Home › Topics › Ketogenic Diet

Ketogenic Diet

GMJ News knowledge hub · last reviewed August 2026 · Georgian Medical Journal

SummaryStatisticsGlossaryGMJ newsFAQDocumentsOrganizationsResearch

The ketogenic diet occupies an unusual position in nutrition science: it has an established, non-negotiable therapeutic indication in drug-resistant epilepsy, particularly in children, where it has been used since the 1920s and is supported by randomised evidence; it produces genuine short-term weight loss and marked improvement in glycaemic control and medication burden in type 2 diabetes; and it is simultaneously surrounded by mechanistic claims that the evidence does not support (WHO). The central contested claim is the carbohydrate-insulin model of obesity — that carbohydrate drives fat storage through insulin independently of energy intake — which has been tested directly in metabolic ward studies and has not been supported: when calories and protein are matched, fat loss does not differ meaningfully by carbohydrate content. Two honest complications deserve statement: long-term adherence is poor and weight regain common, and a minority of lean, metabolically healthy people experience dramatic LDL cholesterol elevation on ketogenic diets, a phenomenon that is real, poorly understood, and should not be dismissed by either side.

Key messages

ESTABLISHED INDICATION: drug-resistant epilepsy, particularly in children
This is not contested and is frequently omitted from popular discussion. The ketogenic diet has been used for refractory epilepsy since the 1920s and is supported by randomised evidence showing meaningful seizure reduction in children who have failed multiple antiseizure medications. It is a genuine medical therapy requiring specialist dietetic supervision, monitoring for growth, lipids, renal stones and micronutrient deficiency, and it has specific indications in glucose transporter type 1 deficiency and pyruvate dehydrogenase deficiency where it is the treatment of choice rather than an alternative.
SUPPORTED: effective for short-term weight loss and glycaemic control in type 2 diabetes
Low-carbohydrate and ketogenic approaches produce clinically meaningful reductions in HbA1c and in glucose-lowering medication requirements, with a substantial proportion of participants in structured programmes achieving remission criteria in the short to medium term. Trial evidence including DiRECT for low-energy approaches and various low-carbohydrate protocols supports this. The mechanism is not mysterious — reducing carbohydrate reduces postprandial glycaemic excursion and, in most cases, energy intake — and the results are real regardless of the mechanistic dispute.
NOT SUPPORTED: the carbohydrate-insulin model of obesity
This is the central contested claim, and it has been tested directly. The model proposes that dietary carbohydrate drives insulin secretion, which drives fat storage and hunger, so that obesity results from carbohydrate rather than from energy excess. Metabolic ward studies matching calories and protein while varying carbohydrate have not found the predicted advantage for low-carbohydrate diets; if anything, fat loss was marginally greater on the lower-fat arm in the most rigorous of these studies. The model has been subject to a formal published debate, and the honest position is that it is a coherent hypothesis that the controlled human data do not support.
GENUINELY OPEN: lean mass hyper-responders and dramatic LDL elevation
A subset of lean, insulin-sensitive, physically active individuals experience extreme LDL cholesterol increases on ketogenic diets — sometimes exceeding 5-8 mmol/L from a normal baseline — a phenomenon described as the lean mass hyper-responder phenotype. This is real, reproducible and poorly understood, plausibly reflecting increased hepatic VLDL secretion to deliver fat as fuel in the absence of dietary carbohydrate. Whether it carries the same atherosclerotic risk as equivalent LDL from other causes is genuinely unresolved and under active study. Neither dismissing it as harmless nor assuming equivalence with familial hypercholesterolaemia is currently justified.
THE PRACTICAL LIMITATION: adherence and regain
Long-term trial data consistently show convergence between low-carbohydrate and other dietary approaches at 12 to 24 months, driven overwhelmingly by declining adherence rather than by any metabolic adaptation specific to the diet. The same is true of every dietary pattern studied, which is the more important general finding: the diet that works is the one that is sustained. Ketogenic diets are socially demanding, difficult in many food cultures, and expensive in some settings — all of which are legitimate considerations rather than failures of willpower.
PRACTICAL SAFETY: what requires monitoring and who should not
Common early effects — fatigue, headache, constipation, cramps, collectively termed keto flu — largely reflect fluid and electrolyte shifts and are manageable. Requiring attention: lipid monitoring, since response is highly variable; renal stones, particularly in children on therapeutic diets; constipation and low fibre intake; micronutrient adequacy. Contraindicated or requiring specialist supervision: fatty acid oxidation disorders and other inborn errors, pancreatitis, severe hepatic impairment, pregnancy, and type 1 diabetes where euglycaemic ketoacidosis risk is real. SGLT2 inhibitors combined with ketogenic diets carry a specific and under-recognised risk of euglycaemic diabetic ketoacidosis.

