired median adult — late 40s, BMI ~30 with visceral adiposity, hypertension, dysglycaemia-leaning (raised triglycerides, low-grade inflammation, toward metabolic syndrome and MASLD). This person is not low-risk, and that fact drives most of what follows.
The two patterns, both run at a matched ~10% caloric deficit:
| Clean Carnivore | Medium-Fat Animal-Based | |
|---|---|---|
| Macros | ~30% protein / ~70% fat / ~0 carb | ~50% protein / ~35% fat / ~15% carb; up to 3.5 g/kg protein |
| Foods | fatty ribeye (high saturated fat) | leaner beef (round roast) + steamed carrots & broccoli; hydration |
Why the matched deficit matters. Holding energy balance equal removes weight loss as a variable — so this is not a contest over which pattern makes you eat less; it isolates the effect of composition.
Start here: the deficit hands both arms the biggest lever, roughly equally
For this profile the single largest win is visceral-fat loss, and a matched ~10% deficit delivers it in either arm. A ~5% loss clears liver fat and 7-10% reverses steatohepatitis (Fatty Liver MASLD and Weight Loss); the same loss improves glycaemia, blood pressure and function. The honest bound: dietary weight loss did not cut cardiovascular events (Look AHEAD null; a 54-RCT meta-analysis confirms), but it does lower all-cause mortality (~6 fewer per 1,000) by a non-cardiovascular route (Does Weight Loss Reduce Cardiovascular Events). So the deficit is worth pursuing for the liver, glycaemia, blood pressure, function and that survival benefit — just not as cardiac protection.
Because the deficit controls weight, the decision is no longer which loses more. It is which composition is better for an already-at-risk person, and which they can sustain — and both high-protein patterns carry a satiety advantage. The live discriminators are the axes below.
Axis 1 — Saturated fat and LDL: the difference that actually matters here
- Clean carnivore maximises saturated fat (70% fat, fatty ribeye) and adds a second LDL-raising push: very-low-carbohydrate/ketogenic eating itself raises LDL on top of the saturated-fat effect (Low-Carbohydrate vs Balanced-Carbohydrate Diets). Reducing saturated fat lowers LDL with high certainty and cumulatively (Saturated Fat Intake and Replacement) — so the carnivore arm moves LDL the wrong way on two counts.
- Medium-fat explicitly cuts saturated fat (35% fat, lean round roast), so LDL runs lower.
- The deficit partly offsets the carnivore’s rise (weight loss lowers LDL), so the net is
deficit-down vs saturated-fat-up in the carnivore arm and both-down in the medium-fat arm
[INFERRED — each direction is held; the net of a combined deficit-plus-high-SFA diet is our synthesis, not a measured value].
How much does the LDL difference matter here? Four steps:
- On its own, small. Cutting saturated fat lowers cardiovascular events by ~RR 0.83 (15 fewer per 1,000) at moderate certainty, with no clear mortality effect (Saturated Fat Intake and Replacement).
- But baseline risk scales it up. Absolute benefit tracks starting risk (Baseline Risk and the Relative-Absolute Split, SCORE2 Baseline Risk and the ESC Treatment Thresholds), and this person is at elevated risk — so the same relative effect is a larger absolute one than for a lean 30-year-old, where the low-risk ceiling makes it genuinely second-order (Cardiometabolic Interventions and Hard CV Outcomes in Low-Risk People).
- And the causal picture is held, not hedged. LDL/apoB causes atherosclerotic disease — genetic, Mendelian-randomization and RCT evidence converging — and risk is cumulative (magnitude × duration, LDL ApoB and Cumulative Exposure). A sustained diet-induced rise is added lifetime exposure, not a moved marker.
- For this profile it’s worse than LDL-C shows. In insulin resistance, LDL-C under-states the atherogenic apoB burden (small, dense LDL packs more particles per unit cholesterol), so the carnivore arm’s LDL-C rise likely understates its apoB rise. Track apoB (or non-HDL), not LDL-C. The genuine uncertainty is how much a given person’s apoB rises (the hyper-responder magnitude), not whether it matters.
Verdict on this axis: for an already-at-risk person the carnivore arm’s LDL/apoB maximisation is a real cost — sharpened, not softened, now the causal transmission is held — and the medium-fat arm sits closer to the evidence. Symmetric standards: the deficit does not excuse a large LDL rise here, and the LDL worry does not erase the shared weight-loss benefit.
Axis 2 — Protein: the medium-fat arm overshoots, but the deficit gives partial cover
- At maintenance, the medium-fat arm’s 3.5 g/kg is gratuitous — muscle gains plateau near ~1.6 g/kg/day (Protein and Resistance Training for Muscle and Strength), above which extra protein buys no more muscle (the effect is modest anyway: ~+2.5 kg on a 1RM, ~+0.3 kg lean mass).
- Under a deficit the rationale improves: ample protein preserves lean mass while fat is lost, and the useful target rises somewhat — but the defensible figure is high, ~1.6-2+ g/kg, still short of 3.5. Both patterns are animal-based, so protein quality is a non-issue (both high on DIAAS, Protein Quality and the DIAAS Score).
