Keto Diet: Evidence, LDL Caveats & How To

Keto is the rare diet that has both a gold-standard clinical use and a genuine scientific disagreement about everything else. For drug-resistant childhood epilepsy, the evidence is grade A: a Cochrane review found that about half of children cut their seizures by more than half. For weight loss and type 2 diabetes, the evidence is solid but time-limited. For heart safety, it is unresolved — in the range of a quarter of low-carb dieters, LDL cholesterol climbs sharply, and no long-term trial has followed the pattern to hard cardiovascular endpoints.

So the honest answer to “is keto healthy?” is not a slogan. It is this: keto can be done healthily, and it can be done badly, and the difference is almost entirely about food quality and monitoring — the same difference that decides whether any diet on the map works. This guide walks through what keto is, what the trials actually measured, where it shines, where it is genuinely risky, and how to build a version worth eating for years rather than weeks.

What keto is — and what it was before it was a diet

The ketogenic diet was invented in 1921 by Dr Russell Wilder at the Mayo Clinic as a medical therapy: a way to mimic the seizure control of fasting while still eating. It went to the margins once anti-epileptic drugs arrived in the 1930s, stayed at a handful of epilepsy centres for sixty years, and re-emerged in the 1990s through the Charlie Foundation and then, spectacularly, in the 2010s through Instagram.

The modern low-carb movement traces back to Dr Robert Atkins — a cardiologist whose 1972 book sold millions — but keto as a distinct pattern is stricter: carbohydrate low enough to push the body into nutritional ketosis, where the liver converts fat into ketone bodies and the brain starts running partly on them.

The numbers that define it:

• Net carbs: under 20–50 g per day (net carbs = total carbohydrates minus fibre)
• Fat: 70–80% of calories
• Protein: 15–25% of calories — roughly 1.2–1.5 g/kg body weight for most people
• Blood ketones: 0.5–3.0 mmol/L — about ten times lower than diabetic ketoacidosis, a point worth remembering when the word ketosis is used to scare you

Variants fan out from there. Modified Atkins is a looser, more practical epilepsy treatment. Targeted keto lets athletes ring a small fast-carb snack around training. Cyclical keto alternates strict days with high-carb days and mostly suits bodybuilders. And “clean keto” versus “dirty keto” is the distinction that will matter most to your health: clean keto is meat, fish, eggs, olive oil and vegetables; dirty keto is the same macro maths delivered through sausage, cheese puff products and sugar-free sweeteners. The research on how much they differ is thin — but every mechanism we do understand says they differ enormously.

How it works

Keto’s claim rests on three mechanisms, and the evidence for each is stronger than the marketing, in some places, and weaker in others.

Ketosis replaces glucose as a major fuel. With glycogen stores emptied within a couple of days, the liver makes beta-hydroxybutyrate and acetoacetate, which cross the blood-brain barrier and substitute for glucose in the brain. That this works at all is why the diet has a ninety-year clinical licence — brains accept ketones, particularly in drug-resistant epilepsy, where ketones shift the brain’s GABA/glutamate balance and lower neuronal excitability.

Insulin drops. Carbohydrate is the main driver of insulin, and meaningful carb restriction lowers fasting and post-meal insulin, improving insulin sensitivity and, in type 2 diabetes, sparing flagging pancreatic beta cells from constant demand. This mechanism is well accepted and is the reason blood-sugar results from keto are real, not a trick.

Appetite falls. Ketogenic diets reliably lower hunger to a degree unusual among diets. Ketones act on appetite pathways, protein and fat increase satiety hormones like GLP-1 and PYY, and studies find participants eat 200–400 fewer calories a day without counting anything. The catch — and this is the point tribal keto forums skip — is contained in the most controlled keto experiment ever run, which looked for the famous “metabolic advantage” and failed to find it.

What the trials found

Alongside those trials sit two uncomfortable cohorts. In the PURE study of 135,000 people across 18 countries, very high carbohydrate intake tracked with higher mortality but fat did not — and in the 2018 Lancet Public Health analysis, the safest carbohydrate window was a moderate 50–55%, with mortality rising below 40%. Sub-group analyses are telling: in that same analysis, low-carb patterns built on animal foods tracked with higher mortality, while low-carb patterns built on plant foods tracked with lower. Observational data is never proof, but the direction is hard to ignore — it argues that a keto diet loaded with vegetables, olive oil and fish is a different biomedical animal from one built on bacon, and that the middle of the macro distribution is where population-level risk is lowest.