Key statistics

Since 1920s
ketogenic diet established for drug-resistant epilepsy, supported by randomised evidence
Cochrane/ILAE
Treatment of choice
in GLUT1 deficiency and pyruvate dehydrogenase deficiency, not merely an alternative
ILAE/metabolic
Not supported
carbohydrate-insulin model tested in metabolic ward studies with calories and protein matched
Cell Metab/AJCN
Convergence
low-carbohydrate and comparison diets converge at 12-24 months, driven by adherence
JAMA/Lancet
Hyper-responders
lean insulin-sensitive individuals may see extreme LDL rises — real and unexplained
JCL/Metabolites
SGLT2i risk
combination with ketogenic diet carries euglycaemic ketoacidosis risk — under-recognised
ADA/EASD

Ketogenic diet — where the disagreement actually lies

Source: Bars show strength of supporting evidence. The therapeutic indications are solid; the obesity mechanism claim is not.

Glossary of key terms

Nutritional ketosis versus ketoacidosis
PhysiologyNutritional ketosis describes blood beta-hydroxybutyrate typically between 0.5 and 3 mmol/L, achieved by restricting carbohydrate sufficiently that hepatic ketogenesis supplies a substantial fraction of energy needs, with blood pH remaining normal because insulin is still present and restrains lipolysis. Diabetic ketoacidosis involves ketone concentrations often above 10-20 mmol/L with metabolic acidosis, occurring when insulin is absent or grossly insufficient. The two are physiologically distinct and conflating them is a common error in both directions — some clinicians treat any ketosis as dangerous, while some enthusiasts dismiss ketoacidosis risk entirely. The genuine overlap is euglycaemic ketoacidosis, which can occur with SGLT2 inhibitors, in type 1 diabetes, in pregnancy and with prolonged fasting or illness on a ketogenic diet, and which is easily missed because glucose is not markedly elevated.
The carbohydrate-insulin model
Obesity scienceThe hypothesis, articulated most prominently by Ludwig and colleagues, that obesity is driven by carbohydrate-stimulated insulin secretion directing energy into adipose storage, producing cellular energy deficit, hunger and reduced expenditure — reversing the conventional causal direction from overeating to fat gain. It has generated the most productive controlled experiments in modern obesity research. Hall and colleagues conducted metabolic ward studies isolating carbohydrate while matching energy and protein, and found no advantage for carbohydrate restriction in fat loss, with fat loss marginally greater on the low-fat arm in one study. A formal debate was published in the American Journal of Clinical Nutrition. The model remains defended in modified forms, and the appropriate summary is that its strong version has been tested and not supported.
Lean mass hyper-responder phenotype
LipidologyA described phenotype in which lean, insulin-sensitive, often athletic individuals develop marked LDL cholesterol elevation on carbohydrate restriction, defined in some proposals as LDL above 5 mmol/L with HDL above 1.6 and triglycerides below 1 mmol/L. The proposed mechanism is the lipid energy model — in the absence of dietary carbohydrate, hepatic VLDL secretion increases to deliver triglyceride to peripheral tissues, with consequent LDL elevation reflecting increased flux rather than impaired clearance. Whether this confers the atherosclerotic risk associated with LDL elevation from other causes is unresolved; the KETO-CTA imaging study was designed to address it and its interpretation is itself contested. The scientifically honest position is that this requires outcome data, not assumption in either direction.
Ketogenic diet in epilepsy practice
NeurologyTherapeutic ketogenic diets are delivered in four main forms: the classical ketogenic diet with a 4:1 or 3:1 ratio of fat to combined protein and carbohydrate by weight; the medium-chain triglyceride diet, which permits more carbohydrate because MCTs are more ketogenic per gram; the modified Atkins diet, which is less restrictive and does not require weighing; and the low glycaemic index treatment. Selection depends on age, feeding route, adherence capacity and setting. Monitoring includes growth in children, lipid profile, renal ultrasound or urinary calcium-to-creatinine ratio for stone risk, bone density on prolonged therapy, selenium and carnitine status, and vitamin and mineral supplementation, which is mandatory. Efficacy is assessed at three months, and the diet is weaned gradually if ineffective or after a sustained seizure-free period.
Low-carbohydrate diets and type 2 diabetes remission
DiabetologyRemission — usually defined as HbA1c below 48 mmol/mol sustained for at least three months without glucose-lowering medication — is achievable through several routes, and the evidence indicates that weight loss is the principal driver rather than macronutrient composition specifically. The DiRECT trial used a total diet replacement low-energy approach and achieved remission in a large proportion at one year with maintenance at two years, proportional to weight lost. Low-carbohydrate approaches achieve comparable glycaemic improvement, sometimes with less weight loss, reflecting a direct effect on postprandial excursion. The essential clinical point is medication safety: sulfonylureas and insulin require immediate dose reduction when carbohydrate is sharply reduced, or severe hypoglycaemia follows, and this must be arranged before the dietary change rather than after.
Fibre, microbiota and long-term considerations
NutritionKetogenic diets are typically low in fibre and in fermentable substrate, with measurable reductions in short-chain fatty acid production and in bifidobacteria in human studies. The clinical significance is not established, but fibre intake has independent associations with cardiovascular and colorectal outcomes strong enough that the reduction warrants attention rather than dismissal. Practical mitigation includes non-starchy vegetables, nuts, seeds, avocado and psyllium supplementation. Other long-term considerations that receive insufficient attention: sustained very low intake of fruit, legumes and whole grains removes food groups with the most consistent outcome evidence in nutritional epidemiology, and the absence of long-term randomised data means that safety beyond two years rests largely on inference.