- One link: raising protein is part of how the medium-fat arm lowers its fat fraction, so Axes 1 and 2 are connected. The upper safety bound of ~3.5 g/kg is a gap we do not hold — named, not invented.
Axis 3 — Carbohydrate, fibre, vegetables
- Carbohydrate restriction helps the dysglycaemic arm — both patterns are low-carb, and restriction raises short-term type-2 diabetes remission (risk difference ~0.32 at 6 months, moderate), but it attenuates by 12 months (~0.10) and collapses toward zero once no medication is required (Carbohydrate Restriction and Type 2 Diabetes Remission). Low- vs balanced-carb differs by ~1 kg, not clinically important (Low-Carbohydrate vs Balanced-Carbohydrate Diets).
- Fibre and vegetables: the carnivore arm has essentially none; the medium-fat arm’s steamed carrots/broccoli add fibre (a modest cohort-level benefit, Whole Grains Refined Grains and Pulses), potassium (Axis 5), and micronutrient diversity. Steaming is low-heat, forming fewer advanced glycation end-products — a small lever (low-AGE diets move some surrogates but are null on weight, glucose and blood pressure, Dietary AGEs and Cooking Method). The brassica goitrogen worry is defused by cooking and normal intake (Antinutrients in Plant Foods).
- Carnivore forgoes the fibre/micronutrient channel entirely; whether muscle-meat-only meets micronutrient needs is a gap we do not hold.
Axis 4 — Red meat: shared, so it decides nothing
Both patterns are beef-heavy, so the contested red-meat question applies equally and cannot separate them: one guideline family says limit red/processed meat, another says current intake is defensible on the same evidence — the disagreement is the standard of proof and whose values set the threshold, not whether meat causes cancer (Should Adults Reduce Red and Processed Meat). One marginal difference: high-temperature searing (ribeye) forms more heterocyclic amines and PAHs than gentler roasting (Red and Processed Meat and Cancer) — a small point for the medium-fat arm’s cooking, not a headline.
Axis 5 — Sodium and potassium: live, because this profile is hypertensive
Reducing sodium lowers blood pressure with high certainty (~3.4 mmHg resting, up to ~5.5 mmHg on ambulatory measurement, Sodium Intake and Blood Pressure), and the effect is larger in the hypertensive. Both patterns are meat-heavy, so salted or processed cuts are the sodium risk; the medium-fat arm’s vegetables also add potassium, which the plant-free carnivore pattern forgoes. Small axis, same direction as Axis 1 — toward the medium-fat pattern.
Where this lands (net, stratum-weighted, not a dogmatic verdict)
- The shared weight loss is the main event, delivered by the matched deficit either way — so whichever pattern the person will sustain pulls the biggest lever available.
- Among the differences, the SFA/LDL axis is the one that genuinely matters here, because this person is at elevated risk; the medium-fat pattern sits closer to the evidence, and the clean carnivore maximises the one exposure with an evidenced (if small) harm.
- The medium-fat arm’s 3.5 g/kg protein is an overshoot — defensible as high for lean-mass preservation at a deficit, not at 3.5.
- Neither pattern has hard-outcome evidence, and no achievable composition difference is large against the shared weight-loss benefit. Read the whole comparison at the margin.
Open questions (surfaced, deliberately NOT adjudicated)
Plausible mechanisms stated so their absence doesn’t mislead; we hold no evidence to settle them and take
no position (routed to ingest-queue.md):
- No-fibre and cholesterol clearance. Does a zero-fibre pattern impair bile-acid/cholesterol excretion so circulating cholesterol runs higher than the SFA effect alone predicts, while the medium-fat arm’s fibre adds a bile-acid-binding LDL-lowering channel? Held only at the effect level (a daily pulse portion moves LDL-C ~0.2 mmol/L); the mechanism is unheld.
- No-fibre and mobilised-toxin clearance during fat loss. During active fat loss, lipophilic compounds stored in adipose enter circulation — does fibre aid faecal clearance, so a zero-fibre pattern permits reuptake? Entirely unheld.
- High dietary fat during fat loss. With adipose mobilising fatty acids, does high dietary fat add to circulating and hepatic lipids relative to a lower-fat arm, even though net loss improves liver fat? Unheld.
What we cannot yet say (routed for later)
- No hard-outcome trial of either pattern — both are assembled from component evidence. The open loop, front and centre.
- The protein upper-safety bound above ~2 g/kg is unheld — we hold the 1.6 g/kg plateau, not a ceiling at 3.5.
- The diet-induced-LDL responder tail on very-low-carb + high-SFA (hyper-responders) — unheld.
- Micronutrient adequacy of a muscle-meat-only diet — unheld.
- Diet tolerability / sustainability at a deficit — the live discriminator we cannot measure.
The honest limits
- This grades the evidence, not the outcome — coherence, not validity.
- It appraises, it does not prescribe: a large diet-induced LDL rise may warrant lipid monitoring or a prescriber’s input, which this cannot give.
- A reference class, not a person: how much the SFA/LDL axis weighs is set by the individual’s actual cardiovascular risk and priorities.
- Symmetric standards, both diets and both directions: neither the carnivore’s ancestral framing nor the medium-fat diet’s SFA-reduction rationale earns a discount; where the evidence is small or absent, this says so.