No long-term RCT with heart attacks or strokes as endpoints has ever been run, in either direction. That gap is the strongest reason keto still carries a grade B overall rather than an A outside epilepsy: we know what happened at six months; we do not yet know what happens at six years.

Where keto is genuinely strong

• Blood sugar control. The mechanism is direct: fewer carbs in, less glucose out. No diet-type debate changes this.
• Short-term fat loss. Ketogenic diets edge other diets at 6–12 months, mostly by making people eat less without trying.
• Triglycerides, HDL, blood pressure. The majority of trial participants improve on all three.
• Metabolic syndrome. All five components — waist, BP, glucose, triglycerides, HDL — improve consistently, with reversal reported at meaningful rates.
• Epilepsy. Grade A, ninety years old, Cochrane-reviewed, used in hospitals worldwide.

The honest trade-offs

LDL cholesterol. In roughly 25–30% of low-carb dieters, LDL rises substantially — sometimes 50–100+ mg/dL — and elevated ApoB/LDL is one of the strongest modifiable cardiovascular risk factors we can measure. The American Heart Association’s 2023 statement lists keto among patterns not recommended for heart health for exactly this reason. Anyone doing keto should do a baseline lipid panel and check again after three months. If LDL surges, the highest-quality response is to swap saturated fat (butter, fatty red meat, coconut oil) toward monounsaturated fat (olive oil, avocado, macadamias) and re-test.

Fibre and micronutrients. Removing grains, legumes and most fruit strips out fibre, folate, magnesium and potassium. On keto you are not “supplementing” these — you are running a deficit unless you deliberately fill it: low-carb vegetables at every meal, nuts, seeds, avocado, leafy greens, and often a magnesium supplement.

Electrolytes and “keto flu”. The headache-fatigue-brain fog of the first fortnight is not ketosis poisoning you; it is sodium and water leaving the body as glycogen drains. Salt food generously, keep potassium and magnesium high, and the transition is far shorter and milder.

Adherence and social life. Keto trial dropout runs 30–50% at a year. Restaurants, dinners with friends and holidays all become logistics. If you love the simplicity of the rules, that is manageable — rule-followers are often keto’s happiest demographic. If you eat for pleasure and company, a looser low-carb pattern at 50–100 g of carbs usually delivers most of the benefit with a fraction of the friction.

Who should not do it without medical supervision. Pregnancy and breastfeeding; anyone with type 1 diabetes without an endocrinologist; people on SGLT2 inhibitors, where keto is contraindicated outright because of euglycemic ketoacidosis risk; patients on insulin or sulfonylureas, who must not change carbohydrate intake unsupervised; people with existing heart disease and an LDL-sensitive picture; and anyone with a history of disordered eating, for whom this level of tracking is a known trigger.

Making keto genuinely healthy

The version of keto worth defending is not the one from the forums. It is this one:

1. Keep the vegetables. You need 6–8 cups of low-carb vegetables a day to hold fibre, folate, magnesium and the microbiome together. This is the single largest difference between clean and dirty keto.
2. Lead with fish and olive oil, not processed meat. Weekly loops of salmon, sardines and vegetables carry the diet on their shoulders.
3. Protect protein. Around 1.5 g/kg keeps muscle, appetite and satiety where they belong.
4. Rebuild electrolytes during adaptation. Salt, potassium and magnesium in the first fortnight.
5. Test, do not assume. Baseline lipids (including ApoB), a re-check at three months, HbA1c if blood sugar is the goal, and kidney function if you have any existing concerns.
6. Rehearse a low-carb fallback. If strict keto makes you dread dinner, drop to 50–100 g of carbs with the same food principles. The evidence says you keep much of the benefit and stay in the game far longer.

And if one partner is keto and the other is not: cook one shared protein-and-vegetable base, plate the insulin-lowering version without the starch, and put rice or potatoes on the other plate only. That is exactly the method SafeTable’s diet guides hub builds on — one meal, two plates, no separate dinners.

Common questions