Latest GMJ coverage

How Diet Shapes Your Gut Microbiome—and Why It Matters for Health
25/07/2026

Frequently asked questions 12 Q&A — structured for Google featured snippets and AI discovery

Knowledge hub: guidelines, conventions and reports

Organizations working in migration and health

Related health topics

Drug-resistant epilepsyType 2 diabetes and remissionIntermittent fastingLDL response and lipid monitoringObesityNutrition

About this hub. Produced by the GMJ News Editorial Team as a public-good service. Every statistic is linked to its primary source. Documents are preserved in the GMJ Repository with full attribution. Georgian Medical Journal · Contact the editorial team
GMJ BriefsView all →
●What Healthcare Providers Need to Know About Sandwich Generation Caregivers3 hours ago●New Data: Unstable Connectivity Amplifies Loneliness More Than No Internet at All3 hours ago●Announcement: Magnesium deficiency linked to multiple chronic diseases: what clinicians need to know3 hours ago●Ultra-Processed Foods Account for Majority of Calories in Latin American Retail35 minutes ago●What Clinicians Must Know: Screening and Managing Immigration-Related Family Separation Trauma35 minutes ago●Announcement: Vitamin C May Reverse Skin Aging at the Genetic Level, New Research Suggests35 minutes ago●Three Clinical Advances in the UK's Updated Influenza Vaccine Framework1 hour ago●Announcement: Bones as Metabolic Organs: How Mechanical Loading Triggers Glucose Control1 hour ago●New Data: Combat Trauma in SOF Creates Cumulative, Multi-Domain Health Burden1 hour ago●Global Adolescent Obesity Climbing—Digital Marketing and Inequality Are Key Culprits3 hours ago