The ketogenic diet was developed for epilepsy, delivered by specialist teams, with a defined indication and measurement throughout. Every element that made it work clinically is the part that does not survive being turned into a lifestyle.
Break your wrist and they’ll put it in a cast.
The cast is a genuinely brilliant intervention. It immobilises the joint, holds the bone in position, and lets the tissue knit back together in a way it simply cannot while the arm is being used. Nobody argues about whether casts work.
Leave it on for two years and the muscle underneath wastes, the joint stiffens, and the bone thins from lack of loading. Same cast. Same mechanism. The thing that healed you is now the thing harming you, and nothing about the cast changed — only how long it stayed on and whether anyone was still checking.
That is nutritional ketosis. And most people using it have no idea they’re wearing a cast.
The ketogenic diet was not developed for weight loss, physique, or longevity. It was developed at Johns Hopkins in the 1920s for epilepsy, at a time when there was almost nothing else to offer a child having dozens of seizures a day.
It worked. It still does. In drug-resistant paediatric epilepsy the ketogenic diet remains a genuine clinical intervention with a real evidence base, delivered by specialist teams with dietetic supervision and regular monitoring. Some children have seizure reductions that no medication achieved.
Note what that clinical use involves: a specific indication, a supervising team, defined duration, and measurement throughout. None of which describes the man in the gym who has been “doing keto” for three years because it made him feel sharp that first fortnight.
If you’re measuring, it’s an intervention. If you’re not, it’s an identity.
Nutritional ketosis has a definition — blood ketones roughly between 0.5 and 3.0 mmol/L. Below that you are not in ketosis regardless of what you are eating. Above it, in a non-diabetic, is unusual and worth understanding rather than celebrating.
A finger-prick meter costs less than a month of the supplements people buy to support a diet they are not verifying. It is the difference between doing something and believing you are doing something.
Your blood glucose in mmol/L divided by your blood ketones in mmol/L. One number describing a metabolic state, rather than two you have to interpret against each other.
It came out of oncology research, where the theoretical interest is in shifting the ratio substantially. Whether it is the right target for you is a separate question — but it illustrates the principle. This is a measurable state, not a food philosophy, and the people using it most seriously measure it most often.
Established
Drug-resistant epilepsy, particularly in children. Strong. Established. Delivered clinically. This is the one that earns the diet its reputation.
Reasonable and growing
Type 2 diabetes and metabolic syndrome. Carbohydrate restriction improves glycaemic markers, with decent trial evidence for significant improvement and in some cases remission. The live debate is less about whether it works and more about whether the effect is specific to ketosis or to carbohydrate reduction and weight loss generally.
Weight loss. Works, and works quickly at first — much of the early drop is glycogen and its associated water. Over twelve months and beyond, trials comparing ketogenic to other approaches with protein and calories matched generally find no meaningful superiority. An effective approach. Not a metabolically magic one.
Preliminary
Neurodegeneration and cognitive function. Genuinely interesting mechanistically, some small trials with MCTs and ketone esters, nothing established. Watch rather than act.
Athletic performance. Fat adaptation improves fat oxidation at lower intensities. It also reliably impairs high-intensity output, because that work is glycogen-dependent and no amount of adaptation changes the chemistry.
Cancer. The preclinical work is interesting. The human evidence does not support ketogenic diets as a cancer treatment, and nobody should delay or replace oncological care on the basis of it.
Unintended weight loss in someone with cancer is a serious problem, not a sign of progress. If this applies to you, it is a conversation for your oncology team.
This is the part almost nobody discusses, and it is where I see the most harm.
A ketogenic diet is metabolically demanding. It requires your liver to run gluconeogenesis and ketogenesis at pace, sustained, for as long as you are on it. It requires enough bile to emulsify a large increase in dietary fat. It requires reasonable pancreatic enzyme output, a thyroid that can cope, and an HPA axis that is not already running on empty.
Anyone with compromised bile flow, or who has had their gallbladder removed. You have just asked for a large increase in dietary fat from a system with reduced capacity to handle it. The result is usually digestive misery, sometimes fat malabsorption, and occasionally people conclude they are “detoxing”.
Anyone with an already-disrupted stress axis. Carbohydrate restriction is a physiological stressor. In someone with a flattened cortisol curve, poor sleep and a history of under-eating, adding another stressor rarely produces the intended result. Very low carbohydrate intake also tends to reduce T3, the active thyroid hormone — adaptive, but not something you want stacked on existing thyroid downregulation.
Anyone with reduced liver function. The organ doing most of the work here needs to be capable of doing it.
The capacity to cope and the design to cope are not the same thing.
A young man with an intact metabolism, good sleep and no stress load can run this and feel excellent. That tells you he has capacity. It does not tell you the diet suits him long-term, and it certainly does not tell you it suits his sister, his mother, or the perimenopausal client who read his post.
A therapeutic protocol with a specific indication, a defined duration and mandatory monitoring has been repackaged as a lifestyle, an aesthetic, and a personality.
The mechanism is depressingly simple. Something works dramatically for a specific group under specific conditions. The dramatic result is the marketable part. The conditions are not. So the conditions get dropped, the result gets promised to everyone, and what began as a clinical intervention becomes a way of describing yourself on the internet.
Every element that made it work clinically — the indication, the supervision, the measurement, the endpoint — is precisely the part that does not survive the transfer.
And it produces a particular kind of casualty: someone eighteen months into a diet they have never verified, who feels worse than when they started, has stopped menstruating or lost their morning energy, and has concluded they are not doing it strictly enough.
An uncomfortable observation from practice: most people who describe themselves as ketogenic are not in ketosis most of the time.
Not because they are dishonest — because sustaining it is genuinely difficult, the threshold is easier to breach than people realise, and without measurement there is no feedback. You can eat in a way that feels ketogenic, believe you are ketogenic, and be metabolically nowhere near it.
Which produces the worst of both worlds: all of the restriction, the social cost, the food anxiety and the expense — and none of the metabolic state the restriction was for.
If you are going to do this, measure. If you will not measure, do something else. A well-constructed lower-carbohydrate diet without the ketogenic threshold delivers a large share of the metabolic benefit with a fraction of the difficulty, and you can actually maintain it.
With an indication. Something specific you are trying to change, that this is a reasonable tool for.
With measurement. Blood ketones, glucose, ideally the GKI. Not urine strips, which stop being informative once you are adapted.
With a defined window. Six weeks, twelve weeks, whatever fits the goal. Written down at the start, not decided later.
With the biochemistry seen first. Thyroid with T3, cortisol pattern, liver function, bile markers, iron.
With an exit. How you come off it, what replaces it, what you monitor afterwards. Almost nobody plans this, which is how three-week interventions become three-year identities.
The diet is not the problem. The cast is not the problem either.
The problem is a powerful, specific, measurable intervention being worn indefinitely by people who never had the indication, never established whether they had the machinery, are not measuring whether it is even happening, and have no idea what would tell them to stop.
If you are using it as a tool, use it properly — with a reason, a measurement and an end date.
If you are using it as an identity, the honest question is not whether keto works. It is when you last checked whether you were actually doing it, and what you expect to be different in a year.
The markers that determine whether someone can run this well are the ones almost nobody checks first: a full thyroid panel including free T3, the cortisol pattern across the day, liver function, bile and pancreatic markers, and iron status. Those are on the TDG Five-Test Programme, and they are considerably more useful before a restrictive diet than after it has stopped working.
Ask AIdan what testing would answer that in your situation, or start with a discovery call.
Ask AIdan → Book a Discovery